1 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3 * i386.c (ix86_vec_cost): New function.
4 (ix86_rtx_costs): Handle vector operations better.
5 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
6 * x86-tune-costs.h: Add new costs to all tables.
8 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
10 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
12 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
13 divsd, sqrtss and sqrtsd
14 * x86-tune-costs.h: Add new entries to all costs.
15 (znver1_cost): Fix to match real instruction latencies.
17 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18 Michael Collison <michael.collison@arm.com>
20 * compare-elim.c: Include emit-rtl.h.
21 (can_merge_compare_into_arith): New function.
22 (try_validate_parallel): Likewise.
23 (try_merge_compare): Likewise.
24 (try_eliminate_compare): Call the above when no previous clobber
26 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
29 2017-10-14 Jakub Jelinek <jakub@redhat.com>
33 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
34 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
38 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
39 to be any operand_equal_p operands. For & (B - 1) require
40 B to be power of 2. Recognize
41 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
43 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
46 * optabs.c (expand_memory_blockage): Fix call of
47 targetm.have_memory_blockage.
49 2017-10-14 Jakub Jelinek <jakub@redhat.com>
52 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
53 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
54 objects to extra_objs instead of overwriting it.
56 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
58 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
59 Use any_fp_register_operand as operand[3] predicate. Simplify
60 equality test for operands[2] and operands[4] memory location.
61 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
62 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
63 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
64 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
65 any_fp_register_operand as operand[1] predicate. Simplify
66 equality test for operands[0] and operands[3] memory location.
67 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
68 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
69 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
71 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
73 * target-insns.def: Add memory_blockage.
74 * optabs.c (expand_memory_blockage): New function.
75 (expand_asm_memory_barrier): Rename ...
76 (expand_asm_memory_blockage): ... to this.
77 (expand_mem_thread_fence): Call expand_memory_blockage
78 instead of expand_asm_memory_barrier.
79 (expand_mem_singnal_fence): Ditto.
80 (expand_atomic_load): Ditto.
81 (expand_atomic_store): Ditto.
82 * doc/md.texi (Standard Pattern Names For Generation):
83 Document memory_blockage instruction pattern.
85 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
87 * config/rl78/rl78.c (rl78_emit_libcall): New function.
88 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
89 * config/rl78/rl78.md: New define_expand "adddi3".
91 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
93 * cfghooks.c (verify_flow_info): Disable check that all probabilities
96 2017-10-13 Jeff Law <law@redhat.com>
98 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
100 2017-10-13 Jakub Jelinek <jakub@redhat.com>
103 * internal-fn.c (expand_mul_overflow): If both operands have
104 the same highpart of -1 or 0 and the topmost bit of lowpart
105 is different, overflow is if res <= 0 rather than res < 0.
107 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
109 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
110 TARGET_P9_VECTOR code for unaligned_load case.
112 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
114 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
116 2017-10-13 Nathan Sidwell <nathan@acm.org>
118 * tree-core.h (tree_contains_struct): Make bool.
119 * tree.c (tree_contains_struct): Likewise.
120 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
121 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
122 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
123 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
124 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
126 2017-10-13 Richard Biener <rguenther@suse.de>
128 * graphite-isl-ast-to-gimple.c
129 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
130 parameters and dominance check.
131 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
132 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
133 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
134 Do not update SSA form here or do intermediate IL verification.
135 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
136 (graphite_initialize): Remove check on the number of loops in
137 the function and inline into graphite_transform_loops.
138 (graphite_finalize): Inline into graphite_transform_loops.
139 (graphite_transform_loops): Perform SSA update and IL verification
141 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
143 2017-10-13 Richard Biener <rguenther@suse.de>
145 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
146 graphite_expression_type_precision): Avoid global constructor
148 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
149 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
150 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
151 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
152 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
153 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
155 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
158 * config/i386/i386.h (ix86_red_zone_size): New.
159 * config/i386/i386.md (push peephole2s): Replace
160 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
162 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
163 Alan Hayward <alan.hayward@arm.com>
164 David Sherwood <david.sherwood@arm.com>
166 * combine.c (can_change_dest_mode): Reject changes in
167 REGMODE_NATURAL_SIZE.
169 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
170 Alan Hayward <alan.hayward@arm.com>
171 David Sherwood <david.sherwood@arm.com>
173 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
174 (expand_debug_source_expr): Likewise.
175 * combine.c (combine_simplify_rtx): Likewise.
176 * cse.c (fold_rtx): Likewise.
177 * fwprop.c (canonicalize_address): Likewise.
178 * targhooks.c (default_shift_truncation_mask): Likewise.
180 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
181 Alan Hayward <alan.hayward@arm.com>
182 David Sherwood <david.sherwood@arm.com>
184 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
185 (widened_mode): Likewise.
186 (expand_unop): Likewise.
187 * ree.c (transform_ifelse): Likewise.
188 (merge_def_and_ext): Likewise.
189 (combine_reaching_defs): Likewise.
190 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
192 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
193 Alan Hayward <alan.hayward@arm.com>
194 David Sherwood <david.sherwood@arm.com>
196 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
197 * combine.c (gen_lowpart_for_combine): Likewise.
198 * dwarf2out.c (rtl_for_decl_location): Likewise.
199 * final.c (alter_subreg): Likewise.
200 * rtlhooks.c (gen_lowpart_general): Likewise.
201 (gen_lowpart_if_possible): Likewise.
203 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
204 Alan Hayward <alan.hayward@arm.com>
205 David Sherwood <david.sherwood@arm.com>
207 * calls.c (expand_call): Use subreg_lowpart_offset.
208 * cse.c (cse_insn): Likewise.
209 * regcprop.c (copy_value): Likewise.
210 (copyprop_hardreg_forward_1): Likewise.
212 2017-10-13 Jakub Jelinek <jakub@redhat.com>
215 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
216 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
217 =Q constraints to +Q and into insn condition add check
218 that operands[0] and operands[1] are equal.
219 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
220 =Q constraints to +Q and into insn condition add check
221 that operands[0] is equal to either operands[1] or operands[2].
224 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
225 instead of handling MINUS_EXPR twice (once for each argument),
226 canonicalize operand order and handle just once, use rtype where
227 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
230 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
231 any values of __C while still being pattern recognizable as a simple
234 2017-10-13 Richard Biener <rguenther@suse.de>
236 PR tree-optimization/82451
238 2017-10-02 Richard Biener <rguenther@suse.de>
240 PR tree-optimization/82355
241 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
242 a mapping for the enclosing loop but avoid generating one for
244 (copy_bb_and_scalar_dependences): Remove premature codegen
245 error on PHIs in blocks duplicated into multiple places.
246 * graphite-scop-detection.c
247 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
248 in the region use it as loop and nest to analyze the DR in.
249 (try_generate_gimple_bb): Likewise.
250 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
251 (add_loop_constraints): For blocks in a loop not in the region
252 create a dimension with a single iteration.
253 * sese.h (gbb_loop_at_index): Remove assert.
255 * cfgloop.c (loop_preheader_edge): For the loop tree root
256 return the single successor of the entry block.
257 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
258 Reset the SCEV hashtable and niters.
259 * graphite-scop-detection.c
260 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
261 assert that we only have POLYNOMIAL_CHREC that vary in loops
262 contained in the region.
263 (scop_detection::graphite_can_represent_expr): Adjust.
264 (scop_detection::stmt_has_simple_data_refs_p): For loops
265 not in the region set loop to NULL. The nest is now the
266 entry edge to the region.
267 (try_generate_gimple_bb): Likewise.
268 * sese.c (scalar_evolution_in_region): Adjust for
269 instantiate_scev change.
270 * tree-data-ref.h (graphite_find_data_references_in_stmt):
271 Make nest parameter the edge into the region.
272 (create_data_ref): Likewise.
273 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
274 entry edge into a region and adjust instantiate_scev calls.
275 (create_data_ref): Likewise.
276 (graphite_find_data_references_in_stmt): Likewise.
277 (find_data_references_in_stmt): Pass the loop preheader edge
278 from the nest argument.
279 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
280 parameter the edge into the region.
281 (instantiate_parameters): Use the loop preheader edge as entry.
282 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
284 (get_instantiated_value_entry): Make instantiate_below parameter
285 the edge into the region.
286 (instantiate_scev_name): Likewise. Adjust dominance checks,
287 when we cannot use loop-based instantiation instantiate by
288 walking use-def chains.
289 (instantiate_scev_poly): Adjust.
290 (instantiate_scev_binary): Likewise.
291 (instantiate_scev_convert): Likewise.
292 (instantiate_scev_not): Likewise.
293 (instantiate_array_ref): Remove.
294 (instantiate_scev_3): Likewise.
295 (instantiate_scev_2): Likewise.
296 (instantiate_scev_1): Likewise.
297 (instantiate_scev_r): Do not blindly handle N-operand trees.
298 Do not instantiate array-refs. Handle all constants and invariants.
299 (instantiate_scev): Make instantiate_below parameter
300 the edge into the region.
301 (resolve_mixers): Use the loop preheader edge for the region
302 parameter to instantiate_scev_r.
303 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
305 2017-10-13 Richard Biener <rguenther@suse.de>
307 PR tree-optimization/82525
308 * graphite-isl-ast-to-gimple.c
309 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
311 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
312 Fail code generation when we cannot represent the isl integer.
313 (binary_op_to_tree): Elide modulo operations that are no-ops
314 in the type we code generate. Remove now superfluous code
317 2017-10-13 Richard Biener <rguenther@suse.de>
319 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
320 (scop_detection::harmful_loop_in_region): Remove premature
322 (scop_detection::graphite_can_represent_scev): We can handle
323 pointer IVs just fine.
325 2017-10-13 Alan Modra <amodra@gmail.com>
327 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
328 "Clobbers and Scratch Registers". Add paragraph on
329 alternative to clobbers for scratch registers and OpenBLAS
332 2017-10-13 Alan Modra <amodra@gmail.com>
334 * doc/extend.texi (Clobbers): Correct vax example. Delete old
335 example of a memory input for a string of known length. Move
336 commentary out of table. Add a number of new examples
337 covering array memory inputs.
339 2017-10-12 Martin Liska <mliska@suse.cz>
341 PR tree-optimization/82493
342 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
343 (test_range_functions): New function.
344 (sbitmap_c_tests): Likewise.
345 * selftest-run-tests.c (selftest::run_tests): Run new tests.
346 * selftest.h (sbitmap_c_tests): New function.
348 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
350 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
352 * config/rs6000/amo.h: Fix spacing issue.
354 2017-10-12 Jakub Jelinek <jakub@redhat.com>
357 * config/i386/i386.md (*ashl<mode>3_mask_1,
358 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
359 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
362 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
364 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
365 (profile_probability): Set max_probability
366 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
368 * profile-count.c (profile_probability::differs_from_p): Do not
369 rely on max_probaiblity == 10000
371 2017-10-12 Jeff Law <law@redhat.com>
373 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
376 2017-10-12 Martin Sebor <msebor@redhat.com>
380 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
381 (handle_alias_pairs): Call it.
382 * common.opt (-Wattribute-alias): New option.
383 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
384 * doc/invoke.texi (-Wattribute-alias): Document.
386 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
389 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
391 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
393 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
394 (make_hard_regno_born, make_hard_regno_dead): Update
395 bb_killed_pseudos and bb_gen_pseudos.
397 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
399 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
401 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
403 * config/alpha/alpha.c (alpha_split_conditional_move):
404 Use std::swap instead of manually swapping.
405 (alpha_stdarg_optimize_hook): Ditto.
406 (alpha_canonicalize_comparison): Ditto.
408 2017-10-12 Bin Cheng <bin.cheng@arm.com>
410 * tree-loop-distribution.c (struct builtin_info): New struct.
411 (struct partition): Refactor fields into struct builtin_info.
412 (partition_free): Free struct builtin_info.
413 (build_size_arg_loc, build_addr_arg_loc): Delete.
414 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
415 information from struct builtin_info.
416 (find_single_drs): New function refactored from classify_partition.
417 Also moved builtin validity checks to this function.
418 (compute_access_range, alloc_builtin): New functions.
419 (classify_builtin_st, classify_builtin_ldst): New functions.
420 (classify_partition): Refactor code into functions find_single_drs,
421 classify_builtin_st and classify_builtin_ldst.
422 (distribute_loop): Don't do runtime alias check when distributing
424 (find_seed_stmts_for_distribution): New function.
425 (pass_loop_distribution::execute): Refactor code finding seed
426 stmts into above function. Support distribution for the innermost
427 two-level loop nest. Adjust dump information.
429 2017-10-12 Bin Cheng <bin.cheng@arm.com>
431 * tree-loop-distribution.c: Adjust the general comment.
432 (NUM_PARTITION_THRESHOLD): New macro.
433 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
434 (classify_partition): Skip builtin pattern of loop nest's inner loop.
435 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
436 in call to build_partition_graph.
437 (finalize_partitions): New parameter. Make loop distribution more
438 conservative by fusing more partitions.
439 (distribute_loop): Don't do runtime alias check in case of loop nest
441 (find_seed_stmts_for_distribution): New function.
442 (prepare_perfect_loop_nest): New function.
443 (pass_loop_distribution::execute): Refactor code finding seed stmts
444 and loop nest into above functions. Support loop nest distribution.
445 Adjust dump information accordingly.
447 2017-10-12 Bin Cheng <bin.cheng@arm.com>
449 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
450 and set PTYPE_SEQUENTIAL for merged partition.
452 2017-10-12 Richard Biener <rguenther@suse.de>
454 PR tree-optimization/69728
456 2017-09-19 Richard Biener <rguenther@suse.de>
458 PR tree-optimization/69728
459 * graphite-sese-to-poly.c (schedule_error): New global.
460 (add_loop_schedule): Handle empty domain by failing the
462 (build_original_schedule): Handle schedule_error.
464 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
465 domain by returning an unchanged schedule.
467 2017-10-12 Jakub Jelinek <jakub@redhat.com>
469 * genrecog.c (validate_pattern): For VEC_SELECT verify that
470 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
472 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
474 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
477 2017-10-12 Jakub Jelinek <jakub@redhat.com>
480 * expr.c (store_field): Don't optimize away bitsize == 0 store
481 from CALL_EXPR with addressable return type.
483 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
485 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
486 * config/rs6000/rs6000.md (sel): Delete mode attribute.
487 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
488 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
489 TARGET_ISEL instead of TARGET_ISEL<sel>.
491 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
493 * config/rs6000/rs6000.c
494 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
496 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
498 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
500 (reg_or_cint_operand): Fix comment.
501 (reg_or_zero_operand): New predicate.
502 * config/rs6000/rs6000-protos.h (output_isel): Delete.
503 * config/rs6000/rs6000.c (output_isel): Delete.
504 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
505 instead of reg_or_cint_operand. Output instruction directly (not via
507 (isel_unsigned_<mode>): Ditto.
508 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
509 gpc_reg_operand. Add an instruction alternative for this. Output
510 instruction directly.
511 (*isel_reversed_unsigned_<mode>): Ditto.
513 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
515 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
516 (TARGET_CANONICALIZE_COMPARISON): Define.
518 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
521 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
522 Check whether the dest is REG before adding REG_EQUIV note.
524 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
527 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
529 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
530 (make_hard_regno_born, make_hard_regno_dead): Update
531 bb_killed_pseudos and bb_gen_pseudos.
533 2017-10-11 Nathan Sidwell <nathan@acm.org>
535 * incpath.h (enum incpath_kind): Name enum, prefix values.
536 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
537 * incpath.c (heads, tails): Use INC_MAX.
538 (add_env_var_paths, add_standard_paths): Use incpath_kind.
539 (merge_include_chains, split_quote_chain,
540 register_include_chains): Update incpath_kind names.
541 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
542 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
544 (add_framework_path, darwin_register_objc_includes): Likewise.
545 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
547 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
549 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
550 Do not use float_operator operator predicate.
551 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
552 * config/i386/predicates.md (float_operator): Remove predicate.
554 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
556 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
557 (*jccxf_i387): Ditto.
558 (*jcc<mode>_i387): Ditto.
559 (*jccu<mode>_i387): Ditto.
560 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
561 (*jcc_*_i387 splitters): Remove.
562 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
563 * config/i386/i386.c (ix86_split_fp_branch): Remove.
564 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
567 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
569 * profile-count.h (slow_safe_scale_64bit): New function.
570 (safe_scale_64bit): New inline.
571 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
572 * profile-count.c: Include wide-int.h
573 (slow_safe_scale_64bit): New.
575 2017-10-11 Nathan Sidwell <nathan@acm.org>
577 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
578 HAS_DECL_ASSEMBLER_NAME_P.
579 * gimple-expr.c (gimple_decl_printable_name: Check
580 HAS_DECL_ASSEMBLER_NAME_P too.
581 * ipa-utils.h (type_in_anonymous_namespace_p): Check
582 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
583 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
584 * passes.c (rest_of_decl_compilation): Check
585 HAS_DECL_ASSEMBLER_NAME_P too.
586 * recog.c (verify_changes): Likewise.
587 * tree-pretty-print.c (dump_decl_name): Likewise.
588 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
590 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
591 (SET_DECL_ASSEMBLER_NAME): Use it.
592 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
593 (COPY_DECL_ASSEMBLER_NAME): Likewise.
594 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
596 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
598 * config.gcc (i386, x86_64): Add extra objects.
599 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
600 (ix86_min_insn_size): Declare.
601 (ix86_issue_rate): Declare.
602 (ix86_adjust_cost): Declare.
603 (ia32_multipass_dfa_lookahead): Declare.
604 (ix86_macro_fusion_p): Declare.
605 (ix86_macro_fusion_pair_p): Declare.
606 (ix86_bd_has_dispatch): Declare.
607 (ix86_bd_do_dispatch): Declare.
608 (ix86_core2i7_init_hooks): Declare.
609 (ix86_atom_sched_reorder): Declare.
610 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
611 (COSTS_N_BYTES): Move to x86-tune-costs.h.
612 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
613 (rip_relative_addr_p): Rename to ...
614 (ix86_rip_relative_addr_p): ... this one; export.
615 (memory_address_length): Update.
616 (ix86_issue_rate): Move to x86-tune-sched.c.
617 (ix86_flags_dependent): Move to x86-tune-sched.c.
618 (ix86_agi_dependent): Move to x86-tune-sched.c.
619 (exact_dependency_1): Move to x86-tune-sched.c.
620 (exact_store_load_dependency): Move to x86-tune-sched.c.
621 (ix86_adjust_cost): Move to x86-tune-sched.c.
622 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
623 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
624 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
625 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
626 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
627 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
628 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
629 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
630 (min_insn_size): Rename to ...
631 (ix86_min_insn_size): ... this one; export.
632 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
633 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
634 (core2i7_first_cycle_multipass_backtrack): Move to
635 x86-tune-sched-core.c.
636 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
637 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
638 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
639 (ix86_avoid_jump_mispredicts): Update.
640 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
641 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
642 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
643 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
644 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
645 (MAX_INSN): Move to ix86-tune-sched-bd.c.
646 (MAX_IMM): Move to ix86-tune-sched-bd.c.
647 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
648 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
649 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
650 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
651 (MAX_STORE): Move to ix86-tune-sched-bd.c.
652 (BIG): Move to ix86-tune-sched-bd.c.
653 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
654 (enum insn_path): Move to ix86-tune-sched-bd.c.
655 (get_mem_group): Move to ix86-tune-sched-bd.c.
656 (is_cmp): Move to ix86-tune-sched-bd.c.
657 (dispatch_violation): Move to ix86-tune-sched-bd.c.
658 (is_branch): Move to ix86-tune-sched-bd.c.
659 (is_prefetch): Move to ix86-tune-sched-bd.c.
660 (init_window): Move to ix86-tune-sched-bd.c.
661 (allocate_window): Move to ix86-tune-sched-bd.c.
662 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
663 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
664 (process_end_window): Move to ix86-tune-sched-bd.c.
665 (allocate_next_window): Move to ix86-tune-sched-bd.c.
666 (find_constant): Move to ix86-tune-sched-bd.c.
667 (get_num_immediates): Move to ix86-tune-sched-bd.c.
668 (has_immediate): Move to ix86-tune-sched-bd.c.
669 (get_insn_path): Move to ix86-tune-sched-bd.c.
670 (get_insn_group): Move to ix86-tune-sched-bd.c.
671 (count_num_restricted): Move to ix86-tune-sched-bd.c.
672 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
673 (add_insn_window): Move to ix86-tune-sched-bd.c.
674 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
675 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
676 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
677 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
678 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
679 (do_dispatch): Move to ix86-tune-sched-bd.c.
680 (has_dispatch): Move to ix86-tune-sched-bd.c.
681 * config/i386/t-i386: Add new object files.
682 * config/i386/x86-tune-costs.h: New file.
683 * config/i386/x86-tune-sched-atom.c: New file.
684 * config/i386/x86-tune-sched-bd.c: New file.
685 * config/i386/x86-tune-sched-core.c: New file.
686 * config/i386/x86-tune-sched.c: New file.
688 2017-10-11 Liu Hao <lh_mouse@126.com>
690 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
691 the generic version below instead.
692 (should_colorize): Recognize Windows consoles as terminals
694 * pretty-print.c [__MINGW32__] (write_all): New function.
695 [__MINGW32__] (find_esc_head): Likewise.
696 [__MINGW32__] (find_esc_terminator): Likewise.
697 [__MINGW32__] (eat_esc_sequence): Likewise.
698 [__MINGW32__] (mingw_ansi_fputs): New function that handles
700 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
703 2017-10-11 Richard Biener <rguenther@suse.de>
705 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
706 Properly call analyze_scalar_evolution with the loop of the stmt.
708 2017-10-11 Richard Biener <rguenther@suse.de>
710 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
711 * tree-core.h (tree_base): Add chrec_var union member.
712 * tree.h (CHREC_VAR): Remove.
713 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
714 * tree-chrec.h (build_polynomial_chrec): Adjust.
715 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
716 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
718 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
720 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
721 * match.pd: ... here.
722 ((T) X == (T) Y): Relax condition.
724 2017-10-11 Bin Cheng <bin.cheng@arm.com>
726 PR tree-optimization/82472
727 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
729 (break_alias_scc_partitions): Update postorder number.
731 2017-10-11 Martin Liska <mliska@suse.cz>
734 * opts.c (parse_no_sanitize_attribute): Do not use error_value
736 * opts.h (parse_no_sanitize_attribute): Remove last argument.
738 2017-10-11 Martin Liska <mliska@suse.cz>
740 * print-rtl.c (print_insn): Move declaration of idbuf
741 to same scope as name.
743 2017-10-11 Martin Liska <mliska@suse.cz>
748 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
751 2017-10-11 Martin Liska <mliska@suse.cz>
754 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
757 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
759 * config.gcc (default_gnu_indirect_function): Default to yes for
760 arm*-*-linux* with glibc.
762 2017-10-11 Richard Biener <rguenther@suse.de>
764 * tree-scalar-evolution.c (get_scalar_evolution): Handle
765 default-defs and types we do not want to analyze.
766 (interpret_loop_phi): Replace unreachable code with an assert.
767 (compute_scalar_evolution_in_loop): Remove and inline ...
768 (analyze_scalar_evolution_1): ... here, replacing condition with
769 what makes the intent clearer. Remove handling of cases
770 get_scalar_evolution now handles.
772 2017-10-10 Jim Wilson <wilson@tuliptree.org>
774 PR rtl-optimization/81434
775 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
776 comment for main loop. In sched_group_found if, also add checks for
777 pass and min_cost_group.
779 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
781 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
782 (rs6000_insn_cost): New function.
783 * config/rs6000/rs6000.md (cost): New attribute.
785 2017-10-10 Jakub Jelinek <jakub@redhat.com>
786 H.J. Lu <hongjiu.lu@intel.com>
790 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
791 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
792 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
793 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
794 to OPTION_MASK_ISA_AVX512VL - builtins that have both
795 OPTION_MASK_ISA_MMX and some other bit set require both
796 mmx and the ISAs without the mmx bit.
797 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
798 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
799 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
800 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
801 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
802 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
803 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
804 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
805 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
806 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
807 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
808 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
809 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
810 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
811 Add OPTION_MASK_ISA_MMX.
813 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
815 * config.gcc (armv7*-*-freebsd*): New target.
816 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
818 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
820 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
821 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
824 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
826 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
827 a template parameter.
828 (WIDE_INT_REF_FOR): Update accordingly.
829 * tree.h (wi::int_traits <const_tree>): Delete.
830 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
831 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
832 (wi::tree_to_wide_ref): New typedef.
833 (wi::to_wide): New function.
834 * calls.c (get_size_range): Use wi::to_wide when operating on
836 * cgraph.c (cgraph_node::create_thunk): Likewise.
837 * config/i386/i386.c (ix86_data_alignment): Likewise.
838 (ix86_local_alignment): Likewise.
839 * dbxout.c (stabstr_O): Likewise.
840 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
841 * expr.c (const_vector_from_tree): Likewise.
842 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
843 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
844 (fold_negate_expr_1, int_const_binop_1, const_binop)
845 (fold_convert_const_int_from_real, optimize_bit_field_compare)
846 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
847 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
848 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
849 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
850 (fold_not_const, round_up_loc): Likewise.
851 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
852 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
853 (alloca_call_type): Likewise.
854 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
855 * godump.c (go_output_typedef): Likewise.
856 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
857 * internal-fn.c (get_min_precision): Likewise.
858 * ipa-cp.c (ipcp_store_vr_results): Likewise.
859 * ipa-polymorphic-call.c
860 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
861 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
862 (ipa_modify_call_arguments): Likewise.
863 * match.pd: Likewise.
864 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
865 * print-tree.c (print_node_brief, print_node): Likewise.
866 * stmt.c (expand_case): Likewise.
867 * stor-layout.c (layout_type): Likewise.
868 * tree-affine.c (tree_to_aff_combination): Likewise.
869 * tree-cfg.c (group_case_labels_stmt): Likewise.
870 * tree-data-ref.c (dr_analyze_indices): Likewise.
871 (prune_runtime_alias_test_list): Likewise.
872 * tree-dump.c (dequeue_and_dump): Likewise.
873 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
874 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
875 * tree-pretty-print.c (dump_generic_node): Likewise.
876 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
877 (simple_iv_with_niters): Likewise.
878 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
879 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
880 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
881 * tree-ssa-loop-niter.c (split_to_var_and_offset)
882 (refine_value_range_using_guard, number_of_iterations_ne_max)
883 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
884 (get_cst_init_from_scev, record_nonwrapping_iv)
885 (scev_var_range_cant_overflow): Likewise.
886 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
887 * tree-ssa-pre.c (compute_avail): Likewise.
888 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
889 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
890 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
891 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
892 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
893 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
894 (dump_case_nodes, try_switch_expansion): Likewise.
895 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
896 (vect_do_peeling): Likewise.
897 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
898 * tree-vect-stmts.c (vectorizable_load): Likewise.
899 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
900 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
901 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
902 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
903 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
904 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
905 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
906 (evrp_dom_walker::before_dom_children): Likewise.
907 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
908 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
909 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
910 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
911 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
912 (get_range_pos_neg): Likewise.
913 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
914 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
915 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
916 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
917 * config/avr/avr.c (avr_fold_builtin): Likewise.
918 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
919 * config/msp430/msp430.c (msp430_attr): Likewise.
920 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
921 * config/powerpcspe/powerpcspe-c.c
922 (altivec_resolve_overloaded_builtin): Likewise.
923 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
924 (rs6000_expand_ternop_builtin): Likewise.
925 * config/rs6000/rs6000-c.c
926 (altivec_resolve_overloaded_builtin): Likewise.
927 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
928 (rs6000_expand_ternop_builtin): Likewise.
929 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
931 2017-10-10 Bin Cheng <bin.cheng@arm.com>
933 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
934 when copying loop nest with only one inner loop.
936 2017-10-10 Richard Biener <rguenther@suse.de>
938 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
939 blocks if SCEV is active.
940 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
942 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
943 (scev_initialize): Assert we are not yet initialized.
945 2017-10-10 Bin Cheng <bin.cheng@arm.com>
947 * tree-loop-distribution.c (generate_loops_for_partition): Remove
948 inner loop's exit stmt by making it always exit the loop, otherwise
949 we would generate an infinite empty loop.
951 2017-10-10 Bin Cheng <bin.cheng@arm.com>
953 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
954 renaming variables in new preheader if it's deleted.
956 2017-10-10 Bin Cheng <bin.cheng@arm.com>
958 * tree-loop-distribution.c (struct partition): Remove unused field
959 loops of the structure.
960 (partition_alloc, partition_free): Ditto.
961 (build_rdg_partition_for_vertex): Ditto.
963 2017-10-09 Jeff Law <law@redhat.com>
965 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
966 return type to match prototype and documentation.
968 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
970 * config/rs6000/rs6000.c (processor_costs): Move to ...
971 * config/rs6000/rs6000.h: ... here.
972 (rs6000_cost): Declare.
974 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
976 * except.c (setjmp_fn): New global variable.
977 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
978 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
979 if DONT_USE_BUILTIN_SETJMP is defined.
981 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
983 * target.def (insn_cost): New hook.
984 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
985 * doc/tm.texi: Regenerate.
986 * rtlanal.c (insn_cost): Use the new hook.
988 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
990 * combine.c (combine_validate_cost): Compute the new insn_cost,
991 not just pattern_cost.
992 (try_combine): Adjust comment.
994 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
996 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
998 * combine.c (uid_insn_cost): Adjust comment.
999 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1001 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1002 * dse.c (find_shift_sequence): Ditto.
1003 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1004 (bb_valid_for_noce_process_p): Use pattern_cost.
1005 * rtl.h (insn_rtx_cost): Delete.
1006 (pattern_cost): New prototype.
1007 (insn_cost): New prototype.
1008 * rtlanal.c (insn_rtx_cost): Rename to...
1009 (pattern_cost): ... this.
1012 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1014 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1015 (*jcc<mode>_0_r_i387): Ditto.
1016 (*jccxf_r_i387): Ditto.
1017 (*jcc<mode>_r_i387): Ditto.
1018 (*jccu<mode>_r_i387): Ditto.
1019 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1020 (*jcc): Rename from *jcc_1.
1022 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1024 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1025 deferred rescans after the lvx/stvx recombination pre-pass.
1027 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1029 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1030 memory operation instruction support.
1031 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1032 (rs6000-ibm-aix[789]*): Likewise.
1033 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1034 Document new functions.
1036 2017-10-09 Richard Biener <rguenther@suse.de>
1038 PR tree-optimization/82397
1039 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1040 equality only for semantically equal trees.
1042 2017-10-09 Richard Biener <rguenther@suse.de>
1044 PR tree-optimization/82449
1045 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1046 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1047 allow constant addresses.
1048 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1051 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1053 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1056 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1059 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1062 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1065 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1067 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1070 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1071 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1073 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1075 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1076 (WI_BINARY_PREDICATE_RESULT): Likewise.
1077 (wi::binary_traits::operator_result): New type.
1078 (wi::binary_traits::predicate_result): Likewise.
1079 (generic_wide_int::operator~, unary generic_wide_int::operator-)
1080 (generic_wide_int::operator==, generic_wide_int::operator!=)
1081 (generic_wide_int::operator&, generic_wide_int::and_not)
1082 (generic_wide_int::operator|, generic_wide_int::or_not)
1083 (generic_wide_int::operator^, generic_wide_int::operator+
1084 (binary generic_wide_int::operator-, generic_wide_int::operator*):
1086 (operator~, unary operator-, operator==, operator!=, operator&)
1087 (operator|, operator^, operator+, binary operator-, operator*): New
1089 * expr.c (get_inner_reference): Use wi::bit_and_not.
1090 * fold-const.c (fold_binary_loc): Likewise.
1091 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
1092 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
1093 (bit_value_binop): Likewise.
1094 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
1095 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
1096 (extract_range_from_binary_expr_1): Likewise.
1097 (masked_increment): Likewise.
1098 (simplify_bit_ops_using_ranges): Likewise.
1100 2017-10-09 Martin Jambor <mjambor@suse.cz>
1103 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
1104 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
1105 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
1107 (hsa_fixup_mov_insn_type): New function.
1108 (hsa_op_with_type::get_in_type): Use it.
1109 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
1110 immediates in an assert.
1111 (hsa_op_with_type::extend_int_to_32bit): New method.
1112 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
1113 types. Convert to dest type if necessary.
1114 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
1115 (reg_for_gimple_ssa): Pass false as min32int to
1116 hsa_type_for_scalar_tree_type.
1117 (gen_hsa_addr): Fixup type when creating addresable temporary.
1118 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
1119 (gen_hsa_unary_operation): Extend operands and convert to dest type if
1120 necessary. Call hsa_fixup_mov_insn_type.
1121 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
1122 extend operands and convert to dest type if necessary.
1123 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
1124 to dest type if necessary.
1125 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
1126 if conversion nt necessary and size matches.
1127 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
1128 to dest type if necessary.
1129 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
1130 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
1132 (gen_hsa_clrsb): Likewise.
1133 (gen_hsa_ffs): Likewise.
1134 (gen_hsa_divmod): Extend operands and convert to dest type if
1136 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
1138 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1140 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
1141 Remove empty default arguments. Use a brace block as output
1143 (conditional return): Ditto.
1145 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
1146 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
1148 (group_ending_nop): Ditto.
1149 (doloop_end): Ditto.
1150 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
1151 (splitters for those): Ditto.
1153 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1155 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
1156 a conditional jump (and the compare for it) so that pc_rtx is the
1158 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
1159 for the deleted and renamed ctr<mode>_internal[234] patterns.
1160 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
1161 Delete second conditional return pattern.
1162 (ctr<mode>_internal2): Delete this second bdnz pattern.
1163 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
1164 (ctr<mode>_internal4): Delete this second bdz pattern.
1166 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1168 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
1169 (always_initialized_rtx_for_ssa_name_p): New predicate.
1170 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
1171 (finish_out_of_ssa): Free new field of SA.
1172 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
1173 * tree-ssa-coalesce.c: Include tree-ssa.h.
1174 (get_parm_default_def_partitions): Remove extern keyword.
1175 (get_undefined_value_partitions): New function.
1176 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
1177 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
1179 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
1181 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1183 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
1185 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1187 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
1188 for targets that preffer 128bit.
1190 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1192 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
1194 2017-10-08 Olivier Hainque <hainque@adacore.com>
1196 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
1197 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
1199 (thumb_set_return_address): Likewise.
1201 2017-10-08 Olivier Hainque <hainque@adacore.com>
1203 * common/config/arm/arm-common.c (arm_except_unwind_info):
1204 Handle DWARF2_UNWIND_INFO.
1206 2017-10-07 Michael Collison <michael.collison@arm.com>
1208 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
1211 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
1213 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
1214 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
1215 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
1216 defined, force the creation of a new block for a dispatch label.
1218 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
1220 * invoke.texi (Wsuggest-attribute=cold): Document.
1221 * common.opt (Wsuggest-attribute=cold): New
1222 * ipa-pure-const.c (warn_function_cold): New function.
1223 * predict.c (compute_function_frequency): Use it.
1224 * predict.h (warn_function_cold): Declare.
1226 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
1228 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
1231 2017-10-06 Martin Liska <mliska@suse.cz>
1233 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
1234 keyword for member functions.
1235 (struct sanopt_tree_couple): New struct.
1236 (struct sanopt_tree_couple_hash): New function.
1237 (struct sanopt_ctx): Add new hash_map.
1238 (has_dominating_ubsan_ptr_check): New function.
1239 (record_ubsan_ptr_check_stmt): Likewise.
1240 (maybe_optimize_ubsan_ptr_ifn): Likewise.
1241 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
1242 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
1244 2017-10-06 Sudakshina Das <sudi.das@arm.com>
1247 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
1248 aarch64_simd_valid_immediate on CONST_VECTORs.
1249 (aarch64_reg_or_bic_imm): Likewise.
1251 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
1253 PR rtl-optimization/82396
1254 * haifa-sched.c (ready_sort_real): Disable qsort checking.
1256 2017-10-06 Sebastian Pop <sebpop@gmail.com>
1258 * graphite-dependences.c (scop_get_reads): Move code to...
1259 (scop_get_must_writes): Move code to...
1260 (scop_get_may_writes): Move code to...
1261 (scop_get_reads_and_writes): ... here.
1262 (scop_get_dependences): Call scop_get_reads_and_writes.
1264 2017-10-06 Jakub Jelinek <jakub@redhat.com>
1266 PR tree-optimization/82434
1267 * fold-const.h (can_native_encode_type_p,
1268 can_native_encode_string_p): Remove.
1269 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
1270 don't encode anything, just return what would be otherwise returned.
1271 (native_encode_fixed, native_encode_complex, native_encode_vector):
1273 (native_encode_string): Likewise. Inline by hand
1274 can_native_encode_string_p.
1275 (can_native_encode_type_p): Remove.
1276 (can_native_encode_string_p): Remove.
1277 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
1278 STRING_CSTs using can_native_encode_string_p, test all
1279 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
1280 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
1281 argument from native_encode_expr.
1282 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
1283 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
1286 2017-10-06 Richard Biener <rguenther@suse.de>
1288 PR tree-optimization/82397
1289 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
1290 operand_equal_p but rely on data_ref_compare_tree for detecting
1292 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
1293 to match up with dr_group_sort_cmp.
1295 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1298 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
1300 * config/s390/s390-builtin-types.def: Regenerate.
1302 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1305 * config/s390/s390-builtin-types.def: Regenerate.
1306 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
1307 Change flag from B_VXE to B_VX.
1308 (s390_vec_min_dbl): Remove B_VXE flag.
1310 2017-10-06 Richard Biener <rguenther@suse.de>
1312 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
1313 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
1314 translate_isl_ast_to_gimple::is_valid_rename,
1315 translate_isl_ast_to_gimple::get_rename,
1316 translate_isl_ast_to_gimple::get_def_bb_for_const,
1317 translate_isl_ast_to_gimple::get_new_name,
1318 translate_isl_ast_to_gimple::collect_all_ssa_names,
1319 translate_isl_ast_to_gimple::copy_loop_phi_args,
1320 translate_isl_ast_to_gimple::collect_all_ssa_names,
1321 translate_isl_ast_to_gimple::copy_loop_phi_args,
1322 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
1323 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
1324 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
1325 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
1326 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
1327 translate_isl_ast_to_gimple::copy_cond_phi_args,
1328 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
1329 translate_isl_ast_to_gimple::edge_for_new_close_phis,
1330 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
1331 translate_isl_ast_to_gimple::rename_uses,
1332 translate_isl_ast_to_gimple::rename_all_uses): Remove.
1333 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
1334 (set_rename_for_each_def): Likewise.
1335 (graphite_copy_stmts_from_block): Handle debug stmt resetting
1336 here. Handle rewriting SCEV analyzable uses here.
1337 (copy_bb_and_scalar_dependences): Generate code for PHI
1339 (graphite_regenerate_ast_isl): Adjust.
1340 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
1341 (add_write, add_read): New functions.
1342 (build_cross_bb_scalars_def): Use it and simplify.
1343 (build_cross_bb_scalars_use): Likewise.
1344 (graphite_find_cross_bb_scalar_vars): Inline into...
1345 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
1346 simulating out-of-SSA. Compute liveout and add dependencies.
1347 (build_scops): Force an empty entry block.
1348 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
1350 (sese_build_liveouts): Declare.
1351 (sese_trivially_empty_bb_p): Likewise.
1352 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
1353 compute liveout and debug_liveout.
1354 (sese_bad_liveouts_use): Remove.
1355 (sese_reset_debug_liveouts_bb): Likewise.
1356 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
1357 (sese_build_liveouts): Build liveout and debug_liveout and store
1359 (new_sese_info): Adjust.
1360 (free_sese_info): Likewise.
1361 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
1362 do not build liveout here.
1363 (move_sese_in_condition): Adjust region entry.
1364 (scev_analyzable_p): Match up with chrec_apply requirements.
1365 (sese_trivially_empty_bb_p): New.
1366 * tree-into-ssa.c (get_reaching_def): Properly support generating
1367 default-defs for incremental rewrite of anonymous names.
1369 2017-10-06 Richard Biener <rguenther@suse.de>
1371 * graphite-sese-to-poly.c (extract_affine): For casts increasing
1372 precision do not perform modulo reduction.
1374 2017-10-06 Richard Biener <rguenther@suse.de>
1376 PR tree-optimization/82436
1377 * tree-vect-slp.c (vect_supported_load_permutation_p): More
1378 conservatively choose the vectorization factor when checking
1379 whether we can perform the required load permutation.
1380 (vect_transform_slp_perm_load): Assert when we may not fail.
1382 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1384 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
1385 message for incompatible -msdata=* and -mcall-* options.
1387 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1389 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
1390 rate for post-reload scheduling.
1392 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1394 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
1396 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1398 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
1399 to improve monte carlo in scimark.
1401 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1403 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
1404 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
1405 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
1406 pentium4_cost, nocona_cost): Set reassociation width to 1.
1407 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
1408 width to 2 for fp operations and 1 otherwise.
1409 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
1411 (atom_cost): Set reassociation width to 2.
1412 (slm_cost, generic_cost): Set fp reassociation width
1413 to 2 and 1 otherwise.
1414 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
1415 (core_cost): Set fp reassociation width to 4 and vector to 2.
1416 (ix86_reassociation_width): Rewrite using cost table; special case
1417 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
1418 and TARGET_AVX128_OPTIMAL.
1419 * config/i386/i386.h (processor_costs): Add
1420 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
1421 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
1422 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
1423 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
1424 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
1425 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
1427 2017-10-05 Nathan Sidwell <nathan@acm.org>
1429 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
1431 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1433 * config/arm/arm.c (arm_test_fpu_data): New.
1434 (arm_run_selftests): Call arm_test_fpu_data.
1436 2017-10-04 Nathan Sidwell <nathan@acm.org>
1438 * toplev.c (toplev::main): Remove excess parens on pretty_printer
1440 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
1442 2017-10-04 Sudakshina Das <sudi.das@arm.com>
1444 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
1445 check type for aarch64_simd_valid_immediate.
1446 (aarch64_output_simd_mov_immediate): Update prototype.
1447 (aarch64_simd_valid_immediate): Update prototype.
1448 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
1449 support for ORR-immediate.
1450 (and<mode>3): modified pattern to add support for BIC-immediate.
1451 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
1452 now checks for valid immediate for BIC and ORR based on new enum
1454 (aarch64_output_simd_mov_immediate): Function now used to output
1455 BIC/ORR imm as well based on new enum argument.
1456 * config/aarch64/constraints.md (Do): New vector immediate constraint.
1458 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
1459 (aarch64_reg_or_bic_imm): Likewise.
1461 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1463 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
1464 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
1466 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1470 PR rtl-optimization/82396
1471 * haifa-sched.c (autopref_multipass_init): Simplify
1473 (autopref_rank_data): Simplify sort order.
1474 * sched-int.h (autopref_multipass_data_): Remove
1475 multi_mem_insn_p, min_offset and max_offset.
1477 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1479 * doc/sourcebuild.texi: Document vect_peeling_profitable.
1481 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1483 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
1486 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1488 * doc/sourcebuild.texi: Document vect_long_mult.
1490 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
1492 PR tree-optimization/82413
1493 * fold-const.c (build_range_check): Use widest_int when comparing
1494 the maximum ETYPE value with HIGH.
1496 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1498 PR rtl-optimization/82396
1499 * haifa-sched.c (autopref_multipass_init): Simplify
1501 (autopref_rank_data): Simplify sort order.
1502 * sched-int.h (autopref_multipass_data_): Remove
1503 multi_mem_insn_p, min_offset and max_offset.
1505 2017-10-04 Jakub Jelinek <jakub@redhat.com>
1507 PR tree-optimization/82381
1508 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
1509 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
1512 PR tree-optimization/82374
1513 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
1514 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
1515 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
1516 current_function_decl to the new decl.
1518 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
1520 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
1521 helper macro for IEEE float128 hardware built-in functions.
1522 (SQRTF128_ODD): Add built-in functions with the round-to-odd
1524 (TRUNCF128_ODD): Likewise.
1525 (ADDF128_ODD): Likewise.
1526 (SUBF128_ODD): Likewise.
1527 (MULF128_ODD): Likewise.
1528 (DIVF128_ODD): Likewise.
1529 (FMAF128_ODD): Likewise.
1530 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
1531 UNSPEC_TRUNC_ROUND_TO_ODD.
1532 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
1533 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
1534 floating point round to odd instructions.
1535 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
1536 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
1537 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
1538 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
1539 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
1540 (trunc<mode>sf2_hw): Change the truncate with round to odd
1541 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
1542 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
1543 to odd hardware instructions.
1544 (sub<mode>3_odd): Likewise.
1545 (mul<mode>3_odd): Likewise.
1546 (div<mode>3_odd): Likewise.
1547 (sqrt<mode>2_odd): Likewise.
1548 (fma<mode>4_odd): Likewise.
1549 (fms<mode>4_odd): Likewise.
1550 (nfma<mode>4_odd): Likewise.
1551 (nfms<mode>4_odd): Likewise.
1552 (trunc<mode>df2_odd): Change the truncate with round to odd
1553 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
1555 * doc/extend.texi (PowerPC built-in functions): Update documentation
1556 for existing IEEE float128-bit built-in functions. Add built-in
1557 functions that generate the IEEE 128-bit floating point round to
1560 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
1562 PR rtl-optimization/77729
1563 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
1564 to (X&(C1&~C2))|C2 transformations.
1566 2017-10-03 Martin Jambor <mjambor@suse.cz>
1568 PR tree-optimization/82363
1569 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
1570 mismatch, mark lacc written regardless of racc.
1572 2017-10-03 Jakub Jelinek <jakub@redhat.com>
1574 PR tree-optimization/82381
1575 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
1576 stmt_to_insert nor wheather SSA_NAMEs are default defs.
1577 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
1578 fallthrough into reassoc_stmt_dominates_stmt_p.
1581 * combine.c (combine_instructions): Don't combine in unreachable
1584 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1587 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
1588 function to not use the have_cpu variable. Do not set cpu_index,
1589 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
1591 (rs6000_valid_attribute_p): Remove duplicate initializations of
1592 old_optimize and func_optimize.
1593 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
1594 (rs6000_activate_target_options): Make global.
1595 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
1598 2017-10-02 Jakub Jelinek <jakub@redhat.com>
1600 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
1601 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
1602 Set *poffset to 0 and *psize and *pmax_size to -1 if
1603 *poffset + *psize overflows in HOST_WIDE_INT.
1605 PR tree-optimization/82387
1606 PR tree-optimization/82388
1607 PR tree-optimization/82389
1608 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
1609 instead of live_bytes non-NULL.
1611 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
1614 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
1617 2017-10-02 Richard Biener <rguenther@suse.de>
1619 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
1620 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
1621 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
1623 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1625 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
1626 requested precision matches the type's.
1627 * calls.c (alloc_max_size): Calculate the new candidate size as
1628 a widest_int and use wi::to_widest when comparing it with the
1629 current candidate size.
1630 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
1631 zero rather than integer_zero_node.
1632 * match.pd: Check for a no-op conversion before using wi::add
1633 rather than after. Use tree_to_uhwi when summing small shift
1634 counts into an unsigned int.
1636 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1637 Alan Hayward <alan.hayward@arm.com>
1638 David Sherwood <david.sherwood@arm.com>
1641 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
1642 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
1643 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
1644 POINTER_AND_FP_REGS.
1646 2017-10-02 Richard Biener <rguenther@suse.de>
1648 PR tree-optimization/82355
1649 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1650 a mapping for the enclosing loop but avoid generating one for
1652 (copy_bb_and_scalar_dependences): Remove premature codegen
1653 error on PHIs in blocks duplicated into multiple places.
1654 * graphite-scop-detection.c
1655 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1656 in the region use it as loop and nest to analyze the DR in.
1657 (try_generate_gimple_bb): Likewise.
1658 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1659 (add_loop_constraints): For blocks in a loop not in the region
1660 create a dimension with a single iteration.
1661 * sese.h (gbb_loop_at_index): Remove assert.
1663 2017-10-01 Kevin Buettner <kevinb@redhat.com>
1665 * omp-expand.c (adjust_context_scope): New function.
1666 (expand_parallel_call): Call adjust_context_scope.
1668 2017-10-01 Jeff Law <law@redhat.com>
1670 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
1671 dom_opt_dom_walker class with direct access to private members.
1672 Add comments. Call test_for_singularity.
1673 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
1674 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
1675 m_dummy_cond anymore.
1676 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
1678 (pass_dominator:execute): Build the dummy_cond here and pass it
1679 to the dom_opt_dom_walker ctor.
1680 (test_for_singularity): New function.
1682 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
1683 Maya Rashish <coypu@sdf.org>
1685 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
1686 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
1687 (alpha*-*-netbsd*) Use nbsd_tm_file.
1688 (arm*-*-netbsdelf*) Likewise.
1689 (i[34567]86-*-netbsdelf*) Likewise.
1690 (x86_64-*-netbsd*) Likewise.
1691 (mips*-*-netbsd*) Likewise.
1692 (powerpc-*-netbsd*) Likewise.
1693 (sh*-*-netbsd*) Likewise.
1694 (sparc-*-netbsdelf*) Likewise.
1695 (sparc64-*-netbsd*) Likewise.
1696 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
1698 (vax-*-netbsdelf*) Likewise.
1699 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
1700 (UINT_FAST8_TYPE) Likewise.
1701 (INT_FAST16_TYPE) Check CHAR_FAST16.
1702 (UINT_FAST16_TYPE) Likewise.
1704 2017-09-30 Jakub Jelinek <jakub@redhat.com>
1707 * config/i386/i386.md
1708 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
1709 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
1710 *divmodsi4_zext_2): New define_insn_and_split.
1711 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
1712 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
1713 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
1714 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
1715 New define_insn_and_split.
1716 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
1717 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
1718 operands[1] having DImode when mode is SImode.
1720 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
1721 always SImode for DIV and MOD in REG_EQUAL notes.
1723 2017-09-29 Yury Gribov <tetra2005@gmail.com>
1726 * match.pd: Fix handling of NaNs in pattern.
1728 2017-09-29 Jeff Law <law@redhat.com>
1730 * sbitmap.c (bitmap_bit_in_range_p): New function.
1731 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
1732 * tree-ssa-dse.c (live_bytes_read): New function.
1733 (dse_classify_store): Ignore reads of dead bytes.
1735 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
1736 typos and whitespace errors.
1737 * config/i386/predicates.md (address_no_seg_operand): Likewise.
1738 * config/s390/s390.c (s390_emit_prologue): Likewise.
1740 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1743 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
1744 with a symbol for LRA.
1746 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1748 PR rtl-optimization/82338
1749 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
1751 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
1753 * genmodes.c (calc_wider_mode): Suppress qsort macro.
1754 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
1755 (qsort_chk): Declare.
1756 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
1757 (qsort_chk): New function.
1759 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1761 PR tree-optimization/82337
1762 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
1763 phi definition if the PHI result appears in an abnormal PHI.
1764 (find_basis_for_base_expr): Don't record a basis if the LHS of the
1765 basis appears in an abnormal PHI.
1767 2017-09-29 Richard Biener <rguenther@suse.de>
1769 * graphite-isl-ast-to-gimple.c
1770 (translate_isl_ast_to_gimple::set_codegen_error): New function.
1771 (binary_op_to_tree): Use it.
1772 (get_rename_from_scev): Likewise.
1773 (copy_loop_phi_nodes): Likewise.
1774 (copy_bb_and_scalar_dependences): Likewise.
1775 (translate_pending_phi_nodes): Likewise.
1777 2017-09-29 Jakub Jelinek <jakub@redhat.com>
1780 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
1781 for movabsq $(i32 << shift), r64.
1783 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
1785 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
1786 index when encoding %esp as %rsp to avoid 0x67 prefix.
1788 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
1790 * config/i386/i386.md (*movsf_internal, *movdf_internal):
1791 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
1793 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
1795 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
1796 Extensions with more than 16 double VFP registers.
1797 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
1798 to_clear_mask and all code related to it. Replace the remaining
1799 entry by a sbitmap and adapt code accordingly.
1801 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
1803 * brig-builtins.def: Change pure attributes to const.
1805 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1807 * config.gcc (default_gnu_indirect_function): Default to yes for
1808 sparc*-*-linux* with glibc.
1810 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1812 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
1813 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
1814 when creating .init_array and .fini_array sections with priority
1817 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
1820 * config/aarch64/aarch64.c
1821 (aarch64_builtin_support_vector_misalignment): Always return false
1822 when misalignment is unknown.
1824 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1826 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
1827 this function to return false if the definition used by the swap
1828 instruction is artificial, or if the memory address from which the
1829 constant value is loaded is not represented by a base address held
1830 in a register or if the base address register is a frame or stack
1831 pointer. Additionally, return false if the base address of the
1832 loaded constant is a SYMBOL_REF but is not considered to be a
1834 (replace_swapped_load_constant): New function.
1835 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
1836 loaded constant vector with a load of a swapped constant vector.
1838 2017-09-27 Carl Love <cel@us.ibm.com>
1840 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
1841 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
1842 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
1845 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
1847 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
1848 first, always call autopref_rank_data otherwise.
1850 2017-09-27 Richard Biener <rguenther@suse.de>
1852 * graphite-scop-detection.c (find_scop_parameters): Move
1853 loop bound handling ...
1854 (gather_bbs::before_dom_children): ... here, avoiding the need
1855 to build scop_info->loop_nest.
1856 (record_loop_in_sese): Remove.
1857 * sese.h (sese_info_t::loop_nest): Remove.
1858 * sese.c (new_sese_info): Do not allocate loop_nest.
1859 (free_sese_info): Do not free loop_nest.
1861 2017-09-27 Jakub Jelinek <jakub@redhat.com>
1864 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
1865 lhs from calls if the lhs has addressable type.
1867 2017-09-27 Richard Biener <rguenther@suse.de>
1869 * graphite.h (scop::max_alias_set): New member.
1870 * graphite-scop-detection.c: Remove references to non-existing
1871 --param in comments.
1872 (build_alias_sets): Record the maximum alias set used for drs.
1873 (build_scops): Support zero as unlimited for
1874 --param graphite-max-arrays-per-scop.
1875 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
1877 (build_poly_sr_1): ... here. Compute alias set based on the
1878 maximum alias set used for drs rather than
1879 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
1881 2017-09-27 Richard Biener <rguenther@suse.de>
1883 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
1884 --param loop-block-tile-size=0 to disable tiling.
1886 2017-09-27 Richard Biener <rguenther@suse.de>
1888 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
1889 (graphite-max-nb-scop-params): Document special value zero.
1890 * domwalk.h (dom_walker::STOP): New symbolical constant.
1891 (dom_walker::dom_walker): Add optional parameter for bb to
1893 (dom_walker::~dom_walker): Declare.
1894 (dom_walker::before_dom_children): Document STOP return value.
1895 (dom_walker::m_user_bb_to_rpo): New member.
1896 (dom_walker::m_bb_to_rpo): Likewise.
1897 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
1898 mapping here if not provided by the user.
1899 (dom_walker::~dom_walker): Free bb to RPO mapping if not
1900 provided by the user.
1901 (dom_walker::STOP): Define.
1902 (dom_walker::walk): Do not compute bb to RPO mapping here.
1903 Support STOP return value from before_dom_children to stop
1905 * graphite-optimize-isl.c (optimize_isl): If the schedule
1906 is the same still generate code if -fgraphite-identity
1907 or -floop-parallelize-all are given.
1908 * graphite-scop-detection.c: Include cfganal.h.
1909 (gather_bbs::gather_bbs): Get and pass through bb to RPO
1911 (gather_bbs::before_dom_children): Return STOP for BBs
1913 (build_scops): Compute bb to RPO mapping and pass it to
1914 the domwalk. Treat --param graphite-max-nb-scop-params=0
1915 as not limiting the number of params.
1916 * graphite.c (graphite_initialize): Remove limit on the
1917 number of basic-blocks in a function.
1918 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
1919 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
1920 default value of 10.
1922 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1924 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
1925 Adjust code to eliminate needing to do the shift right 32-bits
1926 operation after XSCVDPSPN.
1928 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1930 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
1931 ((X / Y) != 0 -> X >= Y): Likewise.
1933 2017-09-26 Carl Love <cel@us.ibm.com>
1935 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
1936 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
1937 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
1938 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
1939 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
1940 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
1941 definitions and overloading.
1942 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
1943 statement for P9V_BUILTIN_XST_LEN_R.
1944 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
1945 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
1946 define_expand and define_insn for the instructions and builtins.
1947 * doc/extend.texi: Update the built-in documentation file for the new
1949 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
1950 define_insn for the instructions
1952 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
1955 * gcc/config/netbsd-protos.h: New file.
1956 * gcc/config/netbsd.c: New file.
1957 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
1958 * gcc/config/t-netbsd: New file.
1959 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
1960 (tmake_file) Add t-netbsd.
1961 (extra_objs) Add netbsd.o.
1963 2017-09-26 Janus Weil <janus@gcc.gnu.org>
1967 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
1969 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1971 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
1972 sign extension from a vector register to a GPR by doing a 32-bit
1973 direct move and then an EXTSW.
1974 (extendsi<mode>2 splitter): Likewise.
1975 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
1976 right or vector extract after doing XSCVDPSPN. Use
1977 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
1979 (movdi_from_sf_zero_ext): Likewise.
1980 (reload_gpr_from_vsxsf): Likewise.
1981 (p8_mfvsrd_4_disf): Delete, no longer used.
1982 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
1983 then needing to move the SFmode to a GPR to use the XSCVDPSP
1984 instruction instead of FRSP and XSCVDPSPN.
1985 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
1986 it is adjacent to the other XSCVSPDP insns.
1987 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
1988 SFmode to be in traditional Altivec registers.
1989 (vsx_xscvdpspn): Eliminate useless alternative constraint.
1990 (vsx_xscvspdpn): Likewise.
1991 (vsx_xscvspdpn_scalar): Likewise.
1993 2017-09-26 Martin Jambor <mjambor@suse.cz>
1995 * tree-sra.c (compare_access_positions): Put integral types first,
1996 stabilize sorting of integral types, remove conditions putting
1997 non-full-precision integers last.
1998 (sort_and_splice_var_accesses): Disable scalarization if a
1999 non-integert would be represented by a non-full-precision integer.
2001 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2003 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2004 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2005 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2006 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2007 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2008 conditionals inside the function instead of around it. Call
2009 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2010 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2012 2017-09-26 Richard Biener <rguenther@suse.de>
2014 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2016 (scop_detection::build_scop_breadth): ... this. Removed.
2017 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2018 (scop_detection::harmful_stmt_in_bb): Likewise.
2019 (scop_detection::graphite_can_represent_stmt): Likewise.
2020 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2021 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2022 (scop_detection::can_represent_loop_1): ... this. Removed.
2023 (scop_detection::harmful_loop_in_region): Simplify after inlining
2024 the above and remove more quadraticness.
2025 (build_scops): Adjust.
2026 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2029 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2032 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2033 REGNO (base) == SP_REG if base is a REG.
2036 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2037 if it is different SSA_NAME.
2038 (optimize_range_tests_cmp_bitwise): New function.
2039 (optimize_range_tests): Call it.
2041 2017-09-26 Richard Biener <rguenther@suse.de>
2043 PR tree-optimization/82321
2044 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2045 for the def being inside the loop.
2047 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2049 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2051 * config/s390/s390-builtins.def: Fix constraint on op4.
2053 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2055 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2056 independent expanders.
2057 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2058 ("vec_ordered", "vec_unordered"): New expanders.
2060 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2062 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2065 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2067 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2068 vec_unpacks_lo_v16qi.
2069 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2071 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2073 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2074 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2075 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2077 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2079 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
2081 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
2083 ("*vec_slb<mode>"): New insn pattern.
2084 ("vec_shr_<mode>"): New expander.
2085 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
2086 and force the shift count operand to V16QImode.
2087 ("vec_srb<mode>"): Set shift count mode to V16QI.
2089 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2091 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
2092 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
2093 ("vec_widen_smult_hi_<mode>"): New expander definitions.
2095 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
2098 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
2100 2017-09-26 Richard Biener <rguenther@suse.de>
2102 PR tree-optimization/82320
2103 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
2106 2017-09-25 Jeff Law <law@redhat.com>
2108 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
2109 prototype for new argument.
2110 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
2111 mostly extracted from rs6000_emit_allocate_stack.
2112 (rs6000_emit_probe_stack_range_stack_clash): New function.
2113 (rs6000_emit_allocate_stack): Call
2114 rs6000_emit_probe_stack_range_stack_clash as needed.
2115 (rs6000_emit_probe_stack_range): Add additional argument
2116 to call to gen_probe_stack_range{si,di}.
2117 (output_probe_stack_range): New.
2118 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
2119 (output_probe_stack_range_stack_clash): New.
2120 (rs6000_emit_prologue): Emit notes into dump file as requested.
2121 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
2122 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
2123 Add additional operand and pass it to output_probe_stack_range.
2125 2017-09-25 Bin Cheng <bin.cheng@arm.com>
2127 PR tree-optimization/82163
2128 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
2129 (checking_verify_loop_closed_ssa): New parameter.
2130 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
2131 (check_loop_closed_ssa_stmt): Delete.
2132 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
2133 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
2134 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
2137 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
2139 * brig-builtins.def: Treat HSAIL barrier builtins as
2140 setjmp/longjump style functions.
2142 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2144 * target.def (constant_alignment): New hook.
2145 * defaults.h (CONSTANT_ALIGNMENT): Delete.
2146 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
2147 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
2148 * doc/tm.texi: Regenerate.
2149 * targhooks.h (default_constant_alignment): Declare.
2150 (constant_alignment_word_strings): Likewise.
2151 * targhooks.c (default_constant_alignment): New function.
2152 (constant_alignment_word_strings): Likewise.
2153 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
2154 instead of CONSTANT_ALIGNMENT.
2155 * varasm.c (align_variable, get_variable_align, build_constant_desc)
2156 (force_const_mem): Likewise.
2157 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
2158 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
2159 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
2160 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2161 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
2163 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
2164 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2165 constant_alignment_word_strings.
2166 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
2167 (CONSTANT_ALIGNMENT): Likewise.
2168 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2169 (arm_constant_alignment): New function.
2170 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
2171 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2172 constant_alignment_word_strings.
2173 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
2174 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2175 constant_alignment_word_strings.
2176 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
2177 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2178 (cris_constant_alignment): New function.
2179 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
2180 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2181 (epiphany_constant_alignment): New function.
2182 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
2183 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2184 constant_alignment_word_strings.
2185 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
2186 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2187 constant_alignment_word_strings.
2188 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
2189 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2190 constant_alignment_word_strings.
2191 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
2192 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
2193 * config/i386/i386.c (ix86_constant_alignment): Make static.
2194 Use the same interface as the target hook.
2195 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2196 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
2197 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2198 constant_alignment_word_strings.
2199 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
2200 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
2201 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2202 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
2203 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2204 constant_alignment_word_strings.
2205 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
2206 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2207 constant_alignment_word_strings.
2208 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
2209 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2210 constant_alignment_word_strings.
2211 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
2212 * config/microblaze/microblaze.c (microblaze_constant_alignment):
2214 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2215 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
2216 * config/mips/mips.c (mips_constant_alignment): New function.
2217 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2218 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
2219 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
2220 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2221 (mmix_constant_alignment): Make static. Use the same interface
2223 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
2224 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2225 constant_alignment_word_strings.
2226 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
2227 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2228 constant_alignment_word_strings.
2229 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
2230 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2231 constant_alignment_word_strings.
2232 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
2233 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2234 (rs6000_constant_alignment): New function.
2235 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
2236 * config/riscv/riscv.c (riscv_constant_alignment): New function.
2237 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2238 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
2239 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2240 (rs6000_constant_alignment): New function.
2241 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
2242 * config/s390/s390.c (s390_constant_alignment): New function.
2243 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2244 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
2245 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2246 constant_alignment_word_strings.
2247 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
2248 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2249 (sparc_constant_alignment): New function.
2250 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
2251 * config/spu/spu.c (spu_constant_alignment): New function.
2252 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2253 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
2254 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2255 constant_alignment_word_strings.
2256 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
2257 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2258 constant_alignment_word_strings.
2259 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
2260 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2261 constant_alignment_word_strings.
2262 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
2263 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2264 (visium_constant_alignment): New function.
2265 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
2266 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2267 (xtensa_constant_alignment): New function.
2268 * system.h (CONSTANT_ALIGNMENT): Poison.
2270 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
2272 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2273 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
2274 (rs6000_builtin_valid_without_lhs): New helper function.
2275 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2276 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
2278 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2280 * target.h (vec_perm_indices): Use unsigned short rather than
2282 (auto_vec_perm_indices): Likewise.
2283 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
2284 Use unsigned int rather than unsigned char.
2285 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
2287 2017-09-25 Richard Biener <rguenther@suse.de>
2289 * cfgloop.h (sort_sibling_loops): Declare.
2290 * cfgloop.c (sort_sibling_loops_cmp): New helper.
2291 (sort_sibling_loops): New function sorting the sibling loop list
2293 * graphite.c (graphite_transform_loops): Sort sibling loops.
2295 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
2297 * target.def (vec_perm_const_ok): Change sel parameter to
2299 * optabs-query.c (can_vec_perm_p): Update accordingly.
2300 * doc/tm.texi: Regenerate.
2301 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
2302 auto_vec_perm_indices and remove separate nelt field.
2303 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
2304 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
2305 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
2306 (aarch64_expand_vec_perm_const): Update accordingly.
2307 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
2308 to vec_perm_indices.
2309 * config/arm/arm.c (expand_vec_perm_d): Change perm to
2310 auto_vec_perm_indices and remove separate nelt field.
2311 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
2312 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
2313 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
2315 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
2316 to vec_perm_indices.
2317 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
2318 sel to vec_perm_indices.
2319 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
2320 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
2321 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
2323 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
2326 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
2329 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
2330 on the FUNCTION_DECL function context if it has a DIE that is a
2333 2017-09-25 Richard Biener <rguenther@suse.de>
2335 PR tree-optimization/82285
2336 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
2339 2017-09-25 Tom de Vries <tom@codesourcery.com>
2343 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
2346 2017-09-25 Richard Biener <rguenther@suse.de>
2348 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
2349 ISL errors other than isl_error_quota happen. Dump if the
2350 schedule is the same.
2351 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
2352 errors instead of aborting inside ISL.
2354 2017-09-25 Iain Sandoe <iain@codesourcery.com>
2357 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
2358 of libgcc_eh for m64.
2359 * config/i386/darwin64.h: Likewise.
2361 2017-09-25 Richard Biener <rguenther@suse.de>
2364 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
2365 attribute for incomplete types nor twice for complete ones.
2367 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
2370 * config/i386/i386.c (ix86_print_operand_address_as): Encode
2371 %esp as %rsp to avoid 0x67 prefix if there is no index or base
2374 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
2377 * config/i386/i386.opt (mprefer-avx256): Use
2378 ix86_target_flags variable.
2379 * config/i386/i386.c (ix86_target_string): Move
2380 -mprefer-avx256 to flag2_opts.
2382 2017-09-22 Jakub Jelinek <jakub@redhat.com>
2385 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
2386 and x != -1 | y != -1 into (x & y) != -1.
2388 2017-09-22 Steve Ellcey <sellcey@cavium.com>
2390 * config.gcc: Add new case statement to set
2391 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
2392 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
2393 s390x-*-linux* case statements. Added aarch64 to the list of
2394 supported architectures.
2396 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2398 PR tree-optimization/82289
2399 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
2400 STMT_VINFO_RELEVANT_P.
2402 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2403 Alan Hayward <alan.hayward@arm.com>
2404 David Sherwood <david.sherwood@arm.com>
2406 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
2407 for VR_RANGE only; don't allow VR_ANTI_RANGE.
2408 (extract_range_from_binary_expr_1): Don't call
2409 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
2411 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2412 Alan Hayward <alan.hayward@arm.com>
2413 David Sherwood <david.sherwood@arm.com>
2415 * target.def (preferred_vector_alignment): New hook.
2416 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
2418 * doc/tm.texi: Regenerate.
2419 * targhooks.h (default_preferred_vector_alignment): Declare.
2420 * targhooks.c (default_preferred_vector_alignment): New function.
2421 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
2423 (DR_TARGET_ALIGNMENT): New macro.
2424 (aligned_access_p): Update commentary.
2425 (vect_known_alignment_in_bytes): New function.
2426 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
2428 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
2429 Calculate the misalignment based on the target alignment rather than
2431 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
2432 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
2433 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
2434 the target alignment, rather than masking the element misalignment
2435 with the number of elements in a vector. Also use the target
2436 alignment when calculating the maximum number of peels.
2437 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
2438 instead of TYPE_ALIGN_UNIT.
2439 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
2440 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
2441 (vect_create_addr_base_for_vector_ref): Update call accordingly.
2442 (vect_create_data_ref_ptr): Likewise.
2443 (vect_setup_realignment): Realign by ANDing with
2444 -DR_TARGET_MISALIGNMENT.
2445 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
2446 the number of peels based on DR_TARGET_ALIGNMENT.
2447 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
2448 with the guaranteed alignment boundary when deciding whether
2450 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
2451 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
2452 (ensure_base_align): Remove stmt_info parameter. Get the
2453 target base alignment from DR_TARGET_ALIGNMENT.
2454 (vectorizable_store): Update call accordingly. Interpret
2455 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
2457 (vectorizable_load): Likewise.
2459 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2460 Alan Hayward <alan.hayward@arm.com>
2461 David Sherwood <david.sherwood@arm.com>
2463 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
2464 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
2465 (vect_enhance_data_refs_alignment): Likewise.
2467 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2469 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2470 error. Only quit immediately if parsing is complete.
2471 (BEGIN): Initialize fatal_err and parse_done.
2472 (begin fpu, end fpu): Check number of arguments.
2473 (begin arch, end arch): Likewise.
2474 (begin cpu, end cpu): Likewise.
2475 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2476 (optalias): Likewise.
2478 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2480 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2481 * config/arm/arm-isa.h: Delete. Move definitions to ...
2482 * arm-cpus.in: ... here. Use new feature and fgroup values.
2483 * config/arm/arm.c (arm_option_override): Use lower case for feature
2485 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2486 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2487 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2489 (print_isa_bits_for): New function.
2490 (gen_isa): New function.
2491 (gen_comm_data): Use print_isa_bits_for.
2492 (define feature): New keyword.
2493 (define fgroup): New keyword.
2494 * config/arm/t-arm (TM_H): Remove.
2495 (GTM_H): Add arm-isa.h.
2496 (arm-isa.h): Add rule to generate file.
2497 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2498 case for feature bit names.
2500 2017-09-22 Richard Biener <rguenther@suse.de>
2502 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
2504 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
2505 print of no dependency loops ...
2506 * graphite.c (graphite_transform_loops): ... here.
2507 (canonicalize_loop_closed_ssa_form): Work from inner to outer
2509 (same_close_phi_node, remove_duplicate_close_phi,
2510 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
2511 (canonicalize_loop_closed_ssa): ... here and simplify.
2512 * graphite-optimize-isl.c: Include tree-vectorizer.h.
2513 (optimize_isl): Use dump_printf_loc to tell when we stopped
2514 optimizing because of an ISL timeout.
2516 2017-09-22 Richard Biener <rguenther@suse.de>
2518 PR tree-optimization/82291
2519 * tree-if-conv.c (predicate_mem_writes): Make sure to
2520 remove writes in blocks predicated with false.
2522 2017-09-22 Richard Biener <rguenther@suse.de>
2524 * sese.c: Include cfganal.h.
2525 (if_region_set_false_region): Remove.
2526 (create_if_region_on_edge): Likewise.
2527 (move_sese_in_condition): Re-implement without destroying
2530 2017-09-22 Richard Biener <rguenther@suse.de>
2532 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
2533 Verify both BBs contain loop PHI nodes before dispatching to
2535 (graphite_regenerate_ast_isl): Do not recompute dominators,
2536 do not verify three times. Restructure for clarity.
2537 * graphite-scop-detection.c (same_close_phi_node,
2538 remove_duplicate_close_phi, make_close_phi_nodes_unique,
2539 defined_in_loop_p, canonicalize_loop_closed_ssa,
2540 canonicalize_loop_closed_ssa_form): Simplify, remove excess
2541 checking and SSA rewrite, move to ...
2542 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
2543 (graphite_initialize): Do not pass in ctx, do not reset the
2544 SCEV cache, compute only dominators.
2545 (graphite_transform_loops): Allocate ISL ctx after
2546 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
2547 Maintain post-dominators only around build_scops.
2548 * sese.c (if_region_set_false_region): Make static. Free
2549 and recompute dominators.
2550 (move_sese_in_condition): Assert we don't get called with
2551 post-dominators computed.
2552 * sese.h (if_region_set_false_region): Remove.
2554 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
2556 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
2557 mode attribute for TARGET_AVX512VL.
2559 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
2561 * config/i386/i386.opt (mprefer-avx256): New option.
2562 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
2564 (ix86_preferred_simd_mode): Return 256-bit AVX modes
2565 for TARGET_PREFER_AVX256.
2566 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
2568 2017-09-21 Jeff Law <law@redhat.com>
2570 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
2571 Fix dump output if the only stack space is for pushed registers.
2573 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2575 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
2578 2017-09-21 Martin Sebor <msebor@redhat.com>
2581 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
2582 code (in C++) or code that triggers warnings.
2584 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
2586 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
2588 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
2590 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
2591 * sched-rgn.c: Ditto.
2592 * sel-sched-ir.c: Ditto.
2594 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
2596 * toplev.h (set_random_seed): Adjust return type.
2597 * toplev.c (init_local_tick): Move eager initialization of random_seed
2598 to get_random_seed. Adjust comment.
2599 (init_random_seed): Inline to get_random_seed, delete.
2600 (get_random_seed): Initialize random_seed lazily.
2601 (set_random_seed): Do not return previous value.
2602 (print_switch_value): Do not call get_random_seed.
2604 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
2606 * cgraph.c (delete_function_version): New, broken out from...
2607 (cgraph_node::delete_function_version): ...here. Rename to
2608 cgraph_node::delete_function_version_by_decl. Update all uses.
2609 (cgraph_node::remove): Call delete_function_version.
2611 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2614 * tree-inline.c (expand_call_inline): Emit clobber stmts for
2615 VAR_DECLs to which addressable non-volatile parameters are mapped
2616 and for id->retvar after the return value assignment. Clear
2617 id->retval and id->retbnd after inlining.
2619 2017-09-21 Richard Biener <rguenther@suse.de>
2621 PR tree-optimization/82276
2622 PR tree-optimization/82244
2623 * tree-vrp.c (build_assert_expr_for): Set
2624 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
2626 (remove_range_assertions): Revert earlier change.
2628 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
2631 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
2633 2017-09-21 Richard Biener <rguenther@suse.de>
2635 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2636 Restore valid IL after code generation errors.
2637 * graphite.c (graphite_transform_loops): Diagnose code
2638 generation issues as MSG_MISSED_OPTIMIZATION and continue
2639 with processing SCOPs.
2641 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2642 Alan Hayward <alan.hayward@arm.com>
2643 David Sherwood <david.sherwood@arm.com>
2645 * calls.c (compute_argument_addresses): Use simplify_gen_binary
2646 rather than choosing between plus_constant and gen_rtx_<CODE>.
2647 * expr.c (emit_push_insn): Likewise.
2648 (expand_expr_real_2): Likewise.
2650 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2651 Alan Hayward <alan.hayward@arm.com>
2652 David Sherwood <david.sherwood@arm.com>
2654 * loop-unroll.c (split_iv): Call copy_rtx on the step.
2656 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2657 Alan Hayward <alan.hayward@arm.com>
2658 David Sherwood <david.sherwood@arm.com>
2660 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
2661 calling tree_to_uhwi.
2663 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2664 Alan Hayward <alan.hayward@arm.com>
2665 David Sherwood <david.sherwood@arm.com>
2667 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
2668 INTEGER_CST rather than a negative test for ADDR_EXPR.
2670 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2671 Alan Hayward <alan.hayward@arm.com>
2672 David Sherwood <david.sherwood@arm.com>
2674 * tree-vrp.c (extract_range_from_binary_expr_1): Check
2675 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
2677 2017-09-21 Richard Biener <rguenther@suse.de>
2679 PR tree-optimization/71351
2680 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
2681 graphite_create_new_loop_guard): Remove, fold remaining parts
2683 (translate_isl_ast_node_for): ... here and simplify.
2685 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2688 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
2689 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
2690 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
2691 alternative always use QI mode, for -Os imov (=R,R) alternative
2692 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
2695 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2696 Jeff Law <law@redhat.com>
2698 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
2699 (allocate_stack_space): New function, partially extracted from
2701 (s390_emit_prologue): Track offset to most recent stack probe.
2702 Code to allocate space moved into allocate_stack_space.
2703 Dump actions when no stack is allocated.
2704 (s390_prologue_plus_offset): New function.
2705 (s390_emit_stack_probe): Likewise.
2707 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
2709 * common.opt (Wa, Wl, Wp, g, gz=): Add
2711 (gno-column-info): Remove.
2712 (gcolumn-info): Drop RejectNegative.
2714 (gno-record-gcc-switches): Remove.
2715 (grecord-gcc-switches): Drop RejectNegative.
2716 (gno-split-dwarf): Remove.
2717 (gsplit-dwarf): Drop RejectNegative.
2718 (gno-strict-dwarf): Remove.
2719 (gstrict-dwarf): Drop RejectNegative.
2720 * config/darwin.opt (gfull, gused): Add RejectNegative.
2721 * dwarf2out.c (gen_producer_string): Drop
2722 gno-record-gcc-switches handler.
2723 * optc-gen.awk: Add g to prefixes with negative forms.
2724 * opts-common.c (remapping_prefix_p): New.
2725 (find_opt): Check it.
2726 (generate_canonical_option): Test g prefix.
2727 (option_map): Add -gno- mapping.
2728 (add_misspelling_candidates): Check remapping_prefix_p.
2730 2017-09-20 Jeff Law <law@redhat.com>
2732 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
2733 thinko in stack clash protection support.
2735 * explow.c (compute_stack_clash_protection_loop_data): Use
2736 CONST_INT_P instead of explicit test. Verify object is a
2737 CONST_INT_P before looking at INTVAL.
2738 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
2739 instead of explicit test.
2741 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2744 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
2745 address instead of to r1 and r11.
2747 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
2749 * config.gcc: Support "knm".
2750 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
2751 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2753 * config/i386/i386.c (m_KNM): Define.
2754 (processor_target_table): Add "knm".
2756 (ix86_option_override_internal): Add "knm".
2757 (ix86_issue_rate): Add PROCESSOR_KNM.
2758 (ix86_adjust_cost): Ditto.
2759 (ia32_multipass_dfa_lookahead): Ditto.
2760 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
2761 (fold_builtin_cpu): Add M_INTEL_KNM.
2762 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
2763 (processor_type): Add PROCESSOR_KNM.
2764 * config/i386/x86-tune.def: Add m_KNM.
2765 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
2767 2017-09-20 Richard Biener <rguenther@suse.de>
2769 PR tree-optimization/80213
2770 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
2771 are allowed in empty BBs as well.
2772 (canonicalize_loop_closed_ssa): Also look for other complex
2774 (scop_detection::get_sese): Include the loop-closed PHI block
2776 (scop_detection::merge_sese): Remove code adding extra blocks.
2777 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
2778 (build_scops): Assert the final returned scop is invalid.
2780 2017-09-20 Richard Biener <rguenther@suse.de>
2782 PR tree-optimization/82264
2783 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
2785 (vn_phi_lookup): Likewise.
2786 (vn_phi_insert): Likewise.
2788 2017-09-20 Jakub Jelinek <jakub@redhat.com>
2790 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
2791 that fits into uhwi or shwi, add DW_AT_const_value regardless
2792 of early_dwarf without going through RTL, using add_AT_unsigned
2795 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
2796 (DEBUG_LTO_ABBREV_SECTION): Likewise.
2797 (DEBUG_LTO_MACINFO_SECTION): Likewise.
2798 (DEBUG_MACRO_SECTION): Likewise.
2799 (DEBUG_LTO_MACRO_SECTION): Likewise.
2800 (DEBUG_STR_DWO_SECTION): Likewise.
2801 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
2802 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
2803 (DEBUG_LTO_DWO_LINE_SECTION): Define.
2804 (DEBUG_LTO_LINE_STR_SECTION): Define.
2805 (init_sections_and_labels): Initialize debug_line_str_section
2806 variable. Initialize debug_loc_section for -gdwarf-5 to
2807 DEBUG_LOCLISTS_SECTION. Formatting fixes.
2809 2017-09-20 Richard Biener <rguenther@suse.de>
2811 * graphite-sese-to-poly.c (extract_affine): Properly handle
2812 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
2814 2017-09-20 Richard Biener <rguenther@suse.de>
2816 PR tree-optimization/81373
2817 * graphite-scop-detection.c (build_cross_bb_scalars_def):
2818 Force SESE live-out defs to be handled even if they are
2821 2017-09-19 Jeff Law <law@redhat.com>
2823 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
2824 nothing for stack adjustments with REG_STACK_CHECK.
2825 * sched-deps.c (parse_add_or_inc): Reject insns with
2826 REG_STACK_CHECK from dependency breaking.
2827 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
2828 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
2829 * reg-notes.def (STACK_CHECK): New note.
2831 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
2832 (ix86_expand_prologue): Dump stack clash info as needed.
2833 Call ix86_adjust_stack_and_probe_stack_clash as needed.
2835 * function.c (dump_stack_clash_frame_info): New function.
2836 * function.h (dump_stack_clash_frame_info): Prototype.
2837 (enum stack_clash_probes): New enum.
2839 * config/alpha/alpha.c (alpha_expand_prologue): Also check
2840 flag_stack_clash_protection.
2841 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
2842 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
2843 (arm_frame_pointer_required): Likewise.
2844 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
2845 (ia64_expand_prologue): Likewise.
2846 * config/mips/mips.c (mips_expand_prologue): Likewise.
2847 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
2848 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2849 (sparc_flat_expand_prologue): Likewise.
2850 * config/spu/spu.c (spu_expand_prologue): Likewise.
2852 * explow.c: Include "params.h".
2853 (anti_adjust_stack_and_probe_stack_clash): New function.
2854 (get_stack_check_protect): Likewise.
2855 (compute_stack_clash_protection_loop_data): Likewise.
2856 (emit_stack_clash_protection_loop_start): Likewise.
2857 (emit_stack_clash_protection_loop_end): Likewise.
2858 (allocate_dynamic_stack_space): Use get_stack_check_protect.
2859 Use anti_adjust_stack_and_probe_stack_clash.
2860 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
2861 (emit_stack_clash_protection_loop_start): Likewise.
2862 (emit_stack_clash_protection_loop_end): Likewise.
2863 * rtl.h (get_stack_check_protect): Prototype.
2864 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
2865 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
2866 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
2868 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
2870 * doc/tm.texi: Rebuilt.
2871 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2872 get_stack_check_protect.
2873 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
2874 * config/arm/arm.c (arm_expand_prologue): Likewise.
2875 (arm_frame_pointer_required): Likewise.
2876 * config/i386/i386.c (ix86_expand_prologue): Likewise.
2877 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
2878 * config/mips/mips.c (mips_expand_prologue): Likewise.
2879 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
2880 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
2881 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2882 (sparc_flat_expand_prologue): Likewise.
2884 * common.opt (-fstack-clash-protection): New option.
2885 * flag-types.h (enum stack_check_type): Note difference between
2886 -fstack-check= and -fstack-clash-protection.
2887 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
2888 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
2889 * toplev.c (process_options): Issue warnings/errors for cases
2890 not handled with -fstack-clash-protection.
2891 * doc/invoke.texi (-fstack-clash-protection): Document new option.
2892 (-fstack-check): Note additional problem with -fstack-check=generic.
2893 Note that -fstack-check is primarily for Ada and refer users
2894 to -fstack-clash-protection for stack-clash-protection.
2895 Document new params for stack clash protection.
2897 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
2899 * config/i386/i386.c (ix86_split_long_move): Do not handle
2900 address used for LEA in a special way.
2902 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2904 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
2906 2017-09-19 Martin Sebor <msebor@redhat.com>
2909 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
2910 of incompatible types.
2912 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
2914 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2915 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
2916 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2917 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
2919 2017-09-19 Richard Biener <rguenther@suse.de>
2921 PR tree-optimization/82244
2922 * tree-vrp.c (remove_range_assertions): Do not propagate
2923 a constant to abnormals but replace the assert with a copy.
2925 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
2927 PR rtl-optimization/57878
2928 PR rtl-optimization/68988
2929 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
2930 avoidance test involving non_reload_pseudos. Move frequency test
2931 below the general fragmentation avoidance test.
2933 2017-09-19 Richard Biener <rguenther@suse.de>
2935 PR tree-optimization/69728
2936 * graphite-sese-to-poly.c (schedule_error): New global.
2937 (add_loop_schedule): Handle empty domain by failing the
2939 (build_original_schedule): Handle schedule_error.
2941 2017-09-19 Richard Biener <rguenther@suse.de>
2943 * graphite-scop-detection.c (scop_detection::can_represent_loop):
2944 Do not iterate to sibling loops but only to siblings of inner
2947 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
2950 * config/m68k/m68k.md (moveq feeding equality comparison): Check
2951 that the registers are different.
2953 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
2955 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
2956 to processor_model and "amdfam17h" to arch_names_table.
2957 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
2959 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2962 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
2964 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2965 Alan Hayward <alan.hayward@arm.com>
2966 David Sherwood <david.sherwood@arm.com>
2968 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
2970 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
2972 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
2973 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
2974 vect_schedule_slp_instance.
2975 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
2976 Update call to vect_slp_analyze_node_operations. Simplify return
2978 (vect_slp_analyze_bb_1): Update call accordingly.
2979 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
2980 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
2981 (vect_schedule_slp): Update call accordingly.
2983 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2984 Alan Hayward <alan.hayward@arm.com>
2985 David Sherwood <david.sherwood@arm.com>
2987 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
2988 with types that aren't in fact scalar.
2990 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2992 * tree-vect-slp.c (vect_record_max_nunits): New function,
2994 (vect_build_slp_tree_1): ...here.
2995 (vect_build_slp_tree_2): Call it for phis too.
2997 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2999 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3000 to vect_get_vec_def_for_operand when getting the mask operand.
3002 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3003 Alan Hayward <alan.hayward@arm.com>
3004 David Sherwood <david.sherwood@arm.com>
3006 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3009 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3010 Alan Hayward <alan.hayward@arm.com>
3011 David Sherwood <david.sherwood@arm.com>
3013 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3014 calculation for vector booleans.
3016 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3017 Alan Hayward <alan.hayward@arm.com>
3018 David Sherwood <david.sherwood@arm.com>
3020 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3022 (vect_transform_stmt): ...here.
3023 (vect_analyze_stmt): Use it instead of calling
3024 vectorizable_live_operation directly.
3026 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3028 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3029 non-SIMT targets in acc vector loops.
3031 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3033 * configure.ac: Add arc and check if assembler supports gdwarf2.
3034 * configure: Regenerate.
3036 2017-09-18 Richard Biener <rguenther@suse.de>
3038 PR tree-optimization/82220
3039 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3040 epilogue niters from the min_profitable_iters compute.
3042 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3045 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3046 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3047 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3049 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3052 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3053 switching to a new text section.
3055 2017-09-18 Richard Biener <rguenther@suse.de>
3057 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3059 (build_alias_set): Reject aliases with no access function.
3061 2017-09-18 Richard Biener <rguenther@suse.de>
3063 PR tree-optimization/79622
3064 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3066 (build_cross_bb_scalars_use): Likewise.
3068 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3070 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3072 2017-09-18 Alan Modra <amodra@gmail.com>
3075 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3076 stack_pointer_rtx for count 0. Update comments. Break up
3077 large rtl expression.
3079 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
3081 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
3082 Increase to 20 bytes.
3083 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
3084 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
3085 or avx version of the stub.
3087 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
3090 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
3091 compute the minimum stack alignment. Also update preferred stack
3092 boundary for leaf functions.
3094 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
3096 PR tree-optimization/82228
3097 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
3100 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3102 * common/config/nds32/nds32-common.c
3103 (nds32_option_optimization_table): Refine formatting.
3104 (nds32_option_optimization_table): Use -fsched-pressure and
3105 -fomit-frame-pointer for specific optimization level.
3107 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3109 * config/nds32/nds32.c: Refine formatting and comments.
3110 * config/nds32/nds32.h: Likewise.
3111 * config/nds32/nds32.md: Likewise.
3112 * config/nds32/nds32-cost.c: Likewise.
3113 * config/nds32/nds32-isr.c: Likewise.
3114 * config/nds32/nds32-md-auxiliary.c: Likewise.
3115 * config/nds32/nds32-multiple.md: Likewise.
3116 * config/nds32/nds32-predicates.c: Likewise.
3118 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
3119 Jakub Jelinek <jakub@redhat.com>
3121 Add support for -std=c++2a.
3122 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
3124 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
3126 2017-09-15 Steve Ellcey <sellcey@cavium.com>
3129 * doc/extend.texi (Common Function Attributes): Add
3130 references to ARM, AArch64, and S/390 specific attributes.
3131 (Function Specific Option Pragmas): Add AArch64 and S/390
3132 to list of back ends that support the target pragma.
3134 2017-09-15 Nathan Sidwell <nathan@acm.org>
3136 * doc/standards.texi: Fix C++17 description. Update URLs for
3139 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3141 * common.opt (Wcast-align=strict): New warning option.
3142 * doc/invoke.texi: Document -Wcast-align=strict.
3144 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
3146 * cgraph.h (cgraph_thunk_info): Add comments.
3147 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
3148 assert for VIRTUAL_* arguments stricter.
3150 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3152 PR tree-optimization/71026
3153 * match.pd: Move RDIV patterns from fold-const.c
3154 * fold-const.c (distribute_real_division): Removed.
3155 (fold_binary_loc): Remove calls to distribute_real_divison.
3157 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3159 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
3160 c++1z and gnu++1z as deprecated. Change other references to
3161 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
3162 Change -Wc++1z-compat to -Wc++17-compat.
3163 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
3164 * dwarf2out.c (highest_c_language): Handle C++17.
3165 (gen_compile_unit_die): Likewise.
3167 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3169 PR rtl-optimization/82192
3170 * combine.c (make_extraction): Don't look through non-paradoxical
3171 SUBREGs or TRUNCATE if pos + len is or might be bigger than
3174 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3175 Alan Hayward <alan.hayward@arm.com>
3176 David Sherwood <david.sherwood@arm.com>
3178 * target.def (function_arg_offset): New hook.
3179 * targhooks.h (default_function_arg_offset): Declare.
3180 * targhooks.c (default_function_arg_offset): New function.
3181 * function.c (locate_and_pad_parm): Use
3182 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
3183 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
3184 (TARGET_FUNCTION_ARG_OFFSET): ...this.
3185 * doc/tm.texi: Regenerate.
3186 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
3187 * config/spu/spu.c (spu_function_arg_offset): New function.
3188 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
3189 * system.h (FUNCTION_ARG_OFFSET): Poison.
3191 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3192 Alan Hayard <alan.hayward@arm.com>
3193 David Sherwood <david.sherwood@arm.com>
3195 * target.def (truly_noop_truncation): New hook.
3196 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
3197 than TRULY_NOOP_TRUNCATION.
3198 * hooks.h (hook_bool_uint_uint_true): Declare.
3199 * hooks.c (hook_bool_uint_uint_true): New function.
3200 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
3201 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
3202 * doc/tm.texi: Regenerate.
3203 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
3204 rather than TRULY_NOOP_TRUNCATION in comments.
3205 (simplify_comparison): Likewise.
3206 (record_truncated_value): Likewise.
3207 * expmed.c (extract_bit_field_1): Likewise.
3208 (extract_split_bit_field): Likewise.
3209 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
3210 instead of TRULY_NOOP_TRUNCATION.
3211 * function.c (assign_parm_setup_block): Likewise.
3212 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
3213 * rtlhooks.c: Include target.h.
3214 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
3215 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
3216 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
3217 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
3218 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
3219 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
3220 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
3221 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
3222 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
3223 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
3224 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
3225 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
3226 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
3227 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
3228 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
3229 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
3230 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
3231 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
3232 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
3233 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
3234 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
3235 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
3236 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
3237 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
3238 * config/mips/mips.c (mips_truly_noop_truncation): New function.
3239 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3240 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3241 rather than TRULY_NOOP_TRUNCATION in comments.
3242 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
3243 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
3244 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
3245 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
3246 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
3247 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
3248 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
3249 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
3250 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
3251 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
3252 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
3253 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3254 rather than TRULY_NOOP_TRUNCATION in comments.
3255 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
3256 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
3257 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
3258 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
3259 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
3260 TRULY_NOOP_TRUNCATION condition.
3261 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
3262 (TRULY_NOOP_TRUNCATION): Delete.
3263 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
3264 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
3265 * config/spu/spu.c (spu_truly_noop_truncation): New function.
3266 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3267 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
3268 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
3269 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
3270 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3271 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3272 rather than TRULY_NOOP_TRUNCATION in comments.
3273 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
3274 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
3275 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
3276 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
3277 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
3278 * system.h (TRULY_NOOP_TRUNCATION): Poison.
3280 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
3283 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
3284 (*cmp_ior): Likewise.
3285 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
3286 (*ior_scc_scc_cmp): Likewise.
3287 (*and_scc_scc): Likewise.
3288 (*and_scc_scc_cmp): Likewise.
3290 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3291 Alan Hayard <alan.hayward@arm.com>
3292 David Sherwood <david.sherwood@arm.com>
3294 * target.def (can_change_mode_class): New hook.
3295 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
3296 (hard_regno_nregs): Likewise.
3297 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
3298 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
3299 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
3300 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
3301 (LOAD_EXTEND_OP): Update accordingly.
3302 * doc/tm.texi: Regenerate.
3303 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
3304 CANNOT_CHANGE_MODE_CLASS.
3305 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
3306 (REG_CAN_CHANGE_MODE_P): ...this new macro.
3307 * combine.c (simplify_set): Update accordingly.
3308 * emit-rtl.c (validate_subreg): Likewise.
3309 * recog.c (general_operand): Likewise.
3310 * regcprop.c (mode_change_ok): Likewise.
3311 * reload1.c (choose_reload_regs): Likewise.
3312 (inherit_piecemeal_p): Likewise.
3313 * rtlanal.c (simplify_subreg_regno): Likewise.
3314 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
3315 instead of CANNOT_CHANGE_MODE_CLASS.
3316 (reload_cse_simplify_operands): Likewise.
3317 * reload.c (push_reload): Use targetm.can_change_mode_class
3318 instead of CANNOT_CHANGE_MODE_CLASS.
3319 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
3320 REG_CANNOT_CHANGE_MODE_P.
3321 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3322 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
3323 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3324 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3325 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3326 (arm_can_change_mode_class): New function.
3327 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3328 than CANNOT_CHANGE_MODE_CLASS in comments.
3329 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3330 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
3331 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
3332 (ix86_can_change_mode_class): ...this new function, inverting the
3333 sense of the return value.
3334 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3335 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3336 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3337 (ia64_can_change_mode_class): New function.
3338 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3339 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
3340 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
3341 (m32c_can_change_mode_class): ...this new function, inverting the
3342 sense of the return value.
3343 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3344 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3345 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
3346 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
3347 (mips_can_change_mode_class): ...this new function, inverting the
3348 sense of the return value.
3349 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3350 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3351 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3352 (msp430_can_change_mode_class): New function.
3353 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3354 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
3355 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3356 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3357 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3358 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
3359 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3360 (pa_cannot_change_mode_class): Replace with...
3361 (pa_can_change_mode_class): ...this new function, inverting the
3362 sense of the return value.
3363 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3364 than CANNOT_CHANGE_MODE_CLASS in comments.
3365 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3366 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
3367 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3368 (pdp11_cannot_change_mode_class): Replace with...
3369 (pdp11_can_change_mode_class): ...this new function, inverting the
3370 sense of the return value.
3371 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3372 * config/powerpcspe/powerpcspe-protos.h
3373 (rs6000_cannot_change_mode_class_ptr): Delete.
3374 * config/powerpcspe/powerpcspe.c
3375 (rs6000_cannot_change_mode_class_ptr): Delete.
3376 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3377 (rs6000_option_override_internal): Assign to
3378 targetm.can_change_mode_class instead of
3379 rs6000_cannot_change_mode_class_ptr.
3380 (rs6000_cannot_change_mode_class): Replace with...
3381 (rs6000_can_change_mode_class): ...this new function, inverting the
3382 sense of the return value.
3383 (rs6000_debug_cannot_change_mode_class): Replace with...
3384 (rs6000_debug_can_change_mode_class): ...this new function.
3385 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3386 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
3387 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3388 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3389 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
3391 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
3392 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3393 (rs6000_option_override_internal): Assign to
3394 targetm.can_change_mode_class instead of
3395 rs6000_cannot_change_mode_class_ptr.
3396 (rs6000_cannot_change_mode_class): Replace with...
3397 (rs6000_can_change_mode_class): ...this new function, inverting the
3398 sense of the return value.
3399 (rs6000_debug_cannot_change_mode_class): Replace with...
3400 (rs6000_debug_can_change_mode_class): ...this new function.
3401 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3402 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
3403 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
3404 (s390_can_change_mode_class): ...this new function, inverting the
3405 sense of the return value.
3406 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3407 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3408 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
3409 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3410 (sh_cannot_change_mode_class): Replace with...
3411 (sh_can_change_mode_class): ...this new function, inverting the
3412 sense of the return value.
3413 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3414 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3415 (sparc_can_change_mode_class): New function.
3416 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3417 * config/spu/spu.c (spu_can_change_mode_class): New function.
3418 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3419 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3420 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3421 (visium_can_change_mode_class): New function.
3422 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
3424 2017-09-15 Richard Biener <rguenther@suse.de>
3426 PR tree-optimization/82217
3427 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
3428 but not undefined case.
3430 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3433 * postreload.c (reload_cse_simplify_operands): Skip
3434 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
3436 2017-09-15 Richard Biener <rguenther@suse.de>
3438 PR tree-optimization/68823
3439 * graphite-scop-detection.c (build_alias_set): If we have a
3440 possible dependence check whether we can handle them by just
3441 looking at the DRs DR_ACCESS_FNs.
3442 (build_scops): If build_alias_set fails, fail the SCOP.
3444 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3446 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
3447 to support float128 built-in functions that require the ISA 3.0
3449 (BU_FLOAT128_3_HW): Likewise.
3450 (SQRTF128): Add support for the IEEE 128-bit square root and fma
3452 (FMAF128): Likewise.
3454 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
3455 support for built-in functions that need the ISA 3.0 IEEE 128-bit
3456 floating point instructions.
3457 (rs6000_invalid_builtin): Likewise.
3458 (rs6000_builtin_mask_names): Likewise.
3459 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
3460 (RS6000_BTM_FLOAT128_HW): Likewise.
3461 (RS6000_BTM_COMMON): Likewise.
3462 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
3464 * doc/extend.texi (RS/6000 built-in functions): Document the
3465 IEEE 128-bit floating point square root and fused multiply-add
3468 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
3470 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
3471 reg (r2) isn't in the set of registers defined in the prologue.
3473 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3474 Alan Hayward <alan.hayward@arm.com>
3475 David Sherwood <david.sherwood@arm.com>
3477 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
3478 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
3479 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
3480 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
3481 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
3483 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
3484 max_vectorization_factor.
3485 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
3487 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3488 Alan Hayward <alan.hayward@arm.com>
3489 David Sherwood <david.sherwood@arm.com>
3491 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
3492 (vect_worthwhile_without_simd_p): Declare.
3493 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
3494 (vectorizable_reduction): Use it.
3495 * tree-vect-stmts.c (vectorizable_shift): Likewise.
3496 (vectorizable_operation): Likewise.
3498 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3499 Alan Hayward <alan.hayward@arm.com>
3500 David Sherwood <david.sherwood@arm.com>
3502 * tree-vectorizer.h (vect_get_num_copies): New function.
3503 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
3504 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3505 (vectorizable_induction): Likewise.
3506 (vectorizable_live_operation): Likewise.
3507 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3508 (vectorizable_bswap): Likewise.
3509 (vectorizable_call): Likewise.
3510 (vectorizable_conversion): Likewise.
3511 (vectorizable_assignment): Likewise.
3512 (vectorizable_shift): Likewise.
3513 (vectorizable_operation): Likewise.
3514 (vectorizable_store): Likewise.
3515 (vectorizable_load): Likewise.
3516 (vectorizable_condition): Likewise.
3517 (vectorizable_comparison): Likewise.
3518 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
3520 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3521 Alan Hayward <alan.hayward@arm.com>
3522 David Sherwood <david.sherwood@arm.com>
3524 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
3525 of vect_init_vector.
3527 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3528 Alan Hayward <alan.hayward@arm.com>
3529 David Sherwood <david.sherwood@arm.com>
3531 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
3532 an inline wrapper that provides a location.
3533 (gimple_build_vector): Likewise.
3534 * gimple-fold.c (gimple_build_vector_from_val): New function.
3535 (gimple_build_vector): Likewise.
3536 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
3537 functions to build the initial value. Always return a gimple value.
3538 (get_initial_defs_for_reduction): Likewise. Only compute
3540 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
3541 vect_init_vector on the results from get_initial_def(s)_for_reduction.
3542 (vectorizable_induction): Use gimple_build_vector rather than
3545 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3546 Alan Hayward <alan.hayward@arm.com>
3547 David Sherwood <david.sherwood@arm.com>
3549 * target.h (vec_perm_indices): New typedef.
3550 (auto_vec_perm_indices): Likewise.
3551 * optabs-query.h: Include target.h
3552 (can_vec_perm_p): Take a vec_perm_indices *.
3553 * optabs-query.c (can_vec_perm_p): Likewise.
3554 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
3555 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3556 * tree-vect-generic.c (lower_vec_perm): Likewise.
3557 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
3558 (vect_grouped_load_supported): Likewise.
3559 (vect_shift_permute_load_chain): Likewise.
3560 (vect_permute_store_chain): Use auto_vec_perm_indices.
3561 (vect_permute_load_chain): Likewise.
3562 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
3563 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
3564 Update uses of can_vec_perm_p.
3565 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
3566 mode with a number of elements. Take a vec_perm_indices *.
3567 (vect_create_epilog_for_reduction): Update accordingly.
3568 Use auto_vec_perm_indices.
3569 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
3570 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
3571 (vect_transform_slp_perm_load): Likewise.
3572 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
3573 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
3574 (vect_gen_perm_mask_checked): Likewise.
3575 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
3576 (vect_gen_perm_mask_checked): Likewise.
3577 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
3578 (vectorizable_store): Likewise.
3579 (vectorizable_load): Likewise.
3580 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
3581 (vectorizable_bswap): Likewise.
3583 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3584 Alan Hayward <alan.hayward@arm.com>
3585 David Sherwood <david.sherwood@arm.com>
3587 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
3588 * tree.c (build_vector): Likewise.
3589 (build_vector_from_ctor): Update accordingly.
3590 (build_vector_from_val): Likewise.
3591 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3592 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3593 * tree-vect-generic.c (add_rshift): Likewise.
3594 (expand_vector_divmod): Likewise.
3595 (optimize_vector_constructor): Likewise.
3596 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3597 (vect_transform_slp_perm_load): Likewise.
3598 (vect_schedule_slp_instance): Likewise.
3599 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
3600 (vectorizable_call): Likewise.
3601 (vect_gen_perm_mask_any): Likewise. Add elements in order.
3602 * expmed.c (make_tree): Likewise.
3603 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
3604 a vector passed to build_vector.
3605 (fold_convert_const): Likewise.
3606 (exact_inverse): Likewise.
3607 (fold_ternary_loc): Likewise.
3608 (fold_relational_const): Likewise.
3609 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
3610 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
3611 (const_unop): Likewise. Store the reduction accumulator in a
3612 variable rather than an array.
3613 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
3614 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
3615 the new vector, rather than constructing it after the input arrays.
3616 (native_interpret_vector): Use auto_vec<tree> when building
3617 a vector passed to build_vector. Add elements in order.
3618 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
3619 auto_vec<tree> when building a vector passed to build_vector.
3620 (vect_create_epilog_for_reduction): Likewise.
3621 (vectorizable_induction): Likewise.
3622 (get_initial_def_for_reduction): Likewise. Fix indentation of
3624 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
3626 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
3627 passed to build_vector.
3629 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3630 Alan Hayward <alan.hayward@arm.com>
3631 David Sherwood <david.sherwood@arm.com>
3633 * tree-core.h (tree_base::u): Add an "nelts" field.
3634 (tree_vector): Use VECTOR_CST_NELTS as the length.
3635 * tree.c (tree_size): Likewise.
3636 (make_vector): Initialize VECTOR_CST_NELTS.
3637 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
3638 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
3639 TYPE_VECTOR_SUBPARTS.
3640 * expr.c (const_vector_mask_from_tree): Consistently use "units"
3641 as the number of units, setting it from VECTOR_CST_NELTS.
3642 (const_vector_from_tree): Likewise.
3643 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
3644 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
3645 (fold_negate_expr_1): Likewise.
3646 (fold_convert_const): Likewise.
3647 (const_binop): Likewise. Differentiate the number of output and
3649 (const_unop): Likewise.
3650 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
3651 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
3652 in cases that did the opposite.
3654 2017-09-14 Richard Biener <rguenther@suse.de>
3656 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
3659 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
3661 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
3663 2017-09-14 Jakub Jelinek <jakub@redhat.com>
3666 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
3667 if and where to split a bb, except for splitting before debug insn
3668 sequences followed by non-label real insn. Delete debug insns
3669 in between basic blocks.
3671 * combine.c (make_compound_operation_int): Formatting fixes.
3673 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
3674 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3675 * config/netbsd.h (LINK_EH_SPEC): Likewise.
3676 * config/sol2.h (LINK_EH_SPEC): Likewise.
3677 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3678 * config/s390/linux.h (LINK_SPEC): Likewise.
3679 * config/freebsd.h (LINK_EH_SPEC): Likewise.
3680 * config/openbsd.h (LINK_EH_SPEC): Likewise.
3681 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3682 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
3683 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
3684 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3685 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
3686 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
3688 2017-09-13 Jakub Jelinek <jakub@redhat.com>
3690 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
3692 (ENDFILE_LINUX_SPEC): Likewise.
3693 (LINK_EH_SPEC): Likewise.
3694 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
3695 (LINK_OS_LINUX_SPEC32): Likewise.
3696 (LINK_OS_LINUX_SPEC64): Likewise.
3697 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
3698 (LINK_OS_LINUX_SPEC): Likewise.
3700 2017-09-13 Martin Liska <mliska@suse.cz>
3703 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
3704 CASE_HIGH is NULL_TREE.
3706 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3707 Alan Hayward <alan.hayward@arm.com>
3708 David Sherwood <david.sherwood@arm.com>
3710 * target.def (secondary_memory_needed): New hook.
3711 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
3712 instead of SECONDARY_MEMORY_NEEDED.
3713 (secondary_memory_needed_mode): Likewise.
3714 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
3715 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
3716 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
3717 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
3718 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
3719 * doc/tm.texi: Regenerate.
3720 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
3721 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
3722 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3723 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
3724 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
3725 * config/i386/i386.c (inline_secondary_memory_needed): Put the
3726 mode argument first and change the reg_class arguments to reg_class_t.
3727 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
3728 Make static. Update the call to inline_secondary_memory_needed.
3729 (ix86_register_move_cost): Update the call to
3730 inline_secondary_memory_needed.
3731 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3732 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
3734 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
3735 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3737 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
3738 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
3739 * config/mips/mips.c (mips_secondary_memory_needed): Make static
3740 and match hook interface. Add comment from mips.h.
3741 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3742 * config/mmix/mmix.md (truncdfsf2): Refer to
3743 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3745 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
3746 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
3747 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3748 (pa_secondary_memory_needed): New function.
3749 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
3750 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
3751 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3752 (pdp11_secondary_memory_needed): Make static and match hook interface.
3753 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
3754 * config/powerpcspe/powerpcspe-protos.h
3755 (rs6000_secondary_memory_needed_ptr): Delete.
3756 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
3758 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3759 (rs6000_option_override_internal): Assign to
3760 targetm.secondary_memory_needed rather than
3761 rs6000_secondary_memory_needed_ptr.
3762 (rs6000_secondary_memory_needed): Match hook interface.
3763 (rs6000_debug_secondary_memory_needed): Likewise.
3764 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
3765 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
3766 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
3767 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3768 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
3769 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
3771 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
3772 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3773 (rs6000_option_override_internal): Assign to
3774 targetm.secondary_memory_needed rather than
3775 rs6000_secondary_memory_needed_ptr.
3776 (rs6000_secondary_memory_needed): Match hook interface.
3777 (rs6000_debug_secondary_memory_needed): Likewise.
3778 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
3779 * config/s390/s390.c (s390_secondary_memory_needed): New function.
3780 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3781 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
3782 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3783 (sparc_secondary_memory_needed): New function.
3784 * lra-constraints.c (check_and_process_move): Refer to
3785 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3787 (curr_insn_transform): Likewise.
3788 (process_alt_operands): Use targetm.secondary_memory_needed
3789 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3790 (check_secondary_memory_needed_p): Likewise.
3791 (choose_split_class): Likewise.
3792 * reload.c: Unconditionally include code that was previously
3793 conditional on SECONDARY_MEMORY_NEEDED.
3794 (push_secondary_reload): Use targetm.secondary_memory_needed
3795 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3796 (push_reload): Likewise.
3797 * reload1.c: Unconditionally include code that was previously
3798 conditional on SECONDARY_MEMORY_NEEDED.
3799 (choose_reload_regs): Use targetm.secondary_memory_needed
3800 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3801 (gen_reload): Likewise.
3802 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
3804 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3805 Alan Hayward <alan.hayward@arm.com>
3806 David Sherwood <david.sherwood@arm.com>
3808 * target.def (secondary_memory_needed_mode): New hook:
3809 * targhooks.c (default_secondary_memory_needed_mode): Declare.
3810 * targhooks.h (default_secondary_memory_needed_mode): New function.
3811 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
3812 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
3813 * doc/tm.texi: Regenerate.
3814 * lra-constraints.c (check_and_process_move): Use
3815 targetm.secondary_memory_needed_mode instead of
3816 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
3817 (curr_insn_transform): Likewise.
3818 * reload.c (get_secondary_mem): Likewise.
3819 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3820 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
3822 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3823 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3824 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
3825 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3826 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
3828 * config/powerpcspe/powerpcspe-protos.h
3829 (rs6000_secondary_memory_needed_mode): Delete.
3830 * config/powerpcspe/powerpcspe.c
3831 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3832 (rs6000_secondary_memory_needed_mode): Make static.
3833 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3834 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
3836 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3838 (rs6000_secondary_memory_needed_mode): Make static.
3839 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3840 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
3841 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3842 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3843 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3845 (sparc_secondary_memory_needed_mode): New function.
3846 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
3848 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
3850 * config/aarch64/constraints.md (Umq): New constraint.
3851 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
3853 (mov<mode>): Update condition.
3855 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3857 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
3858 when bitposition is the same.
3860 2017-09-13 Richard Biener <rguenther@suse.de>
3862 * dwarf2out.c (output_die_symbol): Remove.
3863 (output_die): Do not output a DIEs symbol.
3865 2017-09-13 Richard Biener <rguenther@suse.de>
3868 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
3869 default-def to avoid breaking iterator update with the weird
3870 interaction with cgraph_update_edges_for_call_stmt_node.
3872 2017-09-13 Richard Biener <rguenther@suse.de>
3874 * tree-cfg.c (verify_gimple_assign_binary): Add verification
3875 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
3876 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
3877 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
3879 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3881 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
3882 Disable pc relative literal load irrespective of
3883 TARGET_FIX_ERR_A53_84341 for default.
3885 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
3887 * config/sparc/sparc.c (output_return): Output the source location of
3888 the insn in the delay slot, if any.
3889 (output_sibcall): Likewise.
3891 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
3894 * common.opt (-static-pie): New alias.
3895 (shared): Negate static-pie.
3896 (-no-pie): Update help text.
3898 (static-pie): New option.
3899 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
3900 -static-pie support.
3901 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
3902 (LINK_EH_SPEC): Likewise.
3903 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3904 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3905 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3906 * gcc.c (LINK_COMMAND_SPEC): Likewise.
3907 (init_gcc_specs): Likewise.
3908 (init_spec): Likewise.
3909 (display_help): Update help message for -pie.
3910 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
3913 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
3915 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
3916 (movdi_aarch64): Likewise.
3917 (movti_aarch64): Likewise.
3919 2017-09-12 Simon Wright <simon@pushface.org>
3922 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
3923 calculation of the minor version, always output as 0.
3925 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3928 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3929 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
3930 on it early, rather than manual conversion late. For
3931 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
3932 instead of performing manual conversion.
3934 2017-09-12 Carl Love <cel@us.ibm.com>
3936 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
3937 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
3939 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3940 VMULOSW): Add definitions.
3941 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3942 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3943 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3944 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3945 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3947 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3949 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
3951 (movti_aarch64): Likewise.
3952 (movdf_aarch64): Likewise.
3953 (movtf_aarch64): Likewise.
3954 (load_pairdi): Likewise.
3955 (store_pairdi): Likewise.
3956 (load_pairdf): Likewise.
3957 (store_pairdf): Likewise.
3958 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
3959 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
3960 (ldr_got_small_<mode>): Likewise.
3961 (ldr_got_small_28k_<mode>): Likewise.
3962 (ldr_got_tiny): Likewise.
3963 * config/aarch64/iterators.md (ldst_sz): New.
3964 (ldpstp_sz): Likewise.
3965 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
3967 (thunderx_load): Split load_8 to load_16.
3968 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
3970 (thunderx2t99_storepair_basic): Split store_8 to store_16.
3971 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
3972 (xgene1_store_pair): Split store_8 to store_16.
3973 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
3974 (falkor_st_0_st_sd): Split store_8 to store_16.
3976 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3978 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
3979 and store1/2/3/4 to store_4/8/12/16.
3980 * config/aarch64/aarch64.md: Update for rename.
3981 * config/arm/arm.md: Likewise.: Likewise.
3982 * config/arm/arm.c: Likewise.
3983 * config/arm/thumb1.md: Likewise.
3984 * config/arm/thumb2.md: Likewise.
3985 * config/arm/vfp.md: Likewise.
3986 * config/arm/arm-generic.md: Likewise.
3987 * config/arm/arm1020e.md: Likewise.
3988 * config/arm/arm1026ejs.md: Likewise.
3989 * config/arm/arm1136jfs.md: Likewise.
3990 * config/arm/arm926ejs.md: Likewise.
3991 * config/arm/cortex-a15.md: Likewise.
3992 * config/arm/cortex-a17.md: Likewise.
3993 * config/arm/cortex-a5.md: Likewise.
3994 * config/arm/cortex-a53.md: Likewise.
3995 * config/arm/cortex-a57.md: Likewise.
3996 * config/arm/cortex-a7.md: Likewise.
3997 * config/arm/cortex-a8.md: Likewise.
3998 * config/arm/cortex-a9.md: Likewise.
3999 * config/arm/cortex-m4.md: Likewise.
4000 * config/arm/cortex-m7.md: Likewise.
4001 * config/arm/cortex-r4.md: Likewise.
4002 * config/arm/exynos-m1.md: Likewise.
4003 * config/arm/fa526.md: Likewise.
4004 * config/arm/fa606te.md: Likewise.
4005 * config/arm/fa626te.md: Likewise.
4006 * config/arm/fa726te.md: Likewise.
4007 * config/arm/fmp626.md: Likewise.
4008 * config/arm/iwmmxt.md: Likewise.
4009 * config/arm/ldmstm.md: Likewise.
4010 * config/arm/marvell-pj4.md: Likewise.
4011 * config/arm/xgene1.md: Likewise.
4012 * config/aarch64/thunderx.md: Likewise.
4013 * config/aarch64/thunderx2t99.md: Likewise.
4014 * config/aarch64/falkor.md: Likewise.
4016 2017-09-12 Martin Liska <mliska@suse.cz>
4018 * attribs.c (private_lookup_attribute): New function.
4019 * attribs.h (private_lookup_attribute): Declared here.
4020 (lookup_attribute): Called from this place.
4022 2017-09-12 Richard Biener <rguenther@suse.de>
4024 PR tree-optimization/82157
4025 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4026 stmts with side-effects.
4028 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4029 Alan Hayward <alan.hayward@arm.com>
4030 David Sherwood <david.sherwood@arm.com>
4032 * target.def (hard_regno_nregs): New hook.
4033 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4034 * targhooks.h (default_hard_regno_nregs): Declare.
4035 * targhooks.c (default_hard_regno_nregs): New function.
4036 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4037 (TARGET_HARD_REGNO_NREGS): ...this hook.
4038 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4039 (CLASS_MAX_NREGS): Likewise.
4040 * doc/tm.texi: Regenerate.
4041 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4042 instead of HARD_REGNO_NREGS.
4043 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4044 HARD_REGNO_NREGS in the comment.
4045 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4046 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4047 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4048 Return an unsigned int.
4049 (TARGET_HARD_REGNO_NREGS): Redefine.
4050 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4051 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4052 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4053 (arc_hard_regno_nregs): New function.
4054 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4055 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4056 (arm_hard_regno_nregs): New function.
4057 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4058 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4059 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4060 (TARGET_HARD_REGNO_NREGS): Redefine.
4061 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4062 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4063 (HARD_REGNO_NREGS): Delete.
4064 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4065 (cr16_hard_regno_nregs): New function.
4066 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4067 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4068 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4069 (cris_hard_regno_nregs): New function.
4070 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4071 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4072 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4073 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4074 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4075 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4076 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4077 (frv_hard_regno_nregs): Make static. Take and return an
4079 (frv_class_max_nregs): Remove outdated copy of documentation.
4080 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
4081 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
4082 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
4083 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
4084 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
4085 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
4086 (TARGET_HARD_REGNO_NREGS): Redefine.
4087 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
4088 (CLASS_MAX_NREGS): Update comment.
4089 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
4090 (ia64_hard_regno_nregs): New function.
4091 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
4092 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
4093 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
4094 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
4095 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
4097 (m32c_hard_regno_nregs): Likewise. Make static.
4098 (TARGET_HARD_REGNO_NREGS): Redefine.
4099 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
4100 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
4101 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
4102 (m68k_hard_regno_nregs): New function.
4103 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
4104 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
4105 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
4106 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
4107 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
4108 Take and return an unsigned int.
4109 (TARGET_HARD_REGNO_NREGS): Redefine.
4110 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
4111 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4112 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
4113 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
4114 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
4115 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
4116 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
4117 (msp430_hard_regno_nregs): Make static. Take and return an
4119 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
4120 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
4121 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
4122 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
4123 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
4124 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
4125 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
4126 (TARGET_HARD_REGNO_NREGS): Redefine.
4127 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
4128 (PA_HARD_REGNO_NREGS): ...this.
4129 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
4130 (PA_HARD_REGNO_NREGS): ...this.
4131 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4132 (pa_hard_regno_nregs): New function.
4133 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
4134 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
4135 (pdp11_hard_regno_nregs): New function.
4136 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
4137 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
4138 (rs6000_hard_regno_nregs_hook): New function.
4139 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
4140 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4141 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
4142 Take and return an unsigned int. Move earlier in file.
4143 (TARGET_HARD_REGNO_NREGS): Redefine.
4144 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
4145 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
4146 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
4147 (rl78_hard_regno_nregs): Make static. Take and return an
4149 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
4150 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
4151 (rs6000_hard_regno_nregs_hook): New function.
4152 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
4153 * config/rx/rx.c (rx_hard_regno_nregs): New function.
4154 (TARGET_HARD_REGNO_NREGS): Redefine.
4155 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
4156 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
4157 instead of HARD_REGNO_NREGS.
4158 (s390_hard_regno_nregs): New function.
4159 (s390_hard_regno_mode_ok): Add comment from s390.h.
4160 (TARGET_HARD_REGNO_NREGS): Redefine.
4161 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
4162 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
4163 (sh_hard_regno_nregs): New function.
4164 (sh_pass_in_reg_p): Use it.
4165 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
4166 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4167 (sparc_hard_regno_nregs): New function.
4168 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
4169 * config/spu/spu.c (spu_hard_regno_nregs): New function.
4170 (spu_function_arg_advance): Use it, supplying a valid register number.
4171 (TARGET_HARD_REGNO_NREGS): Redefine.
4172 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
4173 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
4174 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
4175 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
4176 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
4177 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
4178 (CLASS_MAX_NREGS): Remove copy of old documentation.
4179 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
4180 (visium_hard_regno_nregs): New function.
4181 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
4182 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
4183 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4184 xtensa_hard_regno_nregs): New function.
4185 * system.h (HARD_REGNO_NREGS): Poison.
4187 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4189 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
4190 hard_regno_nregs instead of HARD_REGNO_NREGS.
4191 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
4192 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
4193 (c6x_expand_epilogue): Likewise.
4194 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
4195 (frv_read_iacc_argument): Likewise.
4196 * config/sh/sh.c: Include regs.h.
4197 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4198 (regs_used): Likewise.
4199 (output_stack_adjust): Likewise.
4200 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
4201 * expmed.c: Include regs.h.
4202 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4203 * ree.c: Include regs.h.
4204 (combine_reaching_defs): Use hard_regno_nregs instead of
4206 (add_removable_extension): Likewise.
4208 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4210 * regs.h (hard_regno_nregs): Turn into a function.
4211 (end_hard_regno): Update accordingly.
4212 * caller-save.c (setup_save_areas): Likewise.
4213 (save_call_clobbered_regs): Likewise.
4214 (replace_reg_with_saved_mem): Likewise.
4215 (insert_restore): Likewise.
4216 (insert_save): Likewise.
4217 * combine.c (can_change_dest_mode): Likewise.
4218 (move_deaths): Likewise.
4219 (distribute_notes): Likewise.
4220 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
4221 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
4222 (rs6000_split_multireg_move): Likewise.
4223 (rs6000_register_move_cost): Likewise.
4224 (rs6000_memory_move_cost): Likewise.
4225 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
4226 (rs6000_split_multireg_move): Likewise.
4227 (rs6000_register_move_cost): Likewise.
4228 (rs6000_memory_move_cost): Likewise.
4229 * cselib.c (cselib_reset_table): Likewise.
4230 (cselib_lookup_1): Likewise.
4231 * emit-rtl.c (set_mode_and_regno): Likewise.
4232 * function.c (aggregate_value_p): Likewise.
4233 * ira-color.c (setup_profitable_hard_regs): Likewise.
4234 (check_hard_reg_p): Likewise.
4235 (calculate_saved_nregs): Likewise.
4236 (assign_hard_reg): Likewise.
4237 (improve_allocation): Likewise.
4238 (calculate_spill_cost): Likewise.
4239 * ira-emit.c (modify_move_list): Likewise.
4240 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
4241 (ira_hard_reg_in_set_p): Likewise.
4242 * ira.c (setup_reg_mode_hard_regset): Likewise.
4243 (clarify_prohibited_class_mode_regs): Likewise.
4244 (check_allocation): Likewise.
4245 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4246 (lra_setup_reg_renumber): Likewise.
4247 (setup_try_hard_regno_pseudos): Likewise.
4248 (spill_for): Likewise.
4249 (assign_hard_regno): Likewise.
4250 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
4251 * lra-constraints.c (in_class_p): Likewise.
4252 (lra_constraint_offset): Likewise.
4253 (simplify_operand_subreg): Likewise.
4254 (lra_constraints): Likewise.
4255 (split_reg): Likewise.
4256 (split_if_necessary): Likewise.
4257 (invariant_p): Likewise.
4258 (inherit_in_ebb): Likewise.
4259 * lra-lives.c (process_bb_lives): Likewise.
4260 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
4261 (get_hard_regs): Likewise.
4262 (do_remat): Likewise.
4263 * lra-spills.c (assign_spill_hard_regs): Likewise.
4264 * mode-switching.c (create_pre_exit): Likewise.
4265 * postreload.c (reload_combine_recognize_pattern): Likewise.
4266 * recog.c (peep2_find_free_register): Likewise.
4267 * regcprop.c (kill_value_regno): Likewise.
4268 (set_value_regno): Likewise.
4269 (copy_value): Likewise.
4270 (maybe_mode_change): Likewise.
4271 (find_oldest_value_reg): Likewise.
4272 (copyprop_hardreg_forward_1): Likewise.
4273 * regrename.c (check_new_reg_p): Likewise.
4274 (regrename_do_replace): Likewise.
4275 * reload.c (push_reload): Likewise.
4276 (combine_reloads): Likewise.
4277 (find_dummy_reload): Likewise.
4278 (operands_match_p): Likewise.
4279 (find_reloads): Likewise.
4280 (find_equiv_reg): Likewise.
4281 (reload_adjust_reg_for_mode): Likewise.
4282 * reload1.c (count_pseudo): Likewise.
4283 (count_spilled_pseudo): Likewise.
4284 (find_reg): Likewise.
4285 (clear_reload_reg_in_use): Likewise.
4286 (free_for_value_p): Likewise.
4287 (allocate_reload_reg): Likewise.
4288 (choose_reload_regs): Likewise.
4289 (reload_adjust_reg_for_temp): Likewise.
4290 (emit_reload_insns): Likewise.
4291 (delete_output_reload): Likewise.
4292 * rtlanal.c (subreg_get_info): Likewise.
4293 * sched-deps.c (sched_analyze_reg): Likewise.
4294 * sel-sched.c (init_regs_for_mode): Likewise.
4295 (mark_unavailable_hard_regs): Likewise.
4296 (choose_best_reg_1): Likewise.
4297 (verify_target_availability): Likewise.
4298 * valtrack.c (dead_debug_insert_temp): Likewise.
4299 * var-tracking.c (track_loc_p): Likewise.
4300 (emit_note_insn_var_location): Likewise.
4301 * varasm.c (make_decl_rtl): Likewise.
4302 * reginfo.c (choose_hard_reg_mode): Likewise.
4303 (init_reg_modes_target): Refer directly to
4304 this_target_regs->x_hard_regno_nregs.
4306 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4308 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
4309 instead of hard_regno_nregs.
4311 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4313 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
4314 end_hard_regno instead of hard_regno_nregs.
4315 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
4316 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
4317 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
4318 * ira-color.c (improve_allocation): Likewise.
4319 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4320 * lra-lives.c (mark_regno_live): Likewise.
4321 (mark_regno_dead): Likewise.
4322 * lra-remat.c (operand_to_remat): Likewise.
4323 * lra.c (collect_non_operand_hard_regs): Likewise.
4324 * postreload.c (reload_combine_note_store): Likewise.
4325 (move2add_valid_value_p): Likewise.
4326 * reload.c (regno_clobbered_p): Likewise.
4328 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4330 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
4332 * config/v850/v850.c (v850_reorg): Likewise.
4333 * reload.c (refers_to_regno_for_reload_p): Likewise.
4334 (find_equiv_reg): Likewise.
4335 * reload1.c (reload_reg_reaches_end_p): Likewise.
4337 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4339 * caller-save.c (add_used_regs): Use REG_NREGS instead of
4341 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
4342 * config/arm/arm.c (output_move_neon): Likewise.
4343 (arm_attr_length_move_neon): Likewise.
4344 (neon_split_vcombine): Likewise.
4345 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
4346 (c6x_mark_reg_written): Likewise.
4347 (c6x_dwarf_register_span): Likewise.
4348 * config/i386/i386.c (ix86_save_reg): Likewise.
4349 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
4350 (rws_access_reg): Likewise.
4351 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
4352 * mode-switching.c (create_pre_exit): Likewise.
4353 * ree.c (combine_reaching_defs): Likewise.
4354 (add_removable_extension): Likewise.
4355 * regcprop.c (find_oldest_value_reg): Likewise.
4356 (copyprop_hardreg_forward_1): Likewise.
4357 * reload.c (reload_inner_reg_of_subreg): Likewise.
4358 (push_reload): Likewise.
4359 (combine_reloads): Likewise.
4360 (find_dummy_reload): Likewise.
4361 (reload_adjust_reg_for_mode): Likewise.
4362 * reload1.c (find_reload_regs): Likewise.
4363 (forget_old_reloads_1): Likewise.
4364 (reload_reg_free_for_value_p): Likewise.
4365 (reload_adjust_reg_for_temp): Likewise.
4366 (emit_reload_insns): Likewise.
4367 (delete_output_reload): Likewise.
4368 * sel-sched.c (choose_best_reg_1): Likewise.
4369 (choose_best_pseudo_reg): Likewise.
4371 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4372 Alan Hayward <alan.hayward@arm.com>
4373 David Sherwood <david.sherwood@arm.com>
4375 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
4376 * target.def (slow_unaligned_access): New hook.
4377 * targhooks.h (default_slow_unaligned_access): Declare.
4378 * targhooks.c (default_slow_unaligned_access): New function.
4379 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
4380 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
4381 * doc/tm.texi: Regenerate.
4382 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
4383 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
4384 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
4386 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
4387 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
4389 (rs6000_slow_unaligned_access): New function.
4390 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4391 (expand_block_compare): Likewise.
4392 (expand_strn_compare): Likewise.
4393 (rs6000_rtx_costs): Likewise.
4394 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
4395 (riscv_slow_unaligned_access): Likewise.
4396 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
4397 (riscv_slow_unaligned_access_p): ...this and make static.
4398 (riscv_option_override): Update accordingly.
4399 (riscv_slow_unaligned_access): New function.
4400 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4401 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
4402 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4403 (rs6000_slow_unaligned_access): New function.
4404 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4405 (rs6000_rtx_costs): Likewise.
4406 * config/rs6000/rs6000-string.c (expand_block_compare)
4407 (expand_strn_compare): Use targetm.slow_unaligned_access instead
4408 of SLOW_UNALIGNED_ACCESS.
4409 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
4410 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
4411 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
4412 of SLOW_UNALIGNED_ACCESS.
4413 * expmed.c (simple_mem_bitfield_p): Likewise.
4414 * expr.c (alignment_for_piecewise_move): Likewise.
4415 (emit_group_load_1): Likewise.
4416 (emit_group_store): Likewise.
4417 (copy_blkmode_from_reg): Likewise.
4418 (emit_push_insn): Likewise.
4419 (expand_assignment): Likewise.
4420 (store_field): Likewise.
4421 (expand_expr_real_1): Likewise.
4422 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4423 * lra-constraints.c (simplify_operand_subreg): Likewise.
4424 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4425 * gimple-ssa-store-merging.c: Likewise in block comment at start
4427 * tree-ssa-strlen.c: Include target.h.
4428 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
4429 of SLOW_UNALIGNED_ACCESS.
4430 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
4432 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4434 PR rtl-optimization/82185
4435 * expmed.c (emit_store_flag_int): Only test tem if it has been
4438 2017-09-12 Richard Biener <rguenther@suse.de>
4441 * match.pd ((FTYPE) N CMP CST): Fix typo.
4443 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
4445 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
4447 (mips_near_type_p): Add 'short_call' attribute as a synonym
4449 * doc/extend.texi (short_call): Document new function attribute.
4451 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4454 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
4455 assertion check that in the condition.
4456 (get_atomic_generic_size): Likewise. Before testing if parameter
4457 has pointer type, if it has array type, call for C++
4458 default_conversion to perform array-to-pointer conversion.
4460 2017-09-12 Richard Biener <rguenther@suse.de>
4462 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
4463 for operations we cannot scalarize.
4465 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
4467 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
4468 vectors heap vectors. Clean up comments.
4469 Make visited_bbs a reference.
4470 (profitable_jump_thread_path): Make GC
4471 vectors heap vectors. Clean up comments.
4473 (convert_and_register_jump_thread_path): Make GC vectors heap
4475 (check_subpath_and_update_thread_path): Same. Clean up comments.
4476 Make visited_bbs a reference.
4477 (handle_phi): Abstract common code to to
4478 register_jump_thread_path_if_profitable.
4479 Rename VAR_BB to DEF_BB.
4481 Make GC vectors heap vectors.
4482 Make visited_bbs a reference.
4483 (handle_assignment): Same.
4484 (register_jump_thread_path_if_profitable): New.
4485 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
4487 Make GC vectors heap vectors. Clean up comments.
4488 Make visited_bbs a reference.
4489 (find_jump_threads_backwards): Make visited_bbs live in the stack.
4490 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
4493 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
4496 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
4497 words of E_DImode object are reachable by xtensa_uimm8x4 access.
4499 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
4501 Revert r251800 and r251799.
4503 2017-09-11 Martin Jambor <mjambor@suse.cz>
4506 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
4507 arguments in advance.
4508 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
4509 use it to find predecessor edges.
4510 (naive_outof_ssa): Collect vector of predecessors.
4512 2017-09-08 Jason Merrill <jason@redhat.com>
4514 PR c++/70029 - ICE with ref-qualifier and -flto
4515 * langhooks.h (struct lang_hooks_for_types): Add
4516 copy_lang_qualifiers.
4517 * attribs.c (build_type_attribute_qual_variant): Use it.
4518 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
4520 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
4521 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
4523 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
4526 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
4527 (*mulsi3_sp64): New instruction.
4528 (mulsi3): New expander.
4530 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
4532 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
4534 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4536 * sancov.c: Include memmodel.h.
4538 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
4541 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
4544 2017-09-07 Carl Love <cel@us.ibm.com>
4546 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
4547 the sldi instruction.
4549 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
4551 * sancov.c: Include tm_p.h.
4553 2017-09-07 Jakub Jelinek <jakub@redhat.com>
4556 * output.h (switch_to_other_text_partition): New declaration.
4557 * varasm.c (switch_to_other_text_partition): New function.
4558 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
4559 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
4560 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
4561 to the other text partition before emitting LCL label and switch back
4562 after emitting the word after it.
4564 2017-09-07 Richard Biener <rguenther@suse.de>
4566 * passes.def (pass_split_crit_edges): Remove instance before PRE.
4567 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
4568 critical edges here, after loop init.
4569 (pass_data_pre): Remove PROP_no_crit_edges flags.
4570 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
4571 for valueization of call args to avoid leaking VN_TOP.
4572 (visit_use): Assert we do not visit default defs.
4573 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
4574 Use error_mark_node to more easily detect leaking VN_TOP.
4575 All default-defs are varying, not VN_TOP. Mark them visited.
4576 (run_scc_vn): Make code match comment.
4578 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4580 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
4581 OPTION_MASK_FLOAT128_KEYWORD.
4582 (POWERPC_MASKS): Likewise.
4583 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
4584 support for the -mfloat128-type option, and make -mfloat128
4585 default on PowerPC Linux systems. Define or undefine
4586 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
4587 Define __float128 to be __ieee128 if IEEE 128-bit support is
4588 enabled, or undefine it.
4589 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
4590 Delete defining __FLOAT128_TYPE__.
4591 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
4592 -mfloat128-type option and make -mfloat128 default on PowerPC
4594 (TARGET_FLOAT128_TYPE): Likewise.
4595 (-mfloat128-type): Likewise.
4596 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4597 Delete the -mfloat128-type option and make -mfloat128 default on
4598 PowerPC Linux systems. Always use __ieee128 to be the keyword for
4599 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
4600 128-bit floating point is enabled. Change tests from using
4601 -mfloat128-type to -mfloat128.
4602 (rs6000_mangle_type): Use the correct mangling for the __float128
4603 type even if normal long double is restricted to 64-bits.
4604 (floatn_mode): Enable the _Float128 type by default on VSX Linux
4606 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
4607 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
4608 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
4610 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
4611 documentation for -mfloat128.
4613 2017-09-06 Olivier Hainque <hainque@adacore.com>
4615 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
4617 2017-09-06 Wish Wu <wishwu007@gmail.com>
4618 Jakub Jelinek <jakub@redhat.com>
4620 * asan.c (initialize_sanitizer_builtins): Add
4621 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
4622 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
4623 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
4624 BT_FN_VOID_UINT64_PTR variables.
4625 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
4626 (BT_FN_VOID_UINT16_UINT16): Likewise.
4627 (BT_FN_VOID_UINT32_UINT32): Likewise.
4628 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
4629 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
4630 (BT_FN_VOID_UINT64_PTR): Likewise.
4631 * common.opt (flag_sanitize_coverage): New variable.
4632 (fsanitize-coverage=trace-pc): Remove.
4633 (fsanitize-coverage=): Add.
4634 * flag-types.h (enum sanitize_coverage_code): New enum.
4635 * fold-const.c (fold_range_test): Disable non-short-circuit
4636 optimization if flag_sanitize_coverage.
4637 (fold_truth_andor): Likewise.
4638 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4639 * opts.c (COVERAGE_SANITIZER_OPT): Define.
4640 (coverage_sanitizer_opts): New array.
4641 (get_closest_sanitizer_option): Add OPTS argument, handle also
4642 OPT_fsanitize_coverage_.
4643 (parse_sanitizer_options): Adjusted to also handle
4644 OPT_fsanitize_coverage_.
4645 (common_handle_option): Add OPT_fsanitize_coverage_.
4646 * sancov.c (instrument_comparison, instrument_switch): New function.
4647 (sancov_pass): Add trace-cmp support.
4648 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
4649 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
4650 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
4651 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
4652 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
4653 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
4654 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
4655 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
4656 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
4657 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
4659 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4661 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4662 error. Only quit immediately if parsing is complete.
4663 (BEGIN): Initialize fatal_err and parse_done.
4664 (begin fpu, end fpu): Check number of arguments.
4665 (begin arch, end arch): Likewise.
4666 (begin cpu, end cpu): Likewise.
4667 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4668 (optalias): Likewise.
4670 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4672 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4673 * config/arm/arm-isa.h: Delete. Move definitions to ...
4674 * arm-cpus.in: ... here. Use new feature and fgroup values.
4675 * config/arm/arm.c (arm_option_override): Use lower case for feature
4677 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4678 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4679 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4681 (print_isa_bits_for): New function.
4682 (gen_isa): New function.
4683 (gen_comm_data): Use print_isa_bits_for.
4684 (define feature): New keyword.
4685 (define fgroup): New keyword.
4686 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
4687 (arm-isa.h): Add rule to generate file.
4688 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4689 case for feature bit names.
4691 2017-09-06 Richard Biener <rguenther@suse.de>
4693 * tree-ssa-pre.c (NECESSARY): Remove.
4694 (create_expression_by_pieces): Do not touch pass-local flags.
4695 (insert_into_preds_of_block): Likewise.
4696 (do_pre_regular_insertion): Likewise.
4697 (eliminate_insert): Likewise.
4698 (eliminate_dom_walker::before_dom_children): Likewise.
4699 (fini_eliminate): Do not look at inserted_exprs.
4700 (mark_operand_necessary): Remove.
4701 (remove_dead_inserted_code): Replace with simple work-list
4702 algorithm based on inserted_exprs and SSA uses.
4703 (pass_pre::execute): Re-order fini_eliminate and
4704 remove_dead_inserted_code.
4706 2017-09-06 Olivier Hainque <hainque@adacore.com>
4708 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
4709 for VxWorks 7. Adjust surrounding comments.
4711 2017-09-06 Richard Biener <rguenther@suse.de>
4713 * gimple-ssa-strength-reduction.c
4714 (find_candidates_dom_walker::before_dom_children): Also allow
4717 2017-09-06 Richard Biener <rguenther@suse.de>
4719 PR tree-optimization/82108
4720 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
4721 for gap in the non-permutation SLP case.
4723 2017-09-06 Martin Jambor <mjambor@suse.cz>
4725 PR tree-optimization/82078
4726 * tree-sra.c (sort_and_splice_var_accesses): Move call to
4727 add_access_to_work_queue...
4728 (build_accesses_from_assign): ...here.
4729 (propagate_all_subaccesses): Make sure racc is the group
4730 representative, if there is one.
4732 2017-09-06 Jakub Jelinek <jakub@redhat.com>
4735 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
4738 2017-09-06 Richard Biener <rguenther@suse.de>
4740 * gimple-ssa-strength-reduction.c
4741 (find_candidates_dom_walker::before_doom_children): Use a
4742 type and not a mode check.
4744 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4747 * config/arm/predicates.md (arm_general_adddi_operand): Create new
4749 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
4751 2017-09-05 Jeff Law <law@redhat.com>
4753 PR tree-optimization/64910
4754 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
4755 cases where we have 3 or more operands.
4757 2017-09-05 Jakub Jelinek <jakub@redhat.com>
4760 * omp-low.c (lower_omp_for): Recompute tree invariant if
4761 gimple_omp_for_initial/final is ADDR_EXPR.
4764 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
4765 into gimple val before gimplification fo the COND_EXPR.
4767 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
4769 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
4770 REGION_COPY argument.
4771 (thread_through_all_blocks): Remove unused argument to
4772 duplicate_thread_path.
4774 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4775 Alan Hayward <alan.hayward@arm.com>
4776 David Sherwood <david.sherwood@arm.com>
4778 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
4779 Take a scalar_mode rather than a machine_mode.
4780 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4781 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
4782 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4783 (aarch64_gen_adjusted_ldpstp): Likewise.
4784 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
4786 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4787 Alan Hayward <alan.hayward@arm.com>
4788 David Sherwood <david.sherwood@arm.com>
4790 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
4791 Take a scalar_int_mode instead of a machine_mode.
4792 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4793 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4794 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4795 (aarch64_simd_attr_length_rglist): Delete.
4796 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
4797 a scalar_int_mode instead of a machine_mode.
4798 (aarch64_add_offset): Likewise.
4799 (aarch64_internal_mov_immediate): Likewise
4800 (aarch64_add_constant_internal): Likewise.
4801 (aarch64_add_constant): Likewise.
4802 (aarch64_movw_imm): Likewise.
4803 (aarch64_rtx_arith_op_extract_p): Likewise.
4804 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4805 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4806 Remove assert that the mode isn't a vector.
4807 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4808 (aarch64_expand_mov_immediate): Update calls after above changes.
4809 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
4810 (aarch64_and_bitmask_imm): Check for scalar integer modes.
4811 (aarch64_move_imm): Likewise.
4812 (aarch64_can_const_movi_rtx_p): Likewise.
4813 (aarch64_strip_extend): Likewise.
4814 (aarch64_extr_rtx_p): Likewise.
4815 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
4816 a CONST_INT when the mode parameter is VOIDmode.
4817 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
4819 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4821 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
4822 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4823 (bitwise_type_for_mode): Update accordingly.
4825 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4827 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
4828 * stor-layout.c (mode_for_size_tree): Likewise.
4829 (mode_for_array): Update accordingly.
4830 (layout_decl): Likewise.
4831 (compute_record_mode): Likewise. Only set the mode once.
4833 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4835 * target.def (get_mask_mode): Change return type to opt_mode.
4837 * doc/tm.texi: Regenerate.
4838 * targhooks.h (default_get_mask_mode): Return an opt_mode.
4839 * targhooks.c (default_get_mask_mode): Likewise.
4840 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4841 * optabs-query.c (can_vec_mask_load_store_p): Update use of
4842 targetm.get_mask_mode.
4843 * tree.c (build_truth_vector_type): Likewise.
4845 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4847 * machmode.h (mode_for_vector): Return an opt_mode.
4848 * stor-layout.c (mode_for_vector): Likewise.
4849 (mode_for_int_vector): Update accordingly.
4850 (layout_type): Likewise.
4851 * config/i386/i386.c (emit_memmov): Likewise.
4852 (ix86_expand_set_or_movmem): Likewise.
4853 (ix86_expand_vector_init): Likewise.
4854 (ix86_get_mask_mode): Likewise.
4855 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
4857 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
4858 * expmed.c (extract_bit_field_1): Likewise.
4859 * expr.c (expand_expr_real_2): Likewise.
4860 * optabs-query.c (can_vec_perm_p): Likewise.
4861 (can_vec_mask_load_store_p): Likewise.
4862 * optabs.c (expand_vec_perm): Likewise.
4863 * targhooks.c (default_get_mask_mode): Likewise.
4864 * tree-vect-stmts.c (vectorizable_store): Likewise.
4865 (vectorizable_load): Likewise.
4866 (get_vectype_for_scalar_type_and_size): Likewise.
4868 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4870 * machmode.h (mode_for_int_vector): New function.
4871 * stor-layout.c (mode_for_int_vector): Likewise.
4872 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
4873 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
4874 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4875 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4876 (s390_expand_vcond): Likewise.
4878 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4880 * machmode.h (opt_machine_mode): New type.
4881 (opt_mode<T>): Allow construction from anything that can be
4883 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
4884 (mode_for_size): Return an opt_machine_mode.
4885 * stor-layout.c (mode_for_size): Likewise.
4886 (mode_for_size_tree): Update call accordingly.
4887 (bitwise_mode_for_mode): Likewise.
4888 (make_fract_type): Likewise.
4889 (make_accum_type): Likewise.
4890 * caller-save.c (replace_reg_with_saved_mem): Update call
4892 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4893 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4894 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4895 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4896 * expmed.c (extract_bit_field_1): Likewise.
4897 * reload.c (get_secondary_mem): Likewise.
4898 * varasm.c (assemble_integer): Likewise.
4899 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
4902 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4904 * machmode.h (decimal_float_mode_for_size): New function.
4905 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
4906 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
4907 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
4908 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
4909 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
4910 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
4912 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4914 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
4915 (get_builtin_sync_mode): Likewise.
4916 (expand_ifn_atomic_compare_exchange): Likewise.
4917 (expand_builtin_atomic_clear): Likewise.
4918 (expand_builtin_atomic_test_and_set): Likewise.
4919 (fold_builtin_atomic_always_lock_free): Likewise.
4920 * calls.c (compute_argument_addresses): Likewise.
4921 (emit_library_call_value_1): Likewise.
4922 (store_one_arg): Likewise.
4923 * combine.c (combine_instructions): Likewise.
4924 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
4925 * config/arm/arm.c (arm_function_value): Likewise.
4926 (aapcs_allocate_return_reg): Likewise.
4927 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
4928 * config/i386/i386.c (construct_container): Likewise.
4929 (ix86_gimplify_va_arg): Likewise.
4930 (ix86_expand_sse_cmp): Likewise.
4931 (emit_memmov): Likewise.
4932 (emit_memset): Likewise.
4933 (expand_small_movmem_or_setmem): Likewise.
4934 (ix86_expand_pextr): Likewise.
4935 (ix86_expand_pinsr): Likewise.
4936 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
4937 * config/microblaze/microblaze.c (microblaze_block_move_straight):
4939 * config/mips/mips.c (mips_function_value_1) Likewise.
4940 (mips_block_move_straight): Likewise.
4941 (mips_expand_ins_as_unaligned_store): Likewise.
4942 * config/powerpcspe/powerpcspe.c
4943 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4944 (rs6000_darwin64_record_arg_flush): Likewise.
4945 * config/rs6000/rs6000.c
4946 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4947 (rs6000_darwin64_record_arg_flush): Likewise.
4948 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
4949 (sparc_function_value_1): Likewise.
4950 * config/spu/spu.c (adjust_operand): Likewise.
4951 (spu_emit_branch_or_set): Likewise.
4952 (arith_immediate_p): Likewise.
4953 * emit-rtl.c (gen_lowpart_common): Likewise.
4954 * expr.c (expand_expr_real_1): Likewise.
4955 * function.c (assign_parm_setup_block): Likewise.
4956 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4957 * reload1.c (alter_reg): Likewise.
4958 * stor-layout.c (mode_for_vector): Likewise.
4959 (layout_type): Likewise.
4961 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4963 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
4964 (spu_convert_move): Likewise.
4965 * lower-subreg.c (resolve_simple_move): Likewise.
4967 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4970 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
4971 define_insn to a define_expand.
4972 (altivec_vsum2sws_direct): New define_insn.
4973 (altivec_vsumsws): Convert from a define_insn to a define_expand.
4975 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4977 * config/arm/arm.c (arm_option_params_internal): Improve setting of
4980 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
4985 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
4986 realign stack if stack alignment needed is less than incoming
4989 2017-09-05 Marek Polacek <polacek@redhat.com>
4992 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
4995 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
4997 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
4999 2017-09-05 Richard Biener <rguenther@suse.de>
5001 PR tree-optimization/82084
5002 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5004 2017-09-05 Richard Biener <rguenther@suse.de>
5006 PR tree-optimization/82102
5007 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5009 2017-09-05 Martin Liska <mliska@suse.cz>
5011 PR tree-optimization/82032
5012 * tree-cfg.c (generate_range_test): New function.
5013 * tree-cfg.h (generate_range_test): Declared here.
5014 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5015 (cleanup_control_expr_graph): Use it.
5016 * tree-switch-conversion.c (try_switch_expansion): Remove
5018 (emit_case_nodes): Use generate_range_test.
5020 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5023 * config/i386/i386.md (*<btsc><mode>_mask): Add
5024 TARGET_USE_BT to insn constraint.
5025 (*btr<mode>_mask): Ditto.
5027 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5029 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5030 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5032 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5035 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5036 TARGET_NEON and TARGET_IWMMXT.
5037 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5038 TARGET_NEON and TARGET_IWMMXT.
5039 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5041 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5043 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5044 (ix86_rewrite_tls_address): Ditto.
5045 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5046 (ix86_rewrite_tls_address_1): Ditto.
5047 (ix86_rewrite_tls_address): Ditto.
5048 * config/i386/predicates.md (tls_address_pattern): New predicate.
5049 * config/i386/i386.md (TLS address splitter): New splitter.
5051 2017-09-04 Richard Biener <rguenther@suse.de>
5053 PR tree-optimization/82084
5054 * fold-const.h (can_native_encode_string_p): Declare.
5055 * fold-const.c (can_native_encode_string_p): Factor out from ...
5056 (native_encode_string): ... here.
5057 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5058 vectorizing stores from constants we later cannot handle.
5060 2017-09-04 Marek Polacek <polacek@redhat.com>
5063 * doc/invoke.texi: Update -Wtautological-compare documentation.
5065 2017-09-04 Jeff Law <law@redhat.com>
5067 PR tree-optimization/64910
5068 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5069 swap the first and last operand if the last is a constant.
5071 2017-09-04 Marek Polacek <polacek@redhat.com>
5074 * convert.c (do_narrow): When sanitizing signed integer overflows,
5075 bail out for signed types.
5076 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5078 2017-09-04 Richard Biener <rguenther@suse.de>
5080 PR tree-optimization/82060
5081 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5082 Move devirtualization after stmt folding and before EH/AB/noreturn
5083 cleanup to get the stmt refs canonicalized. Use a bool instead
5084 of gimple_modified_p since that doesn't work for NOPs. Schedule
5085 NOPs generated by folding for removal.
5087 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5088 Alan Hayward <alan.hayward@arm.com>
5089 David Sherwood <david.sherwood@arm.com>
5091 * coretypes.h (pad_direction): New enum.
5092 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
5093 (FUNCTION_ARG_PADDING): Likewise.
5094 * target.def (function_arg_padding): New hook.
5095 * targhooks.h (default_function_arg_padding): Declare.
5096 * targhooks.c (default_function_arg_padding): New function.
5097 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
5098 (TARGET_FUNCTION_ARG_PADDING): ...this.
5099 * doc/tm.texi: Regenerate.
5100 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
5101 instead of direction.
5102 (compute_argument_addresses): Likewise.
5103 (load_register_parameters): Likewise.
5104 (emit_library_call_value_1): Likewise.
5105 (store_one_arg): Use targetm.calls.function_arg_padding instead
5106 of FUNCTION_ARG_PADDING.
5107 (must_pass_in_stack_var_size_or_pad): Likewise.
5108 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
5109 (emit_group_store): Likewise.
5110 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
5111 instead of FUNCTION_ARG_PADDING.
5112 (emit_push_insn): Likewise, and propagate enum change throughout
5114 * function.h (direction): Delete.
5115 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
5117 * function.c (assign_parm_find_stack_rtl): Likewise.
5118 (assign_parm_setup_block_p): Likewise.
5119 (assign_parm_setup_block): Likewise.
5120 (gimplify_parameters): Likewise.
5121 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
5122 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
5124 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
5125 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5126 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
5127 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
5128 (aarch64_function_arg_padding): ...this new function.
5129 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
5130 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5131 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
5132 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5133 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
5134 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5135 (arm_pad_arg_upward): Replace with...
5136 (arm_function_arg_padding): ...this new function.
5137 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
5139 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
5140 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
5141 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5142 (ia64_hpux_function_arg_padding): Replace with...
5143 (ia64_function_arg_padding): ...this new function. Use pad_direction
5144 instead of direction. Check for TARGET_HPUX.
5145 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
5146 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5147 (iq2000_function_arg_padding): New function.
5148 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
5149 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
5150 (mips_function_arg_padding): ...this new function.
5151 (mips_pad_reg_upward): Update accordingly.
5152 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5153 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
5154 targetm.calls.function_arg_padding.
5155 (FUNCTION_ARG_PADDING): Delete.
5156 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5157 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
5158 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5159 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
5160 (nios2_block_reg_padding): Return pad_direction instead of direction.
5161 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
5162 instead of direction.
5163 (nios2_function_arg_padding): Likewise. Make static.
5164 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5165 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
5166 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
5167 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
5168 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5169 (pa_function_arg_padding): Make static. Return pad_direction instead
5171 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
5172 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5173 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
5174 instead of direction. Use targetm.calls.function_arg_padding.
5175 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
5176 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5177 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
5178 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
5179 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
5181 (function_arg_padding): Rename to...
5182 (rs6000_function_arg_padding): ...this. Make static. Return
5183 pad_direction instead of direction.
5184 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5185 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
5186 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5187 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
5188 instead of direction. Use targetm.calls.function_arg_padding.
5189 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
5190 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5191 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
5192 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
5193 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5194 (function_arg_padding): Rename to...
5195 (rs6000_function_arg_padding): ...this. Make static. Return
5196 pad_direction instead of direction.
5197 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5198 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
5199 * config/s390/s390.c (s390_function_arg_padding): New function.
5200 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5201 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
5202 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
5203 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5204 (function_arg_padding): Rename to...
5205 (sparc_function_arg_padding): ...this. Make static. Return
5206 pad_direction instead of direction.
5207 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
5208 * config/spu/spu.c (spu_function_arg_padding): New function.
5209 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5210 * system.h (FUNCTION_ARG_PADDING): Poison.
5212 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5213 Alan Hayward <alan.hayward@arm.com>
5214 David Sherwood <david.sherwood@arm.com>
5216 * target.def (modes_tieable_p): New hook.
5217 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
5218 (TARGET_MODES_TIEABLE_P): ...this.
5219 * doc/tm.texi.in: Regenerate.
5220 * hooks.h (hook_bool_mode_mode_true): Declare.
5221 * hooks.c (hook_bool_mode_mode_true): New function.
5222 * combine.c (subst): Use targetm.modes_tieable_p instead of
5224 * dse.c (find_shift_sequence): Likewise.
5225 * expmed.c (extract_low_bits): Likewise.
5226 * lower-subreg.c: Include target.h.
5227 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
5229 * rtlanal.c (rtx_cost): Likewise.
5230 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
5231 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
5232 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
5233 (TARGET_MODES_TIEABLE_P): Redefine.
5234 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
5235 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
5236 (TARGET_MODES_TIEABLE_P): Redefine.
5237 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
5238 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
5239 (arc_modes_tieable_p): New function.
5240 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
5241 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
5242 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
5243 (arm_modes_tieable_p): Make static.
5244 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
5245 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
5246 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
5247 (TARGET_MODES_TIEABLE_P): Redefine.
5248 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
5249 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
5250 (TARGET_MODES_TIEABLE_P): Redefine.
5251 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
5252 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
5253 (cr16_modes_tieable_p): New function.
5254 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
5255 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
5256 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
5257 (TRULY_NOOP_TRUNCATION): Update comment.
5258 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
5259 (TRULY_NOOP_TRUNCATION): Update comment.
5260 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
5261 (frv_modes_tieable_p): New function.
5262 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
5263 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
5264 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
5265 (TARGET_MODES_TIEABLE_P): Redefine.
5266 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
5267 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
5268 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
5269 (TARGET_MODES_TIEABLE_P): Redefine.
5270 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
5271 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
5272 (ia64_modes_tieable_p): New function.
5273 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
5274 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
5275 (iq2000_modes_tieable_p): New function.
5276 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
5277 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
5278 (lm32_modes_tieable_p): New function.
5279 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
5280 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
5281 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
5282 (TARGET_MODES_TIEABLE_P): Redefine.
5283 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
5284 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
5285 (m32r_modes_tieable_p): New function.
5286 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
5287 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
5288 (m68k_modes_tieable_p): New function.
5289 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
5290 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
5291 (mcore_modes_tieable_p): New function.
5292 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
5293 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
5295 (TARGET_MODES_TIEABLE_P): Redefine.
5296 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
5297 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
5298 * config/mips/mips.c (mips_modes_tieable_p): Make static.
5299 (TARGET_MODES_TIEABLE_P): Redefine.
5300 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
5301 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
5302 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
5303 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
5304 (mn10300_modes_tieable_p): ...this and make static.
5305 (TARGET_MODES_TIEABLE_P): Redefine.
5306 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
5307 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
5308 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
5309 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
5310 (msp430_modes_tieable_p): Make static.
5311 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
5312 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
5313 (TARGET_MODES_TIEABLE_P): Redefine.
5314 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
5315 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
5316 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
5317 (TARGET_MODES_TIEABLE_P): Redefine.
5318 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
5319 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
5320 * config/pa/pa.c (pa_modes_tieable_p): Make static.
5321 (TARGET_MODES_TIEABLE_P): Redefine.
5322 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
5323 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
5324 (pdp11_modes_tieable_p): New function.
5325 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
5326 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
5327 (rs6000_modes_tieable_p): New function.
5328 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5329 * config/powerpcspe/powerpcspe.md: Update comment.
5330 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
5331 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
5332 (TARGET_MODES_TIEABLE_P): Redefine.
5333 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
5334 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
5335 (rl78_modes_tieable_p): New function.
5336 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
5337 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
5338 (rs6000_modes_tieable_p): New function.
5339 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5340 * config/rs6000/rs6000.md: Update comment.
5341 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
5342 * config/rx/rx.c (rx_modes_tieable_p): New function.
5343 (TARGET_MODES_TIEABLE_P): Redefine.
5344 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
5345 * config/s390/s390.c (s390_modes_tieable_p): New function.
5346 (TARGET_MODES_TIEABLE_P): Redefine.
5347 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
5348 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
5349 (sh_modes_tieable_p): New function.
5350 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
5351 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
5352 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
5353 (sparc_modes_tieable_p): Make static.
5354 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
5355 * config/spu/spu.c (spu_modes_tieable_p): New function.
5356 (TARGET_MODES_TIEABLE_P): Redefine.
5357 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
5358 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
5359 (TARGET_MODES_TIEABLE_P): Redefine.
5360 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
5361 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
5362 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
5363 * config/v850/v850.c (v850_modes_tieable_p): New function.
5364 (TARGET_MODES_TIEABLE_P): Redefine.
5365 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
5366 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
5367 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
5368 (visium_modes_tieable_p): New function.
5369 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
5370 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
5371 (xtensa_modes_tieable_p): New function.
5372 * system.h (MODES_TIEABLE_P): Poison.
5374 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5375 Alan Hayward <alan.hayward@arm.com>
5376 David Sherwood <david.sherwood@arm.com>
5378 * target.def (hard_regno_mode_ok): New hook.
5379 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
5380 (TARGET_HARD_REGNO_MODE_OK): ...this.
5381 * doc/tm.texi.in: Regenerate.
5382 * hooks.h (hook_bool_uint_mode_true): Declare.
5383 * hooks.c (hook_bool_uint_mode_true): New function.
5384 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
5386 * genpreds.c (write_insn_preds_c): Add an include of target.h.
5387 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
5388 instead of HARD_REGNO_MODE_OK.
5389 * caller-save.c: Include target.h.
5390 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
5392 * combine.c (can_combine_p): Likewise.
5393 (combinable_i3pat): Likewise.
5394 (can_change_dest_mode): Likewise.
5395 * expr.c (init_expr_target): Likewise.
5396 (convert_move): Likewise.
5397 (convert_modes): Likewise.
5398 * ira.c (setup_prohibited_class_mode_regs): Likewise.
5399 (setup_prohibited_mode_move_regs): Likewise.
5400 * ira.h (target_ira): Likewise.
5401 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5402 * lra-constraints.c (process_alt_operands): Likewise.
5403 (split_reg): Likewise.
5404 * recog.c (peep2_find_free_register): Likewise.
5405 * ree.c (combine_reaching_defs): Likewise.
5406 * regcprop.c (maybe_mode_change): Likewise.
5407 * reginfo.c (init_reg_sets_1): Likewise.
5408 (choose_hard_reg_mode): Likewise.
5409 (simplifiable_subregs): Likewise.
5410 * regrename.c (check_new_reg_p): Likewise.
5411 * reload.c (find_valid_class): Likewise.
5412 (find_valid_class_1): Likewise.
5413 (reload_inner_reg_of_subreg): Likewise.
5414 (push_reload): Likewise.
5415 (combine_reloads): Likewise.
5416 (find_dummy_reload): Likewise.
5417 (find_reloads): Likewise.
5418 * reload1.c (find_reg): Likewise.
5419 (set_reload_reg): Likewise.
5420 (allocate_reload_reg): Likewise.
5421 (choose_reload_regs): Likewise.
5422 (reload_adjust_reg_for_temp): Likewise.
5423 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
5424 (simplify_subreg_regno): Likewise.
5425 * sel-sched.c (init_regs_for_mode): Likewise.
5426 * varasm.c (make_decl_rtl): Likewise.
5427 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
5428 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
5430 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
5431 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
5432 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5433 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
5434 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
5435 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5436 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
5437 (arc_mode_class): Delete.
5438 (HARD_REGNO_MODE_OK): Delete.
5439 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5440 (arc_hard_regno_mode_ok): Rename old array to...
5441 (arc_hard_regno_mode_ok_modes): ...this.
5442 (arc_conditional_register_usage): Update accordingly.
5443 (arc_mode_class): Make static.
5444 (arc_hard_regno_mode_ok): New function.
5445 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
5446 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
5447 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5448 (arm_hard_regno_mode_ok): Make static.
5449 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
5451 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
5452 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
5453 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
5455 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5456 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
5457 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
5458 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
5459 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
5460 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5461 * config/bfin/predicates.md (valid_reg_operand): Use
5462 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
5463 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
5464 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
5465 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5466 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
5467 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
5468 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5469 (cr16_hard_regno_mode_ok): Make static and return a bool.
5470 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
5471 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5472 (cris_hard_regno_mode_ok): New function.
5473 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
5474 (epiphany_mode_class): Delete.
5475 (HARD_REGNO_MODE_OK): Delete.
5476 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
5477 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5478 (hard_regno_mode_ok): Rename to...
5479 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
5480 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
5481 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
5483 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
5484 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
5485 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5486 (frv_hard_regno_mode_ok): Make static and return a bool.
5487 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
5489 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
5490 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
5491 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
5492 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
5494 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5495 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
5496 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
5497 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
5499 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5500 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
5501 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5502 (ia64_hard_regno_mode_ok): New function.
5503 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
5504 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5505 (iq2000_hard_regno_mode_ok): New function.
5506 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
5507 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5508 (lm32_hard_regno_mode_ok): New function.
5509 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
5510 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
5511 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
5512 instead of HARD_REGNO_MODE_OK.
5513 (m32c_hard_regno_ok): Rename to...
5514 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
5515 (m32c_cannot_change_mode_class): Update accordingly.
5516 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5517 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
5518 (m32r_mode_class): Delete.
5519 (HARD_REGNO_MODE_OK): Delete.
5520 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5521 (m32r_hard_regno_mode_ok): Rename to...
5522 (m32r_hard_regno_modes): ...this.
5523 (m32r_mode_class): Make static.
5524 (m32r_hard_regno_mode_ok): New function.
5525 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
5526 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
5527 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5528 (m68k_hard_regno_mode_ok): Make static.
5529 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
5530 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5531 (mcore_hard_regno_mode_ok): New function.
5532 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
5533 (HARD_REGNO_MODE_OK): Delete.
5534 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
5536 (microblaze_hard_regno_mode_ok_p): ...this and make static.
5537 (microblaze_hard_regno_mode_ok): New function.
5538 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5539 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
5540 (mips_hard_regno_mode_ok): Delete.
5541 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
5542 (mips_hard_regno_mode_ok_p): ...this and make static.
5543 (mips_hard_regno_mode_ok_p): Rename to...
5544 (mips_hard_regno_mode_ok_uncached): ...this.
5545 (mips_hard_regno_mode_ok): New function.
5546 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
5547 of HARD_REGNO_MODE_OK.
5548 (mips_option_override): Update after above name changes.
5549 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5550 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
5551 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
5552 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
5553 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
5554 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5555 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
5556 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
5557 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
5558 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5559 (msp430_hard_regno_mode_ok): Make static and return a bool.
5560 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
5561 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
5562 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
5564 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5565 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
5566 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
5567 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
5568 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
5569 (PA_HARD_REGNO_MODE_OK): ...this
5570 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
5571 (PA_HARD_REGNO_MODE_OK): ...this.
5572 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5573 (pa_hard_regno_mode_ok): New function.
5574 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
5575 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5576 (pdp11_hard_regno_mode_ok): New function.
5577 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
5578 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
5580 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
5582 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5583 (rs6000_hard_regno_mode_ok): Rename to...
5584 (rs6000_hard_regno_mode_ok_uncached): ...this.
5585 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5586 (rs6000_hard_regno_mode_ok): New function.
5587 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
5588 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
5589 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
5590 (riscv_hard_regno_mode_ok): ...this and make static.
5591 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5592 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
5593 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
5594 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5595 (rl78_hard_regno_mode_ok): Make static and return bool.
5596 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
5597 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
5599 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
5600 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5601 (rs6000_hard_regno_mode_ok): Rename to...
5602 (rs6000_hard_regno_mode_ok_uncached): ...this.
5603 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5604 (rs6000_hard_regno_mode_ok): New function.
5605 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
5606 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
5607 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5608 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
5609 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
5610 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
5611 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5612 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
5613 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
5614 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5615 (sh_hard_regno_mode_ok): Make static.
5616 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
5617 instead of HARD_REGNO_MODE_OK.
5618 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
5619 (sparc_mode_class): Delete.
5620 (HARD_REGNO_MODE_OK): Delete.
5621 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5622 (hard_regno_mode_classes): Make static.
5623 (sparc_mode_class): Likewise.
5624 (sparc_hard_regno_mode_ok): New function.
5625 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
5626 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
5627 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
5629 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5630 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
5631 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
5632 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
5633 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
5634 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5635 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
5636 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
5637 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5638 (visium_hard_regno_mode_ok): New function.
5639 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
5640 instead of HARD_REGNO_MODE_OK.
5641 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
5642 (HARD_REGNO_MODE_OK): Delete.
5643 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
5644 (xtensa_hard_regno_mode_ok_p): ...this and make static.
5645 (xtensa_option_override): Update accordingly.
5646 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5647 (xtensa_hard_regno_mode_ok): New function.
5648 * system.h (HARD_REGNO_MODE_OK): Poison.
5650 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5651 Alan Hayward <alan.hayward@arm.com>
5652 David Sherwood <david.sherwood@arm.com>
5654 * target.def (hard_regno_call_part_clobbered): New hook.
5655 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
5656 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
5657 * doc/tm.texi: Regenerate.
5658 * hooks.h (hook_bool_uint_mode_false): Declare.
5659 * hooks.c (hook_bool_uint_mode_false): New function.
5660 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5661 * cselib.c (cselib_process_insn): Use
5662 targetm.hard_regno_call_part_clobbered instead of
5663 HARD_REGNO_CALL_PART_CLOBBERED.
5664 * ira-conflicts.c (ira_build_conflicts): Likewise.
5665 * ira-costs.c (ira_tune_allocno_costs): Likewise.
5666 * lra-constraints.c (need_for_call_save_p): Likewise.
5667 * lra-lives.c: Include target.h.
5668 (check_pseudos_live_through_calls): Use
5669 targetm.hard_regno_call_part_clobbered instead of
5670 HARD_REGNO_CALL_PART_CLOBBERED.
5671 * regcprop.c: Include target.h.
5672 (copyprop_hardreg_forward_1): Use
5673 targetm.hard_regno_call_part_clobbered instead of
5674 HARD_REGNO_CALL_PART_CLOBBERED.
5675 * reginfo.c (choose_hard_reg_mode): Likewise.
5676 * regrename.c (check_new_reg_p): Likewise.
5677 * reload.c (find_equiv_reg): Likewise.
5678 * reload1.c (emit_reload_insns): Likewise.
5679 * sched-deps.c (deps_analyze_insn): Likewise.
5680 * sel-sched.c (init_regs_for_mode): Likewise.
5681 (mark_unavailable_hard_regs): Likewise.
5682 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
5683 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5684 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
5686 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5687 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5688 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
5690 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
5692 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5693 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5694 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
5696 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5697 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5698 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
5700 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5701 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
5703 * config/powerpcspe/powerpcspe.c
5704 (rs6000_hard_regno_call_part_clobbered): New function.
5705 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5706 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5707 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
5709 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5710 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5711 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
5713 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5714 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5715 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
5717 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5718 Alan Hayward <alan.hayward@arm.com>
5719 David Sherwood <david.sherwood@arm.com>
5721 * rtl.h (subreg_memory_offset): Declare.
5722 * emit-rtl.c (subreg_memory_offset): New function.
5723 * expmed.c (store_bit_field_1): Use it.
5724 * expr.c (undefined_operand_subword_p): Likewise.
5725 * simplify-rtx.c (simplify_subreg): Likewise.
5727 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
5729 PR rtl-optimization/57448
5732 * optabs.c (expand_atomic_load): Place compiler memory barriers if
5733 using atomic_load pattern.
5734 (expand_atomic_store): Likewise.
5736 2017-09-04 Jakub Jelinek <jakub@redhat.com>
5739 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
5740 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
5741 handling. Use replace_call_with_value with NULL instead of
5742 gsi_replace, unlink_stmt_vdef and release_defs.
5744 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
5747 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
5749 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5752 * rtl.h (emit_library_call_value_1): Declare.
5753 (emit_library_call): Replace declaration with a series of overloads.
5754 Remove the parameter count argument.
5755 (emit_library_call_value): Likewise.
5756 * calls.c (emit_library_call_value_1): Make global. Replace varargs
5757 with an "rtx_mode_t *".
5758 (emit_library_call_value): Delete.
5759 (emit_library_call): Likewise.
5760 * asan.c (asan_emit_stack_protection): Update calls accordingly.
5761 (asan_emit_allocas_unpoison): Likewise.
5762 * builtins.c (expand_builtin_powi): Likewise.
5763 (expand_asan_emit_allocas_unpoison): Likewise.
5764 * cfgexpand.c (expand_main_function): Likewise.
5765 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
5766 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
5767 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
5768 * config/arm/arm.c (arm_trampoline_init): Likewise.
5769 (arm_call_tls_get_addr): Likewise.
5770 (arm_expand_divmod_libfunc): Likewise.
5771 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
5772 (smulsi3_highpart): Likewise.
5773 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
5774 (c6x_expand_compare): Likewise.
5775 (c6x_expand_movmem): Likewise.
5776 * config/frv/frv.c (frv_trampoline_init): Likewise.
5777 * config/i386/i386.c (ix86_trampoline_init): Likewise.
5778 (ix86_expand_divmod_libfunc): Likewise.
5779 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
5780 (ia64_expand_compare): Likewise.
5781 (ia64_profile_hook): Likewise.
5782 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
5783 (nonlocal_goto): Likewise.
5784 (restore_stack_nonlocal): Likewise.
5785 * config/m32r/m32r.c (block_move_call): Likewise.
5786 (m32r_trampoline_init): Likewise.
5787 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
5788 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
5789 (m68k_call_m68k_read_tp): Likewise.
5790 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
5791 (microblaze_expand_divide): Likewise.
5792 * config/mips/mips.h (mips_args): Likewise.
5793 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
5794 (MIPS_ICACHE_SYNC): Likewise.
5795 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
5796 (nios2_trampoline_init): Likewise.
5797 * config/pa/pa.c (hppa_tls_call): Likewise.
5798 (pa_trampoline_init): Likewise.
5799 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
5800 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
5801 (expand_strn_compare): Likewise.
5802 (rs6000_generate_compare): Likewise.
5803 (rs6000_expand_float128_convert): Likewise.
5804 (output_profile_hook): Likewise.
5805 (rs6000_trampoline_init): Likewise.
5806 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
5807 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
5808 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
5809 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
5810 (rs6000_generate_compare): Likewise.
5811 (rs6000_expand_float128_convert): Likewise.
5812 (output_profile_hook): Likewise.
5813 (rs6000_trampoline_init): Likewise.
5814 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
5815 * config/sh/sh.c (sh_trampoline_init): Likewise.
5816 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
5817 (sparc_emit_float_lib_cmp): Likewise.
5818 (sparc32_initialize_trampoline): Likewise.
5819 (sparc64_initialize_trampoline): Likewise.
5820 (sparc_profile_hook): Likewise.
5821 * config/spu/spu.c (ea_load_store): Likewise.
5822 * config/spu/spu.md (floatunssidf2): Likewise.
5823 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
5824 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
5825 * config/visium/visium.c (expand_block_move_4): Likewise.
5826 (expand_block_move_2): Likewise.
5827 (expand_block_move_1): Likewise.
5828 (expand_block_set_4): Likewise.
5829 (expand_block_set_2): Likewise.
5830 (expand_block_set_1): Likewise.
5831 (visium_trampoline_init): Likewise.
5832 (visium_profile_hook): Likewise.
5833 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
5834 (xtensa_setup_frame_addresses): Likewise.
5835 (xtensa_trampoline_init): Likewise.
5836 * except.c (sjlj_emit_function_enter): Likewise.
5837 (sjlj_emit_function_exit): Likewise.
5838 * explow.c (allocate_dynamic_stack_space): Likewise.
5839 (probe_stack_range): Likewise.
5840 * expr.c (convert_mode_scalar): Likewise.
5841 * optabs.c (expand_binop): Likewise.
5842 (expand_twoval_binop_libfunc): Likewise.
5843 (expand_unop): Likewise.
5844 (prepare_cmp_insn): Likewise.
5845 (prepare_float_lib_cmp): Likewise.
5846 (expand_float): Likewise.
5847 (expand_fix): Likewise.
5848 (expand_fixed_convert): Likewise.
5849 (maybe_emit_sync_lock_test_and_set): Likewise.
5850 (expand_atomic_compare_and_swap): Likewise.
5851 (expand_mem_thread_fence): Likewise.
5852 (expand_atomic_fetch_op): Likewise.
5854 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
5856 * doc/generic.texi (OpenACC): Adjust URL.
5857 * doc/invoke.texi (C Dialect Options): Ditto.
5859 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
5861 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
5862 predicate for operand 1. Add (m,<S>) constraint.
5863 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
5864 Prevent memory operand 1 with register operand 2.
5866 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
5868 PR rtl-optimization/82024
5869 * combine.c (try_combine): If the combination result is a PARALLEL,
5870 and we only need to retain the SET in there that would be placed
5871 at I2, check that we can place that at I3 instead, before doing so.
5873 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5876 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
5878 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
5879 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
5882 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
5883 Jeff Law <law@redhat.com>
5885 * varasm.c (bss_initializer_p): Do not put constants into .bss
5886 (categorize_decl_for_section): Handle bss_initializer_p returning
5887 false when DECL_INITIAL is NULL.
5889 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5892 * config/s390/s390.c (s390_can_inline_p): New function.
5894 2017-09-01 Jeff Law <law@redhat.com>
5896 PR tree-optimization/82052
5897 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5898 Always initialize the returned slot after a hash table miss
5899 when INSERT is true.
5901 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
5903 * config/s390/s390.md (mem_signal_fence): Remove.
5904 * doc/md.texi (mem_signal_fence): Remove.
5905 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
5907 * target-insns.def (mem_signal_fence): Remove.
5909 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5912 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
5915 * asan.c (create_odr_indicator): Strip name encoding from assembler
5916 name before appending it after __odr_asan_.
5918 2017-09-01 Martin Liska <mliska@suse.cz>
5920 PR tree-optimization/82059
5921 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
5922 frequency only when an edge is redirected.
5924 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5926 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
5927 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
5928 (arc_conditional_register_usage): Remove ARC600 lp_count
5930 (arc_file_start): Emit Tag_ARC_CPU_variation.
5931 (arc_can_use_doloop_p): New conditions to use ZOLs.
5932 (hwloop_fail): New function.
5933 (hwloop_optimize): Likewise.
5934 (hwloop_pattern_reg): Likewise.
5935 (arc_doloop_hooks): New struct, to be used with reorg_loops.
5936 (arc_reorg_loops): New function, calls reorg_loops.
5937 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
5938 (arc600_corereg_hazard): Remove ZOL checking, case handled by
5940 (arc_loop_hazard): Remove function, functionality moved into
5942 (arc_hazard): Remove arc_loop_hazard call.
5943 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
5944 into hwloop_optimize.
5945 (arc_label_align): Remove ZOL handling.
5946 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
5947 * config/arc/arc.md (doloop_begin): Remove pattern.
5948 (doloop_begin_i): Likewise.
5949 (doloop_end_i): Likewise.
5950 (doloop_fallback): Likewise.
5951 (doloop_fallback_m): Likewise.
5952 (doloop_end): Reimplement expand.
5953 (arc_lp): New pattern for LP instruction.
5954 (loop_end): New pattern.
5955 (loop_fail): Likewise.
5956 (decrement_and_branch_until_zero): Likewise.
5957 * config/arc/arc.opt (mlpc-width): New option.
5958 * doc/invoke.texi (mlpc-width): Document option.
5960 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5962 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
5963 (arc_ccfsm_advance): Fix checking for delay slots.
5964 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
5966 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5968 * config/arc/arc.md (movqi_insn): Add stores to save constant long
5970 (movhi_insn): Update store instruction constraint which are saving
5971 6-bit short immediates.
5972 (movsi_insn): Consider also short scaled load operations.
5973 (zero_extendhisi2_i): Use Usd constraint instead of T.
5974 (extendhisi2_i): Add q constraint.
5975 (arc_clzsi2): Add type and length attributes.
5976 (arc_ctzsi2): Likewise.
5977 * config/arc/constraints.md (Usc): Update constraint, the
5978 assembler can parse two relocations for a single instruction.
5980 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5982 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
5983 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
5985 2017-08-31 Olivier Hainque <hainque@adacore.com>
5987 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
5988 match as powerpc-wrs-vxworks*.
5990 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
5992 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
5993 register constraint for by-element operand.
5994 (aarch64_mls_elt_merge<mode>): Likewise.
5996 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
5998 * config/arc/arc.c (arc_can_follow_jump): Check for short
6001 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6003 * config.gcc: Use g.opt for arc.
6004 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6005 functionality moved to ...
6006 (legitimate_scaled_address_p): New function, ...here.
6007 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6008 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6009 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6011 (arc_override_options): Handle G option.
6012 (arc_output_pic_addr_const): Correct function definition.
6013 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6014 (arc_decl_anon_ns_mem_p): Delete.
6015 (arc_in_small_data_p): Overhaul this function to take into
6016 consideration the value given via G option.
6017 (arc_rewrite_small_data_1): Renamed and corrected old
6018 arc_rewrite_small_data function.
6019 (arc_rewrite_small_data): New function.
6020 (small_data_pattern): Don't use pic_offset_table_rtx.
6021 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6022 * config/arc/simdext.md (movmisalignv2hi): Use
6023 prepare_move_operands function.
6025 (movmisalign*): Likewise.
6026 * doc/invoke.texi (ARC options): Document -G option.
6028 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6030 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6032 * config/arc/arc.c (arc_print_operand): Output scalled address for
6033 sdata whenever is possible.
6034 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6035 load/stores are available.
6036 (compact_sda_memory_operand): Check for the alignment required by
6037 code density instructions.
6038 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6040 * config/arc/constraints.md (Usd): Update constraint.
6041 (Us0): New constraint.
6042 (Usc): Update constraint.
6044 2017-08-31 Richard Biener <rguenther@suse.de>
6047 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6050 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6052 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6053 Resize type_signature.
6055 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6056 Alan Hayward <alan.hayward@arm.com>
6057 David Sherwood <david.sherwood@arm.com>
6059 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
6060 subregs whose inner modes can be stored in GPRs.
6061 (aarch64_classify_index): Likewise.
6063 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6064 Alan Hayward <alan.hayward@arm.com>
6065 David Sherwood <david.sherwood@arm.com>
6067 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6068 (V_INT_EQUIV): ...this.
6069 (v_cmp_result): Rename to...
6070 (v_int_equiv): ...this.
6071 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6072 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6073 (copysign<mode>3): Likewise.
6074 (aarch64_simd_bsl<mode>_internal): Likewise.
6075 (aarch64_simd_bsl<mode>): Likewise.
6076 (vec_cmp<mode><mode>): Likewise.
6077 (vcond<mode><mode>): Likewise.
6078 (vcond<v_cmp_mixed><mode>): Likewise.
6079 (vcondu<mode><v_cmp_mixed>): Likewise.
6080 (aarch64_cm<optab><mode>): Likewise.
6081 (aarch64_cmtst<mode>): Likewise.
6082 (aarch64_fac<optab><mode>): Likewise.
6083 (vec_perm_const<mode>): Likewise.
6084 (vcond_mask_<mode><v_cmp_result>): Rename to...
6085 (vcond_mask_<mode><v_int_equiv>): ...this.
6086 (vec_cmp<mode><v_cmp_result>): Rename to...
6087 (vec_cmp<mode><v_int_equiv>): ...this.
6089 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6090 Alan Hayward <alan.hayward@arm.com>
6091 David Sherwood <david.sherwood@arm.com>
6093 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
6095 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
6096 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
6097 (UNSPEC_LD4_DREG): New unspecs.
6098 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
6099 (aarch64_ld2<mode>_dreg_be): Replace with...
6100 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
6102 (aarch64_ld3<mode>_dreg_le)
6103 (aarch64_ld3<mode>_dreg_be): Replace with...
6104 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
6106 (aarch64_ld4<mode>_dreg_le)
6107 (aarch64_ld4<mode>_dreg_be): Replace with...
6108 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
6111 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6113 PR tree-optimization/81987
6114 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
6115 insert an initializer in a location not dominated by the stride
6118 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6120 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
6121 on the entire header of the finally block in the fallthru case.
6123 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6125 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
6127 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
6129 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
6130 rs6000_emit_move_from_cr and call renamed function.
6131 (rs6000_emit_prologue): Call renamed functions.
6132 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
6133 movesi_from_cr, remove volatile CRs.
6135 2017-08-30 Jon Beniston <jon@beniston.com>
6136 Richard Biener <rguenther@suse.de>
6138 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
6139 of VECTOR_MODE_P check.
6140 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
6141 element vector types.
6143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6145 * df.h (df_read_modify_subreg_p): Remove in favor of...
6146 * rtl.h (read_modify_subreg_p): ...this new function. Take a
6147 const_rtx instead of an rtx.
6148 * cprop.c (local_cprop_find_used_regs): Update accordingly.
6149 * df-problems.c (df_word_lr_mark_ref): Likewise.
6150 * ira-lives.c (mark_pseudo_reg_live): Likewise.
6151 (mark_pseudo_reg_dead): Likewise.
6152 (mark_ref_dead): Likewise.
6153 * reginfo.c (init_subregs_of_mode): Likewise.
6154 * sched-deps.c (sched_analyze_1): Likewise.
6155 * df-scan.c (df_def_record_1): Likewise.
6156 (df_uses_record): Likewise.
6157 (df_read_modify_subreg_p): Remove in favor of...
6158 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
6159 const_rtx instead of an rtx.
6161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6162 Alan Hayward <alan.hayward@arm.com>
6163 David Sherwood <david.sherwood@arm.com>
6165 * rtl.h (partial_subreg_p): New function.
6166 * caller-save.c (save_call_clobbered_regs): Use it.
6167 * calls.c (expand_call): Likewise.
6168 * combine.c (combinable_i3pat): Likewise.
6169 (simplify_set): Likewise.
6170 (make_extraction): Likewise.
6171 (make_compound_operation_int): Likewise.
6172 (gen_lowpart_or_truncate): Likewise.
6173 (force_to_mode): Likewise.
6174 (make_field_assignment): Likewise.
6175 (reg_truncated_to_mode): Likewise.
6176 (record_truncated_value): Likewise.
6177 (move_deaths): Likewise.
6178 * cse.c (record_jump_cond): Likewise.
6179 (cse_insn): Likewise.
6180 * cselib.c (cselib_lookup_1): Likewise.
6181 * expmed.c (extract_bit_field_using_extv): Likewise.
6182 * function.c (assign_parm_setup_reg): Likewise.
6183 * ifcvt.c (noce_convert_multiple_sets): Likewise.
6184 * ira-build.c (create_insn_allocnos): Likewise.
6185 * lra-coalesce.c (merge_pseudos): Likewise.
6186 * lra-constraints.c (match_reload): Likewise.
6187 (simplify_operand_subreg): Likewise.
6188 (curr_insn_transform): Likewise.
6189 * lra-lives.c (process_bb_lives): Likewise.
6190 * lra.c (new_insn_reg): Likewise.
6191 (lra_substitute_pseudo): Likewise.
6192 * regcprop.c (mode_change_ok): Likewise.
6193 (maybe_mode_change): Likewise.
6194 (copyprop_hardreg_forward_1): Likewise.
6195 * reload.c (push_reload): Likewise.
6196 (find_reloads): Likewise.
6197 (find_reloads_subreg_address): Likewise.
6198 * reload1.c (alter_reg): Likewise.
6199 (eliminate_regs_1): Likewise.
6200 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6202 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
6204 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
6205 back to if statements, including unpack.
6207 2017-08-30 Martin Liska <mliska@suse.cz>
6210 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
6212 (func_checker::compare_asm_inputs_outputs): ... this function.
6213 (func_checker::compare_gimple_asm): Use the function to compare
6214 also ASM constrains.
6215 * ipa-icf-gimple.h: Rename the function.
6217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6218 Alan Hayward <alan.hayward@arm.com>
6219 David Sherwood <david.sherwood@arm.com>
6221 * coretypes.h (complex_mode): New type.
6222 * gdbhooks.py (build_pretty_printer): Handle it.
6223 * machmode.h (complex_mode): New class.
6224 (complex_mode::includes_p): New function.
6225 (is_complex_int_mode): Likewise.
6226 (is_complex_float_mode): Likewise.
6227 * genmodes.c (get_mode_class): Handle complex mode classes.
6228 * function.c (expand_function_end): Use is_complex_int_mode.
6230 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6231 Alan Hayward <alan.hayward@arm.com>
6232 David Sherwood <david.sherwood@arm.com>
6234 * coretypes.h (scalar_mode_pod): New typedef.
6235 * gdbhooks.py (build_pretty_printer): Handle it.
6236 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
6237 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
6238 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
6239 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
6242 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6243 Alan Hayward <alan.hayward@arm.com>
6244 David Sherwood <david.sherwood@arm.com>
6246 * machmode.h (mode_for_vector): Take a scalar_mode instead
6248 * stor-layout.c (mode_for_vector): Likewise.
6249 * explow.c (promote_mode): Use as_a <scalar_mode>.
6250 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
6252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6253 Alan Hayward <alan.hayward@arm.com>
6254 David Sherwood <david.sherwood@arm.com>
6256 * target.def (preferred_simd_mode): Take a scalar_mode
6257 instead of a machine_mode.
6258 * targhooks.h (default_preferred_simd_mode): Likewise.
6259 * targhooks.c (default_preferred_simd_mode): Likewise.
6260 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
6261 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
6262 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
6263 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
6264 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
6265 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
6266 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
6267 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
6269 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
6270 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
6271 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
6272 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
6273 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
6274 * doc/tm.texi: Regenerate.
6275 * optabs-query.c (can_vec_mask_load_store_p): Return false for
6278 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6279 Alan Hayward <alan.hayward@arm.com>
6280 David Sherwood <david.sherwood@arm.com>
6282 * target.def (scalar_mode_supported_p): Take a scalar_mode
6283 instead of a machine_mode.
6284 * targhooks.h (default_scalar_mode_supported_p): Likewise.
6285 * targhooks.c (default_scalar_mode_supported_p): Likewise.
6286 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
6287 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6288 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
6289 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
6290 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
6291 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
6292 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
6293 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
6294 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
6295 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
6296 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
6297 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
6299 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
6300 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
6301 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6302 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6303 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
6305 * doc/tm.texi: Regenerate.
6307 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6308 Alan Hayward <alan.hayward@arm.com>
6309 David Sherwood <david.sherwood@arm.com>
6311 * coretypes.h (opt_scalar_mode): New typedef.
6312 * gdbhooks.py (build_pretty_printers): Handle it.
6313 * machmode.h (mode_iterator::get_2xwider): Add overload for
6315 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
6317 * expr.c (convert_mode_scalar): Likewise.
6318 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6319 * optabs.c (expand_float): Likewise.
6320 (expand_fix): Likewise.
6321 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6324 Alan Hayward <alan.hayward@arm.com>
6325 David Sherwood <david.sherwood@arm.com>
6327 * optabs.c (expand_float): Explicitly check for scalars before
6328 using a branching expansion.
6329 (expand_fix): Likewise.
6331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6332 Alan Hayward <alan.hayward@arm.com>
6333 David Sherwood <david.sherwood@arm.com>
6335 * expr.c (convert_mode): Split scalar handling out into...
6336 (convert_mode_scalar): ...this new function. Treat the modes
6339 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6340 Alan Hayward <alan.hayward@arm.com>
6341 David Sherwood <david.sherwood@arm.com>
6343 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
6345 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
6347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6348 Alan Hayward <alan.hayward@arm.com>
6349 David Sherwood <david.sherwood@arm.com>
6351 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
6352 rather than a machine_mode.
6353 (fixed_from_string): Likewise.
6354 (fixed_convert): Likewise.
6355 (fixed_convert_from_int): Likewise.
6356 (fixed_convert_from_real): Likewise.
6357 (real_convert_from_fixed): Likewise.
6358 * fixed-value.c (fixed_from_double_int): Likewise.
6359 (fixed_from_string): Likewise.
6360 (fixed_convert): Likewise.
6361 (fixed_convert_from_int): Likewise.
6362 (fixed_convert_from_real): Likewise.
6363 (real_convert_from_fixed): Likewise.
6364 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
6366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6367 Alan Hayward <alan.hayward@arm.com>
6368 David Sherwood <david.sherwood@arm.com>
6370 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
6371 of separate mode class checks. Do not allow vector modes here.
6372 (immed_wide_int_const): Use as_a <scalar_mode>.
6373 * explow.c (trunc_int_for_mode): Likewise.
6374 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
6375 (wi::shwi): Likewise.
6376 (wi::min_value): Likewise.
6377 (wi::max_value): Likewise.
6378 * dwarf2out.c (loc_descriptor): Likewise.
6379 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
6382 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6383 Alan Hayward <alan.hayward@arm.com>
6384 David Sherwood <david.sherwood@arm.com>
6386 * tree.h (SCALAR_TYPE_MODE): New macro.
6387 * expr.c (expand_expr_addr_expr_1): Use it.
6388 (expand_expr_real_2): Likewise.
6389 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
6390 (fold_convert_const_fixed_from_int): Likewise.
6391 (fold_convert_const_fixed_from_real): Likewise.
6392 (native_encode_fixed): Likewise
6393 (native_encode_complex): Likewise
6394 (native_encode_vector): Likewise.
6395 (native_interpret_fixed): Likewise.
6396 (native_interpret_real): Likewise.
6397 (native_interpret_complex): Likewise.
6398 (native_interpret_vector): Likewise.
6399 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
6400 (simd_clone_adjust_argument_types): Likewise.
6401 (simd_clone_init_simd_arrays): Likewise.
6402 (simd_clone_adjust): Likewise.
6403 * stor-layout.c (layout_type): Likewise.
6404 * tree.c (build_minus_one_cst): Likewise.
6405 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6406 * tree-inline.c (estimate_move_cost): Likewise.
6407 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
6408 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
6409 (vectorizable_reduction): Likewise.
6410 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
6411 (vect_recog_mixed_size_cond_pattern): Likewise.
6412 (check_bool_pattern): Likewise.
6413 (adjust_bool_pattern): Likewise.
6414 (search_type_for_mask_1): Likewise.
6415 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6416 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6417 (vectorizable_load): Likewise.
6418 (vectorizable_store): Likewise.
6419 * ubsan.c (ubsan_encode_value): Likewise.
6420 * varasm.c (output_constant): Likewise.
6422 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6423 Alan Hayward <alan.hayward@arm.com>
6424 David Sherwood <david.sherwood@arm.com>
6426 * coretypes.h (scalar_mode): New class.
6427 * machmode.h (scalar_mode): Likewise.
6428 (scalar_mode::includes_p): New function.
6429 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
6430 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
6431 * genmodes.c (get_mode_class): Handle remaining scalar modes.
6432 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
6433 * expmed.c (store_bit_field_1): Likewise.
6434 (extract_bit_field_1): Likewise.
6435 * expr.c (write_complex_part): Likewise.
6436 (read_complex_part): Likewise.
6437 (emit_move_complex_push): Likewise.
6438 (expand_expr_real_2): Likewise.
6439 * function.c (assign_parm_setup_reg): Likewise.
6440 (assign_parms_unsplit_complex): Likewise.
6441 * optabs.c (expand_binop): Likewise.
6442 * rtlanal.c (subreg_get_info): Likewise.
6443 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6444 * varasm.c (output_constant_pool_2): Likewise.
6446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6447 Alan Hayward <alan.hayward@arm.com>
6448 David Sherwood <david.sherwood@arm.com>
6450 * expmed.c (extract_high_half): Use scalar_int_mode and remove
6452 (expmed_mult_highpart_optab): Likewise.
6453 (expmed_mult_highpart): Likewise.
6455 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6456 Alan Hayward <alan.hayward@arm.com>
6457 David Sherwood <david.sherwood@arm.com>
6459 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
6460 instead of a machine_mode.
6461 (builtin_memset_read_str): Likewise.
6462 * builtins.c (c_readstr): Likewise.
6463 (builtin_memcpy_read_str): Likewise.
6464 (builtin_strncpy_read_str): Likewise.
6465 (builtin_memset_read_str): Likewise.
6466 (builtin_memset_gen_str): Likewise.
6467 (expand_builtin_signbit): Use scalar_int_mode for local variables.
6468 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
6469 instead of a machine_mode.
6470 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
6472 (make_extraction): Likewise.
6473 (try_widen_shift_mode): Take and return scalar_int_modes instead
6475 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
6476 a scalar_int_mode instead of a machine_mode.
6477 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
6478 (avr_addr_space_pointer_mode): Likewise.
6479 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
6480 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
6481 (msp430_unwind_word_mode): Likewise.
6482 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
6483 (spu_addr_space_pointer_mode): Likewise.
6484 (spu_addr_space_address_mode): Likewise.
6485 (spu_libgcc_cmp_return_mode): Likewise.
6486 (spu_libgcc_shift_count_mode): Likewise.
6487 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
6488 (rl78_addr_space_pointer_mode): Likewise.
6489 (fl78_unwind_word_mode): Likewise.
6490 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
6492 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
6493 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
6494 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
6495 (mips_valid_pointer_mode): Likewise.
6496 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
6497 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
6498 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
6500 (ft32_addr_space_address_mode): Likewise.
6501 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
6502 scalar_int_mode instead of a machine_mode.
6503 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
6505 (m32c_addr_space_address_mode): Likewise.
6506 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
6507 (rs6000_eh_return_filter_mode): Likewise.
6508 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
6509 (rs6000_eh_return_filter_mode): Likewise.
6510 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
6511 (s390_libgcc_shift_count_mode): Likewise.
6512 (s390_unwind_word_mode): Likewise.
6513 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
6515 * target.def (mode_rep_extended): Likewise.
6516 (valid_pointer_mode): Likewise.
6517 (addr_space.valid_pointer_mode): Likewise.
6518 (eh_return_filter_mode): Return a scalar_int_mode rather than
6520 (libgcc_cmp_return_mode): Likewise.
6521 (libgcc_shift_count_mode): Likewise.
6522 (unwind_word_mode): Likewise.
6523 (addr_space.pointer_mode): Likewise.
6524 (addr_space.address_mode): Likewise.
6525 * doc/tm.texi: Regenerate.
6526 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
6528 (do_jump): Use scalar_int_mode for local variables.
6529 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
6530 rather than a machine_mode.
6531 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
6532 (scompare_loc_descriptor_wide): Likewise.
6533 (scompare_loc_descriptor_narrow): Likewise.
6534 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
6536 * except.c (sjlj_emit_dispatch_table): Likewise.
6537 (expand_builtin_eh_copy_values): Likewise.
6538 * explow.c (convert_memory_address_addr_space_1): Likewise.
6539 Take a scalar_int_mode rather than a machine_mode.
6540 (convert_memory_address_addr_space): Take a scalar_int_mode rather
6541 than a machine_mode.
6542 (memory_address_addr_space): Use scalar_int_mode for local variables.
6543 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
6544 rather than a machine_mode.
6545 * expmed.c (mask_rtx): Likewise.
6546 (init_expmed_one_conv): Likewise.
6547 (expand_mult_highpart_adjust): Likewise.
6548 (extract_high_half): Likewise.
6549 (expmed_mult_highpart_optab): Likewise.
6550 (expmed_mult_highpart): Likewise.
6551 (expand_smod_pow2): Likewise.
6552 (expand_sdiv_pow2): Likewise.
6553 (emit_store_flag_int): Likewise.
6554 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
6556 (extract_low_bits): Likewise.
6557 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
6559 * expr.c (pieces_addr::adjust): Likewise.
6560 (can_store_by_pieces): Likewise.
6561 (store_by_pieces): Likewise.
6562 (clear_by_pieces_1): Likewise.
6563 (expand_expr_addr_expr_1): Likewise.
6564 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
6565 (expand_expr_real_1): Likewise.
6566 (try_casesi): Likewise.
6567 * final.c (shorten_branches): Likewise.
6568 * fold-const.c (fold_convert_const_int_from_fixed): Change the
6569 type of "mode" to machine_mode.
6570 * internal-fn.c (expand_arith_overflow_result_store): Take a
6571 scalar_int_mode rather than a machine_mode.
6572 (expand_mul_overflow): Use scalar_int_mode for local variables.
6573 * loop-doloop.c (doloop_modify): Likewise.
6574 (doloop_optimize): Likewise.
6575 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
6576 than a machine_mode.
6577 (expand_doubleword_shift_condmove): Likewise.
6578 (expand_doubleword_shift): Likewise.
6579 (expand_doubleword_clz): Likewise.
6580 (expand_doubleword_popcount): Likewise.
6581 (expand_doubleword_parity): Likewise.
6582 (expand_absneg_bit): Use scalar_int_mode for local variables.
6583 (prepare_float_lib_cmp): Likewise.
6584 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
6585 rather than a machine_mode.
6586 (convert_memory_address_addr_space): Likewise.
6587 (get_mode_bounds): Likewise.
6588 (get_address_mode): Return a scalar_int_mode rather than a
6590 * rtlanal.c (get_address_mode): Likewise.
6591 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
6592 than a machine_mode.
6593 * targhooks.c (default_mode_rep_extended): Likewise.
6594 (default_valid_pointer_mode): Likewise.
6595 (default_addr_space_valid_pointer_mode): Likewise.
6596 (default_eh_return_filter_mode): Return a scalar_int_mode rather
6597 than a machine_mode.
6598 (default_libgcc_cmp_return_mode): Likewise.
6599 (default_libgcc_shift_count_mode): Likewise.
6600 (default_unwind_word_mode): Likewise.
6601 (default_addr_space_pointer_mode): Likewise.
6602 (default_addr_space_address_mode): Likewise.
6603 * targhooks.h (default_eh_return_filter_mode): Likewise.
6604 (default_libgcc_cmp_return_mode): Likewise.
6605 (default_libgcc_shift_count_mode): Likewise.
6606 (default_unwind_word_mode): Likewise.
6607 (default_addr_space_pointer_mode): Likewise.
6608 (default_addr_space_address_mode): Likewise.
6609 (default_mode_rep_extended): Take a scalar_int_mode rather than
6611 (default_valid_pointer_mode): Likewise.
6612 (default_addr_space_valid_pointer_mode): Likewise.
6613 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
6615 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
6616 rather than a machine_mode.
6617 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
6618 for local variables.
6619 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6620 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
6621 than a machine_mode.
6623 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6624 Alan Hayward <alan.hayward@arm.com>
6625 David Sherwood <david.sherwood@arm.com>
6627 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6628 the mode argument to scalar_int_mode.
6629 (do_jump_by_parts_zero_rtx): Likewise.
6630 (do_jump_by_parts_equality_rtx): Likewise.
6631 (do_jump_by_parts_greater): Take a mode argument.
6632 (do_jump_by_parts_equality): Likewise.
6633 (do_jump_1): Update calls accordingly.
6635 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6636 Alan Hayward <alan.hayward@arm.com>
6637 David Sherwood <david.sherwood@arm.com>
6639 * is-a.h (safe_dyn_cast): New function.
6640 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
6641 (jump_table_for_label): Likewise.
6642 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
6643 instead of an rtx_insn *.
6644 (shorten_branches): Use dyn_cast instead of LABEL_P and
6645 JUMP_TABLE_DATA_P. Use jump_table_for_label and
6646 rtx_jump_table_data::get_data_mode.
6647 (final_scan_insn): Likewise.
6649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6650 Alan Hayward <alan.hayward@arm.com>
6651 David Sherwood <david.sherwood@arm.com>
6653 * combine.c (try_combine): Use is_a <scalar_int_mode> when
6654 trying to combine a full-register integer set with a subreg
6657 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6658 Alan Hayward <alan.hayward@arm.com>
6659 David Sherwood <david.sherwood@arm.com>
6661 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
6662 that is always either address_mode or pointer_mode.
6664 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6665 Alan Hayward <alan.hayward@arm.com>
6666 David Sherwood <david.sherwood@arm.com>
6668 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
6669 when the two are known to be equal.
6671 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6672 Alan Hayward <alan.hayward@arm.com>
6673 David Sherwood <david.sherwood@arm.com>
6675 * simplify-rtx.c (simplify_const_unary_operation): Use
6676 is_a <scalar_int_mode> instead of checking for a nonzero
6677 precision. Forcibly convert op_mode to a scalar_int_mode
6678 in that case. More clearly differentiate the operand and
6679 result modes and use the former when deciding what the value
6680 of a count-bits operation should be. Use is_int_mode instead
6681 of checking for a MODE_INT. Remove redundant check for whether
6682 this mode has a zero precision.
6684 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6685 Alan Hayward <alan.hayward@arm.com>
6686 David Sherwood <david.sherwood@arm.com>
6688 * optabs.c (widen_leading): Change the type of the mode argument
6689 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
6690 (widen_bswap): Likewise.
6691 (expand_parity): Likewise.
6692 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
6693 (expand_ffs): Likewise.
6694 (epand_unop): Check for scalar integer modes before calling the
6697 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6698 Alan Hayward <alan.hayward@arm.com>
6699 David Sherwood <david.sherwood@arm.com>
6701 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
6703 (expand_expr_real_1): Update call accordingly.
6705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6706 Alan Hayward <alan.hayward@arm.com>
6707 David Sherwood <david.sherwood@arm.com>
6709 * expmed.c (store_bit_field_using_insv): Add op0_mode and
6710 value_mode arguments. Use scalar_int_mode internally.
6711 (store_bit_field_1): Rename the new integer mode from imode
6712 to op0_mode and use it instead of GET_MODE (op0). Update calls
6713 to store_split_bit_field, store_bit_field_using_insv and
6714 store_fixed_bit_field.
6715 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
6716 Use scalar_int_mode internally. Use a bit count rather than a mode
6717 when calculating the largest bit size for get_best_mode.
6718 Update calls to store_split_bit_field and store_fixed_bit_field_1.
6719 (store_fixed_bit_field_1): Add mode and value_mode arguments.
6720 Remove assertion that OP0 has a scalar integer mode.
6721 (store_split_bit_field): Add op0_mode and value_mode arguments.
6722 Update calls to extract_fixed_bit_field.
6723 (extract_bit_field_using_extv): Add an op0_mode argument.
6724 Use scalar_int_mode internally.
6725 (extract_bit_field_1): Rename the new integer mode from imode to
6726 op0_mode and use it instead of GET_MODE (op0). Update calls to
6727 extract_split_bit_field, extract_bit_field_using_extv and
6728 extract_fixed_bit_field.
6729 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
6730 to extract_split_bit_field and extract_fixed_bit_field_1.
6731 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
6732 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
6734 (extract_split_bit_field): Add an op0_mode argument. Update call
6735 to extract_fixed_bit_field.
6737 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6738 Alan Hayward <alan.hayward@arm.com>
6739 David Sherwood <david.sherwood@arm.com>
6741 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
6742 * explow.c (hard_function_value): Likewise.
6743 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
6744 convert_to_mode call outside the loop.
6745 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
6746 for the mode iterator. Require the mode specified by max_pieces
6748 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
6750 (copy_blkmode_to_reg): Likewise.
6751 (set_storage_via_setmem): Likewise.
6752 * optabs.c (prepare_cmp_insn): Likewise.
6753 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6754 * stor-layout.c (finish_bitfield_representative): Likewise.
6756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6757 Alan Hayward <alan.hayward@arm.com>
6758 David Sherwood <david.sherwood@arm.com>
6760 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
6761 * expr.c (convert_move): Use them.
6762 (convert_modes): Likewise.
6763 (store_expr_with_bounds): Likewise.
6765 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6766 Alan Hayward <alan.hayward@arm.com>
6767 David Sherwood <david.sherwood@arm.com>
6769 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
6770 parameter for the mode of "x". Remove the "known_x", "known_mode"
6771 and "known_ret" arguments. Change the type of the mode argument
6773 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
6774 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
6775 (reg_num_sign_bit_copies_for_combine): Likewise.
6776 * rtlanal.c (nonzero_bits1): Likewise.
6777 (num_sign_bit_copies1): Likewise.
6778 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
6779 (reg_num_sign_bit_copies_general): Likewise.
6780 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
6781 (reg_nonzero_bits_general): Likewise.
6783 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6784 Alan Hayward <alan.hayward@arm.com>
6785 David Sherwood <david.sherwood@arm.com>
6787 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
6788 than in subroutines. Return 1 for non-integer modes.
6789 (cached_num_sign_bit_copies): Change the type of the mode parameter
6791 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
6792 classes. Handle CONST_INT_P first and then check whether X also
6793 has a scalar integer mode. Check the same thing for inner registers
6794 of a SUBREG and for values that are being extended or truncated.
6796 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6797 Alan Hayward <alan.hayward@arm.com>
6798 David Sherwood <david.sherwood@arm.com>
6800 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
6801 in subroutines. Return the mode mask for non-integer modes.
6802 (cached_nonzero_bits): Change the type of the mode parameter
6804 (nonzero_bits1): Likewise. Remove early exit for other mode
6805 classes. Handle CONST_INT_P first and then check whether X
6806 also has a scalar integer mode.
6808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6809 Alan Hayward <alan.hayward@arm.com>
6810 David Sherwood <david.sherwood@arm.com>
6812 * expr.c (widest_int_mode_for_size): Make the comment match the code.
6813 Return a scalar_int_mode and assert that the size is greater than
6815 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
6816 (op_by_pieces_d::op_by_pieces_d): Likewise.
6817 (op_by_pieces_d::run): Likewise.
6818 (can_store_by_pieces): Likewise.
6820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6821 Alan Hayward <alan.hayward@arm.com>
6822 David Sherwood <david.sherwood@arm.com>
6824 * combine.c (extract_left_shift): Add a mode argument and update
6826 (make_compound_operation_int): Change the type of the mode parameter
6827 to scalar_int_mode and update the call to extract_left_shift.
6829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6830 Alan Hayward <alan.hayward@arm.com>
6831 David Sherwood <david.sherwood@arm.com>
6833 * combine.c (simplify_and_const_int): Change the type of the mode
6834 parameter to scalar_int_mode.
6835 (simplify_and_const_int_1): Likewise. Update recursive call.
6837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6838 Alan Hayward <alan.hayward@arm.com>
6839 David Sherwood <david.sherwood@arm.com>
6841 * combine.c (simplify_compare_const): Check that the mode is a
6842 scalar_int_mode (rather than VOIDmode) before testing its
6844 (simplify_comparison): Move COMPARISON_P handling out of the
6845 loop and restrict the latter part of the loop to scalar_int_modes.
6846 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
6847 and when considering SUBREG_REGs. Use is_int_mode instead of
6848 checking GET_MODE_CLASS against MODE_INT.
6850 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6851 Alan Hayward <alan.hayward@arm.com>
6852 David Sherwood <david.sherwood@arm.com>
6854 * combine.c (try_widen_shift_mode): Move check for equal modes to...
6855 (simplify_shift_const_1): ...here. Use scalar_int_mode for
6856 shift_unit_mode and for modes involved in scalar shifts.
6858 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6859 Alan Hayward <alan.hayward@arm.com>
6860 David Sherwood <david.sherwood@arm.com>
6862 * combine.c (force_int_to_mode): New function, split out from...
6863 (force_to_mode): ...here. Keep xmode up-to-date and use it
6864 instead of GET_MODE (x).
6866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6867 Alan Hayward <alan.hayward@arm.com>
6868 David Sherwood <david.sherwood@arm.com>
6870 * optabs-query.h (extraction_insn::struct_mode): Change type to
6871 opt_scalar_int_mode and update comment.
6872 (extraction_insn::field_mode): Change type to scalar_int_mode.
6873 (extraction_insn::pos_mode): Likewise.
6874 * combine.c (make_extraction): Update accordingly.
6875 * optabs-query.c (get_traditional_extraction_insn): Likewise.
6876 (get_optab_extraction_insn): Likewise.
6877 * recog.c (simplify_while_replacing): Likewise.
6878 * expmed.c (narrow_bit_field_mem): Change the type of the mode
6879 parameter to opt_scalar_int_mode.
6881 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6882 Alan Hayward <alan.hayward@arm.com>
6883 David Sherwood <david.sherwood@arm.com>
6885 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
6886 to a scalar_int_mode instead of a machine_mode.
6887 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
6888 (get_best_mode): Return a boolean and use a pointer argument to store
6889 the selected mode. Replace the limit mode parameter with a bit limit.
6890 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
6891 for the values returned by bit_field_mode_iterator::next_mode.
6892 (store_bit_field): Update call to get_best_mode.
6893 (store_fixed_bit_field): Likewise.
6894 (extract_fixed_bit_field): Likewise.
6895 * expr.c (optimize_bitfield_assignment_op): Likewise.
6896 * fold-const.c (optimize_bit_field_compare): Likewise.
6897 (fold_truth_andor_1): Likewise.
6898 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
6899 Update for new type of m_mode.
6900 (get_best_mode): As above.
6902 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6903 Alan Hayward <alan.hayward@arm.com>
6904 David Sherwood <david.sherwood@arm.com>
6906 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
6907 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
6908 (store_bit_field): Check is_a <scalar_int_mode> before calling
6909 strict_volatile_bitfield_p.
6910 (extract_bit_field): Likewise.
6912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6913 Alan Hayward <alan.hayward@arm.com>
6914 David Sherwood <david.sherwood@arm.com>
6916 * target.def (cstore_mode): Return a scalar_int_mode.
6917 * doc/tm.texi: Regenerate.
6918 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
6919 * targhooks.h (default_cstore_mode): Likewise.
6920 * targhooks.c (default_cstore_mode): Likewise, using a forced
6922 * expmed.c (emit_cstore): Expect the target of the cstore to be
6925 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6926 Alan Hayward <alan.hayward@arm.com>
6927 David Sherwood <david.sherwood@arm.com>
6929 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
6931 (niter_desc): Likewise mode.
6932 (iv_analyze): Add a mode parameter.
6934 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
6935 and change its type to scalar_int_mode.
6936 * loop-iv.c: Update commentary at head of file.
6937 (iv_constant): Pass the mode paraeter before the rtx it describes
6938 and change its type to scalar_int_mode. Remove VOIDmode handling.
6939 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
6940 (iv_extend): Likewise.
6941 (shorten_into_mode): Likewise.
6942 (iv_add): Use scalar_int_mode.
6943 (iv_mult): Likewise.
6944 (iv_shift): Likewise.
6945 (canonicalize_iv_subregs): Likewise.
6946 (get_biv_step_1): Pass the outer_mode parameter before the rtx
6947 it describes and change its mode to scalar_int_mode. Also change
6948 the type of the returned inner_mode to scalar_int_mode.
6949 (get_biv_step): Likewise, turning outer_mode from a pointer
6950 into a direct parameter. Update call to get_biv_step_1.
6951 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
6952 iv_constant and get_biv_step.
6953 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
6954 and change its type to scalar_int_mode. Don't initialise iv->mode
6955 to VOIDmode and remove later checks for its still being VOIDmode.
6956 Update calls to iv_analyze_op and iv_analyze_expr. Check
6957 is_a <scalar_int_mode> when changing the mode under consideration.
6958 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
6959 Update call to iv_analyze_expr.
6960 (iv_analyze_op): Add a mode parameter. Reject subregs whose
6961 inner register is not also a scalar_int_mode. Update call to
6963 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
6964 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
6965 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
6966 separate mode class checks. Update calls to iv_analyze. Remove
6967 fix-up of VOIDmodes after iv_analyze_biv.
6968 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
6969 don't have a scalar_int_mode. Update call to biv_p.
6971 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6972 Alan Hayward <alan.hayward@arm.com>
6973 David Sherwood <david.sherwood@arm.com>
6975 * cfgexpand.c (convert_debug_memory_address): Use
6976 as_a <scalar_int_mode>.
6977 * combine.c (expand_compound_operation): Likewise.
6978 (make_extraction): Likewise.
6979 (change_zero_ext): Likewise.
6980 (simplify_comparison): Likewise.
6981 * cse.c (cse_insn): Likewise.
6982 * dwarf2out.c (minmax_loc_descriptor): Likewise.
6983 (mem_loc_descriptor): Likewise.
6984 (loc_descriptor): Likewise.
6985 * expmed.c (init_expmed_one_mode): Likewise.
6986 (synth_mult): Likewise.
6987 (emit_store_flag_1): Likewise.
6988 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
6989 of a comparison with size.
6990 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
6991 (reduce_to_bit_field_precision): Likewise.
6992 * function.c (expand_function_end): Likewise.
6993 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
6994 * loop-doloop.c (doloop_modify): Likewise.
6995 * optabs.c (expand_binop): Likewise.
6996 (expand_unop): Likewise.
6997 (expand_copysign_absneg): Likewise.
6998 (prepare_cmp_insn): Likewise.
6999 (maybe_legitimize_operand): Likewise.
7000 * recog.c (const_scalar_int_operand): Likewise.
7001 * rtlanal.c (get_address_mode): Likewise.
7002 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7003 (simplify_cond_clz_ctz): Likewise.
7004 * tree-nested.c (get_nl_goto_field): Likewise.
7005 * tree.c (build_vector_type_for_mode): Likewise.
7006 * var-tracking.c (use_narrower_mode): Likewise.
7008 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7009 Alan Hayward <alan.hayward@arm.com>
7010 David Sherwood <david.sherwood@arm.com>
7012 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7013 * builtins.c (expand_builtin_signbit): Use it.
7014 * cfgexpand.c (expand_debug_expr): Likewise.
7015 * dojump.c (do_jump): Likewise.
7016 (do_compare_and_jump): Likewise.
7017 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7018 * expmed.c (make_tree): Likewise.
7019 * expr.c (expand_expr_real_2): Likewise.
7020 (expand_expr_real_1): Likewise.
7021 (try_casesi): Likewise.
7022 * fold-const-call.c (fold_const_call_ss): Likewise.
7023 * fold-const.c (unextend): Likewise.
7024 (extract_muldiv_1): Likewise.
7025 (fold_single_bit_test): Likewise.
7026 (native_encode_int): Likewise.
7027 (native_encode_string): Likewise.
7028 (native_interpret_int): Likewise.
7029 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7030 * internal-fn.c (expand_addsub_overflow): Likewise.
7031 (expand_neg_overflow): Likewise.
7032 (expand_mul_overflow): Likewise.
7033 (expand_arith_overflow): Likewise.
7034 * match.pd: Likewise.
7035 * stor-layout.c (layout_type): Likewise.
7036 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7037 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7038 * tree-ssanames.c (get_range_info): Likewise.
7039 * tree-switch-conversion.c (array_value_type) Likewise.
7040 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7041 (vect_recog_divmod_pattern): Likewise.
7042 (vect_recog_mixed_size_cond_pattern): Likewise.
7043 * tree-vrp.c (extract_range_basic): Likewise.
7044 (simplify_float_conversion_using_ranges): Likewise.
7045 * tree.c (int_fits_type_p): Likewise.
7046 * ubsan.c (instrument_bool_enum_load): Likewise.
7047 * varasm.c (mergeable_string_section): Likewise.
7048 (narrowing_initializer_constant_valid_p): Likewise.
7049 (output_constant): Likewise.
7051 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7052 Alan Hayward <alan.hayward@arm.com>
7053 David Sherwood <david.sherwood@arm.com>
7055 * machmode.h (NARROWEST_INT_MODE): New macro.
7056 * expr.c (alignment_for_piecewise_move): Use it instead of
7057 GET_CLASS_NARROWEST_MODE (MODE_INT).
7058 (push_block): Likewise.
7059 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7061 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7063 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7064 Alan Hayward <alan.hayward@arm.com>
7065 David Sherwood <david.sherwood@arm.com>
7067 * postreload.c (move2add_valid_value_p): Change the type of the
7068 mode parameter to scalar_int_mode.
7069 (move2add_use_add2_insn): Add a mode parameter and use it instead
7071 (move2add_use_add3_insn): Likewise.
7072 (reload_cse_move2add): Update accordingly.
7074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7075 Alan Hayward <alan.hayward@arm.com>
7076 David Sherwood <david.sherwood@arm.com>
7078 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
7080 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
7081 * optabs.c (expand_unop): Likewise.
7083 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7084 Alan Hayward <alan.hayward@arm.com>
7085 David Sherwood <david.sherwood@arm.com>
7087 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
7088 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
7089 (popcount_loc_descriptor): Likewise.
7090 (bswap_loc_descriptor): Likewise.
7091 (rotate_loc_descriptor): Likewise.
7092 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
7093 calling the functions above.
7095 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7096 Alan Hayward <alan.hayward@arm.com>
7097 David Sherwood <david.sherwood@arm.com>
7099 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
7101 (try_combine): Likewise.
7102 (simplify_if_then_else): Likewise.
7103 * cse.c (cse_insn): Likewise.
7104 * dwarf2out.c (mem_loc_descriptor): Likewise.
7105 * emit-rtl.c (gen_lowpart_common): Likewise.
7106 * simplify-rtx.c (simplify_truncation): Likewise.
7107 (simplify_binary_operation_1): Likewise.
7108 (simplify_const_relational_operation): Likewise.
7109 (simplify_ternary_operation): Likewise.
7110 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
7112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7113 Alan Hayward <alan.hayward@arm.com>
7114 David Sherwood <david.sherwood@arm.com>
7116 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
7117 * reload.c (push_reload): Likewise.
7118 (find_reloads): Likewise.
7120 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7121 Alan Hayward <alan.hayward@arm.com>
7122 David Sherwood <david.sherwood@arm.com>
7124 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
7125 (make_compound_operation_int): Likewise.
7126 (change_zero_ext): Likewise.
7127 * expr.c (convert_move): Likewise.
7128 (convert_modes): Likewise.
7129 * fwprop.c (forward_propagate_subreg): Likewise.
7130 * loop-iv.c (get_biv_step_1): Likewise.
7131 * optabs.c (widen_operand): Likewise.
7132 * postreload.c (move2add_valid_value_p): Likewise.
7133 * recog.c (simplify_while_replacing): Likewise.
7134 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7135 (simplify_binary_operation_1): Likewise. Remove redundant
7136 mode equality check.
7138 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7139 Alan Hayward <alan.hayward@arm.com>
7140 David Sherwood <david.sherwood@arm.com>
7142 * combine.c (combine_simplify_rtx): Add checks for
7143 is_a <scalar_int_mode>.
7144 (simplify_if_then_else): Likewise.
7145 (make_field_assignment): Likewise.
7146 (simplify_comparison): Likewise.
7147 * ifcvt.c (noce_try_bitop): Likewise.
7148 * loop-invariant.c (canonicalize_address_mult): Likewise.
7149 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7151 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7152 Alan Hayward <alan.hayward@arm.com>
7153 David Sherwood <david.sherwood@arm.com>
7155 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
7156 is_a <scalar_int_mode> instead of != BLKmode.
7158 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7159 Alan Hayward <alan.hayward@arm.com>
7160 David Sherwood <david.sherwood@arm.com>
7162 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
7163 instead of != VOIDmode.
7164 * combine.c (if_then_else_cond): Likewise.
7165 (change_zero_ext): Likewise.
7166 * dwarf2out.c (mem_loc_descriptor): Likewise.
7167 (loc_descriptor): Likewise.
7168 * rtlanal.c (canonicalize_condition): Likewise.
7169 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
7171 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7172 Alan Hayward <alan.hayward@arm.com>
7173 David Sherwood <david.sherwood@arm.com>
7175 * simplify-rtx.c (simplify_binary_operation_1): Use
7176 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
7178 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7179 Alan Hayward <alan.hayward@arm.com>
7180 David Sherwood <david.sherwood@arm.com>
7182 * wide-int.h (int_traits<unsigned char>) New class.
7183 (int_traits<unsigned short>) Likewise.
7184 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
7185 Use GET_MODE_UNIT_PRECISION and remove redundant test for
7187 * combine.c (set_nonzero_bits_and_sign_copies): Use
7188 is_a <scalar_int_mode>.
7189 (find_split_point): Likewise.
7190 (combine_simplify_rtx): Likewise.
7191 (simplify_logical): Likewise.
7192 (expand_compound_operation): Likewise.
7193 (expand_field_assignment): Likewise.
7194 (make_compound_operation): Likewise.
7195 (extended_count): Likewise.
7196 (change_zero_ext): Likewise.
7197 (simplify_comparison): Likewise.
7198 * dwarf2out.c (scompare_loc_descriptor): Likewise.
7199 (ucompare_loc_descriptor): Likewise.
7200 (minmax_loc_descriptor): Likewise.
7201 (mem_loc_descriptor): Likewise.
7202 (loc_descriptor): Likewise.
7203 * expmed.c (init_expmed_one_mode): Likewise.
7204 * lra-constraints.c (lra_constraint_offset): Likewise.
7205 * optabs.c (prepare_libcall_arg): Likewise.
7206 * postreload.c (move2add_note_store): Likewise.
7207 * reload.c (operands_match_p): Likewise.
7208 * rtl.h (load_extend_op): Likewise.
7209 * rtlhooks.c (gen_lowpart_general): Likewise.
7210 * simplify-rtx.c (simplify_truncation): Likewise.
7211 (simplify_unary_operation_1): Likewise.
7212 (simplify_binary_operation_1): Likewise.
7213 (simplify_const_binary_operation): Likewise.
7214 (simplify_const_relational_operation): Likewise.
7215 (simplify_subreg): Likewise.
7216 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7217 * var-tracking.c (adjust_mems): Likewise.
7218 (prepare_call_arguments): Likewise.
7220 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7221 Alan Hayward <alan.hayward@arm.com>
7222 David Sherwood <david.sherwood@arm.com>
7224 * machmode.h (is_int_mode): New fuction.
7225 * combine.c (find_split_point): Use it.
7226 (combine_simplify_rtx): Likewise.
7227 (simplify_if_then_else): Likewise.
7228 (simplify_set): Likewise.
7229 (simplify_shift_const_1): Likewise.
7230 (simplify_comparison): Likewise.
7231 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
7232 * cse.c (notreg_cost): Likewise.
7233 (cse_insn): Likewise.
7234 * cselib.c (cselib_lookup_1): Likewise.
7235 * dojump.c (do_jump_1): Likewise.
7236 (do_compare_rtx_and_jump): Likewise.
7237 * dse.c (get_call_args): Likewise.
7238 * dwarf2out.c (rtl_for_decl_init): Likewise.
7239 (native_encode_initializer): Likewise.
7240 * expmed.c (emit_store_flag_1): Likewise.
7241 (emit_store_flag): Likewise.
7242 * expr.c (convert_modes): Likewise.
7243 (store_field): Likewise.
7244 (expand_expr_real_1): Likewise.
7245 * fold-const.c (fold_read_from_constant_string): Likewise.
7246 * gimple-ssa-sprintf.c (get_format_string): Likewise.
7247 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
7248 * optabs.c (expand_binop): Likewise.
7249 (expand_unop): Likewise.
7250 (expand_abs_nojump): Likewise.
7251 (expand_one_cmpl_abs_nojump): Likewise.
7252 * simplify-rtx.c (mode_signbit_p): Likewise.
7253 (val_signbit_p): Likewise.
7254 (val_signbit_known_set_p): Likewise.
7255 (val_signbit_known_clear_p): Likewise.
7256 (simplify_relational_operation_1): Likewise.
7257 * tree.c (vector_type_mode): Likewise.
7259 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7260 Alan Hayward <alan.hayward@arm.com>
7261 David Sherwood <david.sherwood@arm.com>
7263 * machmode.h (smallest_mode_for_size): Fix formatting.
7264 (smallest_int_mode_for_size): New function.
7265 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
7266 instead of smallest_mode_for_size.
7267 * combine.c (make_extraction): Likewise.
7268 * config/arc/arc.c (arc_expand_movmem): Likewise.
7269 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
7270 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7271 * config/s390/s390.c (s390_expand_insv): Likewise.
7272 * config/sparc/sparc.c (assign_int_registers): Likewise.
7273 * config/spu/spu.c (spu_function_value): Likewise.
7274 (spu_function_arg): Likewise.
7275 * coverage.c (get_gcov_type): Likewise.
7276 (get_gcov_unsigned_t): Likewise.
7277 * dse.c (find_shift_sequence): Likewise.
7278 * expmed.c (store_bit_field_1): Likewise.
7279 * expr.c (convert_move): Likewise.
7280 (store_field): Likewise.
7281 * internal-fn.c (expand_arith_overflow): Likewise.
7282 * optabs-query.c (get_best_extraction_insn): Likewise.
7283 * optabs.c (expand_twoval_binop_libfunc): Likewise.
7284 * stor-layout.c (layout_type): Likewise.
7285 (initialize_sizetypes): Likewise.
7286 * targhooks.c (default_get_mask_mode): Likewise.
7287 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
7289 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7290 Alan Hayward <alan.hayward@arm.com>
7291 David Sherwood <david.sherwood@arm.com>
7293 * machmode.h (opt_mode::else_blk): New function.
7294 (int_mode_for_mode): Declare.
7295 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
7296 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
7298 * cfgexpand.c (expand_debug_expr): Likewise.
7299 * combine.c (gen_lowpart_or_truncate): Likewise.
7300 (gen_lowpart_for_combine): Likewise.
7301 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
7302 * config/avr/avr.c (avr_to_int_mode): Likewise.
7303 (avr_out_plus_1): Likewise.
7304 (avr_out_plus): Likewise.
7305 (avr_out_round): Likewise.
7306 * config/i386/i386.c (ix86_split_to_parts): Likewise.
7307 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7308 (s390_expand_vcond): Likewise.
7309 * config/spu/spu.c (spu_split_immediate): Likewise.
7310 (spu_expand_mov): Likewise.
7311 * dse.c (get_stored_val): Likewise.
7312 * expmed.c (store_bit_field_1): Likewise.
7313 (convert_extracted_bit_field): Use int_mode_for_mode instead of
7315 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
7316 (extract_low_bits): Likewise.
7317 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
7318 handling rather than repeating the check.
7319 (emit_group_store): Likewise.
7320 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
7321 * optabs.c (expand_absneg_bit): Likewise.
7322 (expand_copysign_absneg): Likewise.
7323 (expand_copysign_bit): Likewise.
7324 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
7325 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
7326 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
7327 * var-tracking.c (prepare_call_arguments): Likewise.
7328 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
7329 int_mode_for_mode instead of mode_for_size.
7330 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7333 Alan Hayward <alan.hayward@arm.com>
7334 David Sherwood <david.sherwood@arm.com>
7336 * machmode.h (int_mode_for_size): New function.
7337 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
7338 instead of mode_for_size.
7339 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
7341 * combine.c (expand_field_assignment): Use int_mode_for_size
7342 instead of mode_for_size.
7343 (make_extraction): Likewise.
7344 (simplify_shift_const_1): Likewise.
7345 (simplify_comparison): Likewise.
7346 * dojump.c (do_jump): Likewise.
7347 * dwarf2out.c (mem_loc_descriptor): Likewise.
7348 * emit-rtl.c (init_derived_machine_modes): Likewise.
7349 * expmed.c (flip_storage_order): Likewise.
7350 (convert_extracted_bit_field): Likewise.
7351 * expr.c (copy_blkmode_from_reg): Likewise.
7352 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
7353 * internal-fn.c (expand_mul_overflow): Likewise.
7354 * lower-subreg.c (simple_move): Likewise.
7355 * optabs-libfuncs.c (init_optabs): Likewise.
7356 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7357 * tree.c (vector_type_mode): Likewise.
7358 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
7359 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
7360 * tree-vect-generic.c (expand_vector_parallel): Likewise.
7361 * tree-vect-stmts.c (vectorizable_load): Likewise.
7362 (vectorizable_store): Likewise.
7364 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7365 Alan Hayward <alan.hayward@arm.com>
7366 David Sherwood <david.sherwood@arm.com>
7368 * coretypes.h (pod_mode): New type.
7369 (scalar_int_mode_pod): New typedef.
7370 * machmode.h (pod_mode): New class.
7371 (int_n_data_t::m): Change type to scalar_int_mode_pod.
7372 * genmodes.c (emit_mode_int_n): Update accordingly.
7373 * lower-subreg.h (target_lower_subreg): Change type to
7374 scalar_int_mode_pod.
7375 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
7376 scalar_int_mode_pod.
7378 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7379 Alan Hayward <alan.hayward@arm.com>
7380 David Sherwood <david.sherwood@arm.com>
7382 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
7383 machine_mode to scalar_int_mode.
7384 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
7385 (rs6000_option_override_internal): Remove cast to int.
7386 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
7387 machine_mode to scalar_int_mode.
7388 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
7389 (rs6000_option_override_internal): Remove cast to int.
7390 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
7391 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
7393 * config/s390/s390.c (s390_expand_builtin): Likewise.
7394 * coretypes.h (scalar_int_mode): New type.
7395 (opt_scalar_int_mode): New typedef.
7396 * machmode.h (scalar_int_mode): New class.
7397 (scalar_int_mode::includes_p): New function.
7398 (byte_mode): Change type to scalar_int_mode.
7399 (word_mode): Likewise.
7400 (ptr_mode): Likewise.
7401 * emit-rtl.c (byte_mode): Likewise.
7402 (word_mode): Likewise.
7403 (ptr_mode): Likewise.
7404 (init_derived_machine_modes): Update accordingly.
7405 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
7406 and MODE_PARTIAL_INT.
7407 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
7408 opt_scalar_int_mode.
7410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7411 Alan Hayward <alan.hayward@arm.com>
7412 David Sherwood <david.sherwood@arm.com>
7414 * target.def (libgcc_floating_mode_supported_p): Take a
7416 * doc/tm.texi: Regenerate.
7417 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
7419 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7420 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
7423 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7424 Alan Hayward <alan.hayward@arm.com>
7425 David Sherwood <david.sherwood@arm.com>
7427 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
7428 * doc/tm.texi: Regenerate.
7429 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
7430 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
7431 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
7432 * targhooks.h (default_floatn_mode): Likewise.
7433 * targhooks.c (default_floatn_mode): Likewise.
7434 * tree.c (build_common_tree_nodes): Update accordingly.
7436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7437 Alan Hayward <alan.hayward@arm.com>
7438 David Sherwood <david.sherwood@arm.com>
7440 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
7441 (mode_iterator::iterate_p): Likewise.
7442 (mode_iterator::get_wider): Likewise.
7443 * expr.c (init_expr_target): Use opt_scalar_float_mode.
7445 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7446 Alan Hayward <alan.hayward@arm.com>
7447 David Sherwood <david.sherwood@arm.com>
7449 * coretypes.h (opt_scalar_float_mode): New typedef.
7450 * machmode.h (float_mode_for_size): New function.
7451 * emit-rtl.c (double_mode): Delete.
7452 (init_emit_once): Use float_mode_for_size.
7453 * stor-layout.c (layout_type): Likewise.
7454 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
7456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7457 Alan Hayward <alan.hayward@arm.com>
7458 David Sherwood <david.sherwood@arm.com>
7460 * output.h (assemble_real): Take a scalar_float_mode.
7461 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
7462 * config/arm/arm.md (consttable_4): Likewise.
7463 (consttable_8): Likewise.
7464 (consttable_16): Likewise.
7465 * config/mips/mips.md (consttable_float): Likewise.
7466 * config/s390/s390.c (s390_output_pool_entry): Likewise.
7467 * varasm.c (assemble_real): Take a scalar_float_mode.
7468 (output_constant_pool_2): Update accordingly.
7469 (output_constant): Likewise.
7471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7472 Alan Hayward <alan.hayward@arm.com>
7473 David Sherwood <david.sherwood@arm.com>
7475 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
7476 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
7477 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
7478 (native_encode_real): Likewise.
7479 (native_interpret_real): Likewise.
7480 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
7481 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7483 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7484 Alan Hayward <alan.hayward@arm.com>
7485 David Sherwood <david.sherwood@arm.com>
7487 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
7488 <scalar_float_mode>. Simplify.
7489 (gen_extend_conv_libfunc): Likewise.
7491 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7492 Alan Hayward <alan.hayward@arm.com>
7493 David Sherwood <david.sherwood@arm.com>
7495 * coretypes.h (scalar_float_mode): New type.
7496 * machmode.h (mode_traits::from_int): Use machine_mode if
7497 USE_ENUM_MODES is defined.
7498 (is_a): New function.
7500 (dyn_cast): Likewise.
7501 (scalar_float_mode): New class.
7502 (scalar_float_mode::includes_p): New function.
7503 (is_float_mode): Likewise.
7504 * gdbhooks.py (MachineModePrinter): New class.
7505 (build_pretty_printer): Use it for scalar_float_mode.
7506 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
7507 (format_helper::format_helper): Turn into a template.
7508 * genmodes.c (get_mode_class): New function.
7509 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
7510 or machine_mode if none.
7511 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
7512 as_a <scalar_float_mode>.
7513 * dwarf2out.c (mem_loc_descriptor): Likewise.
7514 (insert_float): Likewise.
7515 (add_const_value_attribute): Likewise.
7516 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7517 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
7518 (expand_unop): Update accordingly.
7519 (expand_abs_nojump): Likewise.
7520 (expand_copysign_absneg): Take a scalar_float_mode.
7521 (expand_copysign_bit): Likewise.
7522 (expand_copysign): Update accordingly.
7524 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7525 Alan Hayward <alan.hayward@arm.com>
7526 David Sherwood <david.sherwood@arm.com>
7528 * coretypes.h (opt_mode): New class.
7529 * machmode.h (opt_mode): Likewise.
7530 (opt_mode::else_void): New function.
7531 (opt_mode::require): Likewise.
7532 (opt_mode::exists): Likewise.
7533 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
7534 (GET_MODE_2XWIDER_MODE): Likewise.
7535 (mode_iterator::get_wider): Update accordingly.
7536 (mode_iterator::get_2xwider): Likewise.
7537 (mode_iterator::get_known_wider): Likewise, turning into a template.
7538 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
7539 forcing a wider mode to exist.
7540 * config/cr16/cr16.h (LONG_REG_P): Likewise.
7541 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7542 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
7543 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
7544 * lower-subreg.c (init_lower_subreg): Likewise.
7545 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
7546 on the final iteration.
7547 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
7548 a wider mode exists before asking for a move pattern.
7549 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
7550 forcing a wider mode to exist.
7551 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
7552 returning false if no such mode exists.
7553 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
7554 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
7555 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
7556 Avoid checking for a MODE_INT if we already know the mode is not a
7558 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
7559 forcing a wider mode to exist.
7560 (expmed_mult_highpart_optab): Likewise.
7561 (expmed_mult_highpart): Likewise.
7562 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
7564 * lto-streamer-in.c (lto_input_mode_table): Likewise.
7565 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
7566 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7567 * internal-fn.c (expand_mul_overflow): Update use of
7568 GET_MODE_2XWIDER_MODE.
7569 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7570 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
7571 GET_MODE_WIDER_MODE.
7572 (convert_plusminus_to_widen): Likewise.
7573 * tree-switch-conversion.c (array_value_type): Likewise.
7574 * var-tracking.c (emit_note_insn_var_location): Likewise.
7575 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7576 Return false inside rather than outside the loop if no wider mode
7578 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
7579 and GET_MODE_2XWIDER_MODE
7580 (can_compare_p): Use else_void.
7581 * gdbhooks.py (OptMachineModePrinter): New class.
7582 (build_pretty_printer): Use it for opt_mode.
7584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7585 Alan Hayward <alan.hayward@arm.com>
7586 David Sherwood <david.sherwood@arm.com>
7588 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
7589 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
7591 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7592 Alan Hayward <alan.hayward@arm.com>
7593 David Sherwood <david.sherwood@arm.com>
7595 * machmode.h (mode_traits): New structure.
7596 (get_narrowest_mode): New function.
7597 (mode_iterator::start): Likewise.
7598 (mode_iterator::iterate_p): Likewise.
7599 (mode_iterator::get_wider): Likewise.
7600 (mode_iterator::get_known_wider): Likewise.
7601 (mode_iterator::get_2xwider): Likewise.
7602 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
7603 (FOR_EACH_MODE): Likewise.
7604 (FOR_EACH_MODE_FROM): Likewise.
7605 (FOR_EACH_MODE_UNTIL): Likewise.
7606 (FOR_EACH_WIDER_MODE): Likewise.
7607 (FOR_EACH_2XWIDER_MODE): Likewise.
7608 * builtins.c (expand_builtin_strlen): Use new mode iterators.
7609 * combine.c (simplify_comparison): Likewise
7610 * config/i386/i386.c (type_natural_mode): Likewise.
7611 * cse.c (cse_insn): Likewise.
7612 * dse.c (find_shift_sequence): Likewise.
7613 * emit-rtl.c (init_derived_machine_modes): Likewise.
7614 (init_emit_once): Likewise.
7615 * explow.c (hard_function_value): Likewise.
7616 * expmed.c (extract_fixed_bit_field_1): Likewise.
7617 (extract_bit_field_1): Likewise.
7618 (expand_divmod): Likewise.
7619 (emit_store_flag_1): Likewise.
7620 * expr.c (init_expr_target): Likewise.
7621 (convert_move): Likewise.
7622 (alignment_for_piecewise_move): Likewise.
7623 (widest_int_mode_for_size): Likewise.
7624 (emit_block_move_via_movmem): Likewise.
7625 (copy_blkmode_to_reg): Likewise.
7626 (set_storage_via_setmem): Likewise.
7627 (compress_float_constant): Likewise.
7628 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7629 * optabs-query.c (get_best_extraction_insn): Likewise.
7630 * optabs.c (expand_binop): Likewise.
7631 (expand_twoval_unop): Likewise.
7632 (expand_twoval_binop): Likewise.
7633 (widen_leading): Likewise.
7634 (widen_bswap): Likewise.
7635 (expand_parity): Likewise.
7636 (expand_unop): Likewise.
7637 (prepare_cmp_insn): Likewise.
7638 (prepare_float_lib_cmp): Likewise.
7639 (expand_float): Likewise.
7640 (expand_fix): Likewise.
7641 (expand_sfix_optab): Likewise.
7642 * postreload.c (move2add_use_add2_insn): Likewise.
7643 * reg-stack.c (reg_to_stack): Likewise.
7644 * reginfo.c (choose_hard_reg_mode): Likewise.
7645 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7646 * stor-layout.c (mode_for_size): Likewise.
7647 (smallest_mode_for_size): Likewise.
7648 (mode_for_vector): Likewise.
7649 (finish_bitfield_representative): Likewise.
7650 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
7651 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
7652 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7653 * var-tracking.c (prepare_call_arguments): Likewise.
7655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7656 Alan Hayward <alan.hayward@arm.com>
7657 David Sherwood <david.sherwood@arm.com>
7659 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
7660 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
7661 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
7662 * machmode.h (mode_size): Move earlier in file.
7663 (mode_precision): Likewise.
7664 (mode_inner): Likewise.
7665 (mode_nunits): Likewise.
7666 (mode_unit_size): Likewise.
7667 (unit_unit_precision): Likewise.
7668 (mode_wider): Likewise.
7669 (mode_2xwider): Likewise.
7670 (machine_mode): New class.
7671 (mode_to_bytes): New function.
7672 (mode_to_bits): Likewise.
7673 (mode_to_precision): Likewise.
7674 (mode_to_inner): Likewise.
7675 (mode_to_unit_size): Likewise.
7676 (mode_to_unit_precision): Likewise.
7677 (mode_to_nunits): Likewise.
7678 (GET_MODE_SIZE): Use mode_to_bytes.
7679 (GET_MODE_BITSIZE): Use mode_to_bits.
7680 (GET_MODE_PRECISION): Use mode_to_precision.
7681 (GET_MODE_INNER): Use mode_to_inner.
7682 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
7683 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
7684 (GET_MODE_NUNITS): Use mode_to_nunits.
7685 * system.h (ALWAYS_INLINE): New macro.
7686 * config/powerpcspe/powerpcspe-c.c
7687 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
7688 int for arg1_mode and arg2_mode.
7690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7691 Alan Hayward <alan.hayward@arm.com>
7692 David Sherwood <david.sherwood@arm.com>
7694 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7695 Prefix mode names with E_ in case statements.
7696 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7697 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
7698 (aarch64_split_simd_move): Likewise.
7699 (aarch64_gen_storewb_pair): Likewise.
7700 (aarch64_gen_loadwb_pair): Likewise.
7701 (aarch64_gen_store_pair): Likewise.
7702 (aarch64_gen_load_pair): Likewise.
7703 (aarch64_get_condition_code_1): Likewise.
7704 (aarch64_constant_pool_reload_icode): Likewise.
7705 (get_rsqrte_type): Likewise.
7706 (get_rsqrts_type): Likewise.
7707 (get_recpe_type): Likewise.
7708 (get_recps_type): Likewise.
7709 (aarch64_gimplify_va_arg_expr): Likewise.
7710 (aarch64_simd_container_mode): Likewise.
7711 (aarch64_emit_load_exclusive): Likewise.
7712 (aarch64_emit_store_exclusive): Likewise.
7713 (aarch64_expand_compare_and_swap): Likewise.
7714 (aarch64_gen_atomic_cas): Likewise.
7715 (aarch64_emit_bic): Likewise.
7716 (aarch64_emit_atomic_swap): Likewise.
7717 (aarch64_emit_atomic_load_op): Likewise.
7718 (aarch64_evpc_trn): Likewise.
7719 (aarch64_evpc_uzp): Likewise.
7720 (aarch64_evpc_zip): Likewise.
7721 (aarch64_evpc_ext): Likewise.
7722 (aarch64_evpc_rev): Likewise.
7723 (aarch64_evpc_dup): Likewise.
7724 (aarch64_gen_ccmp_first): Likewise.
7725 (aarch64_gen_ccmp_next): Likewise.
7726 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7727 (alpha_emit_xfloating_libcall): Likewise.
7728 (emit_insxl): Likewise.
7729 (alpha_arg_type): Likewise.
7730 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
7731 (arc_preferred_simd_mode): Likewise.
7732 (arc_secondary_reload): Likewise.
7733 (get_arc_condition_code): Likewise.
7734 (arc_print_operand): Likewise.
7735 (arc_legitimate_constant_p): Likewise.
7736 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7737 * config/arc/arc.md (casesi_load): Likewise.
7738 (casesi_compact_jump): Likewise.
7739 * config/arc/predicates.md (proper_comparison_operator): Likewise.
7740 (cc_use_register): Likewise.
7741 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7742 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7743 (arm_init_iwmmxt_builtins): Likewise.
7744 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
7745 (neon_expand_vector_init): Likewise.
7746 (arm_attr_length_move_neon): Likewise.
7747 (maybe_get_arm_condition_code): Likewise.
7748 (arm_emit_vector_const): Likewise.
7749 (arm_preferred_simd_mode): Likewise.
7750 (arm_output_iwmmxt_tinsr): Likewise.
7751 (thumb1_output_casesi): Likewise.
7752 (thumb2_output_casesi): Likewise.
7753 (arm_emit_load_exclusive): Likewise.
7754 (arm_emit_store_exclusive): Likewise.
7755 (arm_expand_compare_and_swap): Likewise.
7756 (arm_evpc_neon_vuzp): Likewise.
7757 (arm_evpc_neon_vzip): Likewise.
7758 (arm_evpc_neon_vrev): Likewise.
7759 (arm_evpc_neon_vtrn): Likewise.
7760 (arm_evpc_neon_vext): Likewise.
7761 (arm_validize_comparison): Likewise.
7762 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
7763 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
7764 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
7765 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
7766 (c6x_preferred_simd_mode): Likewise.
7767 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
7768 (epiphany_rtx_costs): Likewise.
7769 * config/epiphany/predicates.md (proper_comparison_operator):
7771 * config/frv/frv.c (condexec_memory_operand): Likewise.
7772 (frv_emit_move): Likewise.
7773 (output_move_single): Likewise.
7774 (output_condmove_single): Likewise.
7775 (frv_hard_regno_mode_ok): Likewise.
7776 (frv_matching_accg_mode): Likewise.
7777 * config/h8300/h8300.c (split_adds_subs): Likewise.
7778 (h8300_rtx_costs): Likewise.
7779 (h8300_print_operand): Likewise.
7780 (compute_mov_length): Likewise.
7781 (output_logical_op): Likewise.
7782 (compute_logical_op_length): Likewise.
7783 (compute_logical_op_cc): Likewise.
7784 (h8300_shift_needs_scratch_p): Likewise.
7785 (output_a_shift): Likewise.
7786 (compute_a_shift_length): Likewise.
7787 (compute_a_shift_cc): Likewise.
7788 (expand_a_rotate): Likewise.
7789 (output_a_rotate): Likewise.
7790 * config/i386/i386.c (classify_argument): Likewise.
7791 (function_arg_advance_32): Likewise.
7792 (function_arg_32): Likewise.
7793 (function_arg_64): Likewise.
7794 (function_value_64): Likewise.
7795 (ix86_gimplify_va_arg): Likewise.
7796 (ix86_legitimate_constant_p): Likewise.
7797 (put_condition_code): Likewise.
7798 (split_double_mode): Likewise.
7799 (ix86_avx256_split_vector_move_misalign): Likewise.
7800 (ix86_expand_vector_logical_operator): Likewise.
7801 (ix86_split_idivmod): Likewise.
7802 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
7803 (ix86_build_const_vector): Likewise.
7804 (ix86_build_signbit_mask): Likewise.
7805 (ix86_match_ccmode): Likewise.
7806 (ix86_cc_modes_compatible): Likewise.
7807 (ix86_expand_branch): Likewise.
7808 (ix86_expand_sse_cmp): Likewise.
7809 (ix86_expand_sse_movcc): Likewise.
7810 (ix86_expand_int_sse_cmp): Likewise.
7811 (ix86_expand_vec_perm_vpermi2): Likewise.
7812 (ix86_expand_vec_perm): Likewise.
7813 (ix86_expand_sse_unpack): Likewise.
7814 (ix86_expand_int_addcc): Likewise.
7815 (ix86_split_to_parts): Likewise.
7816 (ix86_vectorize_builtin_gather): Likewise.
7817 (ix86_vectorize_builtin_scatter): Likewise.
7818 (avx_vpermilp_parallel): Likewise.
7819 (inline_memory_move_cost): Likewise.
7820 (ix86_tieable_integer_mode_p): Likewise.
7821 (x86_maybe_negate_const_int): Likewise.
7822 (ix86_expand_vector_init_duplicate): Likewise.
7823 (ix86_expand_vector_init_one_nonzero): Likewise.
7824 (ix86_expand_vector_init_one_var): Likewise.
7825 (ix86_expand_vector_init_concat): Likewise.
7826 (ix86_expand_vector_init_interleave): Likewise.
7827 (ix86_expand_vector_init_general): Likewise.
7828 (ix86_expand_vector_set): Likewise.
7829 (ix86_expand_vector_extract): Likewise.
7830 (emit_reduc_half): Likewise.
7831 (ix86_emit_i387_round): Likewise.
7832 (ix86_mangle_type): Likewise.
7833 (ix86_expand_round_sse4): Likewise.
7834 (expand_vec_perm_blend): Likewise.
7835 (canonicalize_vector_int_perm): Likewise.
7836 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
7837 (expand_vec_perm_1): Likewise.
7838 (expand_vec_perm_interleave3): Likewise.
7839 (expand_vec_perm_even_odd_pack): Likewise.
7840 (expand_vec_perm_even_odd_1): Likewise.
7841 (expand_vec_perm_broadcast_1): Likewise.
7842 (ix86_vectorize_vec_perm_const_ok): Likewise.
7843 (ix86_expand_vecop_qihi): Likewise.
7844 (ix86_expand_mul_widen_hilo): Likewise.
7845 (ix86_expand_sse2_abs): Likewise.
7846 (ix86_expand_pextr): Likewise.
7847 (ix86_expand_pinsr): Likewise.
7848 (ix86_preferred_simd_mode): Likewise.
7849 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
7850 * config/i386/sse.md (*andnot<mode>3): Likewise.
7851 (<mask_codefor><code><mode>3<mask_name>): Likewise.
7852 (*<code><mode>3): Likewise.
7853 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
7854 (ia64_expand_atomic_op): Likewise.
7855 (ia64_arg_type): Likewise.
7856 (ia64_mode_to_int): Likewise.
7857 (ia64_scalar_mode_supported_p): Likewise.
7858 (ia64_vector_mode_supported_p): Likewise.
7859 (expand_vec_perm_broadcast): Likewise.
7860 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
7861 (iq2000_function_arg_advance): Likewise.
7862 (iq2000_function_arg): Likewise.
7863 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
7864 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
7865 (m68k_libcall_value): Likewise.
7866 (m68k_function_value): Likewise.
7867 (sched_attr_op_type): Likewise.
7868 * config/mcore/mcore.c (mcore_output_move): Likewise.
7869 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
7871 (microblaze_function_arg): Likewise.
7872 * config/mips/mips.c (mips16_build_call_stub): Likewise.
7873 (mips_print_operand): Likewise.
7874 (mips_mode_ok_for_mov_fmt_p): Likewise.
7875 (mips_vector_mode_supported_p): Likewise.
7876 (mips_preferred_simd_mode): Likewise.
7877 (mips_expand_vpc_loongson_even_odd): Likewise.
7878 (mips_expand_vec_unpack): Likewise.
7879 (mips_expand_vi_broadcast): Likewise.
7880 (mips_expand_vector_init): Likewise.
7881 (mips_expand_vec_reduc): Likewise.
7882 (mips_expand_msa_cmp): Likewise.
7883 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
7884 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
7885 (cc_flags_for_mode): Likewise.
7886 * config/msp430/msp430.c (msp430_print_operand): Likewise.
7887 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
7888 (nds32_output_casesi_pc_relative): Likewise.
7889 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7890 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
7891 (nvptx_gen_unpack): Likewise.
7892 (nvptx_gen_pack): Likewise.
7893 (nvptx_gen_shuffle): Likewise.
7894 (nvptx_gen_wcast): Likewise.
7895 (nvptx_preferred_simd_mode): Likewise.
7896 * config/pa/pa.c (pa_secondary_reload): Likewise.
7897 * config/pa/predicates.md (base14_operand): Likewise.
7898 * config/powerpcspe/powerpcspe-c.c
7899 (altivec_resolve_overloaded_builtin): Likewise.
7900 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
7902 (rs6000_preferred_simd_mode): Likewise.
7903 (output_vec_const_move): Likewise.
7904 (rs6000_expand_vector_extract): Likewise.
7905 (rs6000_split_vec_extract_var): Likewise.
7906 (reg_offset_addressing_ok_p): Likewise.
7907 (rs6000_legitimate_offset_address_p): Likewise.
7908 (rs6000_legitimize_address): Likewise.
7909 (rs6000_emit_set_const): Likewise.
7910 (rs6000_const_vec): Likewise.
7911 (rs6000_emit_move): Likewise.
7912 (spe_build_register_parallel): Likewise.
7913 (rs6000_darwin64_record_arg_recurse): Likewise.
7914 (swap_selector_for_mode): Likewise.
7915 (spe_init_builtins): Likewise.
7916 (paired_init_builtins): Likewise.
7917 (altivec_init_builtins): Likewise.
7918 (do_load_for_compare): Likewise.
7919 (rs6000_generate_compare): Likewise.
7920 (rs6000_expand_float128_convert): Likewise.
7921 (emit_load_locked): Likewise.
7922 (emit_store_conditional): Likewise.
7923 (rs6000_output_function_epilogue): Likewise.
7924 (rs6000_handle_altivec_attribute): Likewise.
7925 (rs6000_function_value): Likewise.
7926 (emit_fusion_gpr_load): Likewise.
7927 (emit_fusion_p9_load): Likewise.
7928 (emit_fusion_p9_store): Likewise.
7929 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
7930 (fusion_gpr_mem_load): Likewise.
7931 (fusion_addis_mem_combo_load): Likewise.
7932 (fusion_addis_mem_combo_store): Likewise.
7933 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
7934 (fusion_gpr_mem_load): Likewise.
7935 (fusion_addis_mem_combo_load): Likewise.
7936 (fusion_addis_mem_combo_store): Likewise.
7937 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7939 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
7940 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7941 (rs6000_preferred_simd_mode): Likewise.
7942 (output_vec_const_move): Likewise.
7943 (rs6000_expand_vector_extract): Likewise.
7944 (rs6000_split_vec_extract_var): Likewise.
7945 (reg_offset_addressing_ok_p): Likewise.
7946 (rs6000_legitimate_offset_address_p): Likewise.
7947 (rs6000_legitimize_address): Likewise.
7948 (rs6000_emit_set_const): Likewise.
7949 (rs6000_const_vec): Likewise.
7950 (rs6000_emit_move): Likewise.
7951 (rs6000_darwin64_record_arg_recurse): Likewise.
7952 (swap_selector_for_mode): Likewise.
7953 (paired_init_builtins): Likewise.
7954 (altivec_init_builtins): Likewise.
7955 (rs6000_expand_float128_convert): Likewise.
7956 (emit_load_locked): Likewise.
7957 (emit_store_conditional): Likewise.
7958 (rs6000_output_function_epilogue): Likewise.
7959 (rs6000_handle_altivec_attribute): Likewise.
7960 (rs6000_function_value): Likewise.
7961 (emit_fusion_gpr_load): Likewise.
7962 (emit_fusion_p9_load): Likewise.
7963 (emit_fusion_p9_store): Likewise.
7964 * config/rx/rx.c (rx_gen_move_template): Likewise.
7965 (flags_from_mode): Likewise.
7966 * config/s390/predicates.md (s390_alc_comparison): Likewise.
7967 (s390_slb_comparison): Likewise.
7968 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
7969 (s390_vector_mode_supported_p): Likewise.
7970 (s390_cc_modes_compatible): Likewise.
7971 (s390_match_ccmode_set): Likewise.
7972 (s390_canonicalize_comparison): Likewise.
7973 (s390_emit_compare_and_swap): Likewise.
7974 (s390_branch_condition_mask): Likewise.
7975 (s390_rtx_costs): Likewise.
7976 (s390_secondary_reload): Likewise.
7977 (__SECONDARY_RELOAD_CASE): Likewise.
7978 (s390_expand_cs): Likewise.
7979 (s390_preferred_simd_mode): Likewise.
7980 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
7981 * config/sh/sh.c (sh_print_operand): Likewise.
7982 (dump_table): Likewise.
7983 (sh_secondary_reload): Likewise.
7984 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7985 * config/sh/sh.md (casesi_worker_1): Likewise.
7986 (casesi_worker_2): Likewise.
7987 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
7988 (fcc_comparison_operator): Likewise.
7989 * config/sparc/sparc.c (sparc_expand_move): Likewise.
7990 (emit_soft_tfmode_cvt): Likewise.
7991 (sparc_preferred_simd_mode): Likewise.
7992 (output_cbranch): Likewise.
7993 (sparc_print_operand): Likewise.
7994 (sparc_expand_vec_perm_bmask): Likewise.
7995 (vector_init_bshuffle): Likewise.
7996 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
7997 (spu_vector_mode_supported_p): Likewise.
7998 (spu_expand_insv): Likewise.
7999 (spu_emit_branch_or_set): Likewise.
8000 (spu_handle_vector_attribute): Likewise.
8001 (spu_builtin_splats): Likewise.
8002 (spu_builtin_extract): Likewise.
8003 (spu_builtin_promote): Likewise.
8004 (spu_expand_sign_extend): Likewise.
8005 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8006 (tilegx_simd_int): Likewise.
8007 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8008 (tilepro_simd_int): Likewise.
8009 * config/v850/v850.c (const_double_split): Likewise.
8010 (v850_print_operand): Likewise.
8011 (ep_memory_offset): Likewise.
8012 * config/vax/vax.c (vax_rtx_costs): Likewise.
8013 (vax_output_int_move): Likewise.
8014 (vax_output_int_add): Likewise.
8015 (vax_output_int_subtract): Likewise.
8016 * config/visium/predicates.md (visium_branch_operator): Likewise.
8017 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8018 (visium_print_operand_address): Likewise.
8019 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8020 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8021 (xtensa_expand_conditional_branch): Likewise.
8022 (xtensa_copy_incoming_a7): Likewise.
8023 (xtensa_output_literal): Likewise.
8024 * dfp.c (decimal_real_maxval): Likewise.
8025 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8027 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8028 Alan Hayward <alan.hayward@arm.com>
8029 David Sherwood <david.sherwood@arm.com>
8031 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8032 (mode_nunits_inline): Likewise.
8033 (mode_inner_inline): Likewise.
8034 (mode_unit_size_inline): Likewise.
8035 (mode_unit_precision_inline): Likewise.
8036 (emit_insn_modes_h): Likewise. Also emit a #define of the
8038 (emit_mode_wider): Add an E_ prefix to mode names.
8039 (emit_mode_complex): Likewise.
8040 (emit_mode_inner): Likewise.
8041 (emit_mode_adjustments): Likewise.
8042 (emit_mode_int_n): Likewise.
8043 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8044 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8045 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8046 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8047 (CRC32_BUILTIN, ENTRY): Likewise.
8048 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8049 (aarch64_pop_regs): Likewise.
8050 (aarch64_process_components): Likewise.
8051 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8052 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8053 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8054 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8055 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8056 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8058 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8059 (pcmpistr): Likewise.
8060 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8061 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8062 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8064 * config/rl78/rl78.c (mduc_regs): Likewise.
8065 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8066 (htm_expand_builtin): Likewise.
8067 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8068 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8069 * config/xtensa/xtensa.c (print_operand): Likewise.
8070 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8071 (NUM_MODE_VECTOR_INT): Likewise.
8072 * genoutput.c (null_operand): Likewise.
8073 (output_operand_data): Likewise.
8074 * genrecog.c (print_parameter_value): Likewise.
8075 * lra.c (debug_operand_data): Likewise.
8077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8079 * dwarf2out.c (scompare_loc_descriptor_wide)
8080 (scompare_loc_descriptor_narrow): New functions, split out from...
8081 (scompare_loc_descriptor): ...here.
8082 * expmed.c (emit_store_flag_int): New function, split out from...
8083 (emit_store_flag): ...here.
8085 2017-08-30 Richard Biener <rguenther@suse.de>
8087 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
8088 (dwarf2out_early_finish): Move setting of AT_pubnames from
8089 early debug output to early finish.
8091 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8093 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
8094 and -mdata-region to the assembler.
8096 2017-08-30 Richard Biener <rguenther@suse.de>
8098 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
8100 (gen_subprogram_die): Add DW_AT_object_pointer only early.
8101 (dwarf2out_early_global_decl): Only generate a DIE for the
8102 abstract origin if it doesn't already exist or is a declaration DIE.
8103 (resolve_addr): Do not add the linkage name twice when
8104 generating a stub DIE for the DW_TAG_GNU_call_site target.
8106 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8108 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8109 Use machine_mode rather than int for arg1_mode.
8111 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8114 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
8115 that the second argument of the built-in functions to unpack
8116 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
8117 switch statement instead a lot of if statements.
8118 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
8119 Allow 64-bit values to be in Altivec registers as well as
8120 traditional floating point registers.
8121 (pack<mode>, FMOVE128_VSX iterator): Likewise.
8123 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
8125 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
8126 MAX_REGS_PER_ADDRESS == 1.
8128 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
8130 * config/i386/i386.opt (flag_fentry): Do not init to -1.
8131 * config/i386/i386.c (ix86_option_override_internal): Simplify
8132 setting of opts->x_flag_entry.
8134 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8135 Jakub Jelinek <jakub@redhat.com>
8136 Richard Biener <rguenther@suse.de>
8138 PR tree-optimization/81503
8139 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
8140 folded constant fits in the target type; reorder tests for clarity.
8142 2017-08-29 Martin Liska <mliska@suse.cz>
8144 * passes.def: Include pass_lower_switch.
8145 * stmt.c (dump_case_nodes): Remove and move to
8146 tree-switch-conversion.
8147 (case_values_threshold): Likewise.
8148 (expand_switch_as_decision_tree_p): Likewise.
8149 (emit_case_decision_tree): Likewise.
8150 (expand_case): Likewise.
8151 (balance_case_nodes): Likewise.
8152 (node_has_low_bound): Likewise.
8153 (node_has_high_bound): Likewise.
8154 (node_is_bounded): Likewise.
8155 (emit_case_nodes): Likewise.
8156 (struct simple_case_node): New struct.
8157 (add_case_node): Remove.
8158 (emit_case_dispatch_table): Use vector instead of case_list.
8159 (reset_out_edges_aux): Remove.
8160 (compute_cases_per_edge): Likewise.
8161 (expand_case): Build list of simple_case_node.
8162 (expand_sjlj_dispatch_table): Use it.
8163 * tree-switch-conversion.c (struct case_node): Moved from
8164 stmt.c and adjusted.
8165 (emit_case_nodes): Likewise.
8166 (node_has_low_bound): Likewise.
8167 (node_has_high_bound): Likewise.
8168 (node_is_bounded): Likewise.
8169 (case_values_threshold): Likewise.
8170 (reset_out_edges_aux): Likewise.
8171 (compute_cases_per_edge): Likewise.
8172 (add_case_node): Likewise.
8173 (dump_case_nodes): Likewise.
8174 (balance_case_nodes): Likewise.
8175 (expand_switch_as_decision_tree_p): Likewise.
8176 (emit_jump): Likewise.
8177 (emit_case_decision_tree): Likewise.
8178 (try_switch_expansion): Likewise.
8179 (do_jump_if_equal): Likewise.
8180 (emit_cmp_and_jump_insns): Likewise.
8181 (fix_phi_operands_for_edge): New function.
8182 (record_phi_operand_mapping): Likewise.
8183 (class pass_lower_switch): New pass.
8184 (pass_lower_switch::execute): New function.
8185 (make_pass_lower_switch): Likewise.
8186 (conditional_probability):
8187 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
8188 * tree-pass.h: Add make_pass_lower_switch.
8190 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8193 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
8196 2017-08-29 Richard Biener <rguenther@suse.de>
8198 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
8199 we do not add a DW_AT_inline attribute twice.
8200 (gen_subprogram_die): Remove code setting DW_AT_inline on
8201 DECL_ABSTRACT_P nodes.
8203 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
8205 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
8206 calls to internal functions.
8207 (gimplify_modify_expr): Likewise.
8208 * tree-call-cdce.c (use_internal_fn): Likewise.
8209 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8210 (convert_to_divmod): Set the nothrow flag.
8211 * tree-if-conv.c (predicate_mem_writes): Likewise.
8212 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8213 (vectorizable_call): Likewise.
8214 (vectorizable_store): Likewise.
8215 (vectorizable_load): Likewise.
8216 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
8217 (vect_recog_mask_conversion_pattern): Likewise.
8219 2017-08-29 Martin Liska <mliska@suse.cz>
8222 * gcc.c (driver_handle_option): Add new argument.
8223 * opts-common.c (handle_option): Pass
8224 target_option_override_hook.
8225 * opts-global.c (lang_handle_option): Add new option.
8226 (set_default_handlers): Add new argument.
8227 (decode_options): Likewise.
8228 * opts.c (target_handle_option): Likewise.
8229 (common_handle_option): Call target_option_override_hook.
8230 * opts.h (struct cl_option_handler_func): Add hook for
8231 target option override.
8232 (struct cl_option_handlers): Likewise.
8233 (set_default_handlers): Add new argument.
8234 (decode_options): Likewise.
8235 (common_handle_option): Likewise.
8236 (target_handle_option): Likewise.
8237 * toplev.c (toplev::main): Pass targetm.target_option.override
8240 2017-08-29 Richard Biener <rguenther@suse.de>
8241 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8243 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
8244 life to the active subtree.
8246 2017-08-28 Jeff Law <law@redhat.com>
8248 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
8249 derive_equivalences.
8250 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
8252 (edge_info::derive_equivalences): New private member function
8254 * tree-ssa-dom.c (class edge_info): Changed from a struct
8255 to a class. Add ctor/dtor, methods and data members.
8256 (edge_info::edge_info): Renamed from allocate_edge_info.
8257 Initialize additional members.
8258 (edge_info::~edge_info): New.
8259 (free_dom_edge_info): Delete the edge info.
8260 (record_edge_info): Use new class & associated member functions.
8261 Tighten forms for testing for edge equivalences.
8262 (record_temporary_equivalences): Iterate over the simple
8263 equivalences rather than assuming there's only one per edge.
8264 (cprop_into_successor_phis): Iterate over the simple
8265 equivalences rather than assuming there's only one per edge.
8266 (optimize_stmt): Use operand_equal_p rather than pointer
8267 equality for mini-DSE code.
8269 2017-08-28 Nathan Sidwell <nathan@acm.org>
8271 * gcc.c (execute): Fold SIGPIPE handling into switch
8272 statement. Adjust internal error message.
8274 2017-08-28 Richard Biener <rguenther@suse.de>
8277 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
8278 Do nothing for removed DIEs.
8280 2017-08-28 Richard Biener <rguenther@suse.de>
8282 PR tree-optimization/81977
8283 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
8286 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
8289 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
8290 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
8291 using targetm.gen_mem_thread_fence.
8293 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
8296 * config/i386/i386.md (*<btsc><mode>): Change operand 2
8297 predicate to register_operand. Reorder operands.
8298 (*btr<mode>): Ditto.
8299 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
8300 (*btr<mode>_mask): Ditto.
8302 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
8304 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
8305 * config/rs6000/xmmintrin.h: New file.
8306 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
8308 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8311 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
8312 parameter and_insn and return it.
8313 (recombine_lvx_pattern): Insert a copy to ensure availability of
8314 the base register of the copied masking operation at the point of
8315 the instruction replacement.
8316 (recombine_stvx_pattern): Likewise.
8318 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
8320 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
8321 undocumented switches.
8322 (-mpower9-dform-vector): Likewise.
8323 (-mpower9-dform): Likewise.
8324 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
8325 comments to delete references to -mpower9-dform* switches.
8326 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
8327 Delete reference to -mpower9-dform* switches, test for
8328 -mpower9-vector instead.
8329 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
8330 (OTHER_P9_VECTOR_MASKS): Likewise.
8331 (POWERPC_MASKS): Likewise.
8332 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
8333 tests against -mpower9-dform* to -mpower9-vector. Delete code
8334 that checked for -mpower9-dform* consistancy with other options.
8335 Add test for -mpower9-misc to enable other power9 switches.
8336 (rs6000_init_hard_regno_mode_ok): Likewise.
8337 (rs6000_option_override_internal): Likewise.
8338 (rs6000_emit_prologue): Likewise.
8339 (rs6000_emit_epilogue): Likewise.
8340 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
8341 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
8342 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
8344 (emit_fusion_p9_store): Likewise.
8345 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
8346 resetting these macros if the assembler does not support ISA 3.0
8348 (TARGET_P9_DFORM_VECTOR): Likewise.
8349 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
8350 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
8352 2017-08-25 Alan Modra <amodra@gmail.com>
8354 PR rtl-optimization/81747
8355 * cse.c (cse_extended_basic_block): Don't attempt to record
8356 equivalences for degenerate conditional branches that branch
8357 to their fall-through.
8359 2017-08-24 Martin Sebor <msebor@redhat.com>
8362 * gimple-fold.c (size_must_be_zero_p): New function.
8363 (gimple_fold_builtin_memory_op): Call it.
8365 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
8367 * config/rs6000/mm_malloc.h: New file.
8369 2017-08-24 Bin Cheng <bin.cheng@arm.com>
8371 PR tree-optimization/81913
8372 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
8373 analysis when either IVs in condition can wrap.
8375 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
8377 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
8378 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
8380 2017-08-24 Richard Biener <rguenther@suse.de>
8383 * targhooks.c (default_target_can_inline_p): Properly
8384 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
8385 is present and always compare.
8386 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
8387 infer -mfpmath=sse from TARGET_SSE_P.
8388 (ix86_can_inline_p): Properly use target_option_default_node when
8389 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
8391 2017-08-24 Richard Biener <rguenther@suse.de>
8394 * dwarf2out.c (output_die): Handle flag_generate_offload like
8396 (output_comp_unit): Likewise.
8397 (gen_array_type_die): Likewise.
8398 (dwarf2out_early_finish): Likewise.
8399 (note_variable_value_in_expr): Likewise.
8400 (dwarf2out_finish): Likewise. Adjust assert.
8401 * cgraphunit.c (symbol_table::compile): Move setting of
8402 flag_generate_offload earlier ...
8403 (symbol_table::finalize_compilation_unit): ... here, before
8404 early debug finalization.
8406 2017-08-24 Richard Biener <rguenther@suse.de>
8408 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
8409 and ipa-fnsummary.h.
8410 (ix86_can_inline_p): When ix86_fpmath flags do not match
8411 check whether the callee uses FP math at all.
8413 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
8416 * tree-ssanames.c (get_nonzero_bits): Use element_precision
8417 instead of TYPE_PRECISION.
8419 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
8420 Alan Hayward <alan.hayward@arm.com>
8421 David Sherwood <david.sherwood@arm.com>
8423 * combine.c (make_extraction): Use subreg_offset_from_lsb.
8425 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
8427 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
8429 (ix86_frame::stack_realign_allocate): New field.
8430 (struct machine_frame_state): Modify comments.
8431 (machine_frame_state::sp_realigned_fp_end): New field.
8432 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
8434 (sp_valid_at): Add assertion to assure no attempt to access invalid
8435 offset of a realigned stack.
8436 (fp_valid_at): Likewise.
8437 (choose_baseaddr): Modify comments.
8438 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
8439 ix86_expand_prologue.
8440 (ix86_expand_prologue): Modify stack realignment and allocation.
8441 (ix86_expand_epilogue): Modify comments.
8442 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
8443 avx2_runtime, avx512f, and avx512f_runtime.
8445 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
8447 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
8448 (mstackrealign): Do not init to -1.
8449 * config/i386/i386.c (ix86_option_override_internal):
8450 Check opts_set, not opts when setting default value of
8451 opts->x_ix86_force_align_arg_pointer.
8453 2017-08-23 Richard Biener <rguenther@suse.de>
8455 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
8456 lang_hooks.decl_printable_name.
8457 * print-rtl-function.c (print_rtx_function): Likewise.
8458 * tree-pretty-print.c (dump_function_header): Likewise.
8460 2017-08-23 Richard Biener <rguenther@suse.de>
8463 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
8464 -g0 at compile-time.
8466 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8469 * doc/sourcebuild.texi (Other hardware attributes):
8472 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8475 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
8476 Fix single-use check.
8478 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
8480 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
8482 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8484 * doc/install.texi: Modify to add more details on running selected
8487 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8489 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
8490 is combined with -mabi=ms.
8491 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
8494 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8496 PR tree-optimization/81488
8497 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
8498 and cached_basis fields.
8499 (MAX_SPREAD): New constant.
8500 (alloc_cand_and_find_basis): Initialize new fields.
8501 (clear_visited): New function.
8502 (create_phi_basis_1): Rename from create_phi_basis, set visited
8503 and cached_basis fields.
8504 (create_phi_basis): New wrapper function.
8505 (phi_add_costs_1): Rename from phi_add_costs, add spread
8506 parameter, set visited field, short-circuit when limits reached.
8507 (phi_add_costs): New wrapper function.
8508 (record_phi_increments_1): Rename from record_phi_increments, set
8510 (record_phi_increments): New wrapper function.
8511 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
8512 (phi_incr_cost): New wrapper function.
8513 (all_phi_incrs_profitable_1): Rename from
8514 all_phi_incrs_profitable, set visited field.
8515 (all_phi_incrs_profitable): New wrapper function.
8517 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8518 Alan Hayward <alan.hayward@arm.com>
8519 David Sherwood <david.sherwood@arm.com>
8521 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
8522 that takes the outer and inner modes.
8523 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
8524 comparison as the canonical test for a paradoxical subreg.
8525 * combine.c (simplify_set): Use paradoxical_subreg_p.
8526 (make_extraction): Likewise.
8527 (force_to_mode): Likewise.
8528 (rtx_equal_for_field_assignment_p): Likewise.
8529 (gen_lowpart_for_combine): Likewise.
8530 (simplify_comparison): Likewise.
8531 * cse.c (equiv_constant): Likewise.
8532 * expmed.c (store_bit_field_1): Likewise.
8533 * final.c (alter_subreg): Likewise.
8534 * fwprop.c (propagate_rtx): Likewise.
8535 (forward_propagate_subreg): Likewise.
8536 * ira-conflicts.c (ira_build_conflicts): Likewise.
8537 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
8538 * lra-constraints.c (curr_insn_transform): Likewise.
8539 (split_reg): Likewise.
8540 * lra-eliminations.c (move_plus_up): Likewise.
8541 (lra_eliminate_regs_1): Likewise.
8542 * recog.c (general_operand): Likewise.
8543 * ree.c (combine_reaching_defs): Likewise.
8544 * reload.c (push_reload): Likewise.
8545 (find_reloads): Likewise.
8546 * reload1.c (elimination_effects): Likewise.
8547 (compute_reload_subreg_offset): Likewise.
8548 (choose_reload_regs): Likewise.
8549 * rtlanal.c (subreg_lsb_1): Likewise.
8550 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8551 (simplify_subreg): Likewise.
8552 * var-tracking.c (track_loc_p): Likewise.
8553 * emit-rtl.c (byte_lowpart_offset): Likewise.
8554 (paradoxical_subreg_p): Delete out-of-line definition.
8556 2017-08-22 Jeff Law <law@redhat.com>
8558 PR tree-optimization/81741
8559 PR tree-optimization/71947
8560 * tree-ssa-dom.c: Include tree-inline.h.
8561 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
8562 equivalences if one is more expensive to compute than the other.
8563 * tree-ssa-scopedtables.h (class const_or_copies): Make
8564 record_const_or_copy_raw method private.
8565 (class avail_exprs_stack): New method simplify_binary_operation.
8566 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
8567 avail_exprs_stack::simplify_binary_operation as needed.
8568 (avail_exprs_stack::simplify_binary_operation): New function.
8570 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
8572 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
8573 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
8574 (DOT_SYMBOLS): Likewise.
8575 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
8576 (RELOCATABLE_NEEDS_FIXUP): Likewise.
8577 (RS6000_ABI_NAME): Likewise.
8578 (TARGET_CMODEL): Likewise.
8579 (TOC_SECTION_ASM_OP): Likewise.
8580 (SET_CMODEL): New macro.
8581 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
8583 2017-08-22 Richard Biener <rguenther@suse.de>
8585 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
8586 to allow for free-lang-data replacements similar to verify_type_variant.
8588 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8589 Alan Hayward <alan.hayward@arm.com>
8590 David Sherwood <david.sherwood@arm.com>
8592 * config/aarch64/aarch64.md (casesi): Use DImode rather than
8593 VOIDmode for the LABEL_REF.
8595 2017-08-22 Richard Biener <rguenther@suse.de>
8597 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
8599 2017-08-22 Richard Biener <rguenther@suse.de>
8601 * common.opt (feliminate-dwarf2-dups): Ignore.
8602 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
8603 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
8604 same_die_p_wrap, compute_section_prefix,
8605 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
8606 (comdat_symbol_id, comdat_symbol_number): Likewise.
8607 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
8609 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
8610 (output_die): Mark unreachable path unreachable.
8611 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
8612 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
8613 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
8614 (dwarf2out_early_finish): Likewise.
8616 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
8618 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
8620 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
8623 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
8624 not VAR_P. Filter attribute warnings with OPT_Wattributes.
8625 (avr_attribute_table) <io, io_low, address>: Initialize
8626 .decl_required with true.
8628 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8630 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
8631 undocumented debugging options.
8632 (-mvsx-scalar-double): Likewise.
8633 (-mallow-df-permute): Likewise.
8634 (-mvectorize-builtins): Likewise.
8635 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8636 (rs6000_builtin_vectorized_function): Likewise.
8637 (rs6000_builtin_md_vectorized_function): Likewise.
8638 (rs6000_opt_vars): Likewise.
8640 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
8643 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
8644 (*btrq_imm): Rename from *btrq.
8645 (*btcq_imm): Rename from *btcq.
8646 (btsc): New code attribute.
8647 (*<btsc><mode>): New insn pattern.
8648 (*btr<mode>): Ditto.
8649 (*<btsc><mode>_mask): New insn_and_split pattern.
8650 (*btr<mode>_mask): Ditto.
8652 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8653 Alan Hayward <alan.hayward@arm.com>
8654 David Sherwood <david.sherwood@arm.com>
8656 * function.c (pad_below): Simplify padding calculation.
8658 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8659 Alan Hayward <alan.hayward@arm.com>
8660 David Sherwood <david.sherwood@arm.com>
8662 * target.def (function_prologue): Remove frame size argument.
8663 (function_epilogue): Likewise.
8664 * doc/tm.texi: Regenerate.
8665 * final.c (final_start_function): Update call to function_prologue.
8666 (final_end_function): Update call to function_epilogue.
8667 (default_function_pro_epilogue): Remove frame size argument.
8668 * output.h (default_function_pro_epilogue): Likewise.
8669 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
8670 (arm_output_function_prologue): Likewise.
8671 * config/frv/frv.c (frv_function_prologue): Likewise.
8672 (frv_function_epilogue): Likewise.
8673 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
8674 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
8675 (ia64_output_function_epilogue): Likewise.
8676 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
8677 (m32r_output_function_epilogue): Likewise.
8678 * config/microblaze/microblaze.c (microblaze_function_prologue)
8679 (microblaze_function_epilogue): Likewise.
8680 * config/mips/mips.c (mips_output_function_prologue): Likewise.
8681 (mips_output_function_epilogue): Likewise.
8682 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
8683 (mmix_target_asm_function_epilogue): Likewise.
8684 * config/msp430/msp430.c (msp430_start_function): Likewise.
8685 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
8686 (nds32_asm_function_epilogue): Likewise.
8687 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
8688 * config/pa/pa.c (pa_output_function_prologue): Likewise.
8689 (pa_output_function_epilogue): Likewise.
8690 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
8691 (rs6000_output_function_epilogue): Likewise.
8692 * config/rl78/rl78.c (rl78_start_function): Likewise.
8693 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
8694 (rs6000_output_function_epilogue): Likewise.
8695 * config/rx/rx.c (rx_output_function_prologue): Likewise.
8696 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
8697 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
8698 (sparc_asm_function_epilogue): Likewise.
8700 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8702 * tree.h (type_has_mode_precision_p): New function.
8703 * convert.c (convert_to_integer_1): Use it.
8704 * expr.c (expand_expr_real_2): Likewise.
8705 (expand_expr_real_1): Likewise.
8706 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
8707 * match.pd: Likewise.
8708 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
8709 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
8710 * tree-tailcall.c (process_assignment): Likewise.
8711 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8712 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
8713 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
8714 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8715 (vectorizable_assignment): Likewise.
8716 (vectorizable_shift): Likewise.
8717 (vectorizable_operation): Likewise.
8718 * tree-vrp.c (register_edge_assert_for_2): Likewise.
8720 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
8722 * match.pd: Add pow (C, x) simplification.
8724 2017-08-21 Richard Biener <rguenther@suse.de>
8726 PR tree-optimization/81900
8727 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
8728 for blocks with abnormal predecessors.
8729 (compute_antic): Do not set visited flag prematurely.
8731 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
8734 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
8736 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8738 * stor-layout.h (vector_type_mode): Move to...
8739 * tree.h (vector_type_mode): ...here.
8740 * stor-layout.c (vector_type_mode): Move to...
8741 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
8743 2017-08-21 Richard Biener <rguenther@suse.de>
8745 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
8746 register_external_die hooks.
8747 (debug_false_tree_charstarstar_uhwistar): Declare.
8748 (debug_nothing_tree_charstar_uhwi): Likewise.
8749 * debug.c (do_nothing_debug_hooks): Adjust.
8750 (debug_false_tree_charstarstar_uhwistar): New do nothing.
8751 (debug_nothing_tree_charstar_uhwi): Likewise.
8752 * dbxout.c (dbx_debug_hooks): Adjust.
8753 (xcoff_debug_hooks): Likewise.
8754 * sdbout.c (sdb_debug_hooks): Likewise.
8755 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8756 * dwarf2out.c (macinfo_label_base): New global.
8757 (dwarf2out_register_external_die): New function for the
8758 register_external_die hook.
8759 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
8760 (dwarf2_debug_hooks): Use them.
8761 (dwarf2_lineno_debug_hooks): Adjust.
8762 (struct die_struct): Add with_offset flag.
8763 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
8764 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
8765 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
8766 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
8767 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
8768 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
8769 defining section names for the early LTO debug variants.
8770 (reset_indirect_string): New helper.
8771 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
8772 (print_dw_val): Add support for offsetted symbol references.
8773 (get_ultimate_context): Split out from is_cxx.
8774 (is_cxx): Use get_ultimate_context.
8775 (is_fortran): Add decl overload.
8776 (compute_comp_unit_symbol): Split out worker from
8777 compute_section_prefix.
8778 (compute_section_prefix): Call compute_comp_unit_symbol and
8779 set comdat_type_p here.
8780 (output_die): Skip DIE symbol output for the LTO added one.
8781 Handle DIE symbol references with offset.
8782 (output_comp_unit): Guard section name mangling properly.
8783 For LTO debug sections emit a symbol at the section beginning
8784 which we use to refer to its DIEs.
8785 (add_abstract_origin_attribute): For DIEs registered via
8786 dwarf2out_register_external_die directly refer to the early
8787 DIE rather than indirectly through the shadow one we created.
8788 Remove obsolete call to dwarf2out_abstract_function for
8789 non-function/block origins.
8790 (gen_array_type_die): When generating early LTO debug do
8791 not emit DW_AT_string_length.
8792 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
8793 late when in LTO. As suggested place a gcc_unreachable for
8794 the DECL_ABSTRACT_P case.
8795 (gen_subprogram_die): Avoid another specification DIE
8796 for early built declarations/definitions for the late LTO case.
8797 (gen_variable_die): Add type references for late duplicated VLA dies
8799 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
8800 we have the abstract instance already.
8801 (process_scope_var): Adjust decl DIE contexts in LTO which
8802 first puts them in limbo.
8803 (gen_decl_die): Do not generate type DIEs late apart from
8804 types for VLAs or for decls we do not yet have a DIE. Do not
8805 call dwarf2out_abstract_function late.
8806 (dwarf2out_early_global_decl): Make sure to create DIEs
8807 for abstract instances of a decl first.
8808 (dwarf2out_late_global_decl): Adjust comment.
8809 (output_macinfo_op): With multiple macro sections use
8810 macinfo_label_base to distinguish labels.
8811 (output_macinfo): Likewise. Update macinfo_label_base.
8812 Pass in the line info label.
8813 (note_variable_value_in_expr): When generating LTO resolve
8814 all variable values here by generating DIEs as needed.
8815 (init_sections_and_labels): Add early LTO debug flag parameter
8816 and generate different sections and names if set. Add generation
8817 counter for the labels so we can have multiple of them.
8818 (reset_dies): Helper to allow DIEs to be output multiple times.
8819 (dwarf2out_finish): When outputting DIEs to the fat part of an
8820 LTO object first reset DIEs.
8821 (dwarf2out_early_finish): Output early DIEs when generating LTO.
8822 (modified_type_die): Check for decl_ultimate_origin being self
8824 (gen_type_die_with_usage): Likewise.
8825 (gen_typedef_die): Allow decl_ultimate_origin being self.
8826 (set_decl_abstract_flags): Remove.
8827 (set_block_abstract_flags): Likewise.
8828 (dwarf2out_abstract_function): Treat the early generated DIEs
8829 as the abstract copy and only add DW_AT_inline and
8830 DW_AT_artificial here and call set_decl_origin_self.
8831 If the DIE has an abstract origin don't do anything.
8832 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
8833 if we have none yet (Go fails to build one, PR78628).
8834 (variably_modified_type_p): Prevent endless recursion for Ada
8835 cyclic pointer types.
8836 * lto-streamer-in.c: Include debug.h.
8837 (dref_queue): New global.
8838 (lto_read_tree_1): Stream in DIE references.
8839 (lto_input_tree): Register DIE references.
8840 (input_function): Stream DECL_DEBUG_ARGS.
8841 * lto-streamer-out.c: Include debug.h.
8842 (lto_write_tree_1): Output DIE references.
8843 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
8844 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
8845 (output_function): Stream DECL_DEBUG_ARGS.
8846 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
8847 Stream DECL_ABSTRACT_ORIGIN.
8848 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
8849 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
8850 DECL_CONTEXT for file-scope decls.
8851 * lto-streamer.h (struct dref_entry): Declare.
8852 (dref_queue): Likewise.
8853 * cfgexpand.c (pass_expand::execute): Do not call the
8854 outlining_inline_function hook here.
8855 * lto-wrapper.c (debug_obj): New global.
8856 (tool_cleanup): Unlink it if required.
8857 (debug_objcopy): New function.
8858 (run_gcc): Handle early debug sections in the IL files by
8859 extracting them to separate files, partially linkin them and
8860 feeding the result back as result to the linker.
8861 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
8862 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
8863 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
8864 sections into a separate segment.
8865 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
8867 (darwin_asm_dwarf_section): Likewise.
8868 (darwin_asm_output_dwarf_offset): Likewise.
8869 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
8871 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8872 Alan Hayward <alan.hayward@arm.com>
8873 David Sherwood <david.sherwood@arm.com>
8875 * read-md.h (md_reader::record_potential_iterator_use): Replace
8876 pointer argument with an rtx and an index.
8877 * read-rtl.c (iterator_group::apply_iterator): Likewise.
8878 (apply_mode_iterator): Likewise.
8879 (apply_code_iterator): Likewise.
8880 (apply_int_iterator): Likewise.
8881 (apply_subst_iterator): Likewise.
8882 (record_iterator_use): Likewise.
8883 (record_attribute_use): Likewise.
8884 (md_reader::record_potential_iterator_use): Likewise. Update calls
8885 to record_iterator_use and apply_iterator.
8886 (iterator_use): Replace ptr with x and index.
8887 (attribute_use): Likewise.
8888 (apply_attribute_uses): Update calls to apply_iterator.
8889 (apply_iterators): Likewise. Update initialization of iterator_use.
8890 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
8891 and record_potential_iterator_use.
8892 (rtx_reader::read_rtx_operand): Likewise.
8894 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8895 Alan Hayward <alan.hayward@arm.com>
8896 David Sherwood <david.sherwood@arm.com>
8898 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
8901 2017-08-21 Richard Biener <rguenther@suse.de>
8904 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
8905 at struct end conservatively when comparing common bases.
8907 2017-08-21 Richard Biener <rguenther@suse.de>
8909 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
8910 (mem_ref_in_stmt): Remove.
8911 (determine_max_movement): Use ref index to get at the reference.
8912 (invariantness_dom_walker::before_dom_children): Deal with
8913 lim data already initialized.
8914 (gather_mem_refs_stmt): Initialize lim data and record ref index.
8916 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8918 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
8919 (TARGET_ISA_ROUND): Ditto.
8920 (TARGET_ROUND): Ditto.
8921 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
8922 * config/i386/i386.md: Ditto.
8923 * config/i386/sse.md: Ditto.
8924 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
8925 with OPTION_MASK_ISA_SSE4_1.
8927 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8930 * doc/extend.texi (x86 Built-in Functions): Correct the name of
8931 __builtin_ia32_lzcnt_u16.
8933 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8936 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
8937 (rs6000_set_current_function): Rewrite function to use it.
8939 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
8942 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
8943 and TYPE_WARN_IF_NOT_ALIGN.
8944 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
8945 (handle_warn_if_not_align): New.
8946 (place_union_field): Call handle_warn_if_not_align.
8947 (place_field): Call handle_warn_if_not_align.
8948 Copy TYPE_WARN_IF_NOT_ALIGN.
8949 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
8950 (layout_type): Likewise.
8951 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
8953 (tree_decl_common): Add warn_if_not_align.
8954 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
8955 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
8956 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
8957 (DECL_WARN_IF_NOT_ALIGN): Likewise.
8958 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
8959 * doc/extend.texi: Document warn_if_not_aligned attribute.
8960 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
8962 2017-08-17 Martin Liska <mliska@suse.cz>
8965 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
8966 (get_data_dependence): Use it as pointer type.
8967 (distribute_loop): Likewise.
8969 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8971 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
8972 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
8973 (p8_vmrgow_<mode>_direct): New define_insn.
8974 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
8975 handle endianness for vmrgew and vmrgow permute patterns.
8977 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
8979 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
8980 * config/rs6000/rs6000-cpus.def: Remove comment.
8981 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
8982 (POWERPC_MASKS): Likewise.
8983 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
8984 use of TARGET_VSX_TIMODE.
8985 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
8986 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
8987 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
8988 (rs6000_option_override_internal): Remove dead code.
8989 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
8990 (rs6000_legitimize_reload_address): Likewise.
8991 (rs6000_legitimate_address_p): Likewise.
8992 (rs6000_opt_masks): Delete "vsx-timode".
8993 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
8994 from function comment.
8995 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
8996 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
8997 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
8999 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9000 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9001 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9004 (define_peephole2): Likewise.
9006 2017-08-17 Martin Sebor <msebor@redhat.com>
9009 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9010 past the end of an array.
9011 (test_pp_format): Add test cases.
9013 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9015 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9016 missing ECF_NOTHROW flags.
9018 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9021 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9022 operands residing in integer registers.
9023 (*vsx_le_perm_load_<mode>): Likewise.
9024 (*vsx_le_perm_store_<mode>): Likewise.
9025 (define_peephole2): Add peepholes to optimize the above.
9027 2017-08-17 Marek Polacek <polacek@redhat.com>
9030 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9031 to mimic what shorten_compare did. Change the return type to bool.
9032 (fold_cond_expr_with_comparison): Update call to
9033 operand_equal_for_comparison_p.
9034 (fold_ternary_loc): Likewise.
9036 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9038 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9040 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9042 2017-08-17 Richard Biener <rguenther@suse.de>
9044 * tree-ssa-structalias.c (solve_graph): When propagating
9045 to successors update the graphs succ edges and avoid duplicate work.
9047 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9050 * config/i386/i386.c (ix86_option_override_internal): Save target
9051 specific options after ix86_stack_protector_guard_reg was changed.
9053 2017-08-17 Richard Biener <rguenther@suse.de>
9055 PR tree-optimization/81827
9056 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9057 (new_var_info): Initialize it conservatively.
9058 (get_call_vi): Mark register vars.
9059 (new_scalar_tmp_constraint_exp): Likewise.
9060 (handle_rhs_call): Likewise.
9061 (handle_const_call): Likewise.
9062 (create_function_info_for): Likewise.
9063 (solve_constraints): Sort varinfos to separate register from
9064 non-register vars to pack points-to solution bitmaps during
9067 2017-08-17 Marek Polacek <polacek@redhat.com>
9069 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9071 2017-08-17 Richard Biener <rguenther@suse.de>
9073 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9074 to true when overflow is undefined and we saturated the result.
9076 2017-08-17 Alan Modra <amodra@gmail.com>
9079 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
9080 Don't use store multiple if only one reg needs saving.
9081 (interesting_frame_related_regno): New function.
9082 (rs6000_frame_related): Don't emit frame info for regs that
9084 (rs6000_emit_epilogue): Likewise.
9086 2017-08-16 Nathan Sidwell <nathan@acm.org>
9088 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
9089 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
9090 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
9091 * tree.c (free_lang_data_in_type): Use else-if chain. Always
9092 clear TYPE_LANG_1. Remove obsolete member-function stripping.
9093 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
9094 (verify_type): Adjust for TYPE_BINFO move.
9095 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
9096 process TYPE_BINFO directly.
9097 (hash_tree): Likewise.
9098 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9100 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9103 2017-08-16 David Malcolm <dmalcolm@redhat.com>
9105 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
9107 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
9110 * config/i386/i386.md (*anddi_1_btr): Change predicates of
9111 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
9112 Add ix86_binary_operator_ok to insn constraint.
9113 (*iordi_1_bts): Ditto.
9114 (*xordi_1_btc): Ditto.
9115 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
9116 Update corresponding peephole2 pattern.
9120 2017-08-16 Bin Cheng <bin.cheng@arm.com>
9122 PR tree-optimization/81832
9123 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
9124 copy loop header which has IFN_LOOP_DIST_ALIAS call.
9126 2017-08-16 Marek Polacek <polacek@redhat.com>
9129 * fold-const.c (fold_indirect_ref_1): Restore original behavior
9130 regarding size_zero_node.
9132 2017-08-16 Martin Liska <mliska@suse.cz>
9135 * config.gcc: Respect previously set extra_objs in case
9138 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
9140 PR tree-optimization/81835
9141 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
9142 the phi SSA_NAME. Check that the condition in a COND_EXPR does
9143 not depend on the phi.
9145 2017-08-16 Alan Modra <amodra@gmail.com>
9147 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
9150 2017-08-16 Alan Modra <amodra@gmail.com>
9152 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
9153 (save_reg_p): ..into this. Update all callers.
9154 (first_reg_to_save): Simplify.
9156 2017-08-16 Alan Modra <amodra@gmail.com>
9158 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
9161 2017-08-15 Joseph Myers <joseph@codesourcery.com>
9165 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
9166 constant count if that count is less than 32.
9168 2017-08-15 Nathan Sidwell <nathan@acm.org>
9170 * gcc.c (execute): Emit friendlier message if inferior is killed
9171 by an external cause.
9173 2017-08-15 Richard Biener <rguenther@suse.de>
9175 PR tree-optimization/81790
9176 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
9177 CONSTRUCTORs from simplifying and VN.
9179 2017-08-14 Martin Sebor <msebor@redhat.com>
9181 * builtin-attrs.def: Add comments.
9183 2017-08-14 Martin Sebor <msebor@redhat.com>
9186 * doc/extend.texi (attribute nonstring): Document new attribute.
9188 2017-08-14 Martin Sebor <msebor@redhat.com>
9191 * tree-diagnostic.c (default_tree_printer): Handle %G.
9192 * gimple-pretty-print.h (percent_G_format): Declare new function.
9193 * gimple-pretty-print.c (percent_G_format): Define.
9194 * tree-pretty-print.c (percent_K_format): Add argument.
9196 2017-08-14 Martin Sebor <msebor@redhat.com>
9198 PR translation/79998
9199 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
9200 Remove a stray space.
9202 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
9205 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
9206 (*iordi_1_bts): Ditto.
9207 (*xordi_1_btc): Ditto.
9209 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9212 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
9213 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9215 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
9216 quoted strings, and make more translator-friendly.
9217 (darwin_rs6000_override_options): Likewise.
9218 (rs6000_option_override_internal): Likewise.
9219 (rs6000_return_in_memory): Fix overlong line.
9220 (init_cmulative_args): Use quoted strings, and make more
9221 translator-friendly.
9222 (rs6000_pass_by_reference): Fix overlong line.
9223 (def_builtin): Use quoted strings.
9224 (altivec_expand_predicate_builtin): Use quoted strings, and make
9225 more translator-friendly.
9226 (htm_expand_builtin): Use quoted strings.
9227 (cpu_expand_builtin): Use quoted strings, and make more
9228 translator-friendly.
9229 (altivec_expand_builtin): Likewise.
9230 (paired_expand_predicate_builtin): Likewise.
9231 (rs6000_invalid_builtin): Likewise.
9232 (builtin_function_type): Use quoted strings.
9233 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
9234 more translator-friendly.
9235 (rs6000_trampoline_init): Likewise.
9236 (rs6000_handle_altivec_attribute): Likewise.
9237 (rs6000_inner_target_options): Use quoted strings.
9238 (rs6000_disable_incompatible_switches): Likewise.
9239 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
9240 strings, and make more translator-friendly.
9241 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
9243 2017-08-14 Bin Cheng <bin.cheng@arm.com>
9245 PR tree-optimization/81799
9246 * tree-loop-distribution.c (version_loop_by_alias_check): Force
9247 cond_expr to simple gimple operand.
9249 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
9252 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
9254 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
9258 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
9259 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
9260 TARGET_GASISR_PROLOGUES.
9261 * config/avr/avr.c (avr_option_override): Same.
9262 (avr_pass_pre_proep::execute): Same.
9264 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
9267 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
9268 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
9270 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
9272 * config/i386/i386.md (*load_tp_<mode>): Redefine as
9273 define_insn_and_split. Split to a memory load from 0 in
9274 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
9275 using PTR mode iterator.
9276 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
9277 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
9278 (*add_tp_<mode>): Redefine as define_insn_and_split.
9279 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
9280 address space. Merge with *add_tp_x32 using PTR mode iterator.
9281 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
9282 Split to an add with a memory load from 0 in
9283 DEFAULT_TLS_SEG_REG address space.
9285 2017-08-12 Andrew Pinski <apinski@cavium.com>
9287 * config/aarch64/aarch64-option-extensions.def (rdma):
9288 Fix feature string to what Linux prints out in /proc/cpuinfo.
9290 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
9293 * dwarf2out.c (modified_type_die): For C typedef types that have
9294 an ultimate origin, process the ultimate origin instead of the
9296 (gen_typedef_die): Assert that input DECLs have no ultimate
9298 (gen_type_die_with_usage): For typedef variants that have an
9299 ultimate origin, just call gen_decl_die on the original DECL.
9300 (process_scope_var): Avoid creating DIEs for local typedefs and
9301 concrete static variables.
9303 2017-08-12 Alan Modra <amodra@gmail.com>
9307 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
9308 match gnu-user.h startfile.
9309 (ENDFILE_LINUX_SPEC): Similarly.
9311 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
9314 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
9316 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
9318 2017-08-11 Tamar Christina <tamar.christina@arm.com>
9319 * config/aarch64/aarch64.md (mov<mode>): Change.
9320 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9321 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
9322 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
9324 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
9326 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
9327 for storage order barriers.
9329 2017-08-11 Martin Liska <mliska@suse.cz>
9332 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
9333 variables of void type.
9335 2017-08-11 Martin Liska <mliska@suse.cz>
9337 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
9338 TARGET_SUPPORTS_ALIASES.
9339 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
9340 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
9341 (optimize_weakref): Likewise.
9342 * symtab.c (symtab_node::noninterposable_alias): Likewise.
9343 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
9344 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
9346 2017-08-11 Martin Liska <mliska@suse.cz>
9349 * config/i386/i386.c (make_resolver_func): Do complete
9350 refactoring of the function.
9352 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9355 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
9356 * config/i386/i386.c (ix86_stack_protect_guard): Use
9357 ix86_stack_protect_guard_symbol_str to generate varible declaration.
9358 * doc/invoke.texi (x86 Options): Document
9359 -mstack-protector-guard-symbol= option.
9361 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9363 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
9364 * config/i386/i386.c (ix86_split_stack_guard): New function.
9365 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
9366 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
9367 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
9368 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
9369 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
9370 (split_stack_space_check): Call ix86_split_stack_guard.
9372 2017-08-10 Martin Sebor <msebor@redhat.com>
9374 * print-tree.c (print_node): Print location using the established
9376 Replace spaces with colons.
9377 (debug_raw, debug): Ditto.
9379 2017-08-10 Martin Sebor <msebor@redhat.com>
9382 * pretty-print.c (pp_format): Correct the handling of %s precision.
9384 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
9387 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9389 (ix86_finalize_stack_frame_flags): This. Also clear
9390 frame_pointer_needed if -fno-omit-frame-pointer is used without
9392 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9393 with ix86_finalize_stack_frame_flags.
9394 (ix86_expand_epilogue): Likewise.
9395 (ix86_expand_split_stack_prologue): Likewise.
9396 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
9398 2017-08-10 Martin Liska <mliska@suse.cz>
9401 * c-attribs.c (handle_target_attribute):
9402 Report warning for an empty string argument of target attribute.
9404 2017-08-09 Jakub Jelinek <jakub@redhat.com>
9407 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
9409 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
9410 or DECL_NONLOCAL labels.
9411 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
9412 or DECL_NONLOCAL labels here.
9414 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
9416 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
9417 to indicate when early gimple folding has been disabled.
9418 (rs6000_gimple_fold_builtin): Add debug content.
9419 (rs6000_invalid_builtin): Fix whitespace.
9420 (rs6000_expand_builtin): Fix whitespace.
9421 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
9423 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
9426 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
9427 SAVE_MULTIPLE if not all the registers that saves, should be saved.
9429 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
9431 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
9432 (qdf24xx): Likewise.
9433 * config/aarch64/aarch64.md: Include falkor.md.
9434 * config/aarch64/falkor.md: New.
9436 2017-08-09 Marek Polacek <polacek@redhat.com>
9439 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
9440 * diagnostic.c (emit_diagnostic): Add a comment.
9441 (emit_diagnostic_valist): New function.
9443 2017-08-09 Marek Polacek <polacek@redhat.com>
9446 * input.c (make_location): New overload.
9447 * input.h (make_location): Declare.
9449 2017-08-08 Alan Modra <amodra@gmail.com>
9450 H.J. Lu <hongjiu.lu@intel.com>
9453 * gcc.c (NO_PIE_SPEC): Delete.
9454 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
9456 (LINK_PIE_SPEC): ..to here.
9457 (LINK_COMMAND_SPEC): Support -no-pie.
9458 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
9459 chain of crtbegin*.o selection, update for PIE_SPEC changes and
9461 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
9462 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
9463 (ENDFILE_CRTEND_SPEC): Similarly.
9465 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
9468 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
9469 (mstack-protector-guard-offset=): Ditto.
9470 * config/i386/i386.c (ix86_option_override): Handle
9471 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
9473 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
9474 ix86_stack_protect_guard_offset variables.
9475 (TARGET_STACK_PROTECT_GUARD): Always define.
9476 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
9477 and -mstack-protector-guard-offset= options.
9479 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9481 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
9482 boundary case for the last candidate.
9484 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9486 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
9487 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
9489 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9492 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
9493 * config/aarch64/aarch64-builtins.c
9494 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
9495 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
9496 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
9498 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9499 Andrew Pinski <pinskia@gmail.com>
9502 * internal-fn.def (XORSIGN): New.
9503 * optabs.def (xorsign_optab): New.
9504 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
9505 (convert_expand_mult_copysign): New.
9506 (pass_optimize_widening_mul::execute): Call
9507 convert_expand_mult_copysign.
9509 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9511 PR tree-optimization/81354
9512 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
9513 Insert on edges rather than explicitly creating landing pads.
9514 (analyze_candidates_and_replace): Commit edge inserts.
9516 2017-08-08 Richard Biener <rguenther@suse.de>
9519 * tree-ssa-loop-niter.c: Include tree-dfa.h.
9520 (expand_simple_operations): Also look through ADDR_EXPRs with
9521 MEM_REF bases treating them as POINTER_PLUS_EXPR.
9523 2017-08-08 Richard Biener <rguenther@suse.de>
9525 PR tree-optimization/81723
9526 * tree-vect-slp.c (struct bst_traits): New hash traits.
9527 (bst_fail): New global.
9528 (vect_build_slp_tree_2): New worker, split out from ...
9529 (vect_build_slp_tree): ... this now wrapping it with using
9530 bst_fail set to cache SLP tree build fails. Properly handle
9532 (vect_analyze_slp_instance): Allocate and free bst_fail.
9534 2017-08-08 Martin Liska <mliska@suse.cz>
9537 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
9538 LABEL_DECLs that can be from a different function.
9540 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9542 PR tree-optimization/81744
9543 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
9544 loop's number of iterations.
9546 2017-08-08 Martin Liska <mliska@suse.cz>
9548 * asan.c: Include header files.
9549 * attribs.c (build_decl_attribute_variant): New function moved
9551 (build_type_attribute_qual_variant): Likewise.
9552 (cmp_attrib_identifiers): Likewise.
9553 (simple_cst_list_equal): Likewise.
9554 (omp_declare_simd_clauses_equal): Likewise.
9555 (attribute_value_equal): Likewise.
9556 (comp_type_attributes): Likewise.
9557 (build_type_attribute_variant): Likewise.
9558 (lookup_ident_attribute): Likewise.
9559 (remove_attribute): Likewise.
9560 (merge_attributes): Likewise.
9561 (merge_type_attributes): Likewise.
9562 (merge_decl_attributes): Likewise.
9563 (merge_dllimport_decl_attributes): Likewise.
9564 (handle_dll_attribute): Likewise.
9565 (attribute_list_equal): Likewise.
9566 (attribute_list_contained): Likewise.
9567 * attribs.h (lookup_attribute): New function moved from tree.[ch].
9568 (lookup_attribute_by_prefix): Likewise.
9569 * bb-reorder.c: Include header files.
9570 * builtins.c: Likewise.
9571 * calls.c: Likewise.
9572 * cfgexpand.c: Likewise.
9573 * cgraph.c: Likewise.
9574 * cgraphunit.c: Likewise.
9575 * convert.c: Likewise.
9576 * dwarf2out.c: Likewise.
9577 * final.c: Likewise.
9578 * fold-const.c: Likewise.
9579 * function.c: Likewise.
9580 * gimple-expr.c: Likewise.
9581 * gimple-fold.c: Likewise.
9582 * gimple-pretty-print.c: Likewise.
9583 * gimple.c: Likewise.
9584 * gimplify.c: Likewise.
9585 * hsa-common.c: Likewise.
9586 * hsa-gen.c: Likewise.
9587 * internal-fn.c: Likewise.
9588 * ipa-chkp.c: Likewise.
9589 * ipa-cp.c: Likewise.
9590 * ipa-devirt.c: Likewise.
9591 * ipa-fnsummary.c: Likewise.
9592 * ipa-inline.c: Likewise.
9593 * ipa-visibility.c: Likewise.
9595 * lto-cgraph.c: Likewise.
9596 * omp-expand.c: Likewise.
9597 * omp-general.c: Likewise.
9598 * omp-low.c: Likewise.
9599 * omp-offload.c: Likewise.
9600 * omp-simd-clone.c: Likewise.
9601 * opts-global.c: Likewise.
9602 * passes.c: Likewise.
9603 * predict.c: Likewise.
9604 * sancov.c: Likewise.
9605 * sanopt.c: Likewise.
9606 * symtab.c: Likewise.
9607 * toplev.c: Likewise.
9608 * trans-mem.c: Likewise.
9609 * tree-chkp.c: Likewise.
9610 * tree-eh.c: Likewise.
9611 * tree-into-ssa.c: Likewise.
9612 * tree-object-size.c: Likewise.
9613 * tree-parloops.c: Likewise.
9614 * tree-profile.c: Likewise.
9615 * tree-ssa-ccp.c: Likewise.
9616 * tree-ssa-live.c: Likewise.
9617 * tree-ssa-loop.c: Likewise.
9618 * tree-ssa-sccvn.c: Likewise.
9619 * tree-ssa-structalias.c: Likewise.
9620 * tree-ssa.c: Likewise.
9621 * tree-streamer-in.c: Likewise.
9622 * tree-vectorizer.c: Likewise.
9623 * tree-vrp.c: Likewise.
9625 * ubsan.c: Likewise.
9626 * varasm.c: Likewise.
9627 * varpool.c: Likewise.
9628 * tree.c: Remove functions moved to attribs.[ch].
9630 * config/aarch64/aarch64.c: Add attrs.h header file.
9631 * config/alpha/alpha.c: Likewise.
9632 * config/arc/arc.c: Likewise.
9633 * config/arm/arm.c: Likewise.
9634 * config/avr/avr.c: Likewise.
9635 * config/bfin/bfin.c: Likewise.
9636 * config/c6x/c6x.c: Likewise.
9637 * config/cr16/cr16.c: Likewise.
9638 * config/cris/cris.c: Likewise.
9639 * config/darwin.c: Likewise.
9640 * config/epiphany/epiphany.c: Likewise.
9641 * config/fr30/fr30.c: Likewise.
9642 * config/frv/frv.c: Likewise.
9643 * config/ft32/ft32.c: Likewise.
9644 * config/h8300/h8300.c: Likewise.
9645 * config/i386/winnt.c: Likewise.
9646 * config/ia64/ia64.c: Likewise.
9647 * config/iq2000/iq2000.c: Likewise.
9648 * config/lm32/lm32.c: Likewise.
9649 * config/m32c/m32c.c: Likewise.
9650 * config/m32r/m32r.c: Likewise.
9651 * config/m68k/m68k.c: Likewise.
9652 * config/mcore/mcore.c: Likewise.
9653 * config/microblaze/microblaze.c: Likewise.
9654 * config/mips/mips.c: Likewise.
9655 * config/mmix/mmix.c: Likewise.
9656 * config/mn10300/mn10300.c: Likewise.
9657 * config/moxie/moxie.c: Likewise.
9658 * config/msp430/msp430.c: Likewise.
9659 * config/nds32/nds32-isr.c: Likewise.
9660 * config/nds32/nds32.c: Likewise.
9661 * config/nios2/nios2.c: Likewise.
9662 * config/nvptx/nvptx.c: Likewise.
9663 * config/pa/pa.c: Likewise.
9664 * config/pdp11/pdp11.c: Likewise.
9665 * config/powerpcspe/powerpcspe.c: Likewise.
9666 * config/riscv/riscv.c: Likewise.
9667 * config/rl78/rl78.c: Likewise.
9668 * config/rx/rx.c: Likewise.
9669 * config/s390/s390.c: Likewise.
9670 * config/sh/sh.c: Likewise.
9671 * config/sol2.c: Likewise.
9672 * config/sparc/sparc.c: Likewise.
9673 * config/spu/spu.c: Likewise.
9674 * config/stormy16/stormy16.c: Likewise.
9675 * config/tilegx/tilegx.c: Likewise.
9676 * config/tilepro/tilepro.c: Likewise.
9677 * config/v850/v850.c: Likewise.
9678 * config/vax/vax.c: Likewise.
9679 * config/visium/visium.c: Likewise.
9680 * config/xtensa/xtensa.c: Likewise.
9682 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9685 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
9686 constraints since the -mupper-regs-* switches have been
9688 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
9689 into a vector from a double word element that was extracted from
9690 another vector, and eliminate extra XXPERMDI instructions.
9691 (vsx_concat_<mode>_2): Likewise.
9692 (vsx_concat_<mode>_3): Likewise.
9693 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
9694 concat to allow optimizing inserts from previous extracts.
9696 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
9698 * config/i386/i386.c (ix86_stack_protect_guard): Generate
9699 memory reference to a SSP offset in TLS address space.
9700 (ix86_print_operand) <case '@'>: Remove.
9701 (ix86_print_operand_punct_valid_p): Remove '@' code.
9702 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
9704 (stack_tls_protect_set_<mode>): Remove.
9705 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
9706 (stack_tls_protect_test_<mode>): Remove.
9707 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
9709 2017-08-07 Olivier Hainque <hainque@adacore.com>
9712 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
9714 2017-08-07 Douglas Rupp <rupp@adacore.com>
9716 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
9717 variable was referenced as multidir in command.
9719 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9722 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
9725 2017-08-07 Martin Liska <mliska@suse.cz>
9727 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
9728 * config/rl78/rl78.c: Add include of attribs.h.
9729 * config/sh/sh.c: Likewise.
9730 * config/v850/v850.c: Likewise.
9732 2017-08-07 Tom de Vries <tom@codesourcery.com>
9735 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
9737 2017-08-07 Martin Liska <mliska@suse.cz>
9739 * config/mips/mips.c: Include attribs.h.
9741 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9744 * doc/invoke.texi: Document change in behvaior for -Ofast for
9747 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
9749 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
9751 (aarch64_pop_regs): Likewise.
9752 (aarch64_gen_load_pair): Likewise.
9753 (aarch64_save_callee_saves): Likewise.
9754 (aarch64_restore_callee_saves): Likewise.
9756 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9758 * config/i386/i386.c: Revert the last change.
9760 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9763 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9765 (ix86_finalize_stack_frame_flags): This. Also clear
9766 frame_pointer_needed if -fno-omit-frame-pointer is used without
9768 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9769 with ix86_finalize_stack_frame_flags.
9770 (ix86_expand_epilogue): Likewise.
9771 (ix86_expand_split_stack_prologue): Likewise.
9773 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9776 * config/i386/i386.c (get_builtin_code_for_version): Set priority
9777 to P_AES for Westmere.
9779 2017-08-07 Jonathan Yong <10walls@gmail.com>
9781 * config/i386/mingw.opt (fset-stack-executable): Removed.
9782 * config/i386/cygming.opt (fset-stack-executable): Moved
9784 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
9786 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
9788 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
9790 2017-08-07 Marek Polacek <polacek@redhat.com>
9793 * fold-const.c (fold_indirect_ref_1): Check type_domain.
9795 2017-08-07 Martin Liska <mliska@suse.cz>
9797 * attribs.h (canonicalize_attr_name): New function.
9798 (cmp_attribs): Move from c-format.c and adjusted.
9799 (is_attribute_p): Moved from tree.h.
9800 * tree-inline.c: Add new includes.
9801 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
9802 (private_is_attribute_p): Remove.
9803 (private_lookup_attribute): Likewise.
9804 (private_lookup_attribute_by_prefix): Simplify.
9805 (remove_attribute): Use is_attribute_p.
9806 * tree.h: Remove removed declarations.
9808 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9811 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
9812 instead of computing it in the function. Formatting fix.
9813 (expand_case): Don't rely on default_edge being the first edge,
9814 clear it if removing it, pass default_edge to
9815 emit_case_dispatch_table.
9816 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
9819 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
9821 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
9822 insn in the function, emit NOP after the insn.
9824 2017-08-06 Tom de Vries <tom@codesourcery.com>
9826 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
9829 2017-08-06 Tom de Vries <tom@codesourcery.com>
9831 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
9834 2017-08-04 Yury Gribov <tetra2005@gmail.com>
9836 PR tree-optimization/57371
9837 * match.pd: New pattern.
9839 2017-08-04 Marek Polacek <polacek@redhat.com>
9842 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
9843 perform the computation in offset_int.
9845 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9847 PR tree-optimization/81136
9848 * tree-vectorizer.h: Include tree-hash-traits.h.
9849 (vec_base_alignments): New typedef.
9850 (vec_info): Add a base_alignments field.
9851 (vect_record_base_alignments): Declare.
9852 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
9854 (DR_IS_CONDITIONAL_IN_STMT): New macro.
9855 (create_data_ref): Add an is_conditional_in_stmt argument.
9856 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
9857 the is_conditional_in_stmt field.
9858 (data_ref_loc): Add an is_conditional_in_stmt field.
9859 (get_references_in_stmt): Set the is_conditional_in_stmt field.
9860 (find_data_references_in_stmt): Update call to create_data_ref.
9861 (graphite_find_data_references_in_stmt): Likewise.
9862 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
9863 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
9864 (vect_record_base_alignment): New function.
9865 (vect_record_base_alignments): Likewise.
9866 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
9867 for nested statements even if we fail to compute a misalignment.
9868 Use pooled base alignments for unconditional references.
9869 (vect_find_same_alignment_drs): Compare base addresses instead
9871 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
9872 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
9874 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9876 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
9877 Add an explicit name for the enum. Use auto_vec for slp_instances
9879 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
9881 (_bb_vec_info): Add a constructor and destructor.
9882 (vinfo_for_stmt): Return NULL for uids of -1 as well.
9883 (destroy_loop_vec_info): Delete.
9884 (vect_destroy_datarefs): Likewise.
9885 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
9886 (vec_info::vec_info): New function.
9887 (vec_info::~vec_info): Likewise.
9888 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
9889 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
9890 destroy_loop_vec_info.
9891 * tree-vect-loop.c (new_loop_vec_info): Replace with...
9892 (_loop_vec_info::_loop_vec_info): ...this.
9893 (destroy_loop_vec_info): Replace with...
9894 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
9895 the stmt_vec_infos. Leave handling of vec_info information to its
9896 destructor. Remove explicit vector releases.
9897 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
9898 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
9899 * tree-vect-slp.c (new_bb_vec_info): Replace with...
9900 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
9901 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
9902 (destroy_bb_vec_info): Replace with...
9903 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
9904 information to its destructor.
9905 (vect_slp_analyze_bb_1): Use new and delete instead of
9906 new_bb_vec_info and destroy_bb_vec_info.
9907 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
9910 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9912 * tree-data-ref.h (subscript): Add access_fn field.
9913 (data_dependence_relation): Add could_be_independent_p.
9914 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
9915 (same_access_functions): Move to tree-data-ref.c.
9916 * tree-data-ref.c (ref_contains_union_access_p): New function.
9917 (access_fn_component_p): Likewise.
9918 (access_fn_components_comparable_p): Likewise.
9919 (dr_analyze_indices): Add a reference to access_fn_component_p.
9920 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
9922 (constant_access_functions): Likewise.
9923 (add_other_self_distances): Likewise.
9924 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
9925 (initialize_data_dependence_relation): Use XCNEW and remove
9926 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
9927 of access functions that have the same type. Allow the
9928 subsequence to end with different bases in some circumstances.
9929 Record the chosen access functions in SUB_ACCESS_FN.
9930 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
9931 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
9932 (subscript_dependence_tester_1): Likewise dra and drb.
9933 (build_classic_dist_vector): Update calls accordingly.
9934 (subscript_dependence_tester): Likewise.
9935 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
9936 DDR_COULD_BE_INDEPENDENT_P.
9937 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
9938 comp_alias_ddrs instead of may_alias_ddrs.
9939 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9941 (vect_analyze_data_ref_dependence): Use it if
9942 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
9943 distance vectors if that fails.
9944 (dependence_distance_ge_vf): New function.
9945 (vect_prune_runtime_alias_test_list): Use it. Don't clear
9946 LOOP_VINFO_MAY_ALIAS_DDRS.
9948 2017-08-04 Richard Biener <rguenther@suse.de>
9951 * fold-const.c (fold_binary_loc): Properly restrict
9952 minus_var0 && minus_var1 case when associating undefined overflow
9955 2017-08-04 Tom de Vries <tom@codesourcery.com>
9957 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
9959 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9961 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9962 Don't start diagnostic messages with a capital letter.
9963 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9965 (rs6000_invalid_builtin): Likewise.
9966 (rs6000_trampoline_init): Likewise.
9968 2017-08-03 Jakub Jelinek <jakub@redhat.com>
9971 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
9972 after setting changeable df flags.
9974 2017-08-03 Richard Biener <rguenther@suse.de>
9976 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
9977 up if the use is in USE - X.
9979 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
9981 * toplev.c (dumpfile.h): New include.
9982 (internal_error_reentered): New static function. Use it...
9983 (internal_error_function): ...here to handle reentered internal_error.
9985 2017-08-03 Richard Biener <rguenther@suse.de>
9988 * fold-const.c (split_tree): Add minus_var and minus_con
9989 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
9990 here but always use minus_*.
9991 (associate_trees): Assert we never associate with MINUS_EXPR
9992 and NULL first operand. Do not recurse for PLUS_EXPR operands
9993 when associating as MINUS_EXPR either.
9994 (fold_binary_loc): Track minus_var and minus_con.
9996 2017-08-03 Tom de Vries <tom@codesourcery.com>
9999 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10000 ACCEL_COMPILER, apply finish_options on
10001 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10003 2017-08-03 Tom de Vries <tom@codesourcery.com>
10006 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10007 function_entry_patch_area_size > 0.
10009 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10012 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10013 instead of 10??LU, perform unit multiplication in wide_int,
10014 don't change alloc_object_size_limit if the limit is larger
10017 PR tree-optimization/81655
10018 PR tree-optimization/81588
10019 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10020 the case when ranges[i].low and high are 1 for unsigned type with
10023 PR middle-end/81052
10024 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10025 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10027 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10029 * tree-vrp.h: Add include guard.
10031 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10034 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10035 (ud2): New insn pattern.
10036 * config/i386/i386.c (ix86_expand_epilogue):
10037 For naked functions, generate ud2 instead of trap insn.
10039 2017-08-02 Marek Polacek <polacek@redhat.com>
10042 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10044 2017-08-02 Tom de Vries <tom@codesourcery.com>
10045 Cesar Philippidis <cesar@codesourcery.com>
10047 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10048 Add missing edge probabilities.
10050 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10052 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10053 Correct endianness.
10055 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10057 PR middle-end/79499
10058 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10059 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10060 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10061 prologue_seq sequences - if any.
10063 2017-08-02 Richard Biener <rguenther@suse.de>
10065 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10066 via vectors if supported, integer extracts via punning if supported
10067 or otherwise vector extracts.
10069 2017-08-02 Richard Biener <rguenther@suse.de>
10071 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10073 (bitmap_insert_into_set): ... this.
10075 2017-08-02 Richard Biener <rguenther@suse.de>
10077 PR tree-optimization/81633
10079 2015-08-17 Alan Hayward <alan.hayward@arm.com>
10081 PR tree-optimization/71752
10082 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
10084 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
10086 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
10087 (machine_function::call_ms2sysv_pad_out): Remove field.
10088 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
10089 (ix86_compute_frame_layout): Likewise.
10091 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
10094 * config/i386/i386.c (ix86_set_func_type): Disallow naked
10095 attribute with interrupt attribute.
10097 2017-08-01 Andrew Pinski <apinski@cavium.com>
10099 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
10100 BIT_INSERT_EXPR's operand 1
10101 to see if the types precision matches.
10103 2017-08-01 Martin Liska <mliska@suse.cz>
10105 PR middle-end/70140
10106 * builtins.c (expand_builtin_memcpy_args): Remove.
10107 (expand_builtin_memcpy): Call newly added function
10108 expand_builtin_memory_copy_args.
10109 (expand_builtin_memcpy_with_bounds): Likewise.
10110 (expand_builtin_mempcpy): Remove last argument.
10111 (expand_builtin_mempcpy_with_bounds): Likewise.
10112 (expand_builtin_memory_copy_args): New function created from
10113 expand_builtin_mempcpy_args with small modifications.
10114 (expand_builtin_mempcpy_args): Remove.
10115 (expand_builtin_stpcpy): Remove unused argument.
10116 (expand_builtin): Likewise.
10117 (expand_builtin_with_bounds): Likewise.
10119 2017-08-01 Martin Liska <mliska@suse.cz>
10122 Make mempcpy more optimal (PR middle-end/70140).
10124 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10127 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10128 __builtin_vec_cmpne verify both arguments are compatible vectors
10129 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
10130 verify arg1_type is a pointer or array type. For __builtin_vec_st,
10131 move computation of aligned to after checking the argument types.
10135 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10136 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
10139 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
10140 Doug Rupp <rupp@adacore.com>
10141 Olivier Hainque <hainque@adacore.com>
10143 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
10144 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
10146 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
10147 for TARGET_OS_CPP_BUILTIN.
10148 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
10149 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
10151 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
10152 passing required abi options to the assembler for EABI configurations.
10153 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
10154 of .text.hot and .text.unlikely sections for kernel modules when
10155 using ARM style exceptions.
10156 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
10157 options. Add EXTRA_CC1_SPEC.
10158 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
10160 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
10162 (ARM_TARGET2_DWARF_FORMAT): Define.
10163 * config/arm/t-vxworks: Adjust multilib control to removal of the
10164 Diab command line options.
10166 2017-08-01 Martin Liska <mliska@suse.cz>
10168 PR gcov-profile/81561
10169 * gcov.c (unblock): Make unblocking safe as we need to preserve
10170 index correspondence of blocks and block_lists.
10172 2017-08-01 Richard Biener <rguenther@suse.de>
10174 PR tree-optimization/81181
10175 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10176 (compute_antic): ... end of iteration here.
10178 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
10180 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
10181 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
10182 (ftree-slp-vectorize): Likewise.
10183 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
10184 can no longer be set independent of flag_tree_loop_vectorize.
10185 * omp-general.c (emp_max_vf): Likewise.
10186 * opts.c (enable_fdo_optimizations): Remove references to
10187 flag_tree_vectorize, these are now implicit.
10188 (common_handle_option): Remove handling for OPT_ftree_vectorize,
10189 and leave it for the options machinery.
10191 2017-08-01 Martin Liska <mliska@suse.cz>
10193 PR middle-end/70140
10194 * builtins.c (expand_builtin_memcpy_args): Remove.
10195 (expand_builtin_memcpy): Call newly added function
10196 expand_builtin_memory_copy_args.
10197 (expand_builtin_memcpy_with_bounds): Likewise.
10198 (expand_builtin_mempcpy): Remove last argument.
10199 (expand_builtin_mempcpy_with_bounds): Likewise.
10200 (expand_builtin_memory_copy_args): New function created from
10201 expand_builtin_mempcpy_args with small modifications.
10202 (expand_builtin_mempcpy_args): Remove.
10203 (expand_builtin_stpcpy): Remove unused argument.
10204 (expand_builtin): Likewise.
10205 (expand_builtin_with_bounds): Likewise.
10207 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10210 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
10211 print "ds:" only for immediates in generic address space.
10213 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10216 * config/i386/i386.c (ix86_funciton_naked): New prototype.
10217 (ix86_function_ok_for_sibcall): Return false for naked functions.
10219 2017-08-01 Richard Biener <rguenther@suse.de>
10221 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
10222 (compute_antic): Seed worklist with exit block predecessors.
10223 * cfganal.c (dfs_find_deadend): For a cycle return the source
10224 of the edge closing it.
10226 2017-08-01 Tamar Christina <tamar.christina@arm.com>
10228 * config/aarch64/aarch64.c
10229 (aarch64_can_const_movi_rtx_p): Move 0 check.
10231 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10233 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
10234 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
10236 * match.pd: Ditto in address comparison pattern.
10238 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10240 PR tree-optimization/81627
10241 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
10242 closed ssa form for store-store chain.
10244 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10246 PR tree-optimization/81620
10247 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
10248 for store-store chain.
10250 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10252 PR tree-optimization/81588
10253 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
10254 ranges[i].in_p, invert comparison code ccode. For >/>=,
10255 swap rhs1 and rhs2 and comparison code unconditionally,
10256 for </<= don't do that. Don't swap rhs1/rhs2 again if
10257 ranges[i].in_p, instead invert comparison code ccode if
10258 opcode or oe->rank is BIT_IOR_EXPR.
10261 * optabs.def (vec_extract_optab, vec_init_optab): Change from
10262 a direct optab to conversion optab.
10263 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
10264 with GET_MODE_INNER as last argument instead of optab_handler.
10265 * expmed.c (extract_bit_field_1): Likewise. Use vector from
10266 vector extraction if possible and optab is available.
10267 * expr.c (store_constructor): Use convert_optab_handler instead
10268 of optab_handler. Use vector initialization from smaller
10269 vectors if possible and optab is available.
10270 * tree-vect-stmts.c (vectorizable_load): Likewise.
10271 * doc/md.texi (vec_extract, vec_init): Document that the optabs
10272 now have two modes.
10273 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
10274 of vec_init from half-sized vectors with the same element mode.
10275 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
10276 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
10277 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
10278 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
10279 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
10280 after mode in gen_vec_extract* calls.
10281 (vec_extract<mode>): Renamed to ...
10282 (vec_extract<mode><ssescalarmodelower>): ... this.
10283 (vec_extract<mode><ssehalfvecmodelower>): New expander.
10284 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
10285 element mode after mode in gen_vec_init* calls.
10286 (VEC_INIT_HALF_MODE): New mode iterator.
10287 (vec_init<mode>): Renamed to ...
10288 (vec_init<mode><ssescalarmodelower>): ... this.
10289 (vec_init<mode><ssehalfvecmodelower>): New expander.
10290 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
10291 (vec_extractv2sfsf): ... this.
10292 (vec_initv2sf): Renamed to ...
10293 (vec_initv2sfsf): ... this.
10294 (vec_extractv2si): Renamed to ...
10295 (vec_extractv2sisi): ... this.
10296 (vec_initv2si): Renamed to ...
10297 (vec_initv2sisi): ... this.
10298 (vec_extractv4hi): Renamed to ...
10299 (vec_extractv4hihi): ... this.
10300 (vec_initv4hi): Renamed to ...
10301 (vec_initv4hihi): ... this.
10302 (vec_extractv8qi): Renamed to ...
10303 (vec_extractv8qiqi): ... this.
10304 (vec_initv8qi): Renamed to ...
10305 (vec_initv8qiqi): ... this.
10306 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
10307 (vec_init<mode>): Renamed to ...
10308 (vec_init<mode><VEC_base_l>): ... this.
10309 (vec_extract<mode>): Renamed to ...
10310 (vec_extract<mode><VEC_base_l>): ... this.
10311 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
10312 (vec_initv2sfsf): ... this.
10313 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
10314 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10315 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
10316 element mode after mode in gen_vec_init* calls.
10317 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
10318 (vec_init<mode><Vel>): ... this.
10319 (vec_extract<mode>): Renamed to ...
10320 (vec_extract<mode><Vel>): ... this.
10321 * config/aarch64/iterators.md (Vel): New mode attribute.
10322 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
10323 Add element mode after mode in gen_vec_extract* calls.
10324 * config/s390/vector.md (non_vec_l): New mode attribute.
10325 (vec_extract<mode>): Renamed to ...
10326 (vec_extract<mode><non_vec_l>): ... this.
10327 (vec_init<mode>): Renamed to ...
10328 (vec_init<mode><non_vec_l>): ... this.
10329 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
10330 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
10332 * config/arm/iterators.md (V_elem_l): New mode attribute.
10333 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
10334 (vec_extract<mode><V_elem_l>): ... this.
10335 (vec_extractv2di): Renamed to ...
10336 (vec_extractv2didi): ... this.
10337 (vec_init<mode>): Renamed to ...
10338 (vec_init<mode><V_elem_l>): ... this.
10339 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
10340 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
10341 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
10342 Add element mode after gen_vec_extract* calls.
10343 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
10344 (vec_init<mode><unitmode>): ... this.
10345 (vec_extract<mode>): Renamed to ...
10346 (vec_extract<mode><unitmode>): ... this.
10347 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
10348 (vec_init<mode><unitmode>): ... this.
10349 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
10350 (vec_initv2sfsf): ... this.
10351 (vec_extractv2sf): Renamed to ...
10352 (vec_extractv2sfsf): ... this.
10353 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
10354 Add element mode after gen_vec_extract* calls.
10355 * config/mips/mips.md (unitmode): New mode iterator.
10356 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
10357 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
10358 * config/spu/spu.md (inner_l): New mode attribute.
10359 (vec_init<mode>): Renamed to ...
10360 (vec_init<mode><inner_l>): ... this.
10361 (vec_extract<mode>): Renamed to ...
10362 (vec_extract<mode><inner_l>): ... this.
10363 * config/sparc/sparc.md (veltmode): New mode iterator.
10364 (vec_init<VMALL:mode>): Renamed to ...
10365 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
10366 * config/ia64/vect.md (vec_initv2si): Renamed to ...
10367 (vec_initv2sisi): ... this.
10368 (vec_initv2sf): Renamed to ...
10369 (vec_initv2sfsf): ... this.
10370 (vec_extractv2sf): Renamed to ...
10371 (vec_extractv2sfsf): ... this.
10372 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
10373 (vec_init<mode>): Renamed to ...
10374 (vec_init<mode><VEC_base_l>): ... this.
10375 (vec_extract<mode>): Renamed to ...
10376 (vec_extract<mode><VEC_base_l>): ... this.
10377 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
10378 (vec_initv2sfsf): ... this.
10379 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
10380 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10381 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
10382 gen_vec_init* calls.
10384 2017-08-01 Richard Biener <rguenther@suse.de>
10386 PR tree-optimization/81297
10387 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
10388 TREE_OVERFLOW from INTEGER_CSTs.
10390 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
10392 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
10394 2017-07-31 Carl Love <cel@us.ibm.com>
10396 * config/rs6000/rs6000-c: Add support for built-in functions
10397 vector signed char vec_xl_be (signed long long, signed char *);
10398 vector unsigned char vec_xl_be (signed long long, unsigned char *);
10399 vector signed int vec_xl_be (signed long long, signed int *);
10400 vector unsigned int vec_xl_be (signed long long, unsigned int *);
10401 vector signed long long vec_xl_be (signed long long, signed long long *);
10402 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
10403 vector signed short vec_xl_be (signed long long, signed short *);
10404 vector unsigned short vec_xl_be (signed long long, unsigned short *);
10405 vector double vec_xl_be (signed long long, double *);
10406 vector float vec_xl_be (signed long long, float *);
10407 * config/rs6000/altivec.h (vec_xl_be): Add #define.
10408 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
10409 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
10411 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
10412 (altivec_expand_builtin): Add switch statement to call
10413 altivec_expand_xl_be for each builtin.
10414 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
10415 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
10416 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
10417 __builtin_vsx_le_be_v16qi.
10418 * doc/extend.texi: Update the built-in documentation file for the
10419 new built-in functions.
10421 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10424 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
10426 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10428 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10430 * config.gcc: Add z14.
10431 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
10432 CPU model numbers for z13s and z14.
10433 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
10435 * config/s390/s390-opts.h (enum processor_type): Rename
10436 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10437 * config/s390/s390.c (processor_table): Add field for CPU name to
10438 be passed to Binutils.
10439 (s390_asm_output_machine_for_arch): Use the new field in
10440 processor_table for Binutils.
10441 (s390_expand_builtin): Replace arch12 with z14.
10442 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10443 (s390_get_sched_attrmask): Likewise.
10444 (s390_get_unit_mask): Likewise.
10445 * config/s390/s390.opt: Add z14 to processor_type enum.
10447 2017-07-31 Martin Jambor <mjambor@suse.cz>
10450 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
10451 regardless of optimization level.
10453 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
10454 Martin Liska <mliska@suse.cz>
10456 * predict.def: Remove old comment and adjust probability.
10457 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
10458 PREDICT statements.
10460 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10463 * config/i386/i386.c (ix86_function_naked): New function.
10464 (ix86_can_use_return_insn_p): Return false for naked functions.
10465 (ix86_expand_prologue): Skip prologue for naked functions.
10466 (ix86_expand_epilogue): Skip epilogue for naked functions
10467 and emit trap instruction.
10468 (ix86_warn_func_return): New function.
10469 (ix86_attribute_table): Add "naked" attribute specification.
10470 (TARGET_WARN_FUNC_RETURN): Define.
10471 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
10473 2017-07-31 Martin Liska <mliska@suse.cz>
10475 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
10476 (dump_gimple_bb_header): Always dump BB info.
10477 (pp_cfg_jump): Do not append info about BB when dumping a jump.
10479 2017-07-31 Martin Liska <mliska@suse.cz>
10482 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
10483 also with current_function_decl non-null equality.
10485 2017-07-31 Jakub Jelinek <jakub@redhat.com>
10488 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
10489 change type to the element type, instead add eltype variable and
10490 use it where we are interested in the element type.
10492 PR tree-optimization/81603
10493 * ipa-polymorphic-call.c
10494 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
10495 offset arithmetic in offset_int, bail out if the resulting bit offset
10496 doesn't fit into shwi.
10498 2017-07-31 Martin Liska <mliska@suse.cz>
10500 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
10501 (gimplify_save_expr): Fix comment.
10503 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
10506 * config/i386/i386.c (ix86_function_arg): Update arguments for
10508 (ix86_compute_frame_layout): Set the initial stack offset to
10509 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
10510 INCOMING_FRAME_SP_OFFSET.
10511 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
10512 stack before exception handler returns.
10513 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
10514 the 'ERROR_CODE' for exception handler.
10516 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
10518 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
10519 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
10520 (ASM_OUTPUT_REG_POP): Ditto.
10521 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
10522 instead of asm_fprintf to output pure string.
10524 2017-07-29 Jakub Jelinek <jakub@redhat.com>
10526 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
10527 to imported_module_or_decl hook.
10528 (debug_nothing_tree_tree_tree_bool): Remove.
10529 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
10530 * debug.c (do_nothing_debug_hooks): Use
10531 debug_nothing_tree_tree_tree_bool_bool instead of
10532 debug_nothing_tree_tree_tree_bool.
10533 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10534 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
10535 * sdbout.c (sdb_debug_hooks): Likewise.
10536 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
10537 (gen_namespace_die): Add DW_AT_export_symbols attribute if
10539 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
10540 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
10541 attribute, don't add anything.
10543 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10545 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
10546 (fold_build2_stat_loc): Likewise.
10547 (fold_build3_stat_loc): Likewise.
10548 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
10549 (fold_build1_loc): Remove macro.
10550 (fold_build2_loc): Likewise.
10551 (fold_build3_loc): Likewise.
10553 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10555 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
10556 (gimple_build_debug_bind_source_stat): Likewise.
10557 * gimple.h (gimple_build_debug_bind): Remove macro.
10558 (gimple_build_debug_bind_source): Likewise.
10560 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10562 * bitmap.c (bitmap_alloc): Adjust.
10563 (bitmap_gc_alloc): Likewise.
10564 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
10566 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10568 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
10569 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
10570 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
10571 (bitmap_gc_alloc_stat): Likewise.
10572 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
10574 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10576 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
10577 * rtl.h (shallow_copy_rtx): Remove macro.
10579 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10581 * emit-rtl.c (gen_raw_REG): Adjust.
10582 * gengenrtl.c (gendef): Likewise.
10583 * rtl.c (rtx_alloc_stat): Remove _stat from name.
10584 * rtl.h (rtx_alloc): Remove macro.
10586 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10588 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
10589 (build_tree_list_stat): Likewise.
10590 * tree.h (build_tree_list): Remove macro.
10591 (build_tree_list_vec): Likewise.
10593 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10595 * tree.c (make_vector_stat): Remove _stat from name.
10596 (build_vector_stat): Likewise.
10597 * tree.h (make_vector_stat): Remove macro.
10598 (build_vector_stat): Likewise.
10600 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10602 * tree.h (build_var_debug_value): Remove prototype.
10604 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10606 * tree.c (tree_cons_stat): Remove _stat from name.
10607 * tree.h (tree_cons): Remove macro.
10609 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10611 * tree.c (build_vl_exp_stat): Remove _stat from name.
10612 * tree.h (build_vl_exp): Remove macro.
10614 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10616 * tree.c (build_decl_stat): Remove _stat from name.
10617 * tree.h (build_decl): Remove macro.
10619 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10621 * gimple.c (gimple_build_with_ops_stat): Adjust.
10622 (gimple_alloc_stat): Remove _stat from name.
10623 * gimple.h (gimple_alloc): Remove macro.
10625 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10627 * tree.c (make_tree_vec_stat): Remove _stat from name.
10628 (grow_tree_vec_stat): Likewise.
10629 * tree.h (make_tree_vec_stat): Adjust prototype.
10630 (grow_tree_vec_stat): Likewise.
10631 (make_tree_vec): Remove macro.
10632 (grow_tree_vec): Likewise.
10634 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10636 * fold-const.c (fold_build1_stat_loc): Adjust.
10637 (fold_build2_stat_loc): Likewise.
10638 (fold_build3_stat_loc): Likewise.
10639 * tree.c (build0_stat): Remove _stat from name.
10640 (build1_stat): Likewise.
10641 (build2_stat): Likewise.
10642 (build3_stat): Likewise.
10643 (build4_stat): Likewise.
10644 (build5_stat): Likewise.
10645 * tree.h (build1_loc): Remove macro, and rename _stat function
10647 (build2_loc): Likewise.
10648 (build3_loc): Likewise.
10649 (build4_loc): Likewise.
10650 (build5_loc): Likewise.
10652 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10654 * tree.c (make_int_cst_stat): Remove _stat from name.
10655 * tree.h (make_int_cst_stat): Adjust prototype.
10656 (make_int_cst): Remove macro.
10658 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10660 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10661 * tree.h (make_tree_binfo_stat): Adjust prototype.
10662 (make_tree_binfo): Remove.
10664 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10666 * tree.c (copy_node_stat): Rename to copy_node.
10667 (build_distinct_type_copy): Adjust.
10668 * tree.h (copy_node_stat): Adjust prototype.
10669 (copy_node): Remove macro.
10671 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10673 * tree.c (make_node_stat): rename to make_node.
10674 (build_tree_list_stat): Adjust.
10675 (build0_stat): Likewise.
10676 (build2_stat): Likewise.
10677 (build3_stat): Likewise.
10678 (build4_stat): Likewise.
10679 (build5_stat): Likewise.
10680 (build_decl_stat): Likewise.
10681 * tree.h (make_node_stat): Adjust prototype.
10682 (make_node): remove macro.
10684 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
10686 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
10687 (PPC_FEATURE2_SCV): Likewise.
10688 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
10690 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10692 * config/aarch64/aarch64.c
10693 (aarch64_internal_mov_immediate): Add new special pattern.
10694 * config/aarch64/aarch64.md (*movdi_aarch64):
10695 Add reg/32bit const mov case.
10697 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10698 Richard Sandiford <richard.sandiford@linaro.org>
10700 * config/aarch64/aarch64.md (mov<mode>): Generalize.
10701 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10702 Add integer and movi cases.
10703 (movi-split-hf-df-sf split, fp16): New.
10704 (enabled): Added TARGET_FP_F16INST.
10705 * config/aarch64/iterators.md (GPF_HF): New.
10706 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
10708 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10710 * config/aarch64/aarch64.c
10711 (aarch64_simd_container_mode): Add prototype.
10712 (aarch64_expand_mov_immediate): Add HI support.
10713 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
10714 (aarch64_can_const_movi_rtx_p): New.
10715 (aarch64_preferred_reload_class):
10716 Remove restrictions of using FP registers for certain SIMD operations.
10717 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
10718 (aarch64_valid_floating_const): Add integer move validation.
10719 (aarch64_simd_imm_scalar_p): Remove.
10720 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
10721 (aarch64_legitimate_constant_p): Expand list of supported cases.
10722 * config/aarch64/aarch64-protos.h
10723 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
10724 (aarch64_reinterpret_float_as_int): New.
10725 (aarch64_simd_imm_scalar_p): Remove.
10726 * config/aarch64/constraints.md (Uvi): New.
10727 (Dd): Split into Ds and new Dd.
10728 * config/aarch64/aarch64.md (*movsi_aarch64):
10730 (*movdi_aarch64): Add SIMD mov case.
10732 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10734 * tree-predcom.c: (struct chain): Handle store-store chain in which
10735 stores for elimination only store loop invariant values.
10736 (execute_pred_commoning_chain): Ditto.
10737 (prepare_initializers_chain_store_elim): Ditto.
10738 (prepare_finalizers): Ditto.
10739 (is_inv_store_elimination_chain): New function.
10740 (initialize_root_vars_store_elim_1): New function.
10742 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10744 * tree-predcom.c: Revise general description of the pass.
10745 (enum chain_type): New enum type for store elimination.
10746 (struct chain): New field supporting store elimination.
10747 (struct component): Ditto.
10748 (dump_chain): Dump store-stores chain.
10749 (release_chain): Release resources.
10750 (split_data_refs_to_components): Compute and create component
10751 contains only stores for elimination.
10752 (get_chain_last_ref_at): New function.
10753 (make_invariant_chain): Initialization.
10754 (make_rooted_chain): Specify chain type in parameter and record it.
10755 (add_looparound_copies): Skip for store-stores chain.
10756 (determine_roots_comp): Compute type of chain and pass it to
10758 (initialize_root_vars_store_elim_2): New function.
10759 (finalize_eliminated_stores): New function.
10760 (remove_stmt): Handle store for elimination.
10761 (execute_pred_commoning_chain): Execute predictive commoning on
10762 store-store chains.
10763 (determine_unroll_factor): Skip unroll for store-stores chain.
10764 (prepare_initializers_chain_store_elim): New function.
10765 (prepare_initializers_chain): Hanlde store-store chain.
10766 (prepare_finalizers_chain, prepare_finalizers): New function.
10767 (tree_predictive_commoning_loop): Return integer value indicating
10768 if loop is unrolled or lcssa form is corrupted.
10769 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
10771 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10773 * tree-predcom.c (initialize_root): Delete.
10774 (execute_pred_commoning_chain): Initialize root vars and replace
10775 reference of non-combined chain directly, rather than call above
10778 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10780 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
10781 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
10783 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10785 * tree-predcom.c (struct chain): New field init_seq.
10786 (release_chain): Release init_seq.
10787 (prepare_initializers_chain): Record intialization stmts in above
10789 (insert_init_seqs): New function.
10790 (tree_predictive_commoning_loop): Call insert_init_seqs.
10792 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10794 * tree-predcom.c (determine_roots_comp): Skip trivial components.
10796 2017-07-28 Richard Biener <rguenther@suse.de>
10798 * match.pd: Remove superfluous :c.
10799 * genmatch.c (simplify::id): Add member.
10800 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
10802 (current_id): New global.
10803 (dt_node::parent): Move from ...
10804 (dt_operand::parent): ... here. Add for_id member.
10805 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
10806 (decision_tree::find_node): Relax order requirement when
10807 merging DT_TRUE nodes to ones inbetween the current simplify
10808 and the one we try to merge with. Add diagnostic whenever
10809 we need to enforce pattern order by not merging.
10810 (decision_tree::insert): Set current_id.
10811 (decision_tree::print_node): Dump parent node and for_id.
10812 (parser::last_id): Add member.
10813 (parser::push_simplify): Assign unique id.
10814 (parser::parser): Initialize last_id.
10816 2017-07-28 Martin Liska <mliska@suse.cz>
10819 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
10820 gimple_build_debug_bind.
10822 2017-07-28 Richard Biener <rguenther@suse.de>
10824 PR tree-optimization/81502
10825 * match.pd: Add pattern combining BIT_INSERT_EXPR with
10827 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
10829 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
10830 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
10831 for BIT_FIELD_REF args.
10832 * fold-const.c (make_bit_field_ref): Likewise.
10833 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
10835 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10838 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
10839 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
10840 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
10841 Or it into SANITIZER_UNDEFINED.
10842 * ubsan.c: Include gimple-fold.h and varasm.h.
10843 (ubsan_expand_ptr_ifn): New function.
10844 (instrument_pointer_overflow): New function.
10845 (maybe_instrument_pointer_overflow): New function.
10846 (instrument_object_size): Formatting fix.
10847 (pass_ubsan::execute): Call instrument_pointer_overflow
10848 and maybe_instrument_pointer_overflow.
10849 * internal-fn.c (expand_UBSAN_PTR): New function.
10850 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
10851 * sanitizer.def (__ubsan_handle_pointer_overflow,
10852 __ubsan_handle_pointer_overflow_abort): New builtins.
10853 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
10854 * internal-fn.def (UBSAN_PTR): New internal function.
10855 * opts.c (sanitizer_opts): Add pointer-overflow.
10856 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
10857 * fold-const.c (build_range_check): Compute pointer range check in
10858 integral type if pointer arithmetics would be needed. Formatting
10861 2017-07-28 Martin Liska <mliska@suse.cz>
10864 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
10865 parameters that are of a variable-length.
10867 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
10869 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
10871 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
10872 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
10873 (CRT_CALL_STATIC_FUNCTION): Likewise.
10874 (ASM_DEFAULT_SPEC): New define.
10875 (ASM_SPEC32): Likewise.
10876 (ASM_SPEC64): Likewise.
10877 (ASM_SPEC_COMMON): Likewise.
10878 (ASM_SPEC): Likewise.
10879 (INVALID_64BIT): Likewise.
10880 (LINK_OS_DEFAULT_SPEC): Likewise.
10881 (LINK_OS_SPEC32): Likewise.
10882 (LINK_OS_SPEC64): Likewise.
10883 (POWERPC_LINUX): Likewise.
10884 (PTRDIFF_TYPE): Likewise.
10885 (RESTORE_FP_PREFIX): Likewise.
10886 (RESTORE_FP_SUFFIX): Likewise.
10887 (SAVE_FP_PREFIX): Likewise.
10888 (SAVE_FP_SUFFIX): Likewise.
10889 (SIZE_TYPE): Likewise.
10890 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
10891 (TARGET_64BIT): Likewise.
10892 (TARGET_64BIT): Likewise.
10893 (TARGET_AIX): Likewise.
10894 (WCHAR_TYPE_SIZE): Likewise.
10895 (WCHAR_TYPE): Undefine.
10896 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
10897 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
10898 (CPP_OS_RTEMS_SPEC): Delete.
10899 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
10900 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
10902 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
10904 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10906 PR tree-optimization/81578
10907 * tree-parloops.c (build_new_reduction): Bail out if
10908 reduction_code isn't one of the standard OpenMP reductions.
10909 Move the details printing after that decision.
10911 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10913 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
10914 related to reload_in_progress.
10915 (splat_input_operand): Likewise.
10916 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
10918 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
10920 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
10921 (TARGET_INSTANTIATE_DECLS): Likewise.
10922 (legitimate_indexed_address_p): Delete reload_in_progress code.
10923 (rs6000_debug_legitimate_address_p): Likewise.
10924 (rs6000_eliminate_indexed_memrefs): Likewise.
10925 (rs6000_emit_le_vsx_store): Likewise.
10926 (rs6000_emit_move_si_sf_subreg): Likewise.
10927 (rs6000_emit_move): Likewise.
10928 (register_to_reg_type): Likewise.
10929 (rs6000_pre_atomic_barrier): Likewise.
10930 (rs6000_machopic_legitimize_pic_address): Likewise.
10931 (rs6000_allocate_stack_temp): Likewise.
10932 (rs6000_address_for_fpconvert): Likewise.
10933 (rs6000_address_for_altivec): Likewise.
10934 (rs6000_secondary_memory_needed_rtx): Delete function.
10935 (rs6000_check_sdmode): Likewise.
10936 (rs6000_alloc_sdmode_stack_slot): Likewise.
10937 (rs6000_instantiate_decls): Likewise.
10938 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
10939 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
10940 Delete reload_in_progress.
10941 (*vec_reload_and_plus_<mptrsize>): Likewise.
10942 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
10943 (vsx_div_v2di): Likewise.
10944 (vsx_udiv_v2di): Likewise.
10946 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10948 * config/rs6000/rs6000.opt (mlra): Replace with stub.
10949 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
10950 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
10951 (rs6000_debug_reg_global): Delete print of LRA status.
10952 (rs6000_option_override_internal): Delete dead LRA related code.
10953 (rs6000_lra_p): Delete function.
10954 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
10956 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
10958 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
10959 * config/riscv/rtems.h: New file.
10961 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10962 Sudakshina Das <sudi.das@arm.com>
10964 * config/aarch64/aarch64.md
10965 (define_split for and<mode>3nr_compare): Move
10966 non aarch64_logical_operand to a register.
10967 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
10968 register immediate operand to a register.
10969 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
10971 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10973 PR middle-end/81564
10974 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
10976 2017-07-27 Richard Biener <rguenther@suse.de>
10978 PR tree-optimization/81573
10979 PR tree-optimization/81494
10980 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
10981 multi defuse cycle case.
10983 2017-07-27 Richard Biener <rguenther@suse.de>
10985 PR tree-optimization/81571
10986 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
10989 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
10991 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
10992 earlier and only if MASK_FPU is set. Adjust formatting.
10994 2017-07-27 Martin Liska <mliska@suse.cz>
10996 * opt-functions.awk: Add validation of value of Init.
10997 * optc-gen.awk: Pass new argument.
10999 2017-07-27 Martin Liska <mliska@suse.cz>
11001 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11002 Fix wrong condition.
11004 2017-07-27 Martin Liska <mliska@suse.cz>
11006 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11007 BBs and edges seen by autoFDO.
11009 2017-07-27 Richard Biener <rguenther@suse.de>
11011 PR tree-optimization/81502
11012 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11013 with incompatible but same sized type.
11014 (execute_update_addresses_taken): Likewise.
11016 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11018 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11019 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11021 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11024 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11025 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11026 Change s_operand to memory_operand.
11028 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11030 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11031 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11032 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11033 Emit instructions rather than returning an expression. Handle TFmode
11034 and KFmode by casting to TImode.
11035 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11036 (rs6000_emit_le_vsx_store): Likewise.
11037 * config/rs6000/vsx.md (VSX_TI): New iterator.
11038 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11039 (*vsx_le_undo_permute_<mode>): Likewise.
11040 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11041 emit the split sequence.
11042 (*vsx_le_perm_store_<mode>): Likewise.
11044 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11046 PR tree-optimization/81555
11047 PR tree-optimization/81556
11048 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11049 if true, force CHANGED for the recursive invocation.
11050 (reassociate_bb): Remember original length of ops array, pass
11051 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11053 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11054 attributes for noipa attribute. For naked attribute use
11055 lookup_attribute first before lookup_attribute_spec.
11056 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11058 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11059 for functions with noipa attribute.
11060 (cgraph_externally_visible_p): Return true for functions with noipa
11062 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11063 for functions with noipa attribute.
11064 * doc/extend.texi: Document noipa function attribute.
11065 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11066 also for functions with noipa attribute.
11067 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11069 2017-07-26 Andrew Pinski <apinski@cavium.com>
11071 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11072 vec_unalign_load_cost and vec_unalign_store_cost.
11074 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11076 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11077 -mvsx-small-integer option.
11078 (ISA_3_0_MASKS_IEEE): Likewise.
11079 (OTHER_VSX_VECTOR_MASKS): Likewise.
11080 (POWERPC_MASKS): Likewise.
11081 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
11082 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
11083 code, only testing for DImode being allowed in non-VSX floating
11085 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
11086 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
11088 (rs6000_option_override_internal): Delete -mvsx-small-integer.
11089 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
11090 TARGET_P8_VECTOR test.
11091 (rs6000_secondary_reload_simple_move): Likewise.
11092 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
11093 since TARGET_P9_VECTOR was already tested.
11094 (rs6000_opt_masks): Remove -mvsx-small-integer.
11095 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
11096 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11098 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
11099 test for TARGET_VEXTRACTUB was used, and that uses
11101 (p9 extract splitter): Likewise.
11102 (vsx_extract_<mode>_di_p9): Likewise.
11103 (vsx_extract_<mode>_store_p9): Likewise.
11104 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
11105 for TARGET_P9_VECTOR was used. Delete code that is now dead with
11106 the elimination of TARGET_VSX_SMALL_INTEGER.
11107 (vsx_extract_<mode>_p8): Likewise.
11108 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
11109 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
11110 (vsx_set_<mode>_p9): Likewise.
11111 (vsx_set_v4sf_p9): Likewise.
11112 (vsx_set_v4sf_p9_zero): Likewise.
11113 (vsx_insert_extract_v4sf_p9): Likewise.
11114 (vsx_insert_extract_v4sf_p9_2): Likewise.
11115 * config/rs6000/rs6000.md (sign extend splitter): Change
11116 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
11117 (floatsi<mode>2_lfiwax_mem): Likewise.
11118 (floatunssi<mode>2_lfiwzx_mem): Likewise.
11119 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
11120 since a test for TARGET_P9_VECTOR was used.
11121 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11122 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
11123 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11124 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11125 TARGET_P8_VECTOR test.
11126 (fix_trunc<mode>si2_stfiwx): Likewise.
11127 (fix_trunc<mode>si2_internal): Likewise.
11128 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
11129 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11131 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11132 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11133 TARGET_P8_VECTOR test.
11134 (fixuns_trunc<mode>si2_stfiwx): Likewise.
11135 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
11136 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11138 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11139 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
11140 since a test for TARGET_P9_VECTOR was used.
11141 (splitter for loading small constants): Likewise.
11143 2017-07-26 Andrew Pinski <apinski@cavium.com>
11145 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
11148 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
11151 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
11152 (fp_valid_at): Likewise.
11154 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11156 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
11157 (qdf24xx_addrcost_table): Likewise.
11158 (cortexa57_tunings): Update to use generic_branch_cost.
11159 (cortexa72_tunings): Likewise.
11160 (cortexa73_tunings): Likewise.
11161 (qdf24xx_tunings): Likewise.
11163 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11165 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
11166 (thunderx2t99_branch_cost): Likewise.
11167 (cortexa35_tunings): Update to use generic_branch_cost.
11168 (cortexa53_tunings): Likewise.
11169 (cortexa57_tunings): Likewise.
11170 (cortexa72_tunings): Likewise.
11171 (cortexa73_tunings): Likewise.
11172 (thunderx2t99_tunings): Likewise.
11174 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11176 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
11177 (sparc_option_override): Honour MASK_FSMULD.
11178 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
11179 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
11180 * config/sparc/sparc.opt (mfsmuld): New option.
11181 * doc/invoke.texi (mfsmuld): Document option.
11183 2017-07-26 Marek Polacek <polacek@redhat.com>
11185 PR middle-end/70992
11186 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
11188 2017-07-26 Richard Biener <rguenther@suse.de>
11190 * gimple-match-head.c (do_valueize): Return OP if valueize
11192 (get_def): New helper to get at the def stmt of a SSA name
11193 if valueize allows.
11194 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
11195 do_valueize to get at the def stmt.
11196 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
11198 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
11200 PR middle-end/46932
11201 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
11203 2017-07-26 Martin Liska <mliska@suse.cz>
11206 * function.c (expand_function_start): Make expansion of
11207 nonlocal_goto_save_area after parm_birth_insn.
11209 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11211 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
11212 from all CPU target flags enable members.
11214 2017-07-26 Richard Biener <rguenther@suse.de>
11216 * genmatch.c (dt_simplify::gen): Make iterator vars const.
11217 (decision_tree::gen): Make 'type' const.
11218 (write_predicate): Likewise.
11220 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11222 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
11223 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
11224 (rs6000_option_override_internal): Likewise.
11225 (rs6000_expand_vector_set): Likewise.
11226 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
11227 (TARGET_UPPER_REGS_SF): Likewise.
11228 (TARGET_UPPER_REGS_DI): Likewise.
11229 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
11230 (TARGET_DIRECT_MOVE_64BIT): Likewise.
11231 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
11232 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11233 (Splitters for DI constants in Altivec registers): Likewise.
11234 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
11235 (vsx_set_v4sf_p9): Likewise.
11236 (vsx_set_v4sf_p9_zero): Likewise.
11237 (vsx_insert_extract_v4sf_p9): Likewise.
11238 (vsx_insert_extract_v4sf_p9_2): Likewise.
11240 2017-07-25 Carl Love <cel@us.ibm.com>
11242 * doc/extend.texi: Update the built-in documentation file for the
11243 existing built-in functions
11244 vector signed char vec_cnttz (vector signed char);
11245 vector unsigned char vec_cnttz (vector unsigned char);
11246 vector signed short vec_cnttz (vector signed short);
11247 vector unsigned short vec_cnttz (vector unsigned short);
11248 vector signed int vec_cnttz (vector signed int);
11249 vector unsigned int vec_cnttz (vector unsigned int);
11250 vector signed long long vec_cnttz (vector signed long long);
11251 vector unsigned long long vec_cnttz (vector unsigned long long);
11253 2017-07-25 Andrew Pinski <apinski@cavium.com>
11255 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
11256 accesses where the use is for the first operand of a BIT_INSERT.
11258 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11261 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11262 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11264 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11266 * config/i386/gstabs.h: Delete.
11267 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
11269 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
11271 * config/i386/i386.c (ix86_decompose_address): Do not check for
11272 register RTX when looking at index_reg or base_reg.
11273 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
11275 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
11277 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
11278 to update EH info here.
11280 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11282 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
11284 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11286 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
11288 2017-07-25 Torsten Duwe <duwe@suse.de>
11290 * common.opt: Introduce -fpatchable-function-entry
11291 command line option, and its variables function_entry_patch_area_size
11292 and function_entry_patch_area_start.
11293 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
11294 including a two-value parser.
11295 * target.def (print_patchable_function_entry): New target hook.
11296 * targhooks.h (default_print_patchable_function_entry): New function.
11297 * targhooks.c (default_print_patchable_function_entry): Likewise.
11298 * toplev.c (process_options): Switch off IPA-RA if
11299 patchable function entries are being generated.
11300 * varasm.c (assemble_start_function): Look at the
11301 patchable-function-entry command line switch and current
11302 function attributes and maybe generate NOP instructions by
11303 calling the print_patchable_function_entry hook.
11304 * doc/extend.texi: Document patchable_function_entry attribute.
11305 * doc/invoke.texi: Document -fpatchable_function_entry
11306 command line option.
11307 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
11309 * doc/tm.texi: Re-generate.
11311 2017-07-25 Jakub Jelinek <jakub@redhat.com>
11314 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
11315 TARGET_AVX512DQ rather than TARGET_AVX512BW.
11317 2017-07-25 Tamar Christina <tamar.christina@arm.com>
11319 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
11321 2017-07-25 Richard Biener <rguenther@suse.de>
11323 PR tree-optimization/81455
11324 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
11325 not walk in cycles when looking for guards.
11327 2017-07-25 Richard Biener <rguenther@suse.de>
11329 PR tree-optimization/81529
11330 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
11331 when optimizing backedge uses.
11333 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
11335 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
11337 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
11338 to dl_section_ref. On AIX, append an expression to subtract
11339 the size of the section length to dl_section_ref.
11341 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
11343 * configure.ac: If any of the config.* scripts fail, exit 1.
11344 * configure: Regenerate.
11346 2017-07-25 Richard Biener <rguenther@suse.de>
11348 PR middle-end/81546
11349 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
11350 of immediate uses, be more verbose on errors.
11352 2017-07-25 Richard Biener <rguenther@suse.de>
11354 PR tree-optimization/81510
11355 * tree-vect-loop.c (vect_is_simple_reduction): When the
11356 reduction stmt is not inside the loop bail out.
11358 2017-07-25 Richard Biener <rguenther@suse.de>
11360 PR tree-optimization/81303
11361 * tree-vect-loop-manip.c (vect_loop_versioning): Build
11362 profitability check against LOOP_VINFO_NITERSM1.
11364 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11366 * domwalk.c (cmp_bb_postorder): Simplify.
11367 (sort_bbs_postorder): New function. Use it...
11368 (dom_walker::walk): ...here to optimize common cases.
11370 2017-07-25 Martin Liska <mliska@suse.cz>
11373 * ipa-visibility.c (function_and_variable_visibility): Make the
11374 redirection just on target that supports aliasing.
11375 Fix GNU coding style.
11377 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
11380 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
11382 * config/arm/rtems.h: Likewise.
11383 * config/bfin/rtems.h: Likewise.
11384 * config/i386/rtemself.h: Likewise.
11385 * config/lm32/rtems.h: Likewise.
11386 * config/m32c/rtems.h: Likewise.
11387 * config/m68k/rtemself.h: Likewise.
11388 * config/microblaze/rtems.h: Likewise.
11389 * config/mips/rtems.h: Likewise.
11390 * config/moxie/rtems.h: Likewise.
11391 * config/nios2/rtems.h: Likewise.
11392 * config/powerpcspe/rtems.h: Likewise.
11393 * config/rs6000/rtems.h: Likewise.
11394 * config/rtems.h: Likewise.
11395 * config/sh/rtems.h: Likewise.
11396 * config/sh/rtemself.h: Likewise.
11397 * config/sparc/rtemself.h: Likewise.
11399 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11402 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
11403 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
11404 * tree-ssa-structalias.c (alias_get_name): Same.
11406 2017-07-25 Bin Cheng <bin.cheng@arm.com>
11409 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
11410 instructions if no du chain is found.
11412 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11414 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
11416 2017-07-25 Richard Biener <rguenther@suse.de>
11418 PR middle-end/81505
11419 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
11422 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11424 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
11425 upper-regs options.
11426 (ISA_2_7_MASKS_SERVER): Likewise.
11427 (ISA_3_0_MASKS_IEEE): Likewise.
11428 (OTHER_P8_VECTOR_MASKS): Likewise.
11429 (OTHER_VSX_VECTOR_MASKS): Likewise.
11430 (POWERPC_MASKS): Likewise.
11431 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
11432 duplicate list of options.
11433 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
11434 explicit -mupper-regs options.
11435 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
11436 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
11437 alias for -mupper-regs-df.
11438 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11439 (rs6000_init_hard_regno_mode_ok): Likewise.
11440 (rs6000_option_override_internal): Likewise.
11441 (rs6000_opt_masks): Likewise.
11442 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
11443 options in terms of whether -mvsx or -mpower8-vector was used.
11444 (TARGET_UPPER_REGS_DI): Likewise.
11445 (TARGET_UPPER_REGS_SF): Likewise.
11446 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
11447 -mupper-regs-* options.
11449 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11451 * passes.c (emergency_dump_function): Print some empty lines and a
11452 header before the RTL dump.
11454 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11456 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
11458 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
11461 * config/aarch64/aarch64.c (aarch64_classify_symbol):
11462 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
11464 2017-07-24 Carl Love <cel@us.ibm.com>
11466 * config/rs6000/rs6000-c.c: Add support for built-in functions
11467 vector float vec_extract_fp32_from_shorth (vector unsigned short);
11468 vector float vec_extract_fp32_from_shortl (vector unsigned short);
11469 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
11470 vec_extract_fp_from_shortl): Add defines for the two builtins.
11471 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
11472 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
11474 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
11475 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
11476 * doc/extend.texi: Update the built-in documentation file for the
11477 new built-in function.
11479 2017-07-24 Jakub Jelinek <jakub@redhat.com>
11482 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
11484 * doc/generic.texi: Likewise.
11485 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11486 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11488 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
11490 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
11491 (aarch64_mls_elt_merge<mode>): Likewise.
11493 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
11495 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11496 having __cxa_atexit.
11498 2017-07-23 Michael Collison <michael.collison@arm.com>
11500 * config/arm/arm.c (arm_option_override): Deprecate
11501 use of -mstructure-size-boundary.
11502 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
11503 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
11505 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11508 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
11509 Reduce cost estimate for direct moves.
11511 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
11514 * config/i386/i386.c (ix86_option_override_internal): Disable
11515 BMI, BMI2 and TBM instructions for -m16.
11517 2017-07-21 Carl Love <cel@us.ibm.com>
11519 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11520 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11521 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11522 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11523 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11524 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11525 VMULOSW): New enum "unspec" values.
11526 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11527 altivec_vmulosw): New patterns.
11528 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11529 VMULOSW): Add definitions.
11531 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
11533 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
11534 (qdf24xx): Likewise.
11535 * config/aarch64/aarch64-options-extensions.def (rdma); New.
11536 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
11537 (AARCH64_FL_V8_1): Renumber.
11538 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
11539 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
11540 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
11541 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
11542 rdma to feature modifiers list.
11544 2017-07-21 Yury Gribov <tetra2005@gmail.com>
11546 PR middle-end/56727
11547 * ipa-visibility (function_and_variable_visibility): Convert
11548 recursive PLT call to direct call if appropriate.
11550 2017-07-21 Andrew Pinski <apinski@cavium.com>
11552 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
11553 operand 1 to see if the types precision matches.
11554 * fold-const.c (operand_equal_p): Likewise.
11556 2017-07-21 Richard Biener <rguenther@suse.de>
11558 PR tree-optimization/81303
11559 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
11560 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
11561 (vect_peeling_hash_get_lowest_cost): Adjust.
11562 (vect_enhance_data_refs_alignment): Likewise. Use
11563 vect_get_peeling_costs_all_drs to compute the penalty for no
11564 peeling to match up costs.
11566 2017-07-21 Richard Biener <rguenther@suse.de>
11568 PR tree-optimization/81500
11569 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
11570 we didn't identify a reduction path.
11572 2017-07-21 Tom de Vries <tom@codesourcery.com>
11573 Cesar Philippidis <cesar@codesourcery.com>
11575 PR gcov-profile/81442
11576 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
11579 2017-07-21 Tom de Vries <tom@codesourcery.com>
11582 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
11584 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
11585 nvptx_override_options_after_change.
11587 2017-07-21 Ulrich Drepper <drepper@redhat.com>
11589 * dwarf2out.c (output_file_names): Avoid double testing for
11590 dwarf_version >= 5.
11592 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
11594 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
11596 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11598 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
11600 (try_crossjump_to_edge): Do not punt on partitioned functions.
11602 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11604 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
11605 Put all BBs reachable only via paths crossing cold region to cold
11607 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
11609 2016-07-21 Richard Biener <rguenther@suse.de>
11611 PR tree-optimization/81303
11612 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
11613 into account prologue and epilogue iterations when raising
11614 min_profitable_iters to sth at least covering one vector iteration.
11616 2017-07-21 Tamar Christina <tamar.christina@arm.com>
11618 * config/arm/arm.c (arm_test_cpu_arch_dat):
11621 2017-07-20 Nathan Sidwell <nathan@acm.org>
11623 Remove TYPE_METHODS.
11624 * tree.h (TYPE_METHODS): Delete.
11625 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
11626 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
11627 (dbxout_type_methods): Scan TYPE_FIELDS.
11628 (dbxout_type): Don't check TYPE_METHODS here.
11629 * function.c (use_register_for_decl): Always ignore register for
11630 class types when not optimizing.
11631 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
11632 * tree.c (free_lang_data_in_type): Stitch out member functions and
11633 templates from TYPE_FIELDS.
11634 (build_distinct_type_copy, verify_type_variant,
11635 verify_type): Member fns are on TYPE_FIELDS.
11636 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
11637 * tree-pretty-print.c (dump_generic_node): Likewise.
11639 2017-07-20 Jakub Jelinek <jakub@redhat.com>
11642 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
11643 V2TImode and V4TImode.
11644 (ix86_expand_vector_extract): Likewise.
11645 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
11646 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
11647 (ssescalarmode): Handle V4TImode and V2TImode.
11648 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
11649 (*vec_extractv2ti, *vec_extractv4ti): New insns.
11650 (VEXTRACTI128_MODE): New mode iterator.
11651 (splitter for *vec_extractv?ti first element): New.
11652 (VEC_INIT_MODE): New mode iterator.
11653 (vec_init<mode>): Consolidate 3 expanders into one using
11654 VEC_INIT_MODE mode iterator.
11656 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11658 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
11659 non_spilled_static_chain_regno_p.
11661 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11663 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
11665 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
11667 * bb-reorder.c (connect_traces): Allow copying of blocks within
11670 2017-07-20 Richard Biener <rguenther@suse.de>
11672 * gimple.h (gimple_phi_result): Add gphi * overload.
11673 (gimple_phi_result_ptr): Likewise.
11674 (gimple_phi_arg): Likewise. Adjust index assert to only
11675 allow actual argument accesses rather than all slots available
11677 (gimple_phi_arg_def): Add gphi * overload.
11678 * tree-phinodes.c (make_phi_node): Initialize only actual
11680 (resize_phi_node): Clear memory not covered by old node,
11681 do not initialize excess argument slots.
11682 (reserve_phi_args_for_new_edge): Initialize new argument slot
11685 2017-07-20 Bin Cheng <bin.cheng@arm.com>
11687 PR tree-optimization/81388
11689 2016-07-21 Bin Cheng <bin.cheng@arm.com>
11691 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
11692 by removing computation of may_be_zero.
11694 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11695 Tom de Vries <tom@codesourcery.com>
11697 PR middle-end/81030
11698 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
11699 when gimple level profile disagrees with what RTL expander did.
11701 2017-07-20 Richard Biener <rguenther@suse.de>
11703 PR tree-optimization/61171
11704 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
11705 (vect_analyze_stmt): Add slp instance parameter.
11706 (vectorizable_reduction): Likewise.
11707 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
11708 (vect_is_simple_reduction): Deal with chains not detected
11709 as SLP reduction chain, specifically not properly associated
11710 chains containing a mix of plus/minus.
11711 (get_reduction_op): Remove.
11712 (get_initial_defs_for_reduction): Simplify, pass in whether
11713 this is a reduction chain, pass in the SLP node for the PHIs.
11714 (vect_create_epilog_for_reduction): Get the SLP instance as
11716 (vectorizable_reduction): Get the SLP instance as arg.
11717 During analysis remember the SLP node with the PHIs in the
11718 instance. Simplify getting at the vectorized reduction PHIs.
11719 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
11720 through SLP instance.
11721 (vect_slp_analyze_operations): Likewise.
11722 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
11723 (vect_transform_stmt): Likewise.
11725 2017-07-20 Tom de Vries <tom@codesourcery.com>
11727 PR tree-optimization/81489
11728 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
11729 read of phi arg location to before loop that modifies phi.
11731 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11733 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
11736 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11738 PR middle-end/81331
11739 * except.c (execute): Fix ordering issue.
11741 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11743 PR rtl-optimization/81423
11744 * combine.c (make_compound_operation_int): Don't try to optimize
11745 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
11747 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11749 PR rtl-optimization/81423
11750 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
11751 with a constant that is -1 in the truncated to mode.
11753 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11755 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
11756 (determine_unlikely_bbs): ... here.
11757 * predict.h (propagate_unlikely_bbs_forward): Declare.
11758 * cfgexpand.c (pass_expand::execute): Use it.
11759 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
11761 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
11762 propagate_unlikely_bbs_forward.
11764 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11766 PR middle-end/81331
11767 * except.c (maybe_add_nop_after_section_switch): New function.
11770 2017-07-19 Tom de Vries <tom@codesourcery.com>
11772 * gimple.h (gimple_phi_set_arg): Make assert more strict.
11774 2017-07-19 Tom de Vries <tom@codesourcery.com>
11776 * gimple.h (gimple_phi_arg): Make assert more strict.
11778 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
11780 * config.gcc (powerpc*-*-*): Add mmintrin.h.
11781 * config/rs6000/mmintrin.h: New file.
11782 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
11784 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11786 PR tree-optimization/81346
11787 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
11789 2017-07-19 Tom de Vries <tom@codesourcery.com>
11791 * config/nvptx/nvptx.md (VECIM): Add V2DI.
11793 2017-07-19 Tom de Vries <tom@codesourcery.com>
11795 * config/nvptx/nvptx-modes.def: Add V2DImode.
11796 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
11797 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
11798 (nvptx_output_mov_insn): Handle lack of mov.b128.
11799 (nvptx_print_operand): Handle 'H' and 'L' codes.
11800 (nvptx_vector_mode_supported): Allow V2DImode.
11801 (nvptx_preferred_simd_mode): New function.
11802 (nvptx_data_alignment): New function.
11803 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
11804 nvptx_preferred_simd_mode.
11805 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
11807 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
11809 2017-07-19 Tom de Vries <tom@codesourcery.com>
11811 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
11812 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
11813 (nvptx_vector_mode_supported): New function. Allow V2SImode.
11814 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
11815 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
11816 (mov<VECIM>_insn): New define_insn.
11817 (define_expand "mov<VECIM>): New define_expand.
11819 2017-07-19 Tom de Vries <tom@codesourcery.com>
11821 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
11823 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11825 PR tree-optimization/81346
11826 * fold-const.h (fold_div_compare, range_check_type): Declare.
11827 * fold-const.c (range_check_type): New function.
11828 (build_range_check): Use range_check_type.
11829 (fold_div_compare): No longer static, rewritten into
11830 a match.pd helper function.
11831 (fold_comparison): Don't call fold_div_compare here.
11832 * match.pd (X / C1 op C2): New optimization using fold_div_compare
11833 as helper function.
11835 2017-07-19 Nathan Sidwell <nathan@acm.org>
11837 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
11838 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
11839 * tree.c (find_decls_types_r, verify_type): Use
11840 TYPE_{MIN,MAX}_VALUE_RAW.
11841 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
11842 (hash_tree): Likewise.
11843 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11845 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11848 2017-07-18 Tom de Vries <tom@codesourcery.com>
11850 PR middle-end/81464
11851 * omp-expand.c (expand_omp_for_static_chunk): Handle
11852 equal-argument loop exit phi.
11854 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
11857 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
11858 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
11859 operand 2 predicate.
11860 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
11861 operand 2 predicate.
11862 (ror,rol -> rorx splitters): Use const_int_operand as
11863 operand 2 predicate.
11865 2017-06-18 Richard Biener <rguenther@suse.de>
11867 PR tree-optimization/81410
11868 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
11869 the gap in the ! slp_perm SLP case after each group.
11871 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11873 PR middle-end/81463
11874 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
11877 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11879 PR middle-end/81462
11880 * predict.c (set_even_probabilities): Cleanup; do not affect
11881 probabilities that are already known.
11882 (combine_predictions_for_bb): Call even when count is set.
11884 2017-07-18 Nathan Sidwell <nathan@acm.org>
11886 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
11889 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11892 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
11893 optimization for loop niter analysis.
11895 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
11898 * config/avr/avr.c (avr_optimize_casesi): Don't use
11899 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
11901 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11903 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
11904 body_cost_vec from _vect_peel_extended_info.
11905 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
11906 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
11909 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11911 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
11913 2017-07-18 Richard Biener <rguenther@suse.de>
11915 PR tree-optimization/80620
11916 PR tree-optimization/81403
11917 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
11918 info when re-using a VN table entry.
11920 2017-07-18 Richard Biener <rguenther@suse.de>
11922 PR tree-optimization/81418
11923 * tree-vect-loop.c (vectorizable_reduction): Properly compute
11924 vectype_in. Verify that with lane-reducing reduction operations
11925 we have a single def-use cycle.
11927 2017-07-17 Carl Love <cel@us.ibm.com>
11929 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
11931 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11932 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11933 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11934 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11935 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11936 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11937 VMULOSW): New enum "unspec" values.
11938 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11939 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11940 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11941 altivec_vmulosw): New patterns.
11942 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11943 VMULOSW): Add definitions.
11944 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
11946 * config/alpha/alpha.c: Include predict.h.
11948 2017-07-17 Yury Gribov <tetra2005@gmail.com>
11950 * tree-vrp.c (compare_assert_loc): Fix comparison function
11951 to return predictable results.
11953 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11955 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
11957 (subdi3): Likewise.
11958 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
11959 * doc/invoke.texi (mexpand-adddi): Update text.
11961 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11963 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
11964 that also clobbers the CC register. The old expand code is moved
11966 (*arc_clzsi2): ... here.
11967 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
11968 the CC register. The old expand code is moved to ...
11969 (arc_ctzsi2): ... here.
11971 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11973 * config/arc/arc.opt (mindexed-loads): Use initial value
11974 TARGET_INDEXED_LOADS_DEFAULT.
11975 (mauto-modify-reg): Use initial value
11976 TARGET_AUTO_MODIFY_REG_DEFAULT.
11977 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11978 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11979 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
11980 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
11982 2017-07-17 Martin Liska <mliska@suse.cz>
11985 * opts.c (finish_options): Do not allow -fgnu-tm
11986 w/ -fsanitize={kernel-,}address. Say sorry.
11988 2017-07-17 Bin Cheng <bin.cheng@arm.com>
11991 * tree-loop-distribution.c (classify_partition): Only assert on
11992 numer of iterations.
11993 (merge_dep_scc_partitions): Delete prameter. Update function call.
11994 (distribute_loop): Remove code handling loop with unknown niters.
11995 (pass_loop_distribution::execute): Skip loop with unknown niters.
11997 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12000 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12001 function sort_partitions_by_post_order.
12003 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12005 PR tree-optimization/81374
12006 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12007 the max index of basic blocks, rather than number of basic blocks.
12009 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12011 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12013 (arc_legitimate_pic_operand_p): Likewise.
12014 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12016 (arc_needs_pcl_p): Likewise.
12017 (arc_legitimate_pc_offset_p): Likewise.
12018 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12019 function is also used in constrains.md.
12020 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12021 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12022 PLUS. Only return true/false in known cases, otherwise assert.
12023 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12024 is already called in arc_legitimate_constant_p.
12025 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12027 (LEGITIMATE_PIC_OPERAND_P): Use
12028 arc_raw_symbolic_reference_mentioned_p function.
12029 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12034 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12035 Andrew Burgess <andrew.burgess@embecosm.com>
12037 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12038 (arc_return_address_register): New function.
12039 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12040 (arc_handle_fndecl_attribute): Add naked attribute.
12041 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12042 (TARGET_WARN_FUNC_RETURN): Likewise.
12043 (arc_allocate_stack_slots_for_args): New function.
12044 (arc_warn_func_return): Likewise.
12045 (machine_function): Change type fn_type.
12046 (arc_compute_function_type): Consider new naked function type,
12047 change function return type.
12048 (arc_must_save_register): Adapt to handle new
12049 arc_compute_function_type's return type.
12050 (arc_expand_prologue): Likewise.
12051 (arc_expand_epilogue): Likewise.
12052 (arc_return_address_regs): Delete.
12053 (arc_return_address_register): New function.
12054 (arc_epilogue_uses): Use above function.
12055 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12056 (arc_function_type): Change encoding, add naked type.
12057 (ARC_INTERRUPT_P): Change to handle the new encoding.
12058 (ARC_FAST_INTERRUPT_P): Likewise.
12059 (ARC_NORMAL_P): Define.
12060 (ARC_NAKED_P): Likewise.
12061 (arc_compute_function_type): Delete prototype.
12062 * config/arc/arc.md (in_ret_delay_slot): Use
12063 arc_return_address_register function.
12064 (simple_return): Likewise.
12065 (p_return_i): Likewise.
12067 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12069 PR tree-optimization/81428
12070 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12071 can't be built for those types.
12073 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12075 Remove stuff dead since r239246.
12077 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12078 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12079 (avr_inform_devices): Remove dead stuff.
12081 2017-07-17 Tamar Christina <tamar.christina@arm.com>
12083 * config/arm/arm_neon.h: Fix softp typo.
12085 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12087 PR tree-optimization/81365
12088 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
12089 aggregate moves onto bb predecessor edges, make sure there are no
12090 loads that could alias the lhs in between the start of bb and the
12093 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12096 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
12097 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
12098 [LSHIFTRT, outer_code = TRUNCATE]: Same.
12100 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12102 PR tree-optimization/81396
12103 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
12104 (init_symbolic_number): Initialize it to 1.
12105 (perform_symbolic_merge): Add n_ops from both operands into the new
12107 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
12108 without base_addr as useless if they need more than one operation.
12109 (bswap_replace): Handle !bswap case for NULL base_addr.
12111 2017-07-17 Tom de Vries <tom@codesourcery.com>
12114 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
12117 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
12119 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
12120 conditional builtin define __FIX_LEON3FT_B2BST.
12122 2017-07-17 Daniel Cederman <cederman@gaisler.com>
12124 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
12125 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
12128 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12130 PR rtl-optimization/81424
12131 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
12132 to remove potential trapping from operands if -fnon-call-exceptions.
12134 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12136 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
12137 profile_proability for scalling.
12138 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
12140 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12142 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
12144 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12146 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
12147 fixpoint arithmetics.
12149 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12151 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
12152 fixpoint arithmetics.
12154 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12156 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
12157 fixpoint arithmetics.
12159 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12161 * profile-count.h (profile_probability::from_reg_br_prob_note,
12162 profile_probability::to_reg_br_prob_note): New functions.
12163 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
12164 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
12165 * predict.c (probability_reliable_p): Update.
12166 (edge_probability_reliable_p): Update.
12167 (br_prob_note_reliable_p): Update.
12168 (invert_br_probabilities): Update.
12169 (add_reg_br_prob_note): New function.
12170 (combine_predictions_for_insn): Update.
12171 * asan.c (asan_clear_shadow): Update.
12172 * cfgbuild.c (compute_outgoing_frequencies): Update.
12173 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12174 (update_br_prob_note): Update.
12175 (rtl_verify_edges): Update.
12176 (purge_dead_edges): Update.
12177 (fixup_reorder_chain): Update.
12178 * emit-rtl.c (try_split): Update.
12179 * ifcvt.c (cond_exec_process_insns): Update.
12180 (cond_exec_process_if_block): Update.
12181 (dead_or_predicable): Update.
12182 * internal-fn.c (expand_addsub_overflow): Update.
12183 (expand_neg_overflow): Update.
12184 (expand_mul_overflow): Update.
12185 * loop-doloop.c (doloop_modify): Update.
12186 * loop-unroll.c (compare_and_jump_seq): Update.
12187 * optabs.c (emit_cmp_and_jump_insn_1): Update.
12188 * predict.h: Update.
12189 * reorg.c (mostly_true_jump): Update.
12191 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
12192 * config/alpha/alpha.c (emit_unlikely_jump): Update.
12193 * config/arc/arc.c: (emit_unlikely_jump): Update.
12194 * config/arm/arm.c: (emit_unlikely_jump): Update.
12195 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
12196 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
12197 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
12198 (ix86_print_operand): Update.
12199 (ix86_split_fp_branch): Update.
12200 (predict_jump): Update.
12201 * config/ia64/ia64.c (ia64_print_operand): Update.
12202 * config/mmix/mmix.c (mmix_print_operand): Update.
12203 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
12204 (rs6000_expand_split_stack_prologue): Update.
12205 * config/rs6000/rs6000.c: Update.
12206 * config/s390/s390.c (s390_expand_vec_strlen): Update.
12207 (s390_expand_vec_movstr): Update.
12208 (s390_expand_cs_tdsi): Update.
12209 (s390_expand_split_stack_prologue): Update.
12210 * config/sh/sh.c (sh_print_operand): Update.
12211 (expand_cbranchsi4): Update.
12212 (expand_cbranchdi4): Update.
12213 * config/sparc/sparc.c (output_v9branch): Update.
12214 * config/spu/spu.c (get_branch_target): Update.
12215 (ea_load_store_inline): Update.
12216 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
12217 * config/tilepro/tilepro.c: Update.
12219 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12221 * gimplify.c (mostly_copy_tree_r): Revert latest change.
12222 (gimplify_save_expr): Likewise.
12224 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12226 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
12228 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12230 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
12232 * timevar.def (TV_IPA_FNSUMMARY): Define.
12234 2017-07-16 Daniel Cederman <cederman@gaisler.com>
12236 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
12237 to back store errata sensitive sequence from being generated.
12238 (sqrtdf2_fix): Likewise.
12240 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12242 * tree-ssa-threadupdate.c (compute_path_counts,
12243 update_joiner_offpath_counts): Use profile_probability.
12245 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
12248 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12250 * config/arm/arm-c.c (arm_cpu_builtins): Define
12251 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12253 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
12255 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12256 array entries to represent __ieee128 versions of the
12257 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
12258 scalar_extract_sig, and scalar_insert_exp built-in functions.
12259 (altivec_resolve_overloaded_builtin): Add special case handling
12260 for the __builtin_scalar_insert_exp function, as represented by
12261 the P9V_BUILTIN_VEC_VSIEDP constant.
12262 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
12263 exponent support for __ieee128 argument.
12264 (VSESQP): Add scalar extract signature support for __ieee128
12266 (VSTDCNQP): Add scalar test negative support for __ieee128
12268 (VSIEQP): Add scalar insert exponent support for __int128 argument
12269 with __ieee128 result.
12270 (VSIEQPF): Add scalar insert exponent support for __ieee128
12271 argument with __ieee128 result.
12272 (VSTDCQP): Add scalar test data class support for __ieee128
12274 (VSTDCNQP): Add overload support for scalar test negative with
12275 __ieee128 argument.
12276 (VSTDCQP): Add overload support for scalar test data class
12277 __ieee128 argument.
12278 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
12279 UNSPEC_VSX_SXSIGDP.
12280 (UNSPEC_VSX_SIEXPQP): New constant.
12281 (xsxexpqp): New insn for VSX scalar extract exponent quad
12283 (xsxsigqp): New insn for VSX scalar extract significand quad
12285 (xsiexpqpf): New insn for VSX scalar insert exponent quad
12286 precision with floating point argument.
12287 (xststdcqp): New expand for VSX scalar test data class quad
12289 (xststdcnegqp): New expand for VSX scalar test negative quad
12291 (xststdcqp): New insn to match expansions for VSX scalar test data
12292 class quad precision and VSX scalar test negative quad precision.
12293 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
12294 special case operand checking to enforce that second operand of
12295 VSX scalar test data class with quad precision argument is a 7-bit
12297 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12298 prototypes and descriptions of __ieee128 versions of
12299 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
12300 scalar_test_data_class, and scalar_test_neg built-in functions.
12302 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12304 PR tree-optimization/81162
12305 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
12306 replace a negate with an add.
12308 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
12310 * doc/invoke.texi (arm/-mcpu): Document +crypto.
12312 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12314 * config/arm/arm-c.c (arm_cpu_builtins): Define
12315 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12317 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12319 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
12320 (armv8-r): Set ARM Cortex-R52 as default CPU.
12321 * config/arm/arm-tables.opt: Regenerate.
12322 * config/arm/arm-tune.md: Regenerate.
12323 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12325 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
12326 extension for -mcpu=cortex-r52.
12328 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12330 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
12331 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
12332 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
12333 (fp-armv8): Define it as FP_ARMv8 only.
12334 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
12335 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
12337 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
12338 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
12339 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
12340 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
12341 than TARGET_FPU_ARMV8.
12342 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
12343 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
12344 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
12346 * config/arm/neon.md (neon_vrint): Likewise.
12347 (neon_vcvt): Likewise.
12348 (neon_<fmaxmin_op><mode>): Likewise.
12349 (<fmaxmin><mode>3): Likewise.
12350 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
12351 * config/arm/predicates.md (arm_cond_move_operator): Check against
12352 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
12354 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
12356 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
12357 to top of function.
12359 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12361 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
12362 loop in comment with memset.
12364 2017-07-14 Martin Liska <mliska@suse.cz>
12366 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
12367 * dwarf2out.c (is_java): Remove the function.
12368 (output_pubname): Remove usage of the function.
12369 (lower_bound_default): Remove usage of DW_LANG_Java.
12370 (gen_compile_unit_die): Likewise.
12371 * gcc.c: Remove compiler defaults for .java and .zip files.
12372 * gimple-expr.c (remove_suffix): Change as there's no longer
12373 extension than 4-letter one.
12374 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
12375 (gimplify_save_expr): Likewise.
12376 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
12377 as it's possible even for other languages than Java.
12378 * langhooks.h (struct lang_hooks): Remove Java from a comment.
12379 * lto-opts.c (lto_write_options): Remove reference to Java.
12380 * opts.c (strip_off_ending): Update file extension handling.
12381 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
12382 * tree-eh.c (lower_resx): Likewise.
12383 * tree.c (free_lang_data_in_type): Remove dead code.
12384 (find_decls_types_r): Likewise.
12385 (build_common_builtin_nodes): Remove Java from a comment.
12386 (verify_type): Remove dead code.
12387 * varasm.c (assemble_external): Remove Java from a comment.
12389 2017-07-14 Martin Liska <mliska@suse.cz>
12391 * opts.c (finish_options): Add quotes.
12392 (common_handle_option): Likewise.
12394 2017-07-14 Martin Liska <mliska@suse.cz>
12396 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
12397 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
12398 Remove N_SO_PASCAL.
12399 * dwarf2out.c (lower_bound_default): Do not handle
12401 (gen_compile_unit_die): Likewise.
12402 * gcc.c: Remove default extension binding for GNU Pascal.
12403 * stmt.c: Remove Pascal language from a comment.
12404 * xcoffout.c: Likewise.
12406 2017-07-13 David Malcolm <dmalcolm@redhat.com>
12409 * diagnostic-show-locus.c (fixit_cmp): New function.
12410 (layout::layout): Sort m_fixit_hints.
12411 (column_range::column_range): Assert that the values are valid.
12412 (struct char_span): New struct.
12413 (correction::overwrite): New method.
12414 (struct source_line): New struct.
12415 (line_corrections::add_hint): Add assertions. Reimplement memcpy
12416 calls in terms of classes source_line and char_span, and
12417 correction::overwrite.
12418 (selftest::test_overlapped_fixit_printing_2): New function.
12419 (selftest::diagnostic_show_locus_c_tests): Call it.
12421 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
12423 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
12424 early if there is no lhs.
12426 2017-07-13 Martin Liska <mliska@suse.cz>
12428 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
12429 (gen_reference_type_die): Likewise.
12430 * stor-layout.c: Remove Pascal-related comment.
12432 2017-07-13 Martin Liska <mliska@suse.cz>
12434 * opts.c (finish_options): Add quotes to error messages.
12435 (parse_sanitizer_options): Likewise.
12437 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
12439 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
12441 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
12443 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
12445 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
12447 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
12449 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
12451 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12454 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
12455 provides the hardware capability bits, define the macro
12456 __BUILTIN_CPU_SUPPORTS__.
12457 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
12458 if GLIBC does not provide the hardware capability bits. Add a
12459 gcc_unreachable call if the built-in cpu function is neither
12460 __builtin_cpu_is nor __builtin_cpu_supports.
12461 (rs6000_get_function_versions_dispatcher): Change the warning
12462 that an old GLIBC is used which does not export the capability
12463 bits to be an error.
12464 * doc/extend.texi (target_clones attribute): Document the
12465 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
12466 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
12467 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
12468 the macros defined by GCC if the newer GLIBC is available.
12470 2017-07-12 Jeff Law <law@redhat.com>
12472 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
12473 remaining includes slightly.
12474 * config/riscv/riscv-builtins.c: Include profile-count.h.
12476 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12479 * config/avr/avr.c (avr_set_current_function): In diagnostic
12480 messages: Quote keywords and (parts of) identifiers.
12481 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
12484 2017-07-12 Carl Love <cel@us.ibm.com>
12486 * config/rs6000/rs6000-c.c: Add support for built-in functions
12487 vector bool char vec_revb (vector bool char);
12488 vector bool short vec_revb (vector short char);
12489 vector bool int vec_revb (vector bool int);
12490 vector bool long long vec_revb (vector bool long long);
12491 * doc/extend.texi: Update the built-in documentation file for the
12492 new built-in functions.
12494 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12496 * config/s390/s390.md: Remove movcc splitter.
12498 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12500 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
12501 load/store on condition.
12503 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12506 * config/avr/avr.c (avr_encode_section_info)
12507 [progmem && !TREE_READONLY]: Error if progmem object needs
12510 2017-07-11 Michael Collison <michael.collison@arm.com>
12512 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
12515 2017-07-11 Carl Love <cel@us.ibm.com>
12517 * config/rs6000/rs6000-c.c: Add support for builtins
12518 vector unsigned int vec_parity_lsbb (vector signed int);
12519 vector unsigned int vec_parity_lsbb (vector unsigned int);
12520 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
12521 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
12522 vector unsigned long long vec_parity_lsbb (vector signed long long);
12523 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
12524 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
12525 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
12526 * doc/extend.texi: Update the built-in documentation file for the
12527 new built-in functions.
12529 2017-07-11 David Malcolm <dmalcolm@redhat.com>
12531 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
12532 (layout::m_primary_loc): New field.
12533 (layout::layout): Initialize new field. Move location filtering
12534 logic from here to...
12535 (layout::maybe_add_location_range): ...this new method. Add
12536 support for filtering to just the lines already specified by other
12538 (layout::will_show_line_p): New method.
12539 (gcc_rich_location::add_location_if_nearby): New method.
12540 (selftest::test_add_location_if_nearby): New test function.
12541 (selftest::diagnostic_show_locus_c_tests): Call it.
12542 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
12545 2017-07-11 Tom de Vries <tom@codesourcery.com>
12547 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
12548 (bb_first_real_insn): New function.
12549 (nvptx_single): Add extra initialization of broadcasted condition
12552 2017-07-11 Nathan Sidwell <nathan@acm.org>
12554 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
12556 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
12558 * doc/extend.texi (AVR Function Attributes): Remove weblink to
12559 Binutils doc as TEXI will mess them up.
12560 * doc/invoke.texi (AVR Options): Same here.
12562 2017-07-11 Daniel Cederman <cederman@gaisler.com>
12564 * config/sparc/sparc.opt (mfix-ut700): New option.
12565 (mfix-gr712rc): Likewise.
12566 (sparc_fix_b2bst): New variable.
12567 * doc/invoke.texi (SPARC options): Document them.
12568 (ARM options): Fix warnings.
12569 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
12570 instructions to prevent sequences that can trigger the store-store
12571 errata for certain LEON3FT processors.
12572 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
12573 (sparc_option_override): Set sparc_fix_b2bst appropriately.
12574 * config/sparc/sparc.md (fix_b2bst): New attribute.
12575 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
12577 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
12580 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
12582 (*rsqrtsf2_sse): Ditto.
12584 (div<mode>3): Macroize insn from divdf3 and divsf3
12585 using MODEF mode iterator.
12587 2017-07-10 Martin Sebor <msebor@redhat.com>
12589 PR tree-optimization/80397
12590 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
12591 instead of testing for equality to INTEGER_TYPE.
12593 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
12595 * config.gcc: Remove uclibc from arc target spec.
12597 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
12599 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
12601 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12604 * lto-wrapper.c (remove_option): New function.
12605 (merge_and_complain): Merge PIC/PIE options more realistically.
12607 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12609 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
12613 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
12614 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
12615 * config.in: Regenerate.
12616 * configure: Regenerate.
12617 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
12618 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
12619 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
12620 (TARGET_GASISR_PROLOGUES): ...target mask.
12621 * common/config/avr/avr-common.c
12622 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
12623 Set -mgas-isr-prologues.
12624 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
12625 INSERT_PASS_BEFORE for it.
12626 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
12627 * config/avr/avr.c (avr_option_override)
12628 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
12629 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
12630 (avr_attribute_table) <no_gccisr>: Add new function attribute.
12631 (avr_set_current_function) <is_no_gccisr>: Init machine field.
12632 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
12634 (make_avr_pass_pre_proep): New function.
12635 (emit_push_sfr) <treg>: Add argument to function and use it
12636 instead of TMP_REG.
12637 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
12638 and set machine->gasisr.yes.
12639 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
12640 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
12641 __gcc_isr.n_pushed to .L__stack_usage.
12642 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
12643 (avr_asm_final_postscan_insn): ...this new static function.
12644 * config/avr/avr.h (machine_function)
12645 <is_no_gccisr, use_L__stack_usage>: New fields.
12646 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
12647 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
12648 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
12649 (gasisr, *gasisr): New expander and insn.
12650 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12651 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
12652 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
12654 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
12656 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
12659 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12661 Move jump-tables out of .text again.
12664 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
12665 (ASM_OUTPUT_ADDR_VEC): New function.
12666 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
12667 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
12668 INSN_ADDRESSes as asm comment.
12669 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
12670 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
12671 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
12672 * config/avr/avr.md (*tablejump): Adjust comment.
12673 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
12674 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
12676 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
12677 (avr_output_addr_vec): New proto.
12678 (avr_log_t) <insn_addresses>: New field.
12680 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
12683 * config/i386/i386.c (ix86_function_arg_advance): Set
12684 outgoing_args_on_stack to true if there are outgoing arguments
12686 (ix86_function_arg): Likewise.
12687 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
12688 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
12689 * config/i386/i386.h (machine_function): Add
12690 outgoing_args_on_stack.
12692 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
12694 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12695 supporting pthreds.
12696 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
12698 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
12700 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
12701 (REAL_H): Remove $(MACHMODE_H).
12702 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
12704 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
12705 $(MACHMODE_H) and double-int.h.
12706 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
12708 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
12711 2017-07-07 Andrew Pinski <apinski@cavium.com>
12713 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
12714 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
12716 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12718 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
12719 Add warning if GCC was not configured to link against a GLIBC that
12720 exports the hardware capability bits.
12721 (make_resolver_func): Make resolver function private and not a
12722 COMDAT function. Create the name with clone_function_name instead
12723 of make_unique_name.
12726 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
12727 correct operand in doing the split.
12729 2017-07-07 Carl Love <cel@us.ibm.com>
12731 * config/rs6000/rs6000-c: Add support for built-in function
12732 vector unsigned short vec_pack_to_short_fp32 (vector float,
12734 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
12735 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
12736 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
12737 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
12738 (convert_4f32_8i16): Add define_expand.
12739 * doc/extend.texi: Update the built-in documentation file for the
12740 new built-in function.
12742 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12744 * config/sparc/m8.md: New file.
12745 * config/sparc/sparc.md: Include m8.md.
12747 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12749 * config/sparc/sparc.opt: New option -mvis4b.
12750 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
12751 (sparc_option_override): Handle VIS4B.
12752 (enum sparc_builtins): Define
12753 SPARC_BUILTIN_DICTUNPACK{8,16,32},
12754 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
12755 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
12756 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
12757 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
12758 (check_constant_argument): New function.
12759 (sparc_vis_init_builtins): Define builtins
12760 __builtin_vis_dictunpack{8,16,32},
12761 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
12762 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
12763 __builtin_vis_fpcmpde{8,16,32}shl and
12764 __builtin_vis_fpcmpur{8,16,32}shl.
12765 (sparc_expand_builtin): Check that the constant operands to
12766 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
12767 constant and in range.
12768 * config/sparc/sparc-c.c (sparc_target_macros): Handle
12770 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
12771 (SPARC_IMM5_P): Likewise.
12772 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
12773 (enabled): Handle vis4b.
12774 (UNSPEC_DICTUNPACK): New unspec.
12775 (UNSPEC_FPCMPSHL): Likewise.
12776 (UNSPEC_FPUCMPSHL): Likewise.
12777 (UNSPEC_FPCMPDESHL): Likewise.
12778 (UNSPEC_FPCMPURSHL): Likewise.
12779 (cpu_feature): New CPU feature `vis4b'.
12780 (dictunpack{8,16,32}): New insns.
12781 (FPCSMODE): New mode iterator.
12782 (fpcscond): New code iterator.
12783 (fpcsucond): Likewise.
12784 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
12785 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
12786 (fpcmpde{8,16,32}{si,di}shl): Likewise.
12787 (fpcmpur{8,16,32}{si,di}shl): Likewise.
12788 * config/sparc/constraints.md: Define constraints `q' for unsigned
12789 2-bit integer constants and `t' for unsigned 5-bit integer
12791 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
12793 (imm5_operand_dictunpack16): Likewise.
12794 (imm5_operand_dictunpack32): Likewise.
12795 (imm2_operand): Likewise.
12796 * doc/invoke.texi (SPARC Options): Document -mvis4b.
12797 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
12798 ditunpack* and fpcmp*shl builtins.
12800 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12802 * config.gcc: Handle m8 in --with-{cpu,tune} options.
12803 * config.in: Add HAVE_AS_SPARC6 define.
12804 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
12806 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
12808 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
12809 (CPP_CPU_SPEC): Handle m8.
12810 (ASM_CPU_SPEC): Likewise.
12811 * config/sparc/sparc-opts.h (enum processor_type): Add
12813 * config/sparc/sparc.c (m8_costs): New struct.
12814 (sparc_option_override): Handle TARGET_CPU_m8.
12815 (sparc32_initialize_trampoline): Likewise.
12816 (sparc64_initialize_trampoline): Likewise.
12817 (sparc_issue_rate): Likewise.
12818 (sparc_register_move_cost): Likewise.
12819 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
12820 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
12821 (ASM_CPU64_DEFAULT_SPEC): Likewise.
12822 (CPP_CPU_SPEC): Handle M8.
12823 (ASM_CPU_SPEC): Likewise.
12824 (AS_M8_FLAG): Define.
12825 * config/sparc/sparc.md: Add m8 to the cpu attribute.
12826 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
12827 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
12829 * configure: Regenerate.
12830 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
12833 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12835 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
12837 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
12838 ("*movdi_insn_sp32"): Do not set v3pipe.
12839 ("*movsi_insn"): Likewise.
12840 ("*movdi_insn_sp64"): Likewise.
12841 ("*movsf_insn"): Likewise.
12842 ("*movdf_insn_sp32"): Likewise.
12843 ("*movdf_insn_sp64"): Likewise.
12844 ("*zero_extendsidi2_insn_sp64"): Likewise.
12845 ("*sign_extendsidi2_insn"): Likewise.
12846 ("*mov<VM32:mode>_insn"): Likewise.
12847 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12848 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12849 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12850 ("<vlop:code><VL:mode>3"): Likewise.
12851 ("*not_<vlop:code><VL:mode>3"): Likewise.
12852 ("*nand<VL:mode>_vis"): Likewise.
12853 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
12854 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
12855 ("one_cmpl<VL:mode>2"): Likewise.
12856 ("faligndata<VM64:mode>_vis"): Likewise.
12857 ("alignaddrsi_vis"): Likewise.
12858 ("alignaddrdi_vis"): Likweise.
12859 ("alignaddrlsi_vis"): Likewise.
12860 ("alignaddrldi_vis"): Likewise.
12861 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12862 ("bmaskdi_vis"): Likewise.
12863 ("bmasksi_vis"): Likewise.
12864 ("bshuffle<VM64:mode>_vis"): Likewise.
12865 ("cmask8<P:mode>_vis"): Likewise.
12866 ("cmask16<P:mode>_vis"): Likewise.
12867 ("cmask32<P:mode>_vis"): Likewise.
12868 ("pdistn<P:mode>_vis"): Likewise.
12869 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12871 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12873 * config/sparc/sparc.md ("subtype"): New insn attribute.
12874 ("*wrgsr_sp64"): Set insn subtype.
12875 ("*rdgsr_sp64"): Likewise.
12876 ("alignaddrsi_vis"): Likewise.
12877 ("alignaddrdi_vis"): Likewise.
12878 ("alignaddrlsi_vis"): Likewise.
12879 ("alignaddrldi_vis"): Likewise.
12880 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12881 ("fexpand_vis"): Likewise.
12882 ("fpmerge_vis"): Likewise.
12883 ("faligndata<VM64:mode>_vis"): Likewise.
12884 ("bshuffle<VM64:mode>_vis"): Likewise.
12885 ("cmask8<P:mode>_vis"): Likewise.
12886 ("cmask16<P:mode>_vis"): Likewise.
12887 ("cmask32<P:mode>_vis"): Likewise.
12888 ("fchksm16_vis"): Likewise.
12889 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
12890 ("fmean16_vis"): Likewise.
12891 ("fp<plusminus_insn>64_vis"): Likewise.
12892 ("<plusminus_insn>v8qi3"): Likewise.
12893 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12894 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
12895 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
12896 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
12897 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
12898 ("*movqi_insn"): Likewise.
12899 ("*movhi_insn"): Likewise.
12900 ("*movsi_insn"): Likewise.
12901 ("movsi_pic_gotdata_op"): Likewise.
12902 ("*movdi_insn_sp32"): Likewise.
12903 ("*movdi_insn_sp64"): Likewise.
12904 ("movdi_pic_gotdata_op"): Likewise.
12905 ("*movsf_insn"): Likewise.
12906 ("*movdf_insn_sp32"): Likewise.
12907 ("*movdf_insn_sp64"): Likewise.
12908 ("*zero_extendhisi2_insn"): Likewise.
12909 ("*zero_extendqihi2_insn"): Likewise.
12910 ("*zero_extendqisi2_insn"): Likewise.
12911 ("*zero_extendqidi2_insn"): Likewise.
12912 ("*zero_extendhidi2_insn"): Likewise.
12913 ("*zero_extendsidi2_insn_sp64"): Likewise.
12914 ("ldfsr"): Likewise.
12915 ("prefetch_64"): Likewise.
12916 ("prefetch_32"): Likewise.
12917 ("tie_ld32"): Likewise.
12918 ("tie_ld64"): Likewise.
12919 ("*tldo_ldub_sp32"): Likewise.
12920 ("*tldo_ldub1_sp32"): Likewise.
12921 ("*tldo_ldub2_sp32"): Likewise.
12922 ("*tldo_ldub_sp64"): Likewise.
12923 ("*tldo_ldub1_sp64"): Likewise.
12924 ("*tldo_ldub2_sp64"): Likewise.
12925 ("*tldo_ldub3_sp64"): Likewise.
12926 ("*tldo_lduh_sp32"): Likewise.
12927 ("*tldo_lduh1_sp32"): Likewise.
12928 ("*tldo_lduh_sp64"): Likewise.
12929 ("*tldo_lduh1_sp64"): Likewise.
12930 ("*tldo_lduh2_sp64"): Likewise.
12931 ("*tldo_lduw_sp32"): Likewise.
12932 ("*tldo_lduw_sp64"): Likewise.
12933 ("*tldo_lduw1_sp64"): Likewise.
12934 ("*tldo_ldx_sp64"): Likewise.
12935 ("*mov<VM32:mode>_insn"): Likewise.
12936 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12937 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12939 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12941 * config/sparc/sparc.md ("type"): New insn type viscmp.
12942 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
12944 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
12945 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
12946 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12947 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
12949 ("n7_vis_logical_11cycle"): Likewise.
12950 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
12951 * config/sparc/niagara2.md ("niag3_vis": Likewise.
12952 * config/sparc/niagara.md ("niag_vis"): Likewise.
12953 * config/sparc/ultra3.md ("us3_fga"): Likewise.
12954 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
12956 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12958 * config/sparc/sparc.md: New instruction type `bmask'.
12959 (bmaskdi_vis): Use the `bmask' type.
12960 (bmasksi_vis): Likewise.
12961 * config/sparc/ultra3.md (us3_array): Likewise.
12962 * config/sparc/niagara7.md (n7_array): Likewise.
12963 * config/sparc/niagara4.md (n4_array): Likewise.
12964 * config/sparc/niagara2.md (niag2_vis): Likewise.
12965 (niag3_vis): Likewise.
12966 * config/sparc/niagara.md (niag_vis): Likewise.
12968 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
12970 * ipa-comdats.c: Remove optimize check from gate.
12971 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
12972 for functions not optimized.
12973 (ipa_fn_summary_read): Skip optimize check.
12974 (ipa_fn_summary_write): Likewise.
12975 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
12977 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
12979 (can_inline_edge_p): Check flag_pcc_struct_return for match.
12980 (check_callers): Give up on caller which is not optimized.
12981 (inline_small_functions): Likewise.
12982 (ipa_inline): Do not give up when not optimizing.
12983 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
12984 away unoptimizes cdtors.
12985 (whole_program_function_and_variable_visibility): Do
12986 ipa_discover_readonly_nonaddressable_vars in LTO mode.
12987 * ipa.c (process_references): Do not check optimize.
12988 (symbol_table::remove_unreachable_nodes): Update optimize check.
12989 (set_writeonly_bit): Update optimize check.
12990 (pass_ipa_cdtor_merge::gate): Do not check optimize.
12991 (pass_ipa_single_use::gate): Remove.
12993 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12995 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
12996 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
12997 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
12998 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
12999 permute_load, permute_store, adjust_extract, adjust_splat,
13000 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13001 replace_swap_with_copy, dump_swap_insn_table,
13002 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13003 recombine_lvx_pattern, recombine_stvx_pattern,
13004 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13005 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13006 to file rs6000-p8swap.c.
13007 * config/rs6000/rs6000-p8swap.c: New file.
13008 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13009 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13010 and rs6000*-*-* targets.
13012 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13014 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13016 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13018 * lto-wrapper.c (merge_and_complain): Do not merge
13019 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13020 fsigned_zeros, ftrapping_math, fwrapv.
13021 (append_compiler_options): Do not track these options.
13022 (append_linker_options): Likewie
13024 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13026 * cgraphunit.c (cgraph_node::finalize_function): When
13027 !flag_toplevel_reorde set no_reorder flag.
13028 (varpool_node::finalize_decl): Likewise.
13029 (symbol_table::compile): Drop no toplevel reorder path.
13031 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13033 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13034 edges; zero probability is not better than uninitialized.
13036 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13038 * asan.h (asan_sanitize_allocas_p): Declare.
13039 * asan.c (asan_sanitize_allocas_p): New function.
13040 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13041 (handle_builtin_alloca): Likewise.
13042 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13043 if !asan_sanitize_allocas_p.
13044 * params.def (asan-instrument-allocas): Add new option.
13045 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13046 * opts.c (common_handle_option): Disable allocas sanitization for
13049 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13051 * asan.c: Include gimple-fold.h.
13052 (get_last_alloca_addr): New function.
13053 (handle_builtin_stackrestore): Likewise.
13054 (handle_builtin_alloca): Likewise.
13055 (asan_emit_allocas_unpoison): Likewise.
13056 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13057 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13058 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13059 (instrument_builtin_call): Pass gimple iterator to
13060 get_mem_refs_of_builtin_call.
13061 (last_alloca_addr): New global.
13062 * asan.h (asan_emit_allocas_unpoison): Declare.
13063 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13064 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13065 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13066 if function calls alloca.
13067 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13068 * gimple-fold.h (replace_call_with_value): Declare.
13069 * internal-fn.c: Include asan.h.
13070 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13071 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13073 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13075 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13076 (C_SELFTEST_FLAGS): New.
13077 (CPP_SELFTEST_FLAGS): New.
13078 (SELFTEST_DEPS): New, from deps of s-selftest.
13079 (C_SELFTEST_DEPS): New, from deps of s-selftest.
13080 (CPP_SELFTEST_DEPS): New.
13081 (selftest): Add dependency on s-selftest-c++.
13082 (s-selftest): Rename to...
13083 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
13084 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
13085 than SELFTEST_FLAGS.
13086 (selftest-gdb): Rename to...
13087 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
13089 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
13090 (selftest-valgrind): Rename to...
13091 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
13093 (selftest-valgrind): Reintroduce as an alias for
13094 selftest-c-valgrind.
13095 (s-selftest-c++): New.
13096 (selftest-c++-gdb): New.
13097 (selftest-c++-valgrind): New.
13099 2017-07-06 Olivier Hainque <hainque@adacore.com>
13101 * gcc.c (process_command): When deciding if undefined variables
13102 should be ignored when processing specs, accept "gcc -v" as well.
13104 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13106 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
13107 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
13109 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13111 * config/arm/arm-cpus.in (armv8-r): Add new entry.
13112 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
13113 * config/arm/arm-tables.opt: Regenerate.
13114 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
13116 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
13118 2017-07-06 Carl Love <cel@us.ibm.com>
13120 * ChangeLog: Clean up from mid air collision
13122 2017-07-06 Carl Love <cel@us.ibm.com>
13124 * config/rs6000/rs6000-c.c: Add support for built-in functions
13125 vector signed int vec_subc (vector signed int, vector signed int);
13126 vector signed __int128 vec_subc (vector signed __int128,
13127 vector signed __int128);
13128 vector unsigned __int128 vec_subc (vector unsigned __int128,
13129 vector unsigned __int128);
13130 vector signed int vec_sube (vector signed int, vector signed int,
13131 vector signed int);
13132 vector unsigned int vec_sube (vector unsigned int,
13133 vector unsigned int,
13134 vector unsigned int);
13135 vector signed __int128 vec_sube (vector signed __int128,
13136 vector signed __int128,
13137 vector signed__int128);
13138 vector unsigned __int128 vec_sube (vector unsigned __int128,
13139 vector unsigned __int128,
13140 vector unsigned __int128);
13141 vector signed int vec_subec (vector signed int, vector signed int,
13142 vector signed int);
13143 vector unsigned int vec_subec (vector unsigned int,
13144 vector unsigned int,
13145 vector unsigned int);
13146 vector signed __int128 vec_subec (vector signed __int128,
13147 vector signed __int128,
13148 vector signed__int128);
13149 vector unsigned __int128 vec_subec (vector unsigned __int128,
13150 vector unsigned __int128,
13151 vector unsigned __int128);
13152 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
13153 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
13154 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
13155 BU_ALTIVEC_OVERLOAD_X definitions.
13156 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
13157 * doc/extend.texi: Update the built-in documentation file for the new
13158 built-in functions.
13160 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13163 * diagnostic-show-locus.c (layout::layout): Use start and finish
13164 spelling location for the start and finish of each range.
13165 * genmatch.c (linemap_client_expand_location_to_spelling_point):
13166 Add unused aspect param.
13167 * input.c (expand_location_1): Add "aspect" param, and use it
13168 to access the correct part of the location.
13169 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
13171 (expand_location_to_spelling_point): Likewise.
13172 (linemap_client_expand_location_to_spelling_point): Add "aspect"
13173 param, and pass it to expand_location_1.
13175 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
13177 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
13178 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
13179 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
13180 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
13181 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
13182 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
13183 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
13184 _mm_maskz_getmant_ss): New intrinsics.
13185 (__builtin_ia32_getexpss128_mask): Changed to ...
13186 __builtin_ia32_getexpss128_round ... this.
13187 (__builtin_ia32_getexpsd128_mask): Changed to ...
13188 __builtin_ia32_getexpsd128_round ... this.
13189 * config/i386/i386-builtin-types.def
13190 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
13191 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
13192 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
13193 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
13194 __builtin_ia32_getmantss_mask_round): New builtins.
13195 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
13196 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
13197 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
13198 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
13199 * config/i386/sse.md
13200 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
13201 avx512f_sgetexp<mode><mask_scalar_name>
13202 <round_saeonly_scalar_name> ... this.
13203 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
13204 %0, %1, %2<round_saeonly_op3>}): Changed to ...
13205 vgetexp<ssescalarmodesuffix>
13206 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13207 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
13208 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
13209 avx512f_vgetmant<mode><mask_scalar_name>
13210 <round_saeonly_scalar_name> ... this.
13211 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
13212 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
13213 vgetmant<ssescalarmodesuffix>
13214 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
13215 %0<mask_scalar_operand4>, %1, %2
13216 <round_saeonly_scalar_mask_op4>, %3} ... this.
13217 * config/i386/subst.md (mask_scalar_operand4,
13218 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
13219 round_saeonly_scalar_nimm_predicate): New subst attributes.
13221 2017-07-06 Julia Koval <julia.koval@intel.com>
13223 * config/i386/i386.c (ix86_erase_embedded_rounding):
13224 Remove code for old rounding pattern.
13226 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
13228 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
13230 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
13232 * doc/sourcebuild.texi (Test Directives, Variants of
13233 dg-require-support): Add documentation for dg-require-stack-check.
13235 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
13237 * config/i386/subst.md (mask_scalar, round_scalar,
13238 round_saeonly_scalar): New meta-templates.
13239 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
13240 round_scalar_mask_operand3, round_scalar_mask_op3,
13241 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
13242 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
13243 round_saeonly_scalar_constraint,
13244 round_saeonly_scalar_prefix): New subst attribute.
13245 * config/i386/sse.md
13246 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
13247 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
13248 <round_scalar_name> ... this.
13249 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
13250 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
13251 <round_scalar_name> ... this.
13252 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
13253 <sse>_vm<code><mode>3<mask_scalar_name>
13254 <round_saeonly_scalar_name> ... this.
13255 (v<plusminus_mnemonic><ssescalarmodesuffix>
13256 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13257 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13258 v<plusminus_mnemonic><ssescalarmodesuffix>
13259 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13260 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13261 (v<multdiv_mnemonic><ssescalarmodesuffix>
13262 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13263 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13264 v<multdiv_mnemonic><ssescalarmodesuffix>
13265 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13266 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13267 (v<maxmin_float><ssescalarmodesuffix>
13268 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
13269 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
13270 v<maxmin_float><ssescalarmodesuffix>
13271 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13272 %0<mask_scalar_operand3>, %1, %<iptr>2
13273 <round_saeonly_scalar_mask_op3>} ... this.
13275 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
13277 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
13278 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
13280 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
13281 Alan Hayward <alan.hayward@arm.com>
13282 David Sherwood <david.sherwood@arm.com>
13284 * combine.c (simplify_if_then_else): Remove "enum" before
13286 * compare-elim.c (can_eliminate_compare): Likewise.
13287 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13289 (aarch64_lookup_simd_builtin_type): Likewise.
13290 (aarch64_simd_builtin_type): Likewise.
13291 (aarch64_init_simd_builtin_types): Likewise.
13292 (aarch64_simd_expand_args): Likewise.
13293 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
13295 (aarch64_reverse_mask): Likewise.
13296 (aarch64_simd_emit_reg_reg_move): Likewise.
13297 (aarch64_gen_adjusted_ldpstp): Likewise.
13298 (aarch64_ccmp_mode_to_code): Likewise.
13299 (aarch64_operands_ok_for_ldpstp): Likewise.
13300 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13301 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
13303 (aarch64_min_divisions_for_recip_mul): Likewise.
13304 (aarch64_reassociation_width): Likewise.
13305 (aarch64_get_condition_code_1): Likewise.
13306 (aarch64_simd_emit_reg_reg_move): Likewise.
13307 (aarch64_simd_attr_length_rglist): Likewise.
13308 (aarch64_reverse_mask): Likewise.
13309 (aarch64_operands_ok_for_ldpstp): Likewise.
13310 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13311 (aarch64_gen_adjusted_ldpstp): Likewise.
13312 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
13314 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
13315 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13316 (arm_lookup_simd_builtin_type): Likewise.
13317 (arm_simd_builtin_type): Likewise.
13318 (arm_init_simd_builtin_types): Likewise.
13319 (arm_expand_builtin_args): Likewise.
13320 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
13321 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
13322 (ft32_setup_incoming_varargs): Likewise.
13323 (ft32_function_arg): Likewise.
13324 (ft32_function_arg_advance): Likewise.
13325 (ft32_pass_by_reference): Likewise.
13326 (ft32_arg_partial_bytes): Likewise.
13327 (ft32_valid_pointer_mode): Likewise.
13328 (ft32_addr_space_pointer_mode): Likewise.
13329 (ft32_addr_space_legitimate_address_p): Likewise.
13330 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
13332 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
13333 (ix86_emit_outlined_ms2sysv_restore): Likewise.
13334 (iamcu_alignment): Likewise.
13335 (canonicalize_vector_int_perm): Likewise.
13336 (ix86_noce_conversion_profitable_p): Likewise.
13337 (ix86_mpx_bound_mode): Likewise.
13338 (ix86_operands_ok_for_move_multiple): Likewise.
13339 * config/microblaze/microblaze-protos.h
13340 (microblaze_expand_conditional_branch_reg): Likewise.
13341 * config/microblaze/microblaze.c
13342 (microblaze_expand_conditional_branch_reg): Likewise.
13343 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13345 (rs6000_reassociation_width): Likewise.
13346 (rs6000_invalid_binary_op): Likewise.
13347 (fusion_p9_p): Likewise.
13348 (emit_fusion_p9_load): Likewise.
13349 (emit_fusion_p9_store): Likewise.
13350 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
13352 (riscv_hard_regno_mode_ok_p): Likewise.
13353 (riscv_address_insns): Likewise.
13354 (riscv_split_symbol): Likewise.
13355 (riscv_legitimize_move): Likewise.
13356 (riscv_function_value): Likewise.
13357 (riscv_hard_regno_nregs): Likewise.
13358 (riscv_expand_builtin): Likewise.
13359 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
13360 (riscv_build_integer): Likewise.
13361 (riscv_split_integer): Likewise.
13362 (riscv_legitimate_constant_p): Likewise.
13363 (riscv_cannot_force_const_mem): Likewise.
13364 (riscv_regno_mode_ok_for_base_p): Likewise.
13365 (riscv_valid_base_register_p): Likewise.
13366 (riscv_valid_offset_p): Likewise.
13367 (riscv_valid_lo_sum_p): Likewise.
13368 (riscv_classify_address): Likewise.
13369 (riscv_legitimate_address_p): Likewise.
13370 (riscv_address_insns): Likewise.
13371 (riscv_load_store_insns): Likewise.
13372 (riscv_force_binary): Likewise.
13373 (riscv_split_symbol): Likewise.
13374 (riscv_force_address): Likewise.
13375 (riscv_legitimize_address): Likewise.
13376 (riscv_move_integer): Likewise.
13377 (riscv_legitimize_const_move): Likewise.
13378 (riscv_legitimize_move): Likewise.
13379 (riscv_address_cost): Likewise.
13380 (riscv_subword): Likewise.
13381 (riscv_output_move): Likewise.
13382 (riscv_canonicalize_int_order_test): Likewise.
13383 (riscv_emit_int_order_test): Likewise.
13384 (riscv_function_arg_boundary): Likewise.
13385 (riscv_pass_mode_in_fpr_p): Likewise.
13386 (riscv_pass_fpr_single): Likewise.
13387 (riscv_pass_fpr_pair): Likewise.
13388 (riscv_get_arg_info): Likewise.
13389 (riscv_function_arg): Likewise.
13390 (riscv_function_arg_advance): Likewise.
13391 (riscv_arg_partial_bytes): Likewise.
13392 (riscv_function_value): Likewise.
13393 (riscv_pass_by_reference): Likewise.
13394 (riscv_setup_incoming_varargs): Likewise.
13395 (riscv_print_operand): Likewise.
13396 (riscv_elf_select_rtx_section): Likewise.
13397 (riscv_save_restore_reg): Likewise.
13398 (riscv_for_each_saved_reg): Likewise.
13399 (riscv_register_move_cost): Likewise.
13400 (riscv_hard_regno_mode_ok_p): Likewise.
13401 (riscv_hard_regno_nregs): Likewise.
13402 (riscv_class_max_nregs): Likewise.
13403 (riscv_memory_move_cost): Likewise.
13404 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
13405 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
13406 (rl78_addr_space_address_mode): Likewise.
13407 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13409 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13410 (rs6000_reassociation_width): Likewise.
13411 (rs6000_invalid_binary_op): Likewise.
13412 (fusion_p9_p): Likewise.
13413 (emit_fusion_p9_load): Likewise.
13414 (emit_fusion_p9_store): Likewise.
13415 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
13416 (ok_for_simple_move_operands): Likewise.
13417 (ok_for_simple_move_strict_operands): Likewise.
13418 (ok_for_simple_arith_logic_operands): Likewise.
13419 (visium_legitimize_reload_address): Likewise.
13420 (visium_select_cc_mode): Likewise.
13421 (output_cbranch): Likewise.
13422 (visium_split_double_move): Likewise.
13423 (visium_expand_copysign): Likewise.
13424 (visium_expand_int_cstore): Likewise.
13425 (visium_expand_fp_cstore): Likewise.
13426 * config/visium/visium.c (visium_pass_by_reference): Likewise.
13427 (visium_function_arg): Likewise.
13428 (visium_function_arg_advance): Likewise.
13429 (visium_libcall_value): Likewise.
13430 (visium_setup_incoming_varargs): Likewise.
13431 (visium_legitimate_constant_p): Likewise.
13432 (visium_legitimate_address_p): Likewise.
13433 (visium_legitimize_address): Likewise.
13434 (visium_secondary_reload): Likewise.
13435 (visium_register_move_cost): Likewise.
13436 (visium_memory_move_cost): Likewise.
13437 (prepare_move_operands): Likewise.
13438 (ok_for_simple_move_operands): Likewise.
13439 (ok_for_simple_move_strict_operands): Likewise.
13440 (ok_for_simple_arith_logic_operands): Likewise.
13441 (visium_function_value_1): Likewise.
13442 (rtx_ok_for_offset_p): Likewise.
13443 (visium_legitimize_reload_address): Likewise.
13444 (visium_split_double_move): Likewise.
13445 (visium_expand_copysign): Likewise.
13446 (visium_expand_int_cstore): Likewise.
13447 (visium_expand_fp_cstore): Likewise.
13448 (visium_split_cstore): Likewise.
13449 (visium_select_cc_mode): Likewise.
13450 (visium_split_cbranch): Likewise.
13451 (output_cbranch): Likewise.
13452 (visium_print_operand_address): Likewise.
13453 * expmed.c (flip_storage_order): Likewise.
13454 * expmed.h (emit_cstore): Likewise.
13455 (flip_storage_order): Likewise.
13456 * genrecog.c (validate_pattern): Likewise.
13457 * hsa-gen.c (gen_hsa_addr): Likewise.
13458 * internal-fn.c (expand_arith_overflow): Likewise.
13459 * ira-color.c (allocno_copy_cost_saving): Likewise.
13460 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13461 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
13462 (process_invariant_for_inheritance): Likewise.
13463 * lra-eliminations.c (move_plus_up): Likewise.
13464 * omp-low.c (lower_oacc_reductions): Likewise.
13465 * simplify-rtx.c (simplify_subreg): Likewise.
13466 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
13467 (TARGET_CHKP_BOUND_MODE): Likewise..
13468 * targhooks.c (default_chkp_bound_mode): Likewise.
13469 (default_setup_incoming_vararg_bounds): Likewise.
13470 * targhooks.h (default_chkp_bound_mode): Likewise.
13471 (default_setup_incoming_vararg_bounds): Likewise.
13472 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
13473 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
13474 (have_whole_vector_shift): Likewise.
13475 * tree-vect-stmts.c (vectorizable_load): Likewise.
13476 * doc/tm.texi: Regenerate.
13478 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13480 Graceful degrade if Binutils PR21472 is not available.
13483 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
13484 .rodata in flash test fails.
13485 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
13486 * confgure: Regenerate.
13487 * config.in: Regenerate.
13488 * config/avr/avr.c (avr_asm_named_section)
13489 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
13490 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
13491 (avr_asm_init_sections): Same.
13493 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13495 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
13496 (fma<VH:mode>4_intrinsic): Likewise.
13497 (*fmsub<VCVTF:mode>4): Likewise.
13498 (*fmsub<VH:mode>4_intrinsic): Likewise.
13500 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13503 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
13504 Don't depend on "optimize > 0".
13505 (out_movhi_r_mr, out_movqi_mr_r): Same.
13506 (out_movhi_mr_r, out_movqi_r_mr): Same.
13507 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
13508 io_address_operand on "optimize > 0".
13510 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13512 * tree-loop-distribution.c: Add general explanantion on the pass.
13513 (generate_loops_for_partition): Mark distributed loop.
13514 (pg_add_dependence_edges): New parameter. Handle alias data
13515 dependence specially and record it in the parameter if asked.
13516 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
13517 (init_partition_graph_vertices, add_partition_graph_edge): New.
13518 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
13519 (free_partition_graph_vdata, build_partition_graph): New.
13520 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
13521 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
13522 (data_ref_segment_size, latch_dominated_by_data_ref): New.
13523 (compute_alias_check_pairs, version_loop_by_alias_check): New.
13524 (version_for_distribution_p, finalize_partitions): New.
13525 (distribute_loop): Handle alias data dependence specially. Factor
13526 out loop fusion code as functions and call these functions.
13528 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13530 * tree-loop-distribution.c (classify_partition): New parameter and
13531 better handle reduction statement.
13532 (rdg_build_partitions): Revise comment.
13533 (distribute_loop): Compute statements in all partitions and pass it
13534 to classify_partition.
13536 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13538 * tree-loop-distribution.c (enum partition_type): New.
13539 (struct partition): New field type.
13540 (partition_merge_into): Add parameter. Update partition type.
13541 (data_dep_in_cycle_p, update_type_for_merge): New functions.
13542 (build_rdg_partition_for_vertex): Compute partition type.
13543 (rdg_build_partitions): Dump partition type.
13544 (distribute_loop): Update calls to partition_merge_into.
13546 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13548 * tree-loop-distribution.c (struct ddr_hasher): New.
13549 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
13551 (classify_partition): Call get_data_dependence.
13552 (pg_add_dependence_edges): Ditto.
13553 (distribute_loop): Release data dependence hash table.
13555 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13557 * tree-loop-distribution.c (ref_base_address): Delete.
13558 (similar_memory_accesses): Rename ...
13559 (share_memory_accesses): ... to this. Check if partitions access
13560 the same memory reference.
13561 (distribute_loop): Call share_memory_accesses.
13563 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13565 * tree-loop-distribution.c (struct partition): New field recording
13566 its data reference.
13567 (partition_alloc, partition_free): Init and release data refs.
13568 (partition_merge_into): Merge data refs.
13569 (build_rdg_partition_for_vertex): Collect data refs for partition.
13570 (pg_add_dependence_edges): Change parameters from vector to bitmap.
13572 (distribute_loop): Remve data refs from vertice data of partition
13575 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13577 * tree-loop-distribution.c (params.h): Include header file.
13578 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
13579 (datarefs_vec): New global var.
13580 (create_rdg_vertices): Use datarefs_vec directly.
13581 (free_rdg): Don't free data references.
13582 (build_rdg): Update use. Don't free data references.
13583 (distribute_loop): Compute global variable for data references.
13584 Bail out if there are too many data references.
13586 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13588 * tree-loop-distribution.c (loop_nest): New global var.
13589 (build_rdg): Use loop directly, rather than loop nest.
13590 (pg_add_dependence_edges): Remove loop nest parameter. Use global
13592 (distribute_loop): Compute global variable loop nest. Update use.
13594 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13596 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
13597 (partition_merge_into): New parameter. Dump reason for fusion.
13598 (distribute_loop): Update use of partition_merge_into.
13600 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13602 * tree-loop-distribution.c (bb_top_order_index): New.
13603 (bb_top_order_index_size, bb_top_order_cmp): New.
13604 (stmts_from_loop): Use topological order.
13605 (pass_loop_distribution::execute): Compute and release topological
13606 order for basic blocks.
13608 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13610 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
13613 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13615 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
13616 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
13617 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
13618 * internal-fn.def (LOOP_DIST_ALIAS): New.
13619 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
13620 (fold_loop_internal_call): ... this.
13621 (vect_loop_dist_alias_call): New function.
13622 (set_uid_loop_bbs): Call fold_loop_internal_call.
13623 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
13626 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13629 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
13630 Require dead FLAGS_REG at the beginning of a peephole.
13632 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13635 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
13636 arguments in the call to __builtin_ia32_sbb_u32.
13637 (_subborrow_u64): Swap _X and _Y arguments in the call to
13638 __builtin_ia32_sbb_u64.
13640 2017-07-04 Jakub Jelinek <jakub@redhat.com>
13643 * tree-vrp.c (compare_assert_loc): Turn into a function template
13644 with stable template parameter. Only test if a->e is NULL,
13645 !a->e == !b->e has been verified already. Use e == NULL or
13646 e != NULL instead of e or ! e tests. If stable is true, don't use
13647 iterative_hash_expr, on the other side allow a or b or both NULL
13648 and sort the NULLs last.
13649 (process_assert_insertions): Sort using compare_assert_loc<false>
13650 instead of compare_assert_loc, later sort using
13651 compare_assert_loc<true> before calling process_assert_insertions_for
13652 in a loop. Use break instead of continue once seen NULL pointer.
13654 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13656 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13657 Cortex-R7 and Cortex-R8 processors.
13659 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13661 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
13662 uninitialized while src is not.
13664 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
13666 * common/config/arm/arm-common.c: Adjust include path for
13668 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
13669 (arm-cpu.h): Create in build directory. Adjust dependency rules.
13670 (arm-cpu-data.h): Likewise.
13671 (arm-cpu-cdata.h): Likewise.
13672 * config/arm/arm-cpu.h: Delete.
13673 * config/arm/arm-cpu-cdata.h: Delete.
13674 * config/arm/arm-cpu-data.h: Delete.
13676 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
13678 * config/arm/arm-cpus.in (cortex-a55): New.
13679 (cortex-a75): Likewise.
13680 (cortex-a75.cortex-a55): Likewise.
13681 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
13683 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
13684 * config/arm/arm-cpu-cdata.h: Regenerate.
13685 * config/arm/arm-cpu-data.h: Regenerate.
13686 * config/arm/arm-cpu.h: Regenerate.
13687 * config/arm/arm-tables.opt: Regenerate.
13688 * config/arm/arm-tune.md: Regenerate.
13690 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13692 * haifa-sched.c (sched_create_recovery_edges): Update profile.
13694 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13696 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
13699 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
13701 PR tree-optimization/81292
13702 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
13703 full_string_p, also call adjust_related_strinfos if the adjustment
13704 is simple, otherwise invalidate related strinfos.
13706 2017-07-04 Martin Liska <mliska@suse.cz>
13709 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
13710 newly created variable as DECL_IGNORED_P.
13712 2017-07-04 Martin Liska <mliska@suse.cz>
13715 * ipa-inline.c (inline_small_functions):
13716 Use xstrdup_for_dump.
13718 2017-07-04 Tom de Vries <tom@codesourcery.com>
13720 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
13722 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13725 * config/darwin.c (darwin_function_switched_text_sections):
13728 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
13730 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
13732 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13734 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
13736 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13738 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
13739 min_profitable_iters, and th as inclusive lower bounds.
13740 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
13741 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
13742 for min_profitable_iters and min_profitable_estimate.
13743 (vect_transform_loop): Treat th as an inclusive lower bound.
13744 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
13746 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13749 * config/darwin.c (darwin_function_switched_text_sections):
13750 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
13751 in two pieces, and suppress the use of buf.
13753 2017-07-03 Nathan Sidwell <nathan@acm.org>
13755 * hash-table.h (hash_table_mod1): Fix indentation.
13757 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13759 PR middle-end/81290
13760 * predict.c (force_edge_cold): Be more careful about propagation
13762 * profile-count.h (profile_probability::guessed,
13763 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
13765 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
13767 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
13769 * doc/invoke.texi (rcpc architecture extension): Document it.
13771 2017-07-03 Richard Biener <rguenther@suse.de>
13773 PR tree-optimization/60510
13774 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
13775 the scalar reduction PHI and use it.
13776 (vectorizable_reduction): Properly guard the single_defuse_cycle
13777 path for non-SLP reduction chains where we cannot use it.
13778 Rework reduc_def/index and vector type deduction. Rework
13779 vector operand gathering during reduction op code-gen.
13780 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
13781 chains dissolve the chain and leave it to non-SLP reduction
13784 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13786 * tree-data-ref.h (dr_alignment): Declare.
13787 * tree-data-ref.c (dr_alignment): New function.
13788 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
13789 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
13791 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
13793 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13795 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
13796 and base_misalignment fields.
13797 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
13798 * tree-data-ref.c: Include builtins.h.
13799 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
13800 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
13801 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
13802 * tree-vect-data-refs.c: Include tree-cfg.h.
13803 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
13804 fields instead of calculating an alignment here.
13805 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
13806 innermost_loop_behavior fields.
13808 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13810 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
13812 (DR_STEP_ALIGNMENT): New macro.
13813 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
13814 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
13815 (create_data_ref): Print it.
13816 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
13817 to tell whether the step preserves vector (mis)alignment.
13818 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13819 Move the check for an integer step and generalise to all INTEGER_CST.
13820 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
13821 Print the outer step alignment.
13823 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13825 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
13826 with offset_alignment.
13827 (DR_ALIGNED_TO): Delete.
13828 (DR_OFFSET_ALIGNMENT): New macro.
13829 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
13830 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
13831 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
13832 (create_data_ref): Likewise.
13833 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13834 (vect_analyze_data_refs): Likewise.
13835 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
13836 creating dummy innermost behavior.
13838 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13840 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
13841 with a "innermost_loop_behavior *" and refeence tree.
13842 * tree-data-ref.c (dr_analyze_innermost): Likewise.
13843 (create_data_ref): Update call accordingly.
13844 * tree-predcom.c (find_looparound_phi): Likewise.
13846 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13848 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
13849 fields with dr_wrt_vec_loop.
13850 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
13851 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
13852 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
13853 (vect_dr_behavior): New function.
13854 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13855 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
13856 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
13857 track whether the step preserves the misalignment.
13858 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13859 Use vect_dr_behavior.
13860 (vect_setup_realignment): Update call accordingly.
13861 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
13862 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
13863 call to vect_create_addr_base_for_vector_ref.
13864 (vect_create_cond_for_align_checks): Likewise.
13865 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
13866 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
13867 (vect_recog_mask_conversion_pattern): Likewise.
13868 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
13869 (new_stmt_vec_info): Remove redundant zeroing.
13871 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13873 * common/config/arm/arm-common.c (arm_be8_option): New function.
13874 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
13875 (ISA_ARMv6): Add isa_bit_be8.
13876 * config/arm/arm.h (arm_be8_option): Add prototype.
13877 (BE8_SPEC_FUNCTION): New define.
13878 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
13879 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
13880 (mlittle-endian): Similarly.
13881 (mbe8, mbe32): New options.
13882 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
13883 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
13885 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13887 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
13889 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13891 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
13892 (cleanup_tree_cfg_bb): Use it.
13893 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
13895 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
13897 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13900 * loop-doloop.c (add_test): Update profile.
13902 2017-07-03 Martin Liska <mliska@suse.cz>
13905 * sanopt.c (rewrite_usage_of_param): New function.
13906 (sanitize_rewrite_addressable_params): Likewise.
13907 (pass_sanopt::execute): Call rewrite_usage_of_param.
13909 2017-07-03 Richard Biener <rguenther@suse.de>
13911 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
13912 back to using VIEW_CONVERT_EXPR.
13914 2017-07-03 Martin Liska <mliska@suse.cz>
13917 * doc/extend.texi: Document when a resolver function is
13918 generated for target_clones.
13920 2017-07-03 Martin Liska <mliska@suse.cz>
13922 * asan.c (asan_emit_stack_protection): Unpoison just red zones
13923 and shadow memory of auto variables which are subject of
13924 use-after-scope sanitization.
13925 (asan_expand_mark_ifn): Add do set only when is_poison.
13927 2016-07-03 Richard Biener <rguenther@suse.de>
13929 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
13931 (vect_force_simple_reduction): Record reduction def -> phi mapping.
13932 (vectorizable_reduction): Perform reduction PHI creation when
13933 visiting a reduction PHI and adjust and simplify code generation
13934 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
13935 (vect_transform_loop): Visit reduction PHIs.
13936 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
13937 defs into the SLP tree.
13938 (vect_build_slp_tree): Reduction defs terminate the recursion.
13939 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
13941 (vect_get_vec_defs_for_stmt_copy): Export.
13942 (vect_get_vec_defs): Likewise.
13943 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
13945 (vect_get_vec_defs_for_stmt_copy): Declare.
13946 (vect_get_vec_defs): Likewise.
13948 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13950 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
13951 parameter with a "loop" parameter and use it instead of the
13952 loop containing DR_STMT. Don't check simple_iv when doing
13953 BB analysis. Describe the two analysis modes in the comment.
13955 2017-07-03 Tom de Vries <tom@codesourcery.com>
13957 PR tree-optimization/69468
13958 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
13959 (find_same_succ_bb): Handle ignore_edge_flags.
13961 2017-07-03 Tom de Vries <tom@codesourcery.com>
13963 PR tree-optimization/81192
13964 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
13966 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
13968 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
13970 2017-07-03 Tom de Vries <tom@codesourcery.com>
13972 PR tree-optimization/81192
13973 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
13974 BB_SAME_SUCC (bb) == NULL.
13976 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13978 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
13981 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13983 * dumpfile.c: Include profile-count.h
13984 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
13986 (insert_cond_bb): Update profile.
13987 * tree-cfg.h (insert_cond_bb): Update prototype.
13988 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
13989 * tree-dump.c: Do not include tree-cfg.
13991 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13993 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
13995 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13997 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14000 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14002 * tree-complex.c (expand_complex_div_wide): update profile.
14004 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14005 Alan Hayward <alan.hayward@arm.com>
14006 David Sherwood <david.sherwood@arm.com>
14008 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14009 (CORETYPES_H): New define.
14010 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14011 (insn-modes-inline.h, s-modes-inline-h): New rules.
14012 (generated_files): Add insn-modes-inline.h.
14013 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14014 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14015 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14016 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14017 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14018 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14019 (build/gencodes.o, build/genconditions.o): Likewise.
14020 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14021 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14022 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14023 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14024 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14025 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14026 * coretypes.h: Include everything up to real.h for generators.
14027 Include insn-modes.h first. Include wide-int-print.h after
14028 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14029 * machmode.h: Don't include insn-modes.h here.
14030 * function-tests.c: Remove includes of signop.h, machmode.h,
14031 double-int.h and wide-int.h.
14033 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14035 * optc-save-gen.awk: Likewise.
14036 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14037 * godump.c: Remove include of wide-int-print.h.
14038 * pretty-print.h: Likewise.
14039 * wide-int-print.cc: Likewise.
14040 * wide-int.cc: Likewise.
14041 * hash-map-tests.c: Remove include of signop.h.
14042 * hash-set-tests.c: Likewise.
14043 * rtl-tests.c: Likewise.
14044 * mkconfig.sh: Remove include of machmode.h.
14045 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14047 (emit_insn_modes_inline_h): ...this new function. Emit the code
14048 into an insn-modes-inline.h header file, adding appropriate
14049 include guards and end comments.
14050 (emit_insn_modes_c_header): Remove include of machmode.h.
14051 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14053 (main): Handle -i flag and call emit_insn_modes_inline_h when
14056 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14058 * tree-ssa-strlen.c (strinfo): Rename the length field to
14059 nonzero_chars. Add a full_string_p field.
14060 (compare_nonzero_chars, zero_length_string_p): New functions.
14061 (get_addr_stridx): Add an offset_out parameter.
14062 Use compare_nonzero_chars.
14063 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14064 (new_strinfo): Update after above changes to strinfo.
14065 (set_endptr_and_length): Set full_string_p.
14066 (get_string_length): Update after above changes to strinfo.
14067 (unshare_strinfo): Update call to new_strinfo.
14068 (maybe_invalidate): Likewise.
14069 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14070 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14071 as a uhwi instead of an shwi. Update after above changes to
14072 strinfo and new_strinfo.
14073 (zero_length_string): Assert that chainsi contains full strings.
14074 Use zero_length_string_p. Update call to new_strinfo.
14075 (adjust_related_strinfos): Update after above changes to strinfo.
14076 Copy full_string_p from origsi.
14077 (adjust_last_stmt): Use zero_length_string_p.
14078 (handle_builtin_strlen): Update after above changes to strinfo and
14079 new_strinfo. Install the lhs as the string length if the previous
14080 entry didn't describe a full string.
14081 (handle_builtin_strchr): Update after above changes to strinfo
14083 (handle_builtin_strcpy): Likewise.
14084 (handle_builtin_strcat): Likewise.
14085 (handle_builtin_malloc): Likewise.
14086 (handle_pointer_plus): Likewise.
14087 (handle_builtin_memcpy): Likewise. Track nonzero characters
14088 that aren't necessarily followed by a nul terminator.
14089 (handle_char_store): Likewise.
14091 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14093 PR tree-optimization/80769
14094 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
14095 for malloc and calloc. Document the new invariant that all related
14096 strinfos have delayed lengths or none do.
14097 (verify_related_strinfos): Move earlier in file.
14098 (set_endptr_and_length): New function, split out from...
14099 (get_string_length): ...here. Also set the lengths of related
14101 (zero_length_string): Assert that chainsi has known (rather than
14103 (adjust_related_strinfos): Likewise.
14105 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14107 PR tree-optimization/81136
14108 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
14109 assert that two references with the same misalignment have the same
14110 compile-time misalignment if those compile-time misalignments
14113 2017-07-01 Andi Kleen <ak@linux.intel.com>
14115 * print-tree.c (print_node): Print all attributes.
14117 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14119 * cfg.c (scale_bbs_frequencies): New function.
14120 * cfg.h (scale_bbs_frequencies): Declare it.
14121 * cfgloopanal.c (single_likely_exit): Cleanup.
14122 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
14124 (scale_loop_profile): Likewise.
14125 (loop_version): Likewise.
14126 (create_empty_loop_on_edge): Update.
14127 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
14128 scale_loop_frequencies, scale_loop_profile, loopify,
14129 loop_version): Update prototypes.
14130 * modulo-sched.c (sms_schedule): Update.
14131 * predict.c (unlikely_executed_edge_p): Also check probability.
14132 (probably_never_executed_edge_p): Fix typo.
14133 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14134 * tree-parloops.c (gen_parallel_loop): Update.
14135 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
14136 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14137 * tree-ssa-loop-split.c (split_loop): Update.
14138 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14139 * tree-vect-loop-manip.c (vect_do_peeling): Update.
14140 (vect_loop_versioning): Update.
14141 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14143 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14145 * trans-mem.c (split_bb_make_tm_edge): Update profile.
14147 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14149 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
14150 to keep profile consistent.
14152 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14154 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
14155 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
14156 * profile-count.h (max_safe_multiplier): Make unsigned.
14157 (profile_count::guessed_zero): New.
14159 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14161 * bb-reorder.c (fix_up_crossing_landing_pad,
14162 fix_crossing_conditional_branches): Use make_single_succ_edge
14163 to keep profile consistent.
14165 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14167 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
14170 2017-07-01 Jakub Jelinek <jakub@redhat.com>
14173 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
14174 the right scopes, make sure cond_jump isn't preserved between multiple
14175 iterations. Search for fallthru edge whenever there are 3+ edges and
14176 use find_fallthru_edge for it.
14178 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14180 Patch by Alexander Monakov <amonakov@ispras.ru>
14181 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
14182 probabilities consistently.
14184 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14186 * pa.c (pa_expand_compare_and_swap_loop): Update call of
14187 emit_cmp_and_jump_insns.
14189 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14192 * tree-inline.c (expand_call_inline): Combine profile statuses.
14194 2017-06-30 Andrew Pinski <apinski@cavium.com>
14196 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
14197 fold_stmt returned true.
14199 2017-06-30 Nathan Sidwell <nathan@acm.org>
14201 * ggc.h (empty_string): Delete.
14202 * cfgexpand.c (expand_asm_stmt): Use plain "".
14203 * optabs.c (expand_asm_memory_barrier): Likewise.
14204 * stringpool.c (empty_string): Delete.
14205 (digit_vector, digit_string): Delete.
14206 (ggc_alloc_string): Use plain "", don't optimize single digit
14207 strings. Use ggc_alloc_atomic.
14209 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
14211 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
14212 comparison set and one other set, use the cost of the non-comparison
14215 2017-06-30 Nathan Sidwell <nathan@acm.org>
14217 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
14220 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
14222 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
14223 loops. Remove now unneeded calls to gimple_switch_set_label() that
14224 just set removed labels to NULL_TREE.
14226 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
14228 * tree-ssanames.c (set_range_info_raw): Abstract from ...
14229 (set_range_info): ...here. Only call set_range_info_raw if domain
14231 (set_nonzero_bits): Call set_range_info_raw.
14232 * tree-ssanames.h (set_range_info_raw): New.
14234 2017-06-30 Jakub Jelinek <jakub@redhat.com>
14237 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
14238 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
14239 of nonimmediate_operand and <store_mask_constraint> instead of m
14240 for the input operand. For V8FI iterator, always split if input
14241 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
14242 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
14243 <store_mask_predicate> instead of register_operand and
14244 <store_mask_constraint> instead of v for the input operand. Make
14245 sure both operands aren't MEMs for if not <mask_applied>.
14247 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
14249 * lto-wrapper.c (copy_file) Close both file descriptors before
14252 2017-06-30 Martin Liska <mliska@suse.cz>
14255 * multiple_target.c (create_dispatcher_calls): Make ifunc
14256 also for function that don't have calls or are not referenced.
14258 2017-06-30 Richard Biener <rguenther@suse.de>
14260 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
14261 analyze the first scalar stmt. Move vector type computation
14262 for the BB case here from ...
14263 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
14264 live operation processing in the SLP case properly.
14266 2017-06-30 Richard Biener <rguenther@suse.de>
14268 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
14270 2017-06-30 Martin Liska <mliska@suse.cz>
14273 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
14274 before BUILT_IN_UNWIND_RESUME when ASAN is used.
14276 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
14278 * doc/invoke.texi (AArch64): Add missing options and remove redundant
14281 2017-06-30 Richard Biener <rguenther@suse.de>
14283 PR tree-optimization/81249
14284 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
14285 condition reduction result to original scalar type.
14287 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14289 * profile-count.h (enum profile_quality): Fix typos and whitespace
14292 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14294 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
14295 type for branch probabilities.
14297 2017-06-29 Julian Brown <julian@codesourcery.com>
14298 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14300 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
14301 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
14302 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
14303 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
14305 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14307 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
14308 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
14309 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
14310 CC usage from generic code to here.
14311 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
14312 CC usage into the target macros.
14314 2017-06-29 Maya Rashish <coypu@sdf.org>
14316 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
14319 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14321 * arm/arm-builtins.c: Include profile-count.h
14322 * except.c (sjlj_emit_function_enter): Use
14323 profile_probability::unlikely.
14325 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14327 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
14328 and tocrel_offset be pointer args rather than implicitly using
14330 (legitimate_constant_pool_address_p, rs6000_emit_move,
14331 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
14332 tocrel_offset and use in toc_relative_expr_p call.
14333 (print_operand, print_operand_address): Use static tocrel_base_oac
14334 and tocrel_offset_oac.
14335 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
14338 2017-06-29 Maya Rashish <coypu@sdf.org>
14340 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
14342 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
14344 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
14345 objects, take into account only the alignment of 'op0' and 'mode1' if
14348 2017-06-29 Steve Ellcey <sellcey@cavium.com>
14350 * ccmp.c (ccmp_tree_comparison_p): New function.
14351 (ccmp_candidate_p): Update to use above function.
14352 (get_compare_parts): New function.
14353 (expand_ccmp_next): Update to use new functions.
14354 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
14356 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
14357 take mode as argument.
14358 * ccmp.h (expand_ccmp_expr): Add mode as argument.
14359 * expr.c (expand_expr_real_1): Pass mode as argument.
14361 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
14363 * combine.c (combine_instructions): Print insns to dump_file, together
14366 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14368 * asan.c (asan_emit_stack_protection): Update.
14369 (create_cond_insert_point): Update.
14370 * auto-profile.c (afdo_propagate_circuit): Update.
14371 * basic-block.h (struct edge_def): Turn probability to
14372 profile_probability.
14373 (EDGE_FREQUENCY): Update.
14374 * bb-reorder.c (find_traces_1_round): Update.
14375 (better_edge_p): Update.
14376 (sanitize_hot_paths): Update.
14377 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
14378 (make_single_succ_edge): Update.
14379 (check_bb_profile): Update.
14380 (dump_edge_info): Update.
14381 (update_bb_profile_for_threading): Update.
14382 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
14383 probabilitycount to 0.
14384 * cfgbuild.c (compute_outgoing_frequencies): Update.
14385 * cfgcleanup.c (try_forward_edges): Update.
14386 (outgoing_edges_match): Update.
14387 (try_crossjump_to_edge): Update.
14388 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
14389 (expand_gimple_tailcall): Update.
14390 (construct_init_block): Use make_single_succ_edge.
14391 (construct_exit_block): Use make_single_succ_edge.
14392 * cfghooks.c (verify_flow_info): Update.
14393 (redirect_edge_succ_nodup): Update.
14394 (split_edge): Update.
14395 (account_profile_record): Update.
14396 * cfgloopanal.c (single_likely_exit): Update.
14397 * cfgloopmanip.c (scale_loop_profile): Update.
14398 (set_zero_probability): Remove.
14399 (duplicate_loop_to_header_edge): Update.
14400 * cfgloopmanip.h (loop_version): Update prototype.
14401 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
14402 (force_nonfallthru_and_redirect): Update.
14403 (update_br_prob_note): Update.
14404 (rtl_verify_edges): Update.
14405 (purge_dead_edges): Update.
14406 (rtl_lv_add_condition_to_bb): Update.
14407 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
14408 * cgraphunit.c (init_lowered_empty_function): Update.
14409 (cgraph_node::expand_thunk): Update.
14410 * cilk-common.c: Include profile-count.h
14411 * dojump.c (inv): Remove.
14412 (jumpifnot): Update.
14413 (jumpifnot_1): Update.
14414 (do_jump_1): Update.
14416 (do_jump_by_parts_greater_rtx): Update.
14417 (do_compare_rtx_and_jump): Update.
14418 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
14419 do_jump_1. do_compare_rtx_and_jump): Update prototype.
14420 * dwarf2cfi.c: Include profile-count.h
14421 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
14422 (sjlj_emit_dispatch_table): Likewise.
14423 * explow.c: Include profile-count.h
14424 * expmed.c (emit_store_flag_force): Update.
14425 (do_cmp_and_jump): Update.
14426 * expr.c (compare_by_pieces_d::generate): Update.
14427 (compare_by_pieces_d::finish_mode): Update.
14428 (emit_block_move_via_loop): Update.
14429 (store_expr_with_bounds): Update.
14430 (store_constructor): Update.
14431 (expand_expr_real_2): Update.
14432 (expand_expr_real_1): Update.
14433 * expr.h (try_casesi, try_tablejump): Update prototypes.
14434 * gimple-pretty-print.c (dump_probability): Update.
14435 (dump_profile): New.
14436 (dump_gimple_label): Update.
14437 (dump_gimple_bb_header): Update.
14438 * graph.c (draw_cfg_node_succ_edges): Update.
14439 * hsa-gen.c (convert_switch_statements): Update.
14440 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
14441 (find_if_case_1): Update.
14442 (find_if_case_2): Update.
14443 * internal-fn.c (expand_arith_overflow_result_store): Update.
14444 (expand_addsub_overflow): Update.
14445 (expand_neg_overflow): Update.
14446 (expand_mul_overflow): Update.
14447 (expand_vector_ubsan_overflow): Update.
14448 * ipa-cp.c (good_cloning_opportunity_p): Update.
14449 * ipa-split.c (split_function): Use make_single_succ_edge.
14450 * ipa-utils.c (ipa_merge_profiles): Update.
14451 * loop-doloop.c (add_test): Update.
14452 (doloop_modify): Update.
14453 * loop-unroll.c (compare_and_jump_seq): Update.
14454 (unroll_loop_runtime_iterations): Update.
14455 * lra-constraints.c (lra_inheritance): Update.
14456 * lto-streamer-in.c (input_cfg): Update.
14457 * lto-streamer-out.c (output_cfg): Update.
14458 * mcf.c (adjust_cfg_counts): Update.
14459 * modulo-sched.c (sms_schedule): Update.
14460 * omp-expand.c (expand_omp_for_init_counts): Update.
14461 (extract_omp_for_update_vars): Update.
14462 (expand_omp_ordered_sink): Update.
14463 (expand_omp_for_ordered_loops): Update.
14464 (expand_omp_for_generic): Update.
14465 (expand_omp_for_static_nochunk): Update.
14466 (expand_omp_for_static_chunk): Update.
14467 (expand_cilk_for): Update.
14468 (expand_omp_simd): Update.
14469 (expand_omp_taskloop_for_outer): Update.
14470 (expand_omp_taskloop_for_inner): Update.
14471 * omp-simd-clone.c (simd_clone_adjust): Update.
14472 * optabs.c (expand_doubleword_shift): Update.
14473 (expand_abs): Update.
14474 (emit_cmp_and_jump_insn_1): Update.
14475 (expand_compare_and_swap_loop): Update.
14476 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
14477 * predict.c (predictable_edge_p): Update.
14478 (edge_probability_reliable_p): Update.
14479 (set_even_probabilities): Update.
14480 (combine_predictions_for_insn): Update.
14481 (combine_predictions_for_bb): Update.
14482 (propagate_freq): Update.
14483 (estimate_bb_frequencies): Update.
14484 (force_edge_cold): Update.
14485 * profile-count.c (profile_count::dump): Add missing space into dump.
14486 (profile_count::debug): Add newline.
14487 (profile_count::differs_from_p): Explicitly convert to unsigned.
14488 (profile_count::stream_in): Update.
14489 (profile_probability::dump): New member function.
14490 (profile_probability::debug): New member function.
14491 (profile_probability::differs_from_p): New member function.
14492 (profile_probability::differs_lot_from_p): New member function.
14493 (profile_probability::stream_in): New member function.
14494 (profile_probability::stream_out): New member function.
14495 * profile-count.h (profile_count_quality): Rename to ...
14496 (profile_quality): ... this one.
14497 (profile_probability): New.
14498 (profile_count): Update.
14499 * profile.c (compute_branch_probabilities): Update.
14500 * recog.c (peep2_attempt): Update.
14501 * sched-ebb.c (schedule_ebbs): Update.
14502 * sched-rgn.c (find_single_block_region): Update.
14503 (compute_dom_prob_ps): Update.
14504 (schedule_region): Update.
14505 * sel-sched-ir.c (compute_succs_info): Update.
14506 * stmt.c (struct case_node): Update.
14507 (do_jump_if_equal): Update.
14508 (get_outgoing_edge_probs): Update.
14509 (conditional_probability): Update.
14510 (emit_case_dispatch_table): Update.
14511 (expand_case): Update.
14512 (expand_sjlj_dispatch_table): Update.
14513 (emit_case_nodes): Update.
14514 * targhooks.c: Update.
14515 * tracer.c (better_p): Update.
14516 (find_best_successor): Update.
14517 * trans-mem.c (expand_transaction): Update.
14518 * tree-call-cdce.c: Update.
14519 * tree-cfg.c (gimple_split_edge): Upate.
14520 (move_sese_region_to_fn): Upate.
14521 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
14522 * tree-eh.c (lower_resx): Upate.
14523 (cleanup_empty_eh_move_lp): Upate.
14524 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14525 * tree-inline.c (copy_edges_for_bb): Update.
14526 (copy_cfg_body): Update.
14527 * tree-parloops.c (gen_parallel_loop): Update.
14528 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
14529 (gimple_gen_time_profiler): Update.
14530 * tree-ssa-dce.c (remove_dead_stmt): Update.
14531 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14532 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
14533 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
14534 (unloop_loops): Update.
14535 (try_peel_loop): Update.
14536 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14537 * tree-ssa-loop-split.c (connect_loops): Update.
14538 (split_loop): Update.
14539 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14540 (hoist_guard): Update.
14541 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
14542 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
14543 (value_replacement): Update.
14544 * tree-ssa-reassoc.c (branch_fixup): Update.
14545 * tree-ssa-tail-merge.c (replace_block_by): Update.
14546 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
14547 (create_edge_and_update_destination_phis): Update.
14548 (compute_path_counts): Update.
14549 (recompute_probabilities): Update.
14550 (update_joiner_offpath_counts): Update.
14551 (freqs_to_counts_path): Update.
14552 (duplicate_thread_path): Update.
14553 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
14554 (struct switch_conv_info): Update.
14555 (gen_inbound_check): Update.
14556 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14557 (vect_do_peeling): Update.
14558 (vect_loop_versioning): Update.
14559 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14560 (optimize_mask_stores): Update.
14561 * ubsan.c (ubsan_expand_null_ifn): Update.
14562 * value-prof.c (gimple_divmod_fixed_value): Update.
14563 (gimple_divmod_fixed_value_transform): Update.
14564 (gimple_mod_pow2): Update.
14565 (gimple_mod_pow2_value_transform): Update.
14566 (gimple_mod_subtract): Update.
14567 (gimple_mod_subtract_transform): Update.
14568 (gimple_ic): Update.
14569 (gimple_stringop_fixed_value): Update.
14570 (gimple_stringops_transform): Update.
14571 * value-prof.h: Update.
14573 2017-06-29 Carl Love <cel@us.ibm.com>
14575 * config/rs6000/rs6000-c.c: Add support for built-in functions
14576 vector signed int vec_signed (vector float);
14577 vector signed long long vec_signed (vector double);
14578 vector signed int vec_signed2 (vector double, vector double);
14579 vector signed int vec_signede (vector double);
14580 vector signed int vec_signedo (vector double);
14581 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
14582 instruction generator.
14583 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14584 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
14585 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
14587 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
14588 vunsignede_v2df): Add define_expands.
14589 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
14590 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
14591 VEC_UNSIGNEDO): Add definitions.
14592 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14593 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
14594 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
14595 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
14596 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
14597 * config/rs6000/altivec.h (vec_signed, vec_signed2,
14598 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
14599 vec_unsignede, vec_unsignedo): Add builtin defines.
14600 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
14602 * doc/extend.texi: Update the built-in documentation file for the
14603 new built-in functions.
14605 2017-06-29 Richard Biener <rguenther@suse.de>
14607 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
14608 reduction chains to LOOP_VINFO_REDUCTIONS.
14609 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
14610 SLP reductions after processing reduction chains.
14612 2017-06-29 Nathan Sidwell <nathan@acm.org>
14614 * builtins.c (fold_builtin_FUNCTION): Use
14615 lang_hooks.decl_printable_name.
14617 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
14619 PR middle-end/81194
14620 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
14621 with only one label.
14622 * stmt.c (expand_case): Assert NCASES is greater than one.
14624 2017-06-29 Richard Biener <rguenther@suse.de>
14626 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
14628 (group_case_labels): Likewise.
14629 (find_taken_edge): Push sanity checking on val to workers...
14630 (find_taken_edge_cond_expr): ... here
14631 (find_taken_edge_switch_expr): ... and here, handle cases
14632 with just a default label.
14633 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
14634 (group_case_labels): Likewise.
14635 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
14636 group_case_labels does anything cleanup the CFG again.
14638 2017-06-29 Bin Cheng <bin.cheng@arm.com>
14640 PR tree-optimization/81196
14641 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
14642 exit condition comparing two IVs.
14644 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
14646 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
14647 profile to the dummy entry at the end of the list of architectures.
14648 * config/arm/arm-cpu-cdata.h: Regenerated.
14650 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14651 Michael Collison <michael.collison@arm.com>
14654 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
14656 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
14657 (*aarch64_reg_<mode>3_minus_mask): New pattern.
14658 (*aarch64_<optab>_reg_di3_mask2): New pattern.
14659 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
14660 of shift when the shift amount is masked with constant equal to
14661 the size of the mode.
14662 * config/aarch64/predicates.md (subreg_lowpart_operator): New
14665 2017-06-29 Martin Liska <mliska@suse.cz>
14667 * config/i386/i386.opt: Change range from [1,5] to [0,5].
14669 2017-06-29 Yury Gribov <tetra2005@gmail.com>
14672 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
14675 (edge_growth_cache_entry::edge_growth_cache_entry): New
14677 (reset_edge_growth_cache): Update to use constructor.
14679 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14681 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
14682 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14683 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14685 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
14687 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
14688 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
14690 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
14692 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
14693 (*-linux-uclibc*): Add t-uclibc tmake_file.
14694 * config/t-musl: New.
14695 * config/t-uclibc: New.
14697 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
14699 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
14701 (gen_comm_data): Emit architectural setting of arch_prof.
14702 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
14704 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
14705 (armv8-m.base, armv8-m.main): Likewise.
14706 * arm-protos.h (arm_build_target): Add profile field.
14707 (arch_option): Likewise.
14708 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
14710 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
14711 arm_active_target.profile.
14713 2017-06-28 Richard Biener <rguenther@suse.de>
14715 PR middle-end/81227
14716 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
14717 TYPE_OVERFLOW_WRAPS.
14718 * match.pd (negate_expr_p): Likewise.
14719 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
14720 fold_build2, not fold_binary.
14722 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14724 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
14725 Convert memory address to Pmode.
14726 (aarch64_print_operand): Assert MEM operands are always Pmode.
14728 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14731 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
14732 Remove redundant if.
14733 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
14734 * config/arm/aarch-common-protos.h
14735 (aarch_forward_to_shift_is_not_shifted_re): Remove.
14736 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
14738 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
14741 * multiple_target.c (create_dispatcher_calls): Set the default
14742 clone to be static, not public.
14744 2017-06-28 Richard Biener <rguenther@suse.de>
14746 * tree-vect-loop.c (vectorizable_reduction): Move special
14747 cond reduction IV var creation ...
14748 (vect_create_epilog_for_reduction): ... here. Remove induction_index
14749 parameter. Use STMT_VINFO_VECTYPE.
14750 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
14753 2017-06-28 Martin Liska <mliska@suse.cz>
14756 * ipa-visibility.c (non_local_p): Handle visibility.
14758 2017-06-28 Martin Liska <mliska@suse.cz>
14761 * common.opt: Add IntegerRange to various options.
14762 * opt-functions.awk (integer_range_info): New function.
14763 * optc-gen.awk: Add integer_range_info to cl_options struct.
14764 * opts-common.c (decode_cmdline_option): Handle
14765 CL_ERR_INT_RANGE_ARG.
14766 (cmdline_handle_error): Likewise.
14767 * opts.c (print_filtered_help): Show valid interval in
14768 when --help is provided.
14769 * opts.h (struct cl_option): Add range_min and range_max fields.
14770 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
14772 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
14774 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
14775 (x * C EQ/NE y * C): New transformation.
14777 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
14779 * genmultilib (combination_space): Accept '+' in option names.
14781 2017-06-28 Martin Liska <mliska@suse.cz>
14784 * asan.c (instrument_derefs): Bail out inner references
14785 that are hard register variables.
14787 2017-06-28 Jakub Jelinek <jakub@redhat.com>
14790 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
14791 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
14793 2017-06-28 Richard Biener <rguenther@suse.de>
14795 * tree-vectorizer.h (vect_get_vec_defs): Remove.
14796 (vect_get_slp_defs): Adjust.
14797 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
14799 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
14801 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
14802 get_initial_defs_for_reduction instead of vect_get_vec_defs.
14803 (vectorizable_reduction): Adjust.
14804 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
14806 (vect_get_slp_defs): Likewise.
14807 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
14808 (vectorizable_bswap): Adjust.
14809 (vectorizable_call): Likewise.
14810 (vectorizable_conversion): Likewise.
14811 (vectorizable_assignment): Likewise.
14812 (vectorizable_shift): Likewise.
14813 (vectorizable_operation): Likewise.
14814 (vectorizable_store): Likewise.
14815 (vectorizable_condition): Likewise.
14816 (vectorizable_comparison): Likewise.
14818 2017-06-28 Michael Collison <michael.collison@arm.com>
14821 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
14823 (arm_gen_movmemqi): Removed unused variable 'i'.
14824 Convert 'for' loop into 'while' loop.
14825 (arm_expand_prologue): Remove last unnecessary set of insn.
14826 (thumb_pop): Remove unused variable 'pushed_words'.
14827 (thumb_exit): Remove last unnecessary set of regs_to_pop.
14829 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14831 * config/s390/predicates.md: Use s390_rel_address_ok_p.
14832 * config/s390/s390-protos.h: Add prototype of
14833 s390_rel_address_ok_p.
14834 * config/s390/s390.c (s390_got_symbol): New function.
14835 (s390_rel_address_ok_p): New function.
14836 (legitimize_pic_address): Use s390_rel_address_ok_p.
14837 (s390_load_got): Use s390_got_symbol.
14838 (s390_option_override): Issue error if
14839 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
14840 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
14842 * config/s390/s390.opt: New option mpic-data-is-text-relative.
14844 2017-06-27 Andrew Pinski <apinski@cavium.com>
14846 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
14847 (X * copysign (1.0, X)): New pattern.
14848 (X * copysign (1.0, -X)): New pattern.
14849 (copysign (-1.0, CST)): New pattern.
14851 2017-06-27 Joseph Myers <joseph@codesourcery.com>
14853 * genmultilib (combination_space): Remove variable.
14854 Validate reuse rules against regular expression for any sequence
14855 of multilib options in any order.
14857 2017-06-27 Michael Collison <michael.collison@arm.com>
14859 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
14860 call aarch64_split_simd_combine.
14861 * (aarch64_combine_internal<mode>): Delete pattern.
14862 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
14863 Allow register and subreg operands.
14865 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14867 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
14868 specific need, just fallback on defaults.
14869 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
14871 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14872 Olivier Hainque <hainque@adacore.com>
14874 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
14876 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
14877 targets. Pick a default if no particular attempt applied.
14878 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
14881 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14883 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
14884 (x86_64-wrs-vxworks7): Likewise.
14886 2017-06-27 Marek Polacek <polacek@redhat.com>
14889 * ubsan.c (instrument_null): Check get_base_address's result for null.
14891 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14893 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
14895 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14897 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
14898 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
14899 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
14900 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
14901 New function types.
14902 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
14903 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
14904 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
14905 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
14906 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
14907 BUILT_IN_FEUPDATEENV): New builtins.
14908 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
14909 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
14910 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
14911 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
14913 (builtin_structptr_types): Adjust size.
14914 * tree.c (builtin_structptr_types): Add four entries.
14916 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14917 Olivier Hainque <hainque@adacore.com>
14919 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
14920 (TLS_SYM): New local macro, forcing reference to __tls__ on
14921 link command lines for VxWorks 7 RTPs, triggering initialization
14923 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
14924 OS features TLS support, true for RTPs on VxWorks 7.
14925 * config/vxworks.c (vxworks_override_options): Setup emutls
14928 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14930 * predict.c (test_prediction_value_range): Use -1U instead of -1
14931 to avoid narrowing conversion warning.
14932 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
14933 to avoid narrowing conversion warning.
14934 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
14936 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
14938 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14940 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
14941 64bit configurations.
14942 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
14943 (SIZE_TYPE): Likewise.
14944 * config/vxworks.c (vxworks_emutls_var_fields): Use
14945 long_unsigned_type_node instead of unsigned_type_node as the offset
14946 field type, which is "pointer" mode in emutls.c.
14948 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14951 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
14953 PR middle-end/81207
14954 * gimple-fold.c (replace_call_with_call_and_fold): Handle
14955 gimple_vuse copying separately from gimple_vdef copying.
14957 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14959 * value-prof.c (free_hist): Remove call to memset and the enclosing if
14962 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14963 Olivier Hainque <hainque@adacore.com>
14965 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
14966 for all vxworks7 targets.
14967 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
14968 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
14969 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
14970 variations for VX6/VX7 and 32/64bits later on in ...
14971 (VXWORKS_LIB_SPEC): Leverage new macros.
14972 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
14973 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
14975 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14977 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
14978 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
14980 2017-06-26 Carl Love <cel@us.ibm.com>
14982 * config/rs6000/rs6000-c.c: Add support for built-in functions
14983 vector bool char vec_reve (vector bool char);
14984 vector signed char vec_reve (vector signed char);
14985 vector unsigned char vec_reve (vector unsigned char);
14986 vector bool int vec_reve (vector bool int);
14987 vector signed int vec_reve (vector signed int);
14988 vector unsigned int vec_reve (vector unsigned int);
14989 vector bool long long vec_reve (vector bool long long);
14990 vector signed long long vec_reve (vector signed long long);
14991 vector unsigned long long vec_reve (vector unsigned long long);
14992 vector bool short vec_reve (vector bool short);
14993 vector signed short vec_reve (vector signed short);
14994 vector double vec_reve (vector double);
14995 vector float vec_reve (vector float);
14996 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
14997 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
14998 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
14999 (altivec_vreve): New pattern.
15000 * config/rs6000/altivec.h (vec_reve): New define.
15001 * doc/extend.texi (vec_rev): Update the built-in documentation file
15002 for the new built-in functions.
15004 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15006 PR tree-optimization/71815
15007 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15009 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15011 (slsr_process_phi): Likewise.
15012 (replace_uncond_cands_and_profitable_phis): Don't replace a
15013 multiply candidate with a stride of 1 (copy or cast).
15014 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15016 (lowest_cost_path): Likewise.
15017 (total_savings): Likewise.
15019 2017-06-26 Richard Biener <rguenther@suse.de>
15022 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15023 Use def_builtin_pure for all gather builtins.
15025 2017-06-26 Richard Biener <rguenther@suse.de>
15027 PR tree-optimization/81203
15028 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15029 non-dominating BBs.
15031 2017-06-26 Marek Polacek <polacek@redhat.com>
15034 * doc/invoke.texi: Document -Wmultistatement-macros.
15036 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15038 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15039 arm_neon_ok_no_float_abi effective target.
15041 2017-06-26 Richard Biener <rguenther@suse.de>
15043 PR tree-optimization/80928
15044 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15045 (copy_bbs): Set BB_DUPLICATED flag early.
15046 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15048 (execute_on_shrinking_pred): Likewise.
15049 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15050 BB_DUPLICATED blocks.
15051 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15052 iterate over all PHIs considering removal of *gsi.
15054 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15056 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15059 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15061 * config/rs6000/rs6000-string.c: (expand_block_clear,
15062 do_load_for_compare, select_block_compare_mode,
15063 compute_current_alignment, expand_block_compare,
15064 expand_strncmp_align_check, expand_strn_compare,
15065 expand_block_move, rs6000_output_load_multiple)
15066 Move functions related to string/block move/compare
15067 to a separate file.
15068 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15069 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15070 for this function which is now used in two files.
15071 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15072 * config.gcc: Add rs6000-string.o to extra_objs for
15073 targets powerpc*-*-* and rs6000*-*-*.
15075 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15078 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15079 32-bit, since indexed is not valid for DImode.
15080 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
15081 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
15082 (define_peephole2 for Altivec d-form load): Add 32-bit support.
15083 (define_peephole2 for Altivec d-form store): Likewise.
15086 * multiple_target.c (create_dispatcher_calls): Only create the
15087 dispatcher call if the function is the default clone of a
15088 versioned function.
15090 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
15092 PR middle-end/80902
15093 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
15094 a call, force the call to not be a tail call.
15096 2017-06-23 Jeff Law <law@redhat.com>
15098 * doc/contrib.texi: Add entry for Steven Pemberton's work on
15101 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
15103 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
15104 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
15105 handling for early expansion of vector shifts (sl,sr,sra,rl).
15106 (builtin_function_type): Add vector shift right instructions
15107 to the unsigned argument list.
15109 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
15111 rtl-optimizatoin/79286
15112 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
15113 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
15114 trap. PIC register plus a const unspec without offset can never trap.
15116 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
15118 * tree.h (builtin_structptr_type): New type.
15119 (builtin_structptr_types): Declare new array.
15120 * tree.c (builtin_structptr_types): New array.
15121 (free_lang_data, build_common_tree_nodes): Use it.
15123 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
15126 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
15129 2017-06-22 Matt Turner <mattst88@gmail.com>
15131 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
15132 Lake models to skylake case. Assume skylake for unknown
15133 models with clflushopt.
15135 2017-06-22 Jeff Law <law@redhat.com>
15137 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
15138 frame sizes that do not satisfy aarch64_uimm12_shift.
15140 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
15142 * profile-count.h (apply_probability,
15143 apply_scale, probability_in): Fix checks for zero.
15145 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15147 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
15148 * doc/cppdiropts.texi (-I @var{dir}): Document it.
15150 2016-06-22 Richard Biener <rguenther@suse.de>
15152 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15153 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
15154 REDUC_MAX_EXPR support.
15155 (vectorizable_reduction): Likewise.
15156 (vect_create_epilog_for_reduction): Likewise.
15158 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
15160 * match.pd (A / (1 << B) -> A >> B): New.
15161 * generic-match-head.c: Include optabs-tree.h.
15162 * gimple-match-head.c: Likewise.
15163 * optabs-tree.h (target_supports_op_p): New.
15164 * optabs-tree.c (target_supports_op_p): New.
15166 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15168 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
15169 $gcc_cv_ld --help output.
15170 (gcc_cv_ld_demangle): Likewise.
15171 (gcc_cv_ld_eh_frame_hdr): Likewise.
15172 (gcc_cv_ld_pie): Likewise.
15173 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
15174 (gcc_cv_ld_buildid): Likewise.
15175 (gcc_cv_ld_sysroot): Likewise.
15176 (ld_bndplt_support): Likewise.
15177 (ld_pushpopstate_support): Likewise.
15178 * configure: Regenerate.
15179 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
15181 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15184 * config/i386/sse.md (round<mode>2): Renumber match_dup and
15185 operands indexes to avoid gap between operands and match_dups.
15187 2017-06-21 Andrew Pinski <apinski@cavium.com>
15189 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
15190 Increment Arith_shift and Arith_shift_reg by 1.
15191 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
15193 * config/aarch64/aarch64.c (thunderx_tunings): Enable
15194 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
15195 (aarch64_strip_extend): Add new argument and test for it.
15196 (aarch64_cheap_mult_shift_p): New function.
15197 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
15198 add a cost if it is true.
15199 Update calls to aarch64_strip_extend.
15200 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
15202 2017-06-21 Andrew Pinski <apinski@cavium.com>
15204 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
15206 (thunderxt88): Likewise.
15207 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
15208 (thunderx_prefetch_tune): New variable.
15209 (thunderx2t99_prefetch_tune): Update for the correct values.
15210 (thunderxt88_tunings): New variable.
15211 (thunderx_tunings): Use thunderx_prefetch_tune instead of
15212 generic_prefetch_tune.
15213 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
15215 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15217 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
15218 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
15219 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
15220 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
15221 (aarch64_atomic_cas<mode>, GPI): Likewise.
15223 2017-06-21 Martin Liska <mliska@suse.cz>
15225 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
15226 statements on cold and hot labels.
15227 * predict.c (tree_estimate_probability_bb): Remove the
15228 prediction from this place.
15230 2017-06-21 Martin Liska <mliska@suse.cz>
15232 PR tree-optimization/79489
15233 * gimplify.c (maybe_add_early_return_predict_stmt): New
15235 (gimplify_return_expr): Call the function.
15236 * predict.c (tree_estimate_probability_bb): Remove handling
15238 * predict.def: Update comment about early return predictor.
15239 * gimple-predict.h (is_gimple_predict): New function.
15240 * predict.def: Change default value of early return to 66.
15241 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
15243 * passes.def: Put pass_strip_predict_hints to the beginning of
15246 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
15248 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
15249 FUNCTION_DECL declarations.
15250 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
15252 (dwaf2out_decl): Likewise.
15253 * godump.c (go_early_global_decl): Skip call to the real debug hook
15254 for FUNCTION_DECL declarations.
15255 * passes.c (rest_of_decl_compilation): Skip call to the
15256 early_global_decl debug hook for FUNCTION_DECL declarations, unless
15257 -fdump-go-spec is passed.
15259 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15261 * config/i386/i386.c (struct builtin_isa): New field pure_p.
15262 Reorder for compactness.
15263 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
15264 (def_builtin_pure, def_builtin_pure2): New functions.
15265 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
15267 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15269 * match.pd (nop_convert): New predicate.
15270 ((A +- CST1) +- CST2): Allow some NOP conversions.
15272 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15275 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
15276 with ctors/dtors if GOVD_SHARED is set.
15278 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15280 * config/aarch64/aarch64.md (movti_aarch64):
15281 Emit mov rather than orr.
15282 (movtf_aarch64): Likewise.
15283 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
15284 Emit mov rather than orr.
15286 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15288 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
15289 Swap alternatives, make integer dup more expensive.
15291 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15293 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
15294 Return true for non-tls symbols.
15296 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
15298 * config/aarch64/aarch64-cores.def (cortex-a55): New.
15299 (cortex-a75): Likewise.
15300 (cortex-a75.cortex-a55): Likewise.
15301 * config/aarch64/aarch64-tune.md: Regenerate.
15302 * doc/invoke.texi (-mtune): Document new values for -mtune.
15304 2017-06-21 Tom de Vries <tom@codesourcery.com>
15306 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
15307 stack_size feature.
15308 (Effective-Target Keywords, Other attributes): Suggest using
15309 dg-add-options stack_size feature to get stack limit in stack_size
15310 effective target documentation.
15312 2017-06-21 Julian Brown <julian@codesourcery.com>
15313 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15315 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
15316 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
15317 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
15319 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
15320 attribute type list for neon_multiply.
15321 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
15322 attribute type list for neon_multiply.
15323 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
15324 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
15325 attribute type list for neon_multiply.
15326 * config/arm/types.md (crypto_pmull): Add.
15327 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
15328 attribute type list.
15330 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
15332 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
15335 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15337 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
15338 to make sure not to dereference a NULL cost_classes_ptr pointer.
15340 2017-06-20 Carl Love <cel@us.ibm.com>
15342 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15343 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15344 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15345 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15346 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15347 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15348 VMULOSW): New enum "unspec" values.
15349 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15350 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15351 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15352 altivec_vmulosw): New patterns.
15353 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15354 VMULOSW): Add definitions.
15356 2017-06-20 Julia Koval <julia.koval@intel.com>
15358 * config/i386/i386.c: Fix rounding expand for new pattern.
15359 * config/i386/subst.md: Fix pattern (parallel -> unspec).
15361 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15363 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
15364 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
15366 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15368 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
15371 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15373 * config/aarch64/aarch64-cores.def: Rearrange to sort by
15374 architecture, then by implementer ID.
15375 * config/aarch64/aarch64-tune.md: Regenerate.
15377 2017-06-20 Richard Biener <rguenther@suse.de>
15379 PR middle-end/81097
15380 * fold-const.c (split_tree): Fold to type before negating.
15382 2017-06-20 David Malcolm <dmalcolm@redhat.com>
15384 * diagnostic-show-locus.c
15385 (selftest::test_fixit_deletion_affecting_newline): New function.
15386 (selftest::diagnostic_show_locus_c_tests): Call it.
15388 2017-06-20 Andreas Schwab <schwab@suse.de>
15391 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
15394 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15396 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
15397 __ARM_FEATURE_COPROC according to support.
15399 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15401 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15402 Rewritten to avoid overflow for > 32-bit pointers.
15405 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
15406 by removing enum keyword.
15407 (ubsan_type_descriptor): Likewise. Formatting fix.
15410 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
15411 splitter): Require TARGET_SSE2 in the condition.
15413 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15416 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
15417 for doing vector set of SFmode on ISA 3.0.
15418 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
15419 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
15421 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
15422 SFmode value into a V4SF variable that was extracted from another
15423 V4SF variable without converting the element to double precision
15424 and back to single precision vector format.
15425 (vsx_insert_extract_v4sf_p9_2): Likewise.
15427 2017-06-19 Jakub Jelinek <jakub@redhat.com>
15429 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
15430 in UWHI to avoid undefined overflow.
15433 * ubsan.h (enum ubsan_encode_value_phase): New.
15434 (ubsan_encode_value): Change second argument to
15435 enum ubsan_encode_value_phase with default value of
15436 UBSAN_ENCODE_VALUE_GENERIC.
15437 * ubsan.c (ubsan_encode_value): Change second argument to
15438 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
15439 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
15440 create_tmp_var_raw instead of create_tmp_var and use a
15442 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
15443 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
15444 ubsan_encode_value callers.
15447 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
15448 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
15449 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
15451 2017-06-19 Richard Biener <rguenther@suse.de>
15453 PR middle-end/81118
15454 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
15455 estimates if we changed anything.
15457 2017-06-19 Richard Biener <rguenther@suse.de>
15459 PR tree-optimization/80887
15460 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
15461 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
15462 simplified lookups, then reset mprts_hook.
15463 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
15465 (try_to_simplify): Likewise.
15467 2017-06-19 Martin Liska <mliska@suse.cz>
15470 * gimplify.c (gimplify_switch_expr):
15471 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
15473 2017-06-19 Martin Liska <mliska@suse.cz>
15475 * doc/install.texi: Document that PGO runs in 4 stages.
15477 2017-06-19 Martin Liska <mliska@suse.cz>
15480 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
15481 to dispatcher function name.
15482 * multiple_target.c (replace_function_decl): New function.
15483 (create_dispatcher_calls): Redirect both edges and references.
15485 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
15487 * profile-count.c (profile_count::dump): Dump quality.
15488 (profile_count::differs_from_p): Update for unsigned val.
15489 * profile-count.h (profile_count_quality): New enum.
15490 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
15492 2017-06-19 Richard Biener <rguenther@suse.de>
15494 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
15495 struct function as arg.
15496 (estimate_numbers_of_iterations): Export overload with loop arg.
15497 (free_numbers_of_iterations_estimates_loop): Use an overload of
15498 free_numbers_of_iterations_estimates instead.
15499 * tree-cfg.c (remove_bb): Adjust.
15500 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
15501 * tree-parloops.c (gen_parallel_loop): Likewise.
15502 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
15504 (tree_unroll_loops_completely): Likewise.
15505 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
15506 Use an overload instead and export.
15507 (estimated_loop_iterations): Adjust.
15508 (max_loop_iterations): Likewise.
15509 (likely_max_loop_iterations): Likewise.
15510 (estimate_numbers_of_iterations): Take struct function as arg
15512 (loop_exits_before_overflow): Adjust.
15513 (free_numbers_of_iterations_estimates_loop): Use an overload.
15514 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
15515 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
15517 2017-06-19 Richard Biener <rguenther@suse.de>
15520 * ipa-prop.c (find_constructor_constant_at_offset): Handle
15521 RANGE_EXPR conservatively.
15523 2017-06-16 Carl Love <cel@us.ibm.com>
15525 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15526 definitions for vec_float, vec_float2, vec_floato,
15527 vec_floate built-ins.
15528 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
15529 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
15531 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
15532 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
15533 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
15534 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
15535 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
15536 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
15537 vec_floato): Add builtin defines.
15538 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
15539 Update the built-in documentation file for the new built-in
15542 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15544 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
15545 (mthumb): Mark as the negative of -marm.
15547 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15549 * doc/invoke.texi (ARM Options, -mcpu): Document supported
15551 (ARM Options, -mtune): Document that this accepts the same
15552 extension options as -mcpu.
15553 (ARM Options, -mfpu): Document addition of -mfpu=auto.
15555 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15557 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
15558 permitted extensions.
15560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15562 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
15563 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
15564 (armv8-m.main): Add option +nodsp.
15565 * config/arm/arm-cpu-cdata.h: Regenerated.
15567 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15569 * config/arm/t-fuchsia: New file.
15570 * config.gcc (arm*-*-fuchsia*): Use it.
15572 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15574 * config/arm/t-symbian: Rewrite for new option infrastructure.
15576 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15578 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
15579 (MULTILIB_REQUIRED): Likewise.
15581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15583 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
15584 (MULTILIB_RESUE): Likewise.
15585 (MULTILIB_MATCHES): Likewise.
15586 (MULTLIB_REQUIRED): Likewise.
15588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15590 * config/arm/t-rtems: Rewrite for new option framework.
15592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15594 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
15595 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
15596 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
15597 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
15598 * config/arm/t-multilib: ... here.
15599 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
15600 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
15601 armv7-a and armv8*-a when A-profile libraries have not been built.
15602 * config/arm/t-rmprofile: Rewrite.
15604 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15606 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
15607 with a backslash. Remove the backslash after substituting unescaped
15609 * doc/fragments.texi (MULTILIB_REUSE): Document it.
15611 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15613 * config.gcc: (arm*-*-*): When building a-profile libraries, force
15614 the driver to pass through the default setting of -mfloat-abi.
15615 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
15617 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
15618 (all_feat_combs): New rule.
15619 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
15621 * config/arm/t-aprofile: Rewrite.
15623 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15625 * config/arm/arm.h (FPUTYPE_AUTO): Define.
15626 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
15627 fpu is not specified by the user/command-line.
15628 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
15629 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
15630 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
15631 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
15632 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
15633 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
15635 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15637 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
15638 * config/arm/t-arm-elf: Rewritten.
15640 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15642 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
15643 have some floating-point instructions.
15644 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
15645 (TARGET_MAYBE_HARD_FLOAT): New macro.
15646 * config/arm/arm-builtins.c (arm_init_builtins): Use
15647 TARGET_MAYBE_HARD_FLOAT.
15648 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
15650 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15652 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
15653 (configargs.h): Include it.
15654 (arm_print_hint_for_fpu_option): New function.
15655 (arm_parse_fpu_option): New function.
15656 (candidate_extension): New class.
15657 (arm_canon_for_multilib): New function.
15658 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
15659 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
15660 (ARCH_CANONICAL_SPECS): New macro.
15661 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
15663 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15665 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
15666 are set after handling multilib fragments. Set target_cpu_default2
15669 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15671 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
15673 (arm*-*-*): Set target_cpu_default2 to a quoted string.
15674 * config/arm/parsecpu.awk (check_cpu): Validate any extension
15676 (check_arch): Likewise.
15677 * config/arm/arm.c (arm_configure_build_target): Handle
15678 TARGET_CPU_DEFAULT being a string constant. Scan any feature
15679 options in the default.
15681 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15683 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
15684 when an option is an alias of another.
15685 * config/arm/parsecpu.awk (optalias): New parser token.
15686 (gen_comm_data): Mark non-alias options as such. Emit entries
15687 for extension aliases.
15688 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
15689 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
15690 (armv6kz, armv6zk, armv6t2): Likewise.
15691 (armv7): Make vfpv3-d16 an alias.
15692 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
15694 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
15695 Sort in canonical order.
15696 (armv8-a): Sort in canonical order.
15697 (armv8.1-a, armv8.2-a): Likewise.
15698 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
15700 (cortex-a9): Sort in canonical order.
15701 * config/arm/arm.c (selftests.h): Include it.
15702 (arm_test_cpu_arch_data): New function.
15703 (arm_run_self_tests): New function.
15704 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
15705 (targetm): Move declaration to the end of the file.
15706 * arm-cpu-cdata.h: Regenerated.
15708 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15710 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
15711 call to target_mode_check describing the type of option passed.
15712 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
15713 (arm_target_thumb_only): Use arm_parse_arch_option_name or
15714 arm_parse_cpu_option_name to match parameters against list of
15716 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
15717 arm_arch_core_flags data structure.
15718 * config/arm/arm-cpu_cdata.h: Regenerated.
15720 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15722 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
15724 (arm_print_hint_for_cpu_option): Likewise.
15725 (arm_print_hint_for_arch_option): Likewise.
15726 (arm_parse_cpu_option_name): Likewise.
15727 (arm_parse_arch_option_name): Likewise.
15728 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
15729 of entries in the all_fpus list.
15730 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
15731 (arm_parse_cpu_option_name): Declare.
15732 (arm_parse_arch_option_name): Declare.
15733 (arm_parse_option_features): Declare.
15734 (arm_intialize_isa): Declare.
15735 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
15737 (gen_comm_data): ... here. Make definitions non-static.
15738 * config/arm/arm-cpu-data.h: Regenerated.
15739 * config/arm/arm-cpu-cdata.h: Regenerated.
15741 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15743 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
15744 (cpu_arch_extension): New structure.
15745 (cpu_arch_option, arch_option, cpu_option): New structures.
15746 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
15747 architecture types.
15748 (gen_data): Generate new format data tables.
15749 * config/arm/arm.c (cpu_tune): New structure.
15750 (cpu_option, processors): Delete.
15751 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
15752 (arm_print_hint_for_cpu_option): ... this and ...
15753 (arm_print_hint_for_arch_option): ... this.
15754 (arm_parse_arch_cpu_name): Delete. Replace with ...
15755 (arm_parse_cpu_option_name): ... this and ...
15756 (arm_parse_arch_option_name): ... this.
15757 (arm_unrecognized_feature): Change type of target parameter to
15759 (arm_parse_arch_cpu_features): Delete. Replace with ...
15760 (arm_parse_option_features): ... this.
15761 (arm_configure_build_target): Rework to use new configuration data
15763 (arm_print_tune_info): Rework for new configuration data tables.
15764 * config/arm/arm-cpu-data.h: Regenerated.
15765 * config/arm/arm-cpu.h: Regenerated.
15767 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15769 * Makefile.in (OBJS): Move sbitmap.o from here ...
15770 (OBJS-libcommon): ... to here.
15772 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15774 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
15775 (ISA_ALL_CRYPTO): New macro.
15776 (ISA_ALL_SIMD): New macro
15777 (ISA_ALL_FP): New macro.
15778 * config/arm/arm.c (fpu_bitlist): Update initializer.
15779 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
15781 (arm9e): Add fpu. Add option for nofp
15782 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
15783 (arm926ej-s, arm1026ej-s): Likewise.
15784 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
15785 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
15786 neon-fp16, neon-vfpv4, nofp and nosimd.
15787 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
15788 (cortex-a8): Add fpu. Add option for nofp.
15789 (cortex-a9): Add fpu. Add options for nosimd and nofp.
15790 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
15791 (cortex-r4f): Add fpu.
15792 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
15793 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
15795 (cortex-r8): Likewise.
15796 (cortex-m4): Add fpu. Add option for nofp.
15797 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
15798 (cortex-a17.cortex-a7): Likewise.
15799 (cortex-a32): Add fpu. Add options for crypto and nofp.
15800 (cortex-a35, cortex-a53): Likewise.
15801 (cortex-a57): Add fpu. Add option for crypto.
15802 (cortex-a72, cortex-a73): Likewise.
15803 (exynos-m1): Likewise.
15804 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
15805 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
15806 (cortex-m33): Add fpu. Add option for nofp.
15807 * config/arm/arm-cpu-cdata.h: Regenerated
15808 * config/arm/arm-cpu-data.h: Regenerated.
15810 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15812 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
15813 (armv5te, armv5tej): Likewise.
15814 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
15815 (armv7): Add options fp and vfpv3-d16.
15816 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
15817 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
15819 (armv7ve): Likewise.
15820 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
15821 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
15822 (armv8-a): Add nocrypto option.
15823 (armv8.1-a, armv8.2-a): Likewise.
15824 (armv8-m.main): add options fp, fp.dp and nofp.
15826 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15828 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
15830 (armv8-a+crc): Delete.
15831 (armv8.1-a): Add options simd, crypto and nofp.
15832 (armv8.2-a): Add options fp16, simd, crypto and nofp.
15833 (armv8.2-a+fp16): Delete.
15834 (armv8-m.main): Add option dsp.
15835 (armv8-m.main+dsp): Delete.
15836 (cortex-a8): Add fpu. Add nofp option.
15837 (cortex-a9): Add fpu. Add nofp and nosimd options.
15838 * config/arm/parsecpu.awk (gen_data): Generate option tables and
15839 link to main cpu and architecture data structures.
15840 (gen_comm_data): Only put isa attributes from the main architecture
15842 (option): New statement for architecture and CPU entries.
15843 * arm.c (struct cpu_option): New structure.
15844 (struct processors): Add entry for options.
15845 (arm_unrecognized_feature): New function.
15846 (arm_parse_arch_cpu_name): Ignore any characters after the first
15848 (arm_parse_arch_cpu_feature): New function.
15849 (arm_configure_build_target): Separate out any CPU and architecture
15850 features and parse separately. Don't error out if -mfpu=auto is
15851 used with only an architecture string.
15852 (arm_print_asm_arch_directives): New function.
15853 (arm_file_start): Call it.
15854 * config/arm/arm-cpu-cdata.h: Regenerated.
15855 * config/arm/arm-cpu-data.h: Likewise.
15856 * config/arm/arm-tables.opt: Likewise.
15858 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15860 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
15861 assembler when it is not -mfpu=auto.
15863 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15865 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
15866 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
15867 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
15868 (ASM_CPU_SPEC): Rewrite.
15869 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
15870 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
15871 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
15872 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
15873 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
15874 copied string is NUL-terminated. Also strip any characters prefixed
15876 (arm_rewrite_selected_arch): New function.
15877 (arm_rewrite_march): New function.
15879 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15881 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
15882 (x_arm_cpu_string, x_arm_tune_string): Likewise.
15883 (march, mcpu, mtune): Convert to string-based options.
15884 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
15885 (arm_parse_arch_cpu_name): New function.
15886 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
15887 identify selected architecture or CPU.
15888 (arm_option_save): New function.
15889 (TARGET_OPTION_SAVE): Redefine.
15890 (arm_option_restore): Restore string options.
15891 (arm_option_print): Print string options.
15893 2017-06-16 Martin Sebor <msebor@redhat.com>
15895 PR tree-optimization/80933
15896 PR tree-optimization/80934
15897 * builtins.c (fold_builtin_3): Do not handle bcmp here.
15898 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
15899 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
15900 (gimple_fold_builtin): Call them.
15902 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15904 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
15905 as unlikely; update profile.
15907 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15909 * predict.c (force_edge_cold): Handle declaring edges impossible
15912 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15914 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
15916 (try_unroll_loop_completely): Fix reporting.
15918 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15920 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
15922 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
15925 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
15926 if given a non-constant argument for an intrinsic which requires a
15929 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15931 * profile.c (compare_freqs): New function.
15932 (branch_prob): Sort edge list.
15933 (find_spanning_tree): Assume that the list is priority sorted.
15935 2017-06-16 Richard Biener <rguenther@suse.de>
15937 PR tree-optimization/81090
15938 * passes.def (pass_record_bounds): Remove.
15939 * tree-pass.h (make_pass_record_bounds): Likewise.
15940 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
15941 make_pass_record_bounds): Likewise.
15942 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
15943 not free niter estimates at the beginning but at the end.
15944 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
15946 2017-06-16 Richard Biener <rguenther@suse.de>
15948 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
15949 initializer to workaround ICE in host GCC 4.8.
15951 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15953 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
15955 (clone_inlined_nodes): Update.
15957 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15959 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
15960 prefetch settings, and enable prefetching by default at -O3.
15962 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15964 * config/aarch64/aarch64.c (aarch64_override_options_internal):
15965 Set flag_prefetch_loop_arrays according to tuning data.
15967 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15969 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
15970 New tune structure.
15971 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
15972 [Unrelated to main purpose of the patch] Place the pointer field last
15973 to enable type checking errors when tune structure are wrongly merged.
15974 * config/aarch64/aarch64.c (generic_prefetch_tune,)
15975 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
15976 (thunderx2t99_prefetch_tune): New tune constants.
15977 (tune_params *_tunings): Update all tunings (no functional change).
15978 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
15979 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
15980 from tunings structures.
15982 2017-06-16 Jakub Jelinek <jakub@redhat.com>
15985 * ubsan.c (instrument_null): Add T argument, use it instead
15986 of computing it based on IS_LHS.
15987 (instrument_object_size): Likewise.
15988 (pass_ubsan::execute): Adjust instrument_null and
15989 instrument_object_size callers to pass gimple_get_lhs or
15990 gimple_assign_rhs1 result to it. Use instrument_null instead of
15991 calling get_base_address and instrument_mem_ref. Handle
15992 aggregate call arguments for object-size sanitization.
15994 2017-06-16 Yury Gribov <tetra2005@gmail.com>
15996 PR tree-optimization/81089
15997 * tree-vrp.c (is_masked_range_test): Validate operands of
16000 2017-06-15 Martin Sebor <msebor@redhat.com>
16003 * dumpfile.c (dump_register): Avoid calling memset to initialize
16004 a class with a default ctor.
16005 * gcc.c (struct compiler): Remove const qualification.
16006 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16007 * hash-table.h: Ditto.
16008 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16010 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16011 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16013 * params.h (struct param_info): Make struct members non-const.
16014 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16015 with default initialization.
16016 * vec.h (vec_copy_construct, vec_default_construct): New helper
16018 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16019 with vec_copy_construct.
16020 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16021 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16022 * doc/invoke.texi (-Wclass-memaccess): Document.
16024 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16026 * emit-rtl.h (is_leaf): Update comment about local
16027 register allocator.
16029 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16032 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16033 for a variable to have a section before checking if the section has a
16035 Set section to.persistent if persistent attribute is set.
16036 Warn if .persistent attribute is used on an automatic variable.
16038 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16040 PR rtl-optimization/80474
16041 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16043 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16045 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16048 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16050 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16051 (casesi): Emit bounds checking as RTL.
16052 (casesi_internal_mips16_<mode>): Remove bounds checking.
16054 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16056 * config/xtensa/xtensa.c (xtensa_option_override): Append
16057 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16058 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16059 xtensa_doloop_hooks): Define unconditionally.
16060 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16062 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16063 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16064 for it in xtensa_option_override.
16065 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16066 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16068 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16070 * doc/cppopts.texi: Document '-' special value to -MF.
16072 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16074 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16075 (cortex_a53_fconst): Likewise.
16076 (cortex_a53_fpmul): Likewise.
16077 (cortex_a53_f_load_64): Likewise.
16078 (cortex_a53_f_load_many): Likewise.
16079 (cortex_a53_advsimd_alu): Likewise.
16080 (cortex_a53_advsimd_alu_q): Likewise.
16081 (cortex_a53_advsimd_mul): Likewise.
16082 (cortex_a53_advsimd_mul_q): Likewise.
16083 (fpmac bypass): Add new bypass for fpmac-fpmac case.
16084 Add missing fmul, r2f_cvt and fconst cases.
16086 2017-06-14 Richard Biener <rguenther@suse.de>
16088 PR middle-end/81088
16089 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
16091 (fold_binary_loc): When associating do not treat pre-existing
16092 TREE_OVERFLOW on literal constants as a reason to allow
16093 TREE_OVERFLOW on associated literal constants.
16095 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16097 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
16098 (MASK_FEATURES): New macro.
16099 * config/sparc/sparc.c (sparc_option_override): Remove the special
16100 handling of -mfpu and generalize it to all MASK_FEATURES switches.
16102 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16104 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
16105 a division of 0 if non-call exceptions are enabled.
16107 2017-06-14 Andrew Pinski <apinski@cavium.com>
16108 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16111 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
16112 Improve vector initialization code gen for only variable case.
16114 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16116 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
16118 2017-06-14 Richard Biener <rguenther@suse.de>
16120 PR tree-optimization/81083
16121 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
16124 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16126 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
16127 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
16128 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
16129 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
16130 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
16131 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
16133 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16135 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
16136 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
16138 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16140 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
16142 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16144 * config/rs6000/t-rtems: Don't handle SPE.
16146 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16148 * config/rs6000/t-linux: Don't handle SPE.
16150 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16152 * config/rs6000/eabispe.h: Delete file.
16154 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16156 * config/rs6000/t-spe: Delete file.
16158 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16160 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
16161 (rs6000_legitimate_offset_address_p): Return false for anything in
16162 V2SImode or V2SFmode.
16164 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16166 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
16167 except V2SF and V2SI. Rearrange the vector modes, and add comments.
16168 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
16170 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
16171 (rs6000_legitimate_offset_address_p): Ditto.
16172 (rs6000_emit_move): Ditto.
16173 (rs6000_init_builtins): Remove V4HI_type_node.
16175 2017-06-13 Martin Liska <mliska@suse.cz>
16178 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
16179 (gate_asan): Likewise.
16180 * asan.h (asan_no_sanitize_address_p): Remove the function.
16181 (sanitize_flags_p): New function.
16182 * builtins.def: Fix coding style.
16183 * common.opt: Use renamed enum value.
16184 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
16185 * doc/extend.texi: Document no_sanitize attribute.
16186 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
16187 to SANITIZE_UNDEFINED_NONDEFAULT.
16188 * gcc.c (sanitize_spec_function): Use the renamed enum value.
16189 * gimple-fold.c (optimize_atomic_compare_exchange_p):
16190 Use sanitize_flags_p.
16191 * gimplify.c (gimplify_function_tree): Likewise.
16192 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
16193 * opts.c (parse_no_sanitize_attribute): New function.
16194 (common_handle_option): Use renamed enum value.
16195 * opts.h (parse_no_sanitize_attribute): Declare.
16196 * tree.c (sanitize_flags_p): New function.
16197 * tree.h: Declared here.
16198 * tsan.c: Use sanitize_flags_p.
16199 * ubsan.c (ubsan_expand_null_ifn): Likewise.
16200 (instrument_mem_ref): Likewise.
16201 (instrument_bool_enum_load): Likewise.
16202 (do_ubsan_in_current_function): Remove the function.
16203 (pass_ubsan::execute): Use sanitize_flags_p.
16204 * ubsan.h: Remove do_ubsan_in_current_function
16205 * tree-cfg.c (print_no_sanitize_attr_value): New function.
16206 (dump_function_to_file): Use it here.
16208 2017-06-13 Martin Jambor <mjambor@suse.cz>
16210 PR tree-optimization/80803
16211 PR tree-optimization/81063
16212 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
16213 (propagate_subaccesses_across_link): Enqueue subtree whenever
16214 necessary instead of relying on the caller.
16216 2017-06-13 Martin Jambor <mjambor@suse.cz>
16218 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
16219 that have a first_link.
16220 (sort_and_splice_var_accesses): Do not check first_link before
16222 (subtree_mark_written_and_enqueue): Likewise.
16223 (propagate_all_subaccesses): Likewise and do not stop at first
16224 parent with a first_link.
16226 2017-06-13 Martin Jambor <mjambor@suse.cz>
16228 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
16231 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16233 * match.pd: New pattern.
16235 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16237 * tree-vrp.c (is_masked_range_test): New function.
16238 (register_edge_assert_for): Determine ranges for
16241 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16243 PR tree-optimization/67328
16244 * fold-const.c (maskable_range_p): New function.
16245 (build_range_check): Generate bittests if possible.
16247 2017-06-13 Martin Liska <mliska@suse.cz>
16249 * gimple-pretty-print.c (dump_probability): Add new argument.
16250 (dump_edge_probability): Dump both probability and count.
16251 (dump_gimple_label): Likewise.
16252 (dump_gimple_bb_header): Likewise.
16254 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
16257 * config/avr/avr-devices.c: Fix indentation.
16258 * config/avr/gen-avr-mmcu-specs.c: Dito.
16260 2017-06-13 Richard Biener <rguenther@suse.de>
16262 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
16263 instead get vector type from stmt_info.
16264 (vectorizable_reduction): Adjust. Remove dead code.
16266 2017-06-13 Richard Biener <rguenther@suse.de>
16268 PR middle-end/81065
16269 * fold-const.c (extract_muldiv_1): Remove bogus distribution
16270 case of C * (x * C2 + C3).
16271 (fold_addr_of_array_ref_difference): Properly fold index difference.
16273 2017-06-12 David S. Miller <davem@davemloft.net>
16276 * config/sparc/sparc.md (return expander): Emit frame blockage if
16277 function uses alloca.
16279 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16281 * combine.c (make_field_assignment): Check len rather than the mode
16282 precision when calling force_to_mode.
16284 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
16286 Support multilibs and devices that see flash in RAM address range.
16289 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
16290 (avr_mcu_t) <flash_pm_offset>: New field.
16291 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
16292 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
16293 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
16294 (AVR_TINY_PM_OFFSET): Remove macro.
16295 * config/avr/avr.opt (-mshort-calls): New option.
16296 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16297 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
16298 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
16299 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
16300 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
16301 instead of avr_arch->have_jmp_call.
16302 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
16303 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
16304 avr_arch->flash_pm_offset to define.
16305 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
16306 new field flash_pm_offset. Add entry for avrxmega3.
16307 (avr_texinfo): Add entry for avrxmega3.
16308 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
16309 attiny212, attiny214,
16310 attiny412, attiny414, attiny416, attiny417,
16311 attiny814, attiny816, attiny817,
16312 attiny1614, attiny1616, attiny1617,
16313 attiny3214, attiny3216, attiny3217.
16314 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
16315 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
16316 (avr_print_operand_address) [AVR_TINY]: Same.
16317 (avr_asm_init_sections) <readonly_data_section>: Only patch
16318 callback if avr_arch->flash_pm_offset = 0.
16319 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
16320 for rodata if avr_arch->flash_pm_offset != 0.
16321 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
16322 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
16323 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
16324 (m_options): Append opt_rcall.
16325 (m_dirnames): Append dir_rcall.
16326 * config/avr/t-multilib: Regenerate.
16328 * configure.ac [target=avr]: Check whether avrxmega3 default
16329 linker description file works as needed.
16330 * configure: Regenerate.
16331 * doc/avr-mmcu.texi: Regenerate.
16332 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
16333 <__AVR_ARCH__>: Document avrxmega3 and 103.
16334 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
16335 <__AVR_SHORT_CALLS__>: Document it.
16336 <__AVR_PM_BASE_ADDRESS__>: Document it.
16337 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
16338 (AVR Variable Attributes) <progmem>: Document this is
16339 not needed for avrxmega3.
16340 (AVR Named Address Spaces) <__flash>: Dito.
16342 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
16344 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
16346 2017-06-12 Doug Rupp <rupp@adacore.com>
16348 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
16349 Append vxworks-stdint.h to the tm_file list.
16350 * config/vxworks-stdint.h: New file.
16352 2017-06-12 Martin Liska <mliska@suse.cz>
16354 PR tree-optimization/81041
16355 * tree-profile.c (gimple_gen_ic_func_profiler):
16356 Create an extra BB in profile-generate
16357 (gimple_gen_time_profiler): Likewise.
16359 2017-06-12 Jakub Jelinek <jakub@redhat.com>
16361 PR tree-optimization/81003
16362 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
16363 (update_range_test): Use it instead of force_gimple_operand_gsi.
16365 2017-06-12 Richard Biener <rguenther@suse.de>
16367 PR tree-optimization/81053
16368 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
16369 with backedge value not defined in loop. Simplify def stmt
16372 2017-06-11 Tom de Vries <tom@codesourcery.com>
16375 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
16377 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
16378 nvptx_cannot_force_const_mem.
16380 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16382 * opts.c (finish_options): Move test for flag_split_stack after
16383 it has been initialized.
16385 2017-06-11 Jason Merrill <jason@redhat.com>
16387 * tree.h (id_equal): New.
16388 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
16389 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
16390 instead of strcmp of IDENTIFIER_POINTER.
16392 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16394 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
16395 (mark_all_inlined_calls_cdtor): Fix formating.
16396 (inline_transform): Rescale profile before inlining.
16398 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16400 * cgraph.h (cgraph_edge::clone): Update prototype.
16401 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
16402 (cgraph_node::create_clone): Update.
16403 (cgraph_node::create_version_clone): Update.
16404 * tree-inline.c (copy_bb): Update.
16405 (expand_call_inline): Update.
16407 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16409 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
16410 factored out from ...
16411 (rs6000_emit_prologue): ... here.
16413 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16415 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
16416 factored out from ...
16417 (rs6000_emit_prologue): ... here.
16419 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16421 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
16423 (handle_missing_profiles): Fix computation of tp_first_run.
16424 (counts_to_freqs): Do not touch freqs when count is 0.
16426 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16428 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
16431 2017-06-10 Tom de Vries <tom@codesourcery.com>
16433 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
16434 attributes): Document signal effective target.
16436 2017-06-10 Tom de Vries <tom@codesourcery.com>
16438 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
16439 Document effective target stack_size.
16441 2017-06-09 David Malcolm <dmalcolm@redhat.com>
16443 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
16444 to the edit_context if they can be auto-applied.
16446 2017-06-9 Ian Lance Taylor <iant@golang.org>
16448 * opts.c (finish_options): If -fsplit-stack, disable implicit
16449 -forder-blocks-and-partition.
16450 * doc/invoke.texi (Optimize Options): Document that when using
16451 -fsplit-stack -forder-blocks-and-partition is not implicitly
16454 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16456 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
16457 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
16458 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
16459 * builtins.def (abort, trap, unreachable): Declare cold.
16460 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
16461 * tree-core.h (ECF_COLD): New.
16462 * tree.c (set_call_expr_flags): Handle ECF_COLD.
16463 (build_common_builtin_nodes): Mark unreachable and abort as cold.
16465 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16467 * predict.c (unlikely_executed_stmt_p): Cleanup.
16469 2017-06-09 Richard Biener <rguenther@suse.de>
16471 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
16472 model if the ref is always written to.
16474 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16476 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
16478 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16480 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
16483 2017-06-09 Tom de Vries <tom@codesourcery.com>
16486 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
16487 "target cannot support label values" when encountering LABEL_REF.
16489 2017-06-09 Martin Liska <mliska@suse.cz>
16491 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
16492 (gimple_gen_ic_func_profiler): Emit direct comparison
16493 of __gcov_indirect_call_callee with NULL.
16494 (gimple_gen_time_profiler): Change probability from
16495 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
16497 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16499 * profile.c (edge_gcov_counts): Turn to pointer.
16500 (compute_branch_probabilities, compute_branch_probabilities): Update.
16501 (branch_prob): Do not clear edge_gcov_count.
16502 * profile.h (edge_gcov_counts): Turn to pointer.
16503 (edge_gcov_count): Update.
16505 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16507 * gimple.h (gimple_check_failed): Mark cold.
16509 2017-06-09 Richard Biener <rguenther@suse.de>
16511 PR tree-optimization/66623
16512 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
16513 refactor check_reduction into two parts, properly computing
16514 whether we have to check reduction validity for outer loop
16517 2017-06-09 Richard Biener <rguenther@suse.de>
16519 PR tree-optimization/79483
16520 * graphite-scop-detection.c (order): New global.
16521 (get_order): Compute bb to order mapping that satisfies code
16522 generation constraints.
16523 (cmp_pbbs): New helper.
16524 (build_scops): Start domwalk at entry block, sort generated
16527 2017-06-09 Richard Biener <rguenther@suse.de>
16529 PR middle-end/81007
16530 * ipa-polymorphic-call.c
16531 (ipa_polymorphic_call_context::restrict_to_inner_class):
16532 Skip FIELD_DECLs with error_mark_node type.
16533 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
16536 2017-06-09 Martin Liska <mliska@suse.cz>
16538 * predict.c (struct branch_predictor): New struct.
16539 (test_prediction_value_range): New test.
16540 (predict_c_tests): New function.
16541 * selftest-run-tests.c (selftest::run_tests): Run the function.
16542 * selftest.h: Declare new tests.
16544 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
16547 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
16548 gen_add3_insn did not fail.
16549 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
16550 r0, construct that number in a temporary reg and add that reg to r0.
16551 If asked to put the result in r0 as well, fail.
16553 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
16555 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16556 for early expansion of vec_eqv.
16558 2017-06-08 Jakub Jelinek <jakub@redhat.com>
16560 PR middle-end/81005
16561 * ubsan.c (instrument_null): Avoid pointless code temporary.
16562 (pass_ubsan::execute): Instrument aggregate arguments of calls.
16564 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
16568 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
16571 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
16574 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16576 * predict.c (unlikely_executed_edge_p): Move ahead.
16577 (probably_never_executed_edge_p): Use it.
16579 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16581 PR middle-end/79988
16582 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
16583 gimple_call_builtin_p call.
16585 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16587 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
16588 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
16589 rtl_check_failed_type2, rtl_check_failed_code1,
16590 rtl_check_failed_code2, rtl_check_failed_code_mode,
16591 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
16592 rtvec_check_failed_bounds, rtl_check_failed_flag,
16593 _fatal_insn_not_found, _fatal_insn): Likewise.
16594 * tree.h (tree_contains_struct_check_failed,
16595 tree_check_failed, tree_not_check_failed,
16596 tree_class_check_failed, tree_range_check_failed,
16597 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
16598 tree_vec_elt_check_failed, phi_node_elt_check_failed,
16599 tree_operand_check_failed, omp_clause_check_failed,
16600 omp_clause_operand_check_failed, omp_clause_range_check_failed):
16603 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16605 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
16606 flag_branch_probabilities.
16607 * ipa-inline.c (edge_badness): Likewise.
16608 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16609 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
16610 * predict.c (maybe_hot_frequency_p): Likewise.
16611 (probably_never_executed): Likewise.
16612 * sched-ebb.c (schedule_ebbs): Likewise.
16613 * sched-rgn.c (find_single_block_region): Likewise.
16614 * tracer.c (tail_duplicate): Likewise.
16616 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16618 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
16619 longer requires x_flag_profile_use.
16621 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16623 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
16624 instead of flag_reorder_blocks_and_partition.
16625 * dbxout.c (dbxout_function_end): Likewise.
16626 * dwarf2out.c (gen_subprogram_die): Likewise.
16627 * haifa-sched.c (sched_create_recovery_edges): Likewise.
16628 * hw-doloop.c (reorg_loops): Likewise.
16629 * varasm.c (assemble_start_function,
16630 assemble_end_function): Likewise.
16631 (decide_function_section): Do not check for
16632 flag_reorder_blocks_and_partition.
16634 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16636 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16638 (chkp_get_hard_register_fake_addr_expr): Ditto.
16639 (chkp_build_addr_expr): Add check for hard reg case.
16640 (chkp_parse_array_and_component_ref): Ditto.
16641 (chkp_find_bounds_1): Ditto.
16642 (chkp_process_stmt): Don't generate bounds store for
16645 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16647 * predict.c (maybe_hot_bb_p): Do not check profile status.
16648 (maybe_hot_edge_p): Likewise.
16649 (probably_never_executed): Check for zero counts even if profile
16651 (unlikely_executed_edge_p): New function.
16652 (unlikely_executed_stmt_p): New function.
16653 (unlikely_executed_bb_p): New function.
16654 (set_even_probabilities): Use unlikely predicates.
16655 (combine_predictions_for_bb): Likewise.
16656 (predict_paths_for_bb): Likewise.
16657 (predict_paths_leading_to_edge): Likewise.
16658 (determine_unlikely_bbs): New function.
16659 (estimate_bb_frequencies): Use it.
16660 (compute_function_frequency): Use zero counts even if profile is
16662 * profile-count.h: Fix typo.
16664 2017-08-08 Julia Koval <julia.koval@intel.com>
16666 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
16667 _mm512_mask_cvtsepi16_storeu_epi8,
16668 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
16669 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
16670 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
16671 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
16672 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
16673 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
16674 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
16675 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
16676 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
16677 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
16678 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
16679 __builtin_ia32_pmovuswb256mem_mask,
16680 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
16681 __builtin_ia32_pmovwb512mem_mask): New builtins.
16683 2017-08-08 Julia Koval <julia.koval@intel.com>
16685 PR target/73350,80862
16686 * config/i386/subst.md (round): Fix round pattern.
16687 * config/i386/i386.c (ix86_erase_embedded_rounding):
16688 Fix erasing rounding for the fixed pattern.
16690 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16692 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
16694 2017-06-08 Martin Liska <mliska@suse.cz>
16696 PR gcov-profile/80911
16697 * gcov.c (block_info::block_info): New constructor.
16699 2017-06-07 Carl Love <cel@us.ibm.com>
16701 * config/rs6000/rs6000-c: The return type of the following
16702 built-in functions was implemented as int not long long. Fix sign
16703 of return value for the unsigned version of vec_mulo and vec_mule.
16704 vector unsigned long long vec_bperm (vector unsigned long long,
16705 vector unsigned char)
16706 vector signed long long vec_mule (vector signed int,
16708 vector unsigned long long vec_mule (vector unsigned int,
16709 vector unsigned int)
16710 vector signed long long vec_mulo (vector signed int,
16712 vector unsigned long long vec_mulo (vector unsigned int,
16713 vector unsigned int)
16714 * doc/extend.texi: Fix the documentation for the built-in
16717 2017-06-07 Carl Love <cel@us.ibm.com>
16720 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
16723 2017-06-07 Carl Love <cel@us.ibm.com>
16725 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
16726 support, Generate doublehv for signed int/float for BE case only.
16728 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
16730 * doc/invoke.texi (mcx16): Rewrite.
16732 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16734 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
16735 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
16736 *mov<mode>_softfloat, and an anonymous splitter): Use
16737 nonimmediate_operand instead of rs6000_nonimmediate_operand.
16739 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16741 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
16743 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16744 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
16745 (rs6000_debug_reg_global): Adjust.
16746 (rs6000_init_hard_regno_mode_ok): Adjust.
16747 (rs6000_dbx_register_number): Adjust.
16748 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
16749 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16750 Remove SPE_ACC and SPEFSCR.
16751 (REG_ALLOC_ORDER): Ditto.
16752 (FRAME_POINTER_REGNUM): Change to 111.
16753 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
16754 (REG_CLASS_NAMES): Ditto.
16755 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
16756 (REGISTER_NAMES): Ditto.
16757 (ADDITIONAL_REG_NAMES): Ditto.
16758 (rs6000_reg_names): Ditto.
16759 * config/rs6000/rs6000.md: Renumber some register number
16762 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16764 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
16766 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16767 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
16769 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
16770 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16771 Delete the SPE high registers.
16772 (REG_ALLOC_ORDER): Ditto.
16773 (enum reg_class): Remove SPE_HIGH_REGS.
16774 (REG_CLASS_NAMES): Ditto.
16775 (REG_CLASS_CONTENTS): Delete the SPE high registers.
16776 (REGISTER_NAMES): Ditto.
16777 (rs6000_reg_names): Ditto.
16778 * doc/tm.texi.in: Remove SPE as example.
16779 * doc/tm.texi: Regenerate.
16781 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16783 * config/rs6000/8540.md (ppc8540_brinc): Delete.
16784 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
16785 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
16786 * config/rs6000/rs6000.md (type): Remove "brinc".
16788 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16790 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
16791 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
16792 * config/rs6000/linuxspe.h: Delete file.
16793 * config/rs6000/rs6000.md: Don't include spe.md.
16794 * config/rs6000/spe.h: Delete file.
16795 * config/rs6000/spe.md: Delete file.
16796 * config/rs6000/t-rs6000: Remove spe.md.
16798 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16800 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
16801 (reg_or_none500mem_operand): Delete.
16802 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
16803 instead of reg_or_none500mem_operand.
16805 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16807 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16808 handling of SPE flags.
16809 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
16811 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16813 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
16815 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
16816 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
16817 paired_divv2sf3): Similar.
16818 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
16819 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
16820 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
16822 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
16823 Rename the paired_* instruction patterns.
16824 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
16826 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
16827 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
16828 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
16829 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
16830 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
16831 PAIRED_VECTOR_MODE.
16832 (struct machine_function): Delete field spe_insn_chain_scanned_p.
16833 (spe_func_has_64bit_regs_p): Delete.
16834 (spe_expand_predicate_builtin): Delete.
16835 (spe_expand_evsel_builtin): Delete.
16836 (TARGET_DWARF_REGISTER_SPAN): Do not define.
16837 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
16838 (invalid_e500_subreg): Delete.
16839 (rs6000_legitimize_address): Always force_reg op2 as well, for
16840 paired single memory accesses.
16841 (rs6000_member_type_forces_blk): Delete.
16842 (rs6000_spe_function_arg): Delete.
16843 (rs6000_expand_unop_builtin): Delete SPE handling.
16844 (rs6000_expand_binop_builtin): Ditto.
16845 (spe_expand_stv_builtin): Delete.
16846 (bdesc_2arg_spe): Delete.
16847 (spe_expand_builtin): Delete.
16848 (spe_expand_predicate_builtin): Delete.
16849 (spe_expand_evsel_builtin): Delete.
16850 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
16851 (spe_init_builtins): Delete.
16852 (spe_func_has_64bit_regs_p): Delete.
16853 (savres_routine_name): Delete "info" parameter. Adjust callers.
16854 (rs6000_emit_stack_reset): Ditto.
16855 (rs6000_dwarf_register_span): Delete.
16856 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
16857 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
16858 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
16860 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
16862 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
16863 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
16864 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
16865 mulv2sf3, divv2sf3): Delete expanders.
16867 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16869 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
16871 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16873 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
16874 * config/rs6000/rs6000.c: Ditto.
16876 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16878 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
16879 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
16880 comparison_operator.
16882 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16884 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
16885 * config/rs6000/rs6000.opt: Ditto.
16886 * config/rs6000/t-rtems: Ditto.
16888 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16890 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
16891 TARGET_E500_SINGLE by 0, simplify.
16892 * config/rs6000/rs6000.c: Ditto.
16893 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
16894 (spe_build_register_parallel): Delete.
16895 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
16896 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
16897 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
16898 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
16899 (E500_CONVERT): Delete.
16900 * config/rs6000/spe.md: Remove many patterns and all define_constants.
16902 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16904 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
16905 * config/rs6000/dfp.md: Ditto.
16906 (negdd2, *negdd2_fpr): Merge.
16907 (absdd2, *absdd2_fpr): Merge.
16908 (negtd2, *negtd2_fpr): Merge.
16909 (abstd2, *abstd2_fpr): Merge.
16910 * config/rs6000/e500.h: Delete file.
16911 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
16912 TARGET_FPRS by 1 and simplify.
16913 * config/rs6000/rs6000-c.c: Ditto.
16914 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
16915 TARGET_DF_SPE by 0.
16916 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
16918 * config/rs6000/rs6000.md: Ditto.
16919 (floatdidf2, *floatdidf2_fpr): Merge.
16920 (move_from_CR_gt_bit): Delete.
16921 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
16922 (E500_CR_IOR_COMPARE): Delete.
16923 (All patterns that require !TARGET_FPRS): Delete.
16924 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
16926 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16928 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
16930 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16932 * graphds.c (add_edge): Intitialize edge's attached data.
16933 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
16934 pointer parameter. Call pointed function on each edge during
16935 graph traversing. Skip traversing the edge when the function
16937 (graphds_dfs, graphds_scc): Ditto.
16938 (for_each_edge): New parameter. Pass the new parameter to callback
16940 * graphds.h (skip_edge_callback): New function pointer type.
16941 (graphds_dfs, graphds_scc): New function pointer parameter.
16942 (graphds_edge_callback, for_each_edge): New parameter.
16944 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16946 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
16947 out code checking if runtime alias check is possible to below ...
16948 Call the new function.
16949 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
16950 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
16952 2017-06-07 Marek Polacek <polacek@redhat.com>
16955 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
16956 TYPE_OVERFLOW_WRAPS check.
16958 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16960 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
16961 if versioning is required.
16962 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
16963 peeling with the check for versioning.
16965 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16967 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
16968 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
16969 Set true to new parameter if new ssa variable is defined.
16970 (vect_gen_vector_loop_niters): Refactor. Set range information
16971 for the new vector loop bound variable.
16972 (vect_do_peeling): Ditto.
16974 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16976 * tree-affine.c (ssa.h): Include header file.
16977 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
16978 has wrapping overflow behavior.
16980 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16982 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
16984 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16986 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
16987 (tree_to_aff_combination): ... here.
16989 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16991 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
16992 reg_pressure model function.
16993 (ivopts_global_cost_for_size): Delete.
16994 (determine_set_costs, iv_ca_recount_cost): Call new model function
16995 ivopts_estimate_reg_pressure.
16997 2017-06-07 Tamar Christina <tamar.christina@arm.com>
16999 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17000 expensive than udiv. Remove floating point cases from mod.
17002 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17004 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17005 Increase idiv cost.
17007 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17009 * config/aarch64/aarch64.md
17010 (copysignsf3): Fix mask generation.
17012 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17014 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17016 (class dump_manager): Add register_dumps method.
17017 * dumpfile.c: Include langhooks.h.
17018 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17019 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17020 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17021 (dump_manager::dump_register): Start with 512 entries instead of 32.
17022 (dump_manager::register_dumps): New method.
17023 * toplev.c (general_init): Instead of invoking register_dumps
17024 langhook, invoke register_dumps method on the dump manager.
17025 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17028 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17030 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17031 Say that # requires an associated define_split to exist, and that
17032 the define_split must be suitable for use after register allocation.
17034 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17036 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17037 (compute_outgoing_frequencies): Also initialize zero counts.
17038 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17039 around loops; preserve more of profile when nothing changes.
17041 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17043 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17045 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17046 * config/arm/arm-cpu-cdata.h: Regenerate.
17047 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17048 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17049 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17050 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17051 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17053 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17055 * config/arm/t-rmprofile: Likewise.
17056 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17058 2017-06-06 David S. Miller <davem@davemloft.net>
17061 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17062 blockage if function uses alloca.
17064 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17066 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17067 New "uid" fields to hold pretty-print IDs of group and ref.
17068 Memory references are now identified as <group_id>:<ref_id>
17069 instead of using [random] addresses.
17070 (dump_mem_details): Simplify, no functional change.
17071 (dump_mem_ref): Simplify and make output more concise.
17072 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17073 (find_or_create_group): Initialize group uid.
17074 (record_ref): Initialize ref uid. Improve debug output.
17075 (prune_group_by_reuse, should_issue_prefetch_p,)
17076 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17077 (mark_nontemporal_store, determine_loop_nest_reuse):
17078 Improve debug output.
17080 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17082 * dbgcnt.def (prefetch): New debug counter.
17083 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
17084 (schedule_prefetches): Stop issueing prefetches if debug counter
17087 2017-06-06 Tom de Vries <tom@codesourcery.com>
17089 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
17090 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
17092 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17094 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
17095 Use aarch64_reg_or_zero predicate for operand 4.
17096 (aarch64_compare_and_swap<mode> define_insn_and_split):
17097 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
17098 (aarch64_store_exclusive<mode>): Likewise for operand 2.
17100 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17102 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
17103 (arm_compute_save_core_reg_mask): This.
17104 (thumb1_compute_save_reg_mask): Rename into ...
17105 (thumb1_compute_save_core_reg_mask): This.
17106 (arm_compute_save_reg0_reg12_mask): Adapt comment.
17107 (arm_compute_frame_layout): Likewise.
17109 2017-06-06 Richard Biener <rguenther@suse.de>
17111 PR tree-optimization/80974
17112 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
17113 keep or clear leaders SSA info.
17115 2017-06-06 Tom de Vries <tom@codesourcery.com>
17117 * config/nvptx/nvptx.c (split_mode_p): New function.
17118 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
17120 2017-06-06 Tom de Vries <tom@codesourcery.com>
17122 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
17124 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17127 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
17130 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17132 * shrink-wrap.c (handle_simple_exit): Update profile.
17133 (try_shrink_wrapping): Upate profile.
17135 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17137 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
17138 (tree_guess_outgoing_edge_probabilities): New.
17139 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
17140 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
17142 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17144 * ipa-split.c (split_function): Initialize return bb profile.
17146 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17148 * profile.c (compute_branch_probabilities): Also initialize
17149 EXIT_BLOCK profile.
17151 2017-06-06 Richard Biener <rguenther@suse.de>
17153 PR tree-optimization/80928
17154 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
17155 (vect_analyze_loop_operations): Properly guard analysis for
17157 (vect_transform_loop): Likewise.
17158 (vect_analyze_loop_2): Also reset SLP type on PHIs.
17159 (vect_model_induction_cost): Do not cost for pure SLP.
17160 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
17161 of induction in inner loop vectorization.
17162 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
17163 (vect_get_and_check_slp_defs): Handle vect_induction_def.
17164 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
17166 (vect_analyze_slp_cost_1): Cost induction.
17167 (vect_detect_hybrid_slp_stmts): Handle PHIs.
17168 (vect_get_slp_vect_defs): Likewise.
17169 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
17170 (vect_transform_stmt): Handle SLP reductions.
17171 * tree-vectorizer.h (vectorizable_induction): Adjust.
17173 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17175 * config/rs6000/rs6000.c (make_resolver_func): Update
17176 init_lowered_empty_function call.
17178 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
17180 * doc/invoke.texi: Document the -fprofile-abs-path option.
17181 * common.opt (fprofile-abs-path): New option.
17182 * gcov-io.h (gcov_write_filename): Declare.
17183 * gcov-io.c (gcov_write_filename): New function.
17184 * coverage.c (coverage_begin_function): Use gcov_write_filename.
17185 * profile.c (output_location): Likewise.
17187 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17189 * shring-wrap.c: Revert accidental commit.
17191 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
17193 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
17195 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17197 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
17199 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
17200 profile in callgraph edge.
17201 * profile-count.h (apply_probability): If THIS is 0, then result is 0
17202 (apply_scale): Likewise.
17203 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
17204 Also scale profile when inlining function with zero profile.
17205 (initialize_cfun): Update exit block profile even when it is zero.
17206 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
17207 when profile is read.
17209 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17211 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
17212 (CLONE_*): New constants to define the processors we can generate
17213 code for with the target_clone attribute.
17214 (rs6000_clone_map): New array to identify which clone processors
17215 the current program is running on.
17216 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
17217 target_clone attribute.
17218 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
17219 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
17220 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
17221 (cpu_expand_builtin): Add support for target_clone attribute.
17222 (rs6000_valid_attribute_p): Allow "default" attribute.
17223 (get_decl_name): New debug function to simplify printing the
17224 current function name in debugging statements.
17225 (rs6000_clone_priority): New functions to support the target_clone
17226 attribute, and be able to generate code to switch between ISA 2.05
17227 through ISA 3.0 (power6 through power9).
17228 (rs6000_compare_version_priority): Likewise.
17229 (rs6000_get_function_versions_dispatcher): Likewise.
17230 (make_resolver_func): Likewise.
17231 (add_condition_to_bb): Likewise.
17232 (dispatch_function_versions): Likewise.
17233 (rs6000_generate_version_dispatcher_body): Likewise.
17234 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
17235 (fusion_gpr_load_p): Fix a spacing issue.
17236 * doc/extend.texi (Common Function Attributes): Document that the
17237 PowerPC supports the target_clone attribute.
17239 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17241 * config/arm/arm.h: explain F symbol found in description of ARM
17242 register allocation in its legend.
17244 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17246 * config/mips/frame-header-opt.c: Include profile-count.h.
17247 * config/riscv/riscv.c: Include profile-count.h
17249 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17251 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
17253 (sm_set_flag_if_changed): Add bbs field.
17254 (execute_sm_if_changed_flag_set): Pass BBS.
17255 (execute_sm): Update.
17257 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17259 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
17262 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17264 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
17265 (peephole2): New peephole2 to emit the above.
17266 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
17268 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17270 * config/aarch64/aarch64.c (define_peephole2 above
17271 *sub_<shift>_<mode>): New peephole.
17273 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17275 * config/i386/i386.c (make_resolver_func): Update.
17276 * Makefile.in: Add profile-count.h and profile-count.o
17277 * auto-profile.c (afdo_indirect_call): Update to new API.
17278 (afdo_set_bb_count): Update.
17279 (afdo_propagate_edge): Update.
17280 (afdo_propagate_circuit): Update.
17281 (afdo_calculate_branch_prob): Update.
17282 (afdo_annotate_cfg): Update.
17283 * basic-block.h: Include profile-count.h
17284 (struct edge_def): Turn count to profile_count.
17285 (struct basic_block_def): Likewie.
17286 (REG_BR_PROB_BASE): Move to profile-count.h
17287 (RDIV): Move to profile-count.h
17288 * bb-reorder.c (max_entry_count): Turn to profile_count.
17289 (find_traces): Update.
17290 (rotate_loop):Update.
17291 (connect_traces):Update.
17292 (sanitize_hot_paths):Update.
17293 * bt-load.c (migrate_btr_defs): Update.
17294 * cfg.c (RDIV): Remove.
17295 (init_flow): Use alloc_block.
17296 (alloc_block): Uninitialize count.
17297 (unchecked_make_edge): Uninitialize count.
17298 (check_bb_profile): Update.
17299 (dump_edge_info): Update.
17300 (dump_bb_info): Update.
17301 (update_bb_profile_for_threading): Update.
17302 (scale_bbs_frequencies_int): Update.
17303 (scale_bbs_frequencies_gcov_type): Update.
17304 (scale_bbs_frequencies_profile_count): New.
17305 * cfg.h (update_bb_profile_for_threading): Update.
17306 (scale_bbs_frequencies_profile_count): Declare.
17307 * cfgbuild.c (compute_outgoing_frequencies): Update.
17308 (find_many_sub_basic_blocks): Update.
17309 * cfgcleanup.c (try_forward_edges): Update.
17310 (try_crossjump_to_edge): Update.
17311 * cfgexpand.c (expand_gimple_tailcall): Update.
17312 (construct_exit_block): Update.
17313 * cfghooks.c (verify_flow_info): Update.
17314 (dump_bb_for_graph): Update.
17315 (split_edge): Update.
17316 (make_forwarder_block): Update.
17317 (duplicate_block): Update.
17318 (account_profile_record): Update.
17319 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
17320 (get_estimated_loop_iterations): Update.
17321 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
17322 (single_likely_exit): Update.
17323 * cfgloopmanip.c (scale_loop_profile): Update.
17325 (set_zero_probability): Update.
17326 (lv_adjust_loop_entry_edge): Update.
17327 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17328 (purge_dead_edges): Update.
17329 (rtl_account_profile_record): Update.
17330 * cgraph.c (cgraph_node::create): Uninitialize count.
17331 (symbol_table::create_edge): Uninitialize count.
17332 (cgraph_update_edges_for_call_stmt_node): Update.
17333 (cgraph_edge::dump_edge_flags): Update.
17334 (cgraph_node::dump): Update.
17335 (cgraph_edge::maybe_hot_p): Update.
17336 * cgraph.h: Include profile-count.h
17337 (create_clone), create_edge, create_indirect_edge): Update.
17338 (cgraph_node): Turn count to profile_count.
17339 (cgraph_edge0: Likewise.
17340 (make_speculative, clone): Update.
17341 (create_edge): Update.
17342 (init_lowered_empty_function): Update.
17343 * cgraphclones.c (cgraph_edge::clone): Update.
17344 (duplicate_thunk_for_node): Update.
17345 (cgraph_node::create_clone): Update.
17346 * cgraphunit.c (cgraph_node::analyze): Update.
17347 (cgraph_node::expand_thunk): Update.
17348 * final.c (dump_basic_block_info): Update.
17349 * gimple-streamer-in.c (input_bb): Update.
17350 * gimple-streamer-out.c (output_bb): Update.
17351 * graphite.c (print_global_statistics): Update.
17352 (print_graphite_scop_statistics): Update.
17353 * hsa-brig.c: Include basic-block.h.
17354 * hsa-dump.c: Include basic-block.h.
17355 * hsa-gen.c (T sum_slice): Update.
17356 (convert_switch_statements):Update.
17357 * hsa-regalloc.c: Include basic-block.h.
17358 * ipa-chkp.c (chkp_produce_thunks): Update.
17359 * ipa-cp.c (struct caller_statistics): Update.
17360 (init_caller_stats): Update.
17361 (gather_caller_stats): Update.
17362 (ipcp_cloning_candidate_p): Update.
17363 (good_cloning_opportunity_p): Update.
17364 (get_info_about_necessary_edges): Update.
17365 (dump_profile_updates): Update.
17366 (update_profiling_info): Update.
17367 (update_specialized_profile): Update.
17368 (perhaps_add_new_callers): Update.
17369 (decide_about_value): Update.
17370 (ipa_cp_c_finalize): Update.
17371 * ipa-devirt.c (struct odr_type_warn_count): Update.
17372 (struct decl_warn_count): Update.
17373 (struct final_warning_record): Update.
17374 (possible_polymorphic_call_targets): Update.
17375 (ipa_devirt): Update.
17376 * ipa-fnsummary.c (redirect_to_unreachable): Update.
17377 * ipa-icf.c (sem_function::merge): Update.
17378 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
17379 * ipa-inline.c (compute_uninlined_call_time): Update.
17380 (compute_inlined_call_time): Update.
17381 (want_inline_small_function_p): Update.
17382 (want_inline_self_recursive_call_p): Update.
17383 (edge_badness): Update.
17384 (lookup_recursive_calls): Update.
17385 (recursive_inlining): Update.
17386 (inline_small_functions): Update.
17387 (dump_overall_stats): Update.
17388 (dump_inline_stats): Update.
17389 * ipa-profile.c (ipa_profile_generate_summary): Update.
17390 (ipa_propagate_frequency): Update.
17391 (ipa_profile): Update.
17392 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17393 * ipa-utils.c (ipa_merge_profiles): Update.
17394 * loop-doloop.c (doloop_modify): Update.
17395 * loop-unroll.c (report_unroll): Update.
17396 (unroll_loop_runtime_iterations): Update.
17397 * lto-cgraph.c (lto_output_edge): Update.
17398 (lto_output_node): Update.
17399 (input_node): Update.
17400 (input_edge): Update.
17401 (merge_profile_summaries): Update.
17402 * lto-streamer-in.c (input_cfg): Update.
17403 * lto-streamer-out.c (output_cfg): Update.
17404 * mcf.c (create_fixup_graph): Update.
17405 (adjust_cfg_counts): Update.
17406 (sum_edge_counts): Update.
17407 * modulo-sched.c (sms_schedule): Update.
17408 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
17409 * predict.c (maybe_hot_count_p): Update.
17410 (probably_never_executed): Update.
17411 (dump_prediction): Update.
17412 (combine_predictions_for_bb): Update.
17413 (propagate_freq): Update.
17414 (handle_missing_profiles): Update.
17415 (counts_to_freqs): Update.
17416 (rebuild_frequencies): Update.
17417 (force_edge_cold): Update.
17418 * predict.h: Include profile-count.h
17419 (maybe_hot_count_p, counts_to_freqs): UPdate.
17420 * print-rtl-function.c: Do not include cfg.h
17421 * print-rtl.c: Include basic-block.h
17422 * profile-count.c: New file.
17423 * profile-count.h: New file.
17424 * profile.c (is_edge_inconsistent): Update.
17425 (correct_negative_edge_counts): Update.
17426 (is_inconsistent): Update.
17427 (set_bb_counts): Update.
17428 (read_profile_edge_counts): Update.
17429 (compute_frequency_overlap): Update.
17430 (compute_branch_probabilities): Update; Initialize and deinitialize
17432 (branch_prob): Update.
17433 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
17434 (edge_gcov_count): New.
17435 (bb_gcov_count): New.
17436 * shrink-wrap.c (try_shrink_wrapping): Update.
17437 * tracer.c (better_p): Update.
17438 * trans-mem.c (expand_transaction): Update.
17439 (ipa_tm_insert_irr_call): Update.
17440 (ipa_tm_insert_gettmclone_call): Update.
17441 * tree-call-cdce.c: Update.
17442 * tree-cfg.c (gimple_duplicate_sese_region): Update.
17443 (gimple_duplicate_sese_tail): Update.
17444 (gimple_account_profile_record): Update.
17445 (execute_fixup_cfg): Update.
17446 * tree-inline.c (copy_bb): Update.
17447 (copy_edges_for_bb): Update.
17448 (initialize_cfun): Update.
17449 (freqs_to_counts): Update.
17450 (copy_cfg_body): Update.
17451 (expand_call_inline): Update.
17452 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17453 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
17454 (try_unroll_loop_completely): Update.
17455 (try_peel_loop): Update.
17456 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17457 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
17458 * tree-ssa-loop-split.c (connect_loops): Update.
17459 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
17460 * tree-ssa-reassoc.c (branch_fixup): Update.
17461 * tree-ssa-tail-merge.c (replace_block_by): Update.
17462 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
17463 (compute_path_counts): Update.
17464 (update_profile): Update.
17465 (recompute_probabilities): Update.
17466 (update_joiner_offpath_counts): Update.
17467 (estimated_freqs_path): Update.
17468 (freqs_to_counts_path): Update.
17469 (clear_counts_path): Update.
17470 (ssa_fix_duplicate_block_edges): Update.
17471 (duplicate_thread_path): Update.
17472 * tree-switch-conversion.c (case_bit_test_cmp): Update.
17473 (struct switch_conv_info): Update.
17474 * tree-tailcall.c (decrease_profile): Update.
17475 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17476 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17477 * value-prof.c (check_counter): Update.
17478 (gimple_divmod_fixed_value): Update.
17479 (gimple_mod_pow2): Update.
17480 (gimple_mod_subtract): Update.
17481 (gimple_ic_transform): Update.
17482 (gimple_stringop_fixed_value): Update.
17483 * value-prof.h (gimple_ic): Update.
17485 2017-06-02 Carl Love <cel@us.ibm.com>
17487 * config/rs6000/rs6000-c: Add support for built-in functions
17488 vector double vec_doublee (vector signed int);
17489 vector double vec_doublee (vector unsigned int);
17490 vector double vec_doublee (vector float);
17491 vector double vec_doubleh (vector signed int);
17492 vector double vec_doubleh (vector unsigned int);
17493 vector double vec_doubleh (vector float);
17494 vector double vec_doublel (vector signed int);
17495 vector double vec_doublel (vector unsigned int);
17496 vector double vec_doublel (vector float);
17497 vector double vec_doubleo (vector signed int);
17498 vector double vec_doubleo (vector unsigned int);
17499 vector double vec_doubleo (vector float);.
17500 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
17501 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
17503 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
17504 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
17505 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
17507 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
17508 vec_doublel, vec_doubleh.
17509 * doc/extend.texi: Update the built-in documentation file for the
17510 new built-in functions.
17512 2017-06-02 David Malcolm <dmalcolm@redhat.com>
17515 * ipa-inline-analysis.c (free_growth_caches): Set
17516 edge_removal_hook_holder to NULL after removing it.
17518 2017-06-02 Sudakshina Das <sudi.das@arm.com>
17520 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
17521 comparision with zero.
17523 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17524 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17525 for early expansion of vec_min and vec_max builtins.
17526 (builtin_function_type): Add min/max unsigned variants to those
17527 identified as having unsigned arguments.
17529 2017-06-02 Olivier Hainque <hainque@adacore.com>
17531 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
17533 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17535 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
17536 Use VALL_F16 iterator rather than VALL.
17538 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17540 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
17541 Emit CBNZ inside loop when doing a strong exchange and comparing
17542 against zero. Generate the CC flags after the loop.
17544 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
17546 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
17547 (dl_section_ref): New.
17548 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
17549 On AIX, append an expression to subtract the size of the
17550 section length to dl_section_ref.
17552 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17554 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17555 for early expansion of vector absolute builtins.
17557 2017-06-02 Richard Biener <rguenther@suse.de>
17559 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
17560 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
17562 2017-06-02 Richard Biener <rguenther@suse.de>
17564 PR tree-optimization/80948
17565 * tree-tailcall.c (find_tail_calls): Track stmts to move in
17566 stmt order as well.
17568 2017-06-02 Richard Biener <rguenther@suse.de>
17570 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
17572 * tree-vect-stmts.c (process_use): Do not mark backedge defs
17573 for inductions as relevant.
17575 2017-06-02 Richard Biener <rguenther@suse.de>
17577 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
17578 (vectorizable_induction): ... this. Remove dead code.
17580 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
17582 * builtins. (expand_builtin_alloca): Remove second parameter and
17583 infer its value from the first parameter instead.
17584 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
17586 2017-06-02 Jakub Jelinek <jakub@redhat.com>
17588 PR rtl-optimization/80903
17589 * loop-doloop.c (add_test): Unshare sequence.
17591 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
17593 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
17595 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
17597 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
17599 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
17600 xlogue_layout::get_instance, logue_layout::xlogue_layout,
17601 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
17602 (xlogue_layout::get_stub_rtx): Make static.
17603 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
17604 (xlogue_layout::compute_stub_managed_regs): Rename to...
17605 (xlogue_layout::count_stub_managed_regs): ...this.
17606 (xlogue_layout::is_stub_managed_reg): New function.
17607 (xlogue_layout::m_stub_names): Rename to...
17608 (xlogue_layout::s_stub_names): ...this, make static.
17609 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
17610 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
17611 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
17612 xlogue_layout::s_stub_names): Instantiate statics.
17613 (stub_managed_regs): Remove.
17614 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
17615 (disable_call_ms2sysv_xlogues): Rename to...
17616 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
17617 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
17619 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
17620 change after reload_completed.
17621 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
17623 (ix86_expand_prologue): Likewise.
17624 (ix86_expand_epilogue): Likewise.
17625 (ix86_expand_split_stack_prologue): Likewise.
17626 (ix86_compute_frame_layout): Remove frame parameter ...
17627 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
17628 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
17630 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
17631 (ix86_frame): Move from here ...
17632 * config/i386/i386.h (ix86_frame): ... to here.
17633 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
17634 complete ix86_frame data structure instead. Remove some_ld_name.
17636 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17638 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
17639 symbols that hold a DECL_VALUE_EXPR.
17641 2017-06-01 Martin Jambor <mjambor@suse.cz>
17643 PR tree-optimization/80898
17644 * tree-sra.c (process_subtree_disqualification): Removed.
17645 (disqualify_candidate): Do not acll
17646 process_subtree_disqualification.
17647 (subtree_mark_written_and_enqueue): New function.
17648 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
17649 RHS has been disqualified and re-queue LHS if necessary. Apart
17650 from that, ignore disqualified RHS.
17652 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17654 * config/s390/s390.c (s390_emit_epilogue): Disable early return
17655 address fetch for z10 or later.
17657 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17659 * config/arc/arc.md (tst_movb): Add guard when splitting.
17661 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17663 * config/arc/arc.c (arc_can_eliminate): Test against
17664 arc_frame_pointer_needed.
17666 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17668 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
17669 to prevent store reordering.
17670 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
17671 (type): Add block type.
17672 (stack_tie): Define special instruction to be used in
17675 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17677 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
17678 constraint. It is not valid for the pattern.
17679 (noncommutative_binary_comparison): Likewise.
17681 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17683 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
17686 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17688 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
17689 be used by the reg-alloc.
17691 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17693 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
17694 reg-alloc when having mul64 or mul32x16 instructions.
17695 (mulsidi3): Likewise.
17696 (umulsidi3): Likewise.
17697 (mulsi32x16): New pattern.
17698 (mulsi64): Likewise.
17699 (mulsidi64): Likewise.
17700 (umulsidi64): Likewise.
17701 (MUL32x16_REG): Define.
17702 (mul64_600): Use MUL32x16_REG.
17703 (mac64_600): Likewise.
17704 (umul64_600): Likewise.
17705 (umac64_600): Likewise.
17707 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17709 * config/arc/arc.md (mulsi3_700): Make it commutative.
17711 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
17713 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
17715 (*sign_extendsidi2_insn): Likewise for movstosw.
17717 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17719 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
17720 the type of the input discriminant value. Convert the
17721 discriminant value of signedness vary.
17723 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
17725 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
17726 Add to -Wall section.
17728 2017-06-01 Richard Biener <rguenther@suse.de>
17730 PR middle-end/66313
17731 * fold-const.c (fold_plusminus_mult_expr): If the factored
17732 factor may be zero use a wrapping type for the inner operation.
17733 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
17734 and handle moved defs.
17735 (process_assignment): Properly guard the unary op case. Return a
17736 tri-state indicating that moving the stmt before the call may allow
17737 to continue. Pass through to_move.
17738 (find_tail_calls): Handle moving unrelated defs before
17741 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
17744 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
17745 splitter result in the canonical way.
17747 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
17749 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
17750 also for 32bit target. Update insn attributes.
17751 (zero-extendsidi2 splitter): Allow all registers for operand 1.
17753 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
17755 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
17756 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
17757 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
17758 (_mm_maskz_min_ss): New intrinsics.
17760 2017-05-31 Martin Liska <mliska@suse.cz>
17762 * tree-vect-loop.c (vect_create_epilog_for_reduction):
17763 Change comment style to one we normally use.
17764 (vectorizable_reduction): Likewise.
17765 (vectorizable_induction): Likewise.
17766 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17767 (vectorizable_call): Likewise.
17768 (vectorizable_simd_clone_call): Likewise.
17769 (vectorizable_conversion): Likewise.
17770 (vectorizable_assignment): Likewise.
17771 (vectorizable_shift): Likewise.
17772 (vectorizable_operation): Likewise.
17773 (vectorizable_store): Likewise.
17774 (vectorizable_load): Likewise.
17775 * tree-vectorizer.h: Likewise.
17777 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
17779 * passes.c (emergency_dump_function): New.
17780 * tree-pass.h (emergency_dump_function): Declare.
17781 * plugin.c (plugins_internal_error_function): Remove.
17782 * plugin.h (plugins_internal_error_function): Remove declaration.
17783 * toplev.c (internal_error_function): New static function. Use it...
17784 (general_init): ...here.
17786 2017-05-31 Graham Markall <graham.markall@embecosm.com>
17788 * config/arc/arc.c (arc_print_operand): Handle constant operands.
17789 (arc_rtx_costs): Add costs for new patterns.
17790 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
17791 * config/arc/predicates.md: Add _1_2_3_operand predicate.
17793 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17795 * tree-ssa-strlen.c (get_next_strinfo): New function.
17796 (get_stridx_plus_constant): Use it.
17797 (zero_length_string): Likewise.
17798 (adjust_related_strinfos): Likewise.
17799 (adjust_last_stmt): Likewise.
17801 2017-05-31 Richard Biener <rguenther@suse.de>
17804 * config/i386/i386.c (ix86_expand_builtin): Remove assert
17805 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
17807 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17809 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
17810 loop_vinfo argument and use of dependence distance vectors.
17811 Check instead whether the two references differ only in their
17812 initial value and assume that they have the same alignment if the
17813 difference is a multiple of the vector alignment.
17814 (vect_analyze_data_refs_alignment): Update call accordingly.
17816 2017-05-31 Martin Liska <mliska@suse.cz>
17819 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
17821 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17823 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
17824 (create_intersect_range_checks): Move from ...
17825 * tree-data-ref.c (create_intersect_range_checks_index)
17826 (create_intersect_range_checks): ... to here.
17827 (create_runtime_alias_checks): New function factored from ...
17828 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
17829 here. Call above function.
17830 * tree-data-ref.h (create_runtime_alias_checks): New function.
17832 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17834 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
17835 segment length for dr_b and compute it in wide_int.
17837 2017-05-31 Richard Biener <rguenther@suse.de>
17839 PR tree-optimization/80906
17840 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
17841 and pass through iv_map.
17842 (copy_bb_and_scalar_dependences): Adjust.
17843 (translate_pending_phi_nodes): Likewise.
17844 (copy_loop_close_phi_args): Handle code-generating IVs instead
17847 2017-05-30 David Malcolm <dmalcolm@redhat.com>
17849 * diagnostic-color.c (color_dict): Add "type-diff".
17850 (parse_gcc_colors): Update comment.
17851 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
17852 -fdiagnostics-show-template-tree and -fno-elide-type.
17853 (GCC_COLORS): Add type-diff to example.
17855 (-fdiagnostics-show-template-tree): New.
17856 (-fno-elide-type): New.
17857 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
17858 the pp_format_decoder callback. Call any m_format_postprocessor's
17860 (pretty_printer::pretty_printer): Initialize
17861 m_format_postprocessor.
17862 (pretty_printer::~pretty_printer): Delete any
17863 m_format_postprocessor.
17864 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
17865 (class format_postprocessor): New class.
17866 (struct pretty_printer::format_decoder): Document the new parameters.
17867 (struct pretty_printer::m_format_postprocessor): New field.
17868 * tree-diagnostic.c (default_tree_printer): Update for new
17869 bool and const char ** params.
17870 * tree-diagnostic.h (default_tree_printer): Likewise.
17872 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
17874 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
17875 (lwa_operand): Delete rs6000_gen_cell_microcode test.
17876 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17877 rs6000_gen_cell_microcode code.
17878 (rs6000_final_prescan_insn): Delete.
17879 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
17880 "warn-cell-microcode" entries.
17881 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
17882 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
17883 throughout. Change cc_reg_not_micro_cr0_operand to
17884 cc_reg_not_cr0_operand throughout.
17885 (*extendhi<mode>2_noload): Delete.
17886 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
17887 (mwarn-cell-microcode): Delete.
17888 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
17889 -mgen-cell-microcode and -mwarn-cell-microcode.
17891 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
17894 * config/i386/constraints.md (Yd): New constraint.
17896 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
17897 and (?Yd, r) alternatives. Update insn attributes.
17898 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
17899 and (?*Yd, r) alternatives. Update insn attributes.
17900 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
17902 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
17904 * gimplify.c (gimplify_modify_expr): Don't create a
17905 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
17908 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17910 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
17912 2017-05-30 Richard Biener <rguenther@suse.de>
17914 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
17915 and reduc_def fields.
17916 (STMT_VINFO_REDUC_TYPE): New define.
17917 (STMT_VINFO_REDUC_DEF): Likewise.
17918 (vect_force_simple_reduction): Adjust prototype.
17919 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
17920 (vect_is_simple_reduction): Remove check_reduction argument.
17921 (vect_force_simple_reduction): Adjust and set
17922 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17923 (vectorizable_reduction): Do not re-do reduction analysis
17924 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17925 * tree-parloops.c (gather_scalar_reductions): Adjust.
17927 2017-05-30 Richard Biener <rguenther@suse.de>
17929 PR middle-end/80901
17930 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
17933 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
17935 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
17936 Introduce unknown_misalignment parameter and remove vf.
17937 (vect_peeling_hash_get_lowest_cost):
17938 Pass unknown_misalignment parameter.
17939 (vect_enhance_data_refs_alignment):
17940 Fix unsupportable data ref treatment.
17942 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17944 * tree-vect-data-refs.c (vect_get_data_access_cost):
17945 Workaround for SLP handling.
17946 (vect_enhance_data_refs_alignment):
17947 Compute costs for doing no peeling at all, compare to the best
17948 peeling costs so far and avoid peeling if cheaper.
17950 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17952 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
17953 Return peeling info and set costs to zero for unlimited cost
17955 (vect_enhance_data_refs_alignment): Also inspect all datarefs
17956 with unknown misalignment. Compute and costs for unknown
17957 misalignment, compare them to the costs for known misalignment
17958 and choose the cheapest for peeling.
17960 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17962 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
17963 (vect_get_peeling_costs_all_drs): Create function.
17964 (vect_peeling_hash_get_lowest_cost):
17965 Use vect_get_peeling_costs_all_drs.
17966 (vect_peeling_supportable): Create function.
17967 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
17969 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17971 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
17972 DR_HAS_NEGATIVE_STEP.
17973 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
17974 (vect_enhance_data_refs_alignment): Use.
17975 (vect_duplicate_ssa_name_ptr_info): Use.
17976 * tree-vectorizer.h (dr_misalignment): Use.
17977 (known_alignment_for_access_p): Use.
17979 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
17982 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
17984 (has_section_name): New function.
17986 2017-05-30 Martin Liska <mliska@suse.cz>
17989 * auto-profile.c (get_function_decl_from_block): Fix
17992 2017-05-30 Richard Biener <rguenther@suse.de>
17994 PR middle-end/80876
17995 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
17997 2017-05-30 Martin Liska <mliska@suse.cz>
17999 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18000 * dumpfile.h (struct dump_file_info): Remove ctors.
18002 2017-05-30 Martin Liska <mliska@suse.cz>
18004 * predict.def: Fix GNU coding style.
18006 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18008 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18009 Mark 'to' argument with ATTRIBUTE_UNUSED.
18011 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18013 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18014 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18015 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18018 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18020 * doc/install.texi (Options specification): Restore entry of
18021 --enable-sjlj-exceptions.
18023 2017-05-27 Michael Eager <eager@eagercon.com>
18026 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18028 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18030 * config/microblaze/microblaze.h
18031 (FIXED_REGISTERS): Update in macro.
18032 (CALL_USED_REGISTERS): Update in macro.
18034 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18036 * doc/install.texi: Add links to macOS binary distributions.
18038 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18042 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18044 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18046 2017-05-26 Martin Liska <mliska@suse.cz>
18048 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18050 2017-05-26 Martin Liska <mliska@suse.cz>
18052 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18053 always leading ';; '.
18054 (dump_bb_info): Likewise.
18055 (brief_dump_cfg): Likewise.
18056 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18057 * dumpfile.c: Remove usage of TDF_VERBOSE.
18058 * dumpfile.h (enum dump_kind): Likewise.
18059 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18060 * print-tree.c (debug_verbose): Remove.
18061 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18062 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18063 * tree-diagnostic.c (default_tree_printer): Replace
18064 TDF_DIAGNOSTIC with TDF_SLIM.
18066 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18068 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18069 in parameter loop, rather than loop_vinfo.
18070 (create_intersect_range_checks): Ditto.
18071 (vect_create_cond_for_alias_checks): Update call to above functions.
18073 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18075 PR tree-optimization/80815
18076 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18077 for merging runtime alias checks. Handle negative DR_STEPs.
18079 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18081 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
18083 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
18084 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
18085 out code pruning runtime alias checks.
18086 * tree-data-ref.c (prune_runtime_alias_test_list): New function
18087 factored out from above.
18088 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18090 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18092 (prune_runtime_alias_test_list): New decalaration.
18094 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18096 * tree-vect-data-refs.c (compare_tree): Rename and move ...
18097 * tree-data-ref.c (data_ref_compare_tree): ... to here.
18098 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
18099 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
18100 (operator==, comp_dr_with_seg_len_pair): Ditto.
18101 (vect_prune_runtime_alias_test_list): Ditto.
18103 2017-05-26 Martin Liska <mliska@suse.cz>
18106 * params.def: Bound partial-inlining-entry-probability param.
18108 2017-05-26 Marek Polacek <polacek@redhat.com>
18111 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
18114 2017-05-26 Richard Biener <rguenther@suse.de>
18116 PR tree-optimization/80842
18117 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
18120 2017-05-26 Richard Biener <rguenther@suse.de>
18122 PR tree-optimization/80844
18123 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
18125 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
18127 * doc/md.texi (Machine Constraints): Update x86 family
18128 machine constraints section to match 'config/i386/constraints.md'.
18130 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
18132 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
18134 2017-05-25 Nathan Sidwell <nathan@acm.org>
18136 * doc/invoke.texi (--enable-languages): Update documentation.
18138 2017-05-25 Martin Liska <mliska@suse.cz>
18140 * dumpfile.c: Add TDF_FOLDING.
18141 * dumpfile.h (enum dump_kind): Likewise.
18142 * genmatch.c (dt_simplify::gen_1): Use it.
18144 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18146 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
18148 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18150 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18151 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
18153 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18155 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
18156 * match.pd (X == C): Rewrite it here.
18157 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
18158 with_certain_nonzero_bits2): New predicates.
18159 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
18161 2017-05-24 Nathan Sidwell <nathan@acm.org>
18163 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
18166 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
18169 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18171 * config/powerpcspe: New port. Files are copied from the rs6000
18172 port, with "rs6000" in filenames replaced by "powerpcspe".
18174 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
18176 PR rtl-optimization/80754
18177 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
18179 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
18181 * config/sparc/sparc.md (length): Return the correct value for -mflat
18182 sibcalls to match output_sibcall.
18184 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18188 * config/rs6000/rs6000.c (struct machine_function): Add new field
18190 (rs6000_get_separate_components): Init that field, use it.
18191 (rs6000_components_for_bb): Use the field.
18193 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
18195 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
18197 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
18199 PR middle-end/80823
18200 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
18202 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18205 * config/s390/s390.c (s390_check_qrst_address): Check incoming
18206 address against address_operand predicate.
18207 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
18209 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
18211 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
18212 parameters passed indirectly.
18214 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
18216 * config/i386/i386.md (*movdi_internal): Remove SSE4
18217 alternative 18 (?r, *v). Update insn attributes.
18218 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
18219 Update insn attributes.
18220 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
18221 Update insn attributes.
18222 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
18223 alternative 1 (r, v). Remove isa attribute.
18224 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
18225 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
18226 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18228 2017-05-23 Tom de Vries <tom@codesourcery.com>
18230 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
18233 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18235 * cgraphunit.c (symbol_table::process_new_functions): Update.
18236 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
18237 (inline_generate_summary): Rename to ...
18238 (ipa_fn_summary_generate): ... this one.
18239 (inline_read_summary): Rename to ...
18240 (ipa_fn_summary_read): ... this one.
18241 (inline_write_summary): Rename to ...
18242 (ipa_fn_summary_write): ... this one.
18243 (inline_free_summary): Rename to ...
18244 (ipa_free_fn_summary): ... this one.
18245 (pass_data_local_fn_summary, pass_local_fn_summary,
18246 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
18247 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
18248 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
18249 make_pass_ipa_fn_summary): New.
18250 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
18251 inline_write_summary, inline_free_summary): Remove.
18252 (ipa_free_fn_summary) : New.
18253 * ipa-inline.c (ipa_inline): Update.
18254 (pass_ipa_inline): Do not generate summaries.
18255 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
18257 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
18258 and add pass_ipa_fn_summary.
18259 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
18261 (make_pass_inline_parameters): Remove.
18263 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
18265 * omp-low.c (struct omp_context): Remove "default_kind" member.
18268 * omp-offload.c (execute_oacc_device_lower): Remove the
18269 parallelism dimensions function attributes for unparallelized
18270 OpenACC kernels constructs.
18272 2017-05-23 Martin Liska <mliska@suse.cz>
18274 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
18276 (cgraph_edge::make_speculative): Likewise.
18277 (cgraph_edge::resolve_speculation): Likewise.
18278 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
18279 (cgraph_node::dump): Likewise.
18280 * cgraph.h: Likewise.
18281 * cgraphunit.c (analyze_functions): Likewise.
18282 (symbol_table::compile): Likewise.
18283 * ipa-cp.c (print_all_lattices): Likewise.
18284 (determine_versionability): Likewise.
18285 (initialize_node_lattices): Likewise.
18286 (ipcp_verify_propagated_values): Likewise.
18287 (estimate_local_effects): Likewise.
18288 (update_profiling_info): Likewise.
18289 (create_specialized_node): Likewise.
18290 (perhaps_add_new_callers): Likewise.
18291 (decide_about_value): Likewise.
18292 (decide_whether_version_node): Likewise.
18293 (identify_dead_nodes): Likewise.
18294 (ipcp_store_bits_results): Likewise.
18295 * ipa-devirt.c (dump_targets): Likewise.
18296 (ipa_devirt): Likewise.
18297 * ipa-icf.c (sem_item::dump): Likewise.
18298 (sem_function::equals): Likewise.
18299 (sem_variable::equals): Likewise.
18300 (sem_item_optimizer::read_section): Likewise.
18301 (sem_item_optimizer::execute): Likewise.
18302 (congruence_class::dump): Likewise.
18303 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
18304 (dump_inline_summary): Likewise.
18305 (estimate_node_size_and_time): Likewise.
18306 (inline_analyze_function): Likewise.
18307 * ipa-inline-transform.c (inline_call): Likewise.
18308 * ipa-inline.c (report_inline_failed_reason): Likewise.
18309 (want_early_inline_function_p): Likewise.
18310 (edge_badness): Likewise.
18311 (update_edge_key): Likewise.
18312 (inline_small_functions): Likewise.
18313 * ipa-profile.c (ipa_profile): Likewise.
18314 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
18315 (ipa_make_edge_direct_to_target): Likewise.
18316 (remove_described_reference): Likewise.
18317 (ipa_impossible_devirt_target): Likewise.
18318 (propagate_controlled_uses): Likewise.
18319 (ipa_print_node_params): Likewise.
18320 (ipcp_transform_function): Likewise.
18321 * ipa-pure-const.c (pure_const_read_summary): Likewise.
18322 (propagate_pure_const): Likewise.
18323 * ipa-reference.c (generate_summary): Likewise.
18324 (read_write_all_from_decl): Likewise.
18325 (propagate): Likewise.
18326 (ipa_reference_read_optimization_summary): Likewise.
18327 * ipa-utils.c (ipa_merge_profiles): Likewise.
18328 * ipa.c (walk_polymorphic_call_targets): Likewise.
18329 (symbol_table::remove_unreachable_nodes): Likewise.
18330 (ipa_single_use): Likewise.
18331 * passes.c (execute_todo): Likewise.
18332 * predict.c (drop_profile): Likewise.
18333 * symtab.c (symtab_node::get_dump_name): New function.
18334 (symtab_node::dump_name): Likewise.
18335 (symtab_node::dump_asm_name): Likewise.
18336 (symtab_node::dump_references): Likewise.
18337 (symtab_node::dump_referring): Likewise.
18338 (symtab_node::dump_base): Likewise.
18339 (symtab_node::debug_symtab): Likewise.
18340 * tree-sra.c (convert_callers_for_node): Likewise.
18341 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
18342 * value-prof.c (init_node_map): Likewise.
18344 2017-05-23 Martin Liska <mliska@suse.cz>
18346 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
18347 and symtab_node::debug_symtab to symbol_table::debug.
18348 * cgraphunit.c (analyze_functions): Use the renamed function.
18349 (symbol_table::compile): Likewise.
18350 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
18351 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
18352 * passes.c (execute_todo): Likewise.
18353 * symtab.c (symbol_table::dump): New function.
18354 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
18356 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18358 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
18359 that nonconst implies exec.
18361 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18363 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18364 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18365 (inline_edge_summary_vec): Turn into ...
18366 (ipa_call_summaries): ... this one.
18367 (redirect_to_unreachable, edge_set_predicate,
18368 evaluate_properties_for_edge, inline_summary_alloc,
18369 reset_ipa_call_summary, reset_inline_summary,
18370 inline_summary_t::duplicate): Update.
18371 (inline_edge_duplication_hook): Turn to ...
18372 (ipa_call_summary_t::duplicate): ... this one.
18373 (inline_edge_removal_hook): Turn to ...
18374 (ipa_call_summary_t::remove): ... this one.
18375 (dump_inline_edge_summary): Turn to ...
18376 (dump_ipa_call_summary): ... this one.
18377 (estimate_function_body_sizes): Update.
18378 (inline_update_callee_summaries): Update.
18379 (remap_edge_change_prob): Update.
18380 (remap_edge_summaries): Update.
18381 (inline_merge_summary): Update.
18382 (do_estimate_edge_time): Update.
18383 (inline_generate_summary): Update.
18384 (inline_read_section): Update.
18385 (inline_read_summary): Update.
18386 (inline_free_summary): Update.
18387 * ipa-inline.c (can_inline_edge_p): Update.
18388 (compute_inlined_call_time): Update.
18389 (want_inline_small_function_p): Update.
18390 (edge_badness): Update.
18391 (early_inliner): Update.
18392 * ipa-inline.h (inline_edge_summary): Turn to ...
18393 (ipa_call_summary): ... this one.
18394 (ipa_call_summary_t): New class.
18395 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18396 (ipa_call_summaries): New.
18397 (inline_edge_summary): Remove.
18398 (estimate_edge_growth): Update.
18399 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18400 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18401 * ipa-split.c (execute_split_functions): Update.
18402 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18404 2017-05-23 Tom de Vries <tom@codesourcery.com>
18406 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18407 attributes): Document rdrand effective target.
18409 2017-05-23 Tom de Vries <tom@codesourcery.com>
18411 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18412 attributes): Sort alphabetically.
18414 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
18416 * config/avr/genmultilib.awk: Use gsub instead of gensub.
18418 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
18421 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
18422 V2DF/V2DI splat into two separate patterns, one that handles
18423 registers, and the other that only handles memory. Drop support
18424 for splatting from a GPR on ISA 2.07 and then splitting the
18425 splat into direct move and splat.
18426 (vsx_splat_<mode>_reg): Likewise.
18427 (vsx_splat_<mode>_mem): Likewise.
18429 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
18431 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
18433 2017-05-22 Jakub Jelinek <jakub@redhat.com>
18435 PR middle-end/80809
18436 * omp-low.c (finish_taskreg_remap): New function.
18437 (finish_taskreg_scan): If unit size of ctx->record_type
18438 is non-constant, unshare the size expression and replace
18439 decls in it with possible outer var refs.
18441 PR middle-end/80809
18442 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
18443 GOVD_SHARED rather than GOVD_PRIVATE with it.
18444 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
18445 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
18447 PR middle-end/80853
18448 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
18449 as last argument to build_outer_var_ref for pointer bases of array
18450 section reductions.
18452 2017-05-19 Martin Sebor <msebor@redhat.com>
18454 * print-tree.c (print_node): Print DECL_READ_P flag.
18456 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18458 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
18459 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
18460 * cgraph.c: Likewise.
18461 * cgraphunit.c: Likewise.
18462 * gengtype.c: Likewise.
18463 * ipa-cp.c: Likewise.
18464 * ipa-devirt.c: Likewise.
18465 * ipa-icf.c: Likewise.
18466 * ipa-predicate.c: Likewise.
18467 * ipa-profile.c: Likewise.
18468 * ipa-prop.c: Likewise.
18469 * ipa-split.c: Likewise.
18471 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
18472 edge_predicate_pool, dump_inline_hints,
18473 inline_summary::account_size_time, redirect_to_unreachable,
18474 edge_set_predicate, set_hint_predicate,
18475 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
18476 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
18477 inline_summary_t::remove, remap_hint_predicate_after_duplication,
18478 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
18479 ipa_call_summary_t::remove, initialize_growth_caches,
18480 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
18481 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
18482 mark_modified, unmodified_parm_1, unmodified_parm,
18483 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
18484 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
18485 compute_bb_predicates, will_be_nonconstant_expr_predicate,
18486 will_be_nonconstant_predicate, record_modified_bb_info,
18487 get_minimal_bb, record_modified, param_change_prob,
18488 phi_result_unknown_predicate, predicate_for_phi_result,
18489 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
18490 estimate_function_body_sizes, compute_inline_parameters,
18491 compute_inline_parameters_for_curren, pass_data_inline_parameters,
18492 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
18493 inline_update_callee_summaries, remap_edge_change_prob,
18494 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
18495 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
18496 inline_analyze_function, inline_summary_t::insert,
18497 inline_generate_summary, read_ipa_call_summary, inline_read_section,
18498 inline_read_summary, write_ipa_call_summary, inline_write_summary,
18499 inline_free_summary): Move to ipa-fnsummary.h
18500 (predicate_t): Remove.
18501 * ipa-fnsummary.c: New file.
18502 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
18503 (enum inline_hints_vals, inline_hints, agg_position_info,
18504 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
18505 inline_summaries, ipa_call_summary, ipa_call_summary_t,
18506 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
18507 dump_inline_summary, dump_inline_hints, inline_generate_summary,
18508 inline_read_summary, inline_write_summary, inline_free_summary,
18509 inline_analyze_function, initialize_inline_failed,
18510 inline_merge_summary, inline_update_overall_summary,
18511 compute_inline_parameters): Move to ipa-fnsummary.h
18512 * ipa-fnsummary.h: New file.
18513 * ipa-inline-transform.h: Include ipa-inline.h.
18514 * ipa-inline.c: LIkewise.
18516 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18518 * ipa-inline.c (edge_badness): Use inlined_time instead of
18519 inline_summaries->get.
18521 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18523 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
18525 2017-05-22 Nathan Sidwell <nathan@acm.org>
18527 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
18528 (fdump-lang): Document 'raw' option.
18529 * dumpfile.h (TDI_tu): Delete.
18530 * dumpfile.c (dump_files): Remove translation-unit.
18531 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
18533 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
18535 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
18536 command option from $(AWK) call.
18537 * config/avr/genmultilib.awk: Simplify and rewrite so that it
18538 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
18539 [FORMAT]: Remove handling of variable.
18540 * config/avr/t-multilib: Regenerate.
18542 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18544 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
18546 (dump_inline_summary): Do not print self_time.
18547 (estimate_function_body_sizes): Do not set self_time.
18548 (compute_inline_parameters): Likewise.
18549 (inline_read_section, inline_write_summary): Do not stream self_time.
18550 * ipa-inline.h (inline_summary): Drop self_time.
18552 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18554 * ipa-inline-analysis.c (account_size_time): Rename to ...
18555 (inline_summary::account_size_time): ... this one.
18556 (reset_ipa_call_summary): Turn to ...
18557 (ipa_call_summary::reset): ... this one.
18558 (reset_inline_summary): Turn to ...
18559 (inline_summary::reset): ... this one.
18560 (inline_summary_t::remove): Update.
18561 (inline_summary_t::duplicate): Update.
18562 (ipa_call_summary_t::remove): Update.
18563 (dump_inline_summary): Update.
18564 (estimate_function_body_sizes): Update.
18565 (compute_inline_parameters): Update.
18566 (estimate_node_size_and_time): Update.
18567 (inline_merge_summary): Update.
18568 (inline_update_overall_summary): Update.
18569 (inline_read_section): Update.
18570 (inline_write_summary): Update.
18571 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
18572 add account_size_time and reset member functions.
18573 (ipa_call_summary): Add reset function.
18574 * ipa-predicate.h (predicate::operator &): Constify.
18576 2017-05-22 Richard Biener <rguenther@suse.de>
18578 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
18580 2017-05-19 Jason Merrill <jason@redhat.com>
18582 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
18584 2017-05-19 Marek Polacek <polacek@redhat.com>
18587 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
18588 TYPE_OVERFLOW_WRAPS checks.
18590 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
18592 * tree-core.h (enum omp_clause_default_kind): Add
18593 "OMP_CLAUSE_DEFAULT_PRESENT".
18594 * tree-pretty-print.c (dump_omp_clause): Handle it.
18595 * gimplify.c (enum gimplify_omp_var_data): Add
18596 "GOVD_MAP_FORCE_PRESENT".
18597 (gimplify_adjust_omp_clauses_1): Map it to
18598 "GOMP_MAP_FORCE_PRESENT".
18599 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
18601 * gimplify.c (oacc_default_clause): Clarify.
18603 2017-05-19 Nathan Sidwell <nathan@acm.org>
18605 LANG_HOOK_REGISTER_DUMPS
18606 * toplev.c (general_init): Call register dump lang hook.
18607 * doc/invoke.texi: Document -fdump-lang option family.
18608 * dumpfile.c (dump_files): Remove class dump here.
18609 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
18610 * dumpfile.h (tree_dump_index): Remove TDI_class.
18611 * langhooks-def.h (lhd_register_dumps): Declare.
18612 (LANG_HOOKS_REGISTER_DUMPS): Define.
18613 (LANG_HOOKS_INITIALIZER): Add it.
18614 * langhooks.c (lhd_register_dumps): Define.
18615 * langhooks.h (struct lang_hooks): Add register_dumps.
18617 2017-05-19 Nathan Sidwell <nathan@acm.org>
18619 * context.h (context::set_passes): New.
18620 * context.c (context::context): Do not create pass manager.
18621 * toplev.c (general_init): Create pass manager here.
18623 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
18625 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
18626 use this splitter if two add or or instructions would also work for
18627 the constant we want to generate.
18629 2017-05-19 Richard Biener <rguenther@suse.de>
18632 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
18633 predicate evaluation.
18635 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18637 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
18639 * ipa-inline.c (want_inline_small_function_p): Do not cast to
18642 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18644 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18645 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18646 (inline_edge_summary_vec): Turn into ...
18647 (ipa_call_summaries): ... this one.
18648 (redirect_to_unreachable, edge_set_predicate,
18649 evaluate_properties_for_edge, inline_summary_alloc,
18650 reset_ipa_call_summary, reset_inline_summary,
18651 inline_summary_t::duplicate): Update.
18652 (inline_edge_duplication_hook): Turn to ...
18653 (ipa_call_summary_t::duplicate): ... this one.
18654 (inline_edge_removal_hook): Turn to ...
18655 (ipa_call_summary_t::remove): ... this one.
18656 (dump_inline_edge_summary): Turn to ...
18657 (dump_ipa_call_summary): ... this one.
18658 (estimate_function_body_sizes): Update.
18659 (inline_update_callee_summaries): Update.
18660 (remap_edge_change_prob): Update.
18661 (remap_edge_summaries): Update.
18662 (inline_merge_summary): Update.
18663 (do_estimate_edge_time): Update.
18664 (inline_generate_summary): Update.
18665 (inline_read_section): Update.
18666 (inline_read_summary): Update.
18667 (inline_free_summary): Update.
18668 * ipa-inline.c (can_inline_edge_p): Update.
18669 (compute_inlined_call_time): Update.
18670 (want_inline_small_function_p): Update.
18671 (edge_badness): Update.
18672 (early_inliner): Update.
18673 * ipa-inline.h (inline_edge_summary): Turn to ...
18674 (ipa_call_summary): ... this one.
18675 (ipa_call_summary_t): New class.
18676 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18677 (ipa_call_summaries): New.
18678 (inline_edge_summary): Remove.
18679 (estimate_edge_growth): Update.
18680 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18681 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18682 * ipa-split.c (execute_split_functions): Update.
18683 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18685 2017-05-19 Richard Biener <rguenther@suse.de>
18687 PR middle-end/80764
18688 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
18690 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
18692 * config/rs6000/rs6000.c (struct machine_function): Add field
18693 fpr_is_wrapped_separately.
18694 (rs6000_get_separate_components): Use 64 components. Handle the
18695 new FPR components.
18696 (rs6000_components_for_bb): Handle the FPR components.
18697 (rs6000_emit_prologue_components): Handle the FPR components.
18698 (rs6000_emit_epilogue_components): Handle the FPR components.
18699 (rs6000_set_handled_components): Handle the FPR components.
18700 (rs6000_emit_prologue): Don't output prologue code for those FPRs
18701 that are already separately shrink-wrapped.
18702 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
18703 that are already separately shrink-wrapped.
18705 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18708 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
18711 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
18712 (define_peephole2 for Altivec d-form load): Add peepholes to catch
18713 cases where the register allocator uses a move and an offsettable
18714 memory operation to/from a FPR register on ISA 2.06/2.07.
18715 (define_peephole2 for Altivec d-form store): Likewise.
18717 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
18720 * config/i386/mmx.md (*mov<mode>_internal): Enable
18721 alternatives 11, 12, 13 and 14 also for 32bit targets.
18722 Remove alternatives 15, 16, 17 and 18.
18723 * config/i386/sse.md (vec_concatv2di): Change
18724 alternative (!x, *y) to (x, ?!*Yn).
18726 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
18728 * dumpfile.h (enum dump_kind): Remove stray comma.
18730 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18732 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
18733 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
18734 predicate::num_conditions
18735 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
18736 (CHANGED): turn into predicate::changed.
18737 (agg_position_info): Move to ipa-predicate.h
18738 (add_condition, predicate::add_clause, predicate::operator &=,
18739 predicate::or_with, predicate::evaluate, predicate::probability,
18740 dump_condition, dump_clause, predicate::dump,
18741 predicate::remap_after_duplication, predicate::remap_after_inlining,
18742 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
18743 (evaluate_conditions_for_known_args): Update.
18744 (set_cond_stmt_execution_predicate): Update.
18745 * ipa-inline.h: Include ipa-predicate.h
18746 (condition, inline_param_summary, conditions, agg_position_info,
18747 predicate): Move to ipa-predicate.h
18748 * ipa-predicate.c: New file.
18749 * ipa-predicate.h: New file.
18751 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
18753 * final.c (leaf_function_p): Check we are not in a sequence.
18755 2017-05-18 Martin Liska <mliska@suse.cz>
18757 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
18758 * dumpfile.c (dump_register): Use new enum dump_kind.
18759 (get_dump_file_name): Likewise.
18760 (dump_enable_all): Likewise.
18761 (dump_switch_p_1): Likewise.
18762 (enable_rtl_dump_file): Remove usage of TDF_RTL.
18763 * dumpfile.h (enum dump_kind): New enum type.
18764 (struct dump_file_info): Create constructor and
18765 format fields and comments.
18766 * passes.c (pass_manager::register_one_dump_file):
18768 * statistics.c (statistics_early_init): Likewise.
18769 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
18770 TDF_TREE with TDF_SLIM.
18771 (gather_memory_references_ref): Likewise.
18773 2017-05-18 Martin Liska <mliska@suse.cz>
18775 * vec.h (struct vnull): Use it.
18777 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18779 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
18780 (true_predicate, false_predicate, true_predicate_p,
18781 false_predicate_p): Remove.
18782 (single_cond_predicate, not_inlined_predicate): Turn to member function
18784 (add_condition): Update.
18785 (add_clause): Turn to...
18786 (predicate::add_clause): ... this one; update; allow passing NULL
18788 (and_predicates): Turn to ...
18789 (predicate::operator &=): ... this one.
18790 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
18791 (or_predicates): Turn to ...
18792 (predicate::or_with): ... this one.
18793 (evaluate_predicate): Turn to ...
18794 (predicate::evaluate): ... this one.
18795 (predicate_probability): Turn to ...
18796 (predicate::probability): ... this one.
18797 (dump_condition): Update.
18798 (dump_predicate): Turn to ...
18799 (predicate::dump): ... this one.
18800 (account_size_time): Update.
18801 (edge_set_predicate): Update.
18802 (set_hint_predicate): UPdate.
18803 (evaluate_conditions_for_known_args): Update.
18804 (evaluate_properties_for_edge): Update.
18805 (remap_predicate_after_duplication): Turn to...
18806 (predicate::remap_after_duplication): ... this one.
18807 (remap_hint_predicate_after_duplication): Update.
18808 (inline_summary_t::duplicate): UPdate.
18809 (dump_inline_edge_summary): Update.
18810 (dump_inline_summary): Update.
18811 (set_cond_stmt_execution_predicate): Update.
18812 (set_switch_stmt_execution_predicate): Update.
18813 (compute_bb_predicates): Update.
18814 (will_be_nonconstant_expr_predicate): Update.
18815 (will_be_nonconstant_predicate): Update.
18816 (phi_result_unknown_predicate): Update.
18817 (predicate_for_phi_result): Update.
18818 (array_index_predicate): Update.
18819 (estimate_function_body_sizes): Update.
18820 (estimate_node_size_and_time): Update.
18821 (estimate_ipcp_clone_size_and_time): Update.
18822 (remap_predicate): Rename to ...
18823 (predicate::remap_after_inlining): ... this one.
18824 (remap_hint_predicate): Update.
18825 (inline_merge_summary): Update.
18826 (inline_update_overall_summary): Update.
18827 (estimate_size_after_inlining): Update.
18828 (read_predicate): Rename to ...
18829 (predicate::stream_in): ... this one.
18830 (read_inline_edge_summary): Update.
18831 (write_predicate): Rename to ...
18832 (predicate::stream_out): ... this one.
18833 (write_inline_edge_summary): Update.
18834 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
18835 (clause_t): Turn to uint32_t
18836 (predicate): Turn to class; implement constructor and operators
18838 (size_time_entry): Update.
18839 (inline_summary): Update.
18840 (inline_edge_summary): Update.
18842 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
18844 * fold-const.c (fold_binary_loc): Move transformation...
18845 * match.pd (C - X CMP X): ... here.
18847 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
18849 * config/sparc/sparc.c (sparc_option_override): Set function
18850 alignment for -mcpu=niagara7 to 64 to match the I$ line.
18851 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
18853 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
18855 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
18857 2017-05-18 Marek Polacek <polacek@redhat.com>
18860 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
18861 (pass_ubsan::execute): Call gimple_assign_single_p instead of
18862 gimple_assign_load_p.
18864 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
18866 PR middle-end/80692
18867 * real.c (do_compare): Give decimal_do_compare preference over
18868 comparing just the signs.
18870 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
18872 * doc/md.texi (Canonicalization of Instructions): Describe the
18873 canonical form of instructions that inherently set a condition
18876 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
18878 PR middle-end/80775
18879 * tree-cfg.c: Move deletion of unreachable case statements to after
18880 the merging of consecutive case labels.
18882 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
18884 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
18885 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
18886 restoring of callee-saved registers.
18888 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
18890 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
18891 * config/visium/visium.c (single_set_and_flags): Likewise.
18892 * config/visium/visium.md (Substitutions): Likewise.
18894 2017-05-17 Martin Liska <mliska@suse.cz>
18896 * cfg.c: Introduce dump_flags_t type and
18897 use it instead of int type.
18899 * cfghooks.c: Likewise.
18900 * cfghooks.h (struct cfg_hooks): Likewise.
18901 * cfgrtl.c: Likewise.
18902 * cfgrtl.h: Likewise.
18903 * cgraph.c (cgraph_node::get_body): Likewise.
18904 * coretypes.h: Likewise.
18905 * domwalk.c: Likewise.
18906 * domwalk.h: Likewise.
18907 * dumpfile.c (struct dump_option_value_info): Likewise.
18908 (dump_enable_all): Likewise.
18909 (dump_switch_p_1): Likewise.
18910 (opt_info_switch_p): Likewise.
18911 * dumpfile.h (enum tree_dump_index): Likewise.
18912 (struct dump_file_info): Likewise.
18913 * genemit.c: Likewise.
18914 * generic-match-head.c: Likewise.
18915 * gengtype.c (open_base_files): Likewise.
18916 * gimple-pretty-print.c: Likewise.
18917 * gimple-pretty-print.h: Likewise.
18918 * graph.c (print_graph_cfg): Likewise.
18919 * graphite-scop-detection.c (dot_all_sese): Likewise.
18920 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
18921 * loop-unroll.c (report_unroll): Likewise.
18922 * passes.c (pass_manager::register_one_dump_file): Likewise.
18923 * print-tree.c: Likewise.
18924 * statistics.c: Likewise.
18925 * tree-cfg.c: Likewise.
18926 * tree-cfg.h: Likewise.
18927 * tree-dfa.c: Likewise.
18928 * tree-dfa.h: Likewise.
18929 * tree-dump.c (dump_function): Likewise.
18930 * tree-dump.h (struct dump_info): Likewise.
18931 * tree-pretty-print.c: Likewise.
18932 * tree-pretty-print.h: Likewise.
18933 * tree-ssa-live.c: Likewise.
18934 * tree-ssa-live.h: Likewise.
18935 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
18936 * tree-vect-loop.c: Likewise.
18937 * tree-vect-slp.c: Likewise.
18939 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
18940 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18942 PR tree-optimization/80457
18943 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
18944 of all arguments to a statement as scalar_to_vec operations.
18945 (vectorizable_call): Adjust call to vect_model_simple_cost for
18947 (vectorizable_conversion): Likewise.
18948 (vectorizable_assignment): Likewise.
18949 (vectorizable_shift): Likewise.
18950 (vectorizable_operation): Likewise.
18951 (vectorizable_comparison): Likewise.
18952 (vect_is_simple_cond): Record the def types for operands.
18953 (vectorizable_condition): Likewise, call vect_model_simple_cost.
18954 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
18955 for statement argument count.
18957 2017-05-16 Carl Love <cel@us.ibm.com>
18959 * config/rs6000/rs6000-c: Add support for built-in functions
18960 vector unsigned long long vec_bperm (vector unsigned long long,
18961 vector unsigned char)
18962 vector signed long long vec_mule (vector signed int,
18964 vector unsigned long long vec_mule (vector unsigned int,
18965 vector unsigned int)
18966 vector signed long long vec_mulo (vector signed int,
18968 vector unsigned long long vec_mulo (vector unsigned int,
18969 vector unsigned int)
18970 vector signed char vec_sldw (vector signed char,
18971 vector signed char,
18973 vector unsigned char vec_sldw (vector unsigned char,
18974 vector unsigned char,
18976 vector signed short vec_sldw (vector signed short,
18977 vector signed short,
18979 vector unsigned short vec_sldw (vector unsigned short,
18980 vector unsigned short,
18982 vector signed int vec_sldw (vector signed int,
18985 vector unsigned int vec_sldw (vector unsigned int,
18986 vector unsigned int,
18988 vector signed long long vec_sldw (vector signed long long,
18989 vector signed long long,
18991 vector unsigned long long vec_sldw (vector unsigned long long,
18992 vector unsigned long long,
18994 * config/rs6000/rs6000-c: Add support for built-in functions
18995 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
18996 * config/rs6000/altivec.h: Add defintion for vec_sldw.
18997 * doc/extend.texi: Update the built-in documentation for the
18998 new built-in functions.
19000 2017-05-16 Marek Polacek <polacek@redhat.com>
19004 * tree.c (save_expr): Don't fold the expression.
19006 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19008 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19009 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19010 and (?*y,m). Update insn attributes.
19012 2017-05-16 Martin Liska <mliska@suse.cz>
19014 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19015 flags argument of print_gimple_stmt, print_gimple_expr,
19016 print_generic_stmt and print_generic_expr.
19017 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19018 * coretypes.h: Likewise.
19019 * except.c (dump_eh_tree): Likewise.
19020 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19021 * gimple-pretty-print.h: Likewise.
19022 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19023 (backprop::push_to_worklist): Likewise.
19024 (backprop::pop_from_worklist): Likewise.
19025 (backprop::process_use): Likewise.
19026 (backprop::intersect_uses): Likewise.
19027 (note_replacement): Likewise.
19028 * gimple-ssa-store-merging.c
19029 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19030 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19031 (pass_store_merging::execute): Likewise.
19032 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19033 (ssa_base_cand_dump_callback): Likewise.
19034 (dump_incr_vec): Likewise.
19035 (replace_refs): Likewise.
19036 (replace_mult_candidate): Likewise.
19037 (create_add_on_incoming_edge): Likewise.
19038 (create_phi_basis): Likewise.
19039 (insert_initializers): Likewise.
19040 (all_phi_incrs_profitable): Likewise.
19041 (introduce_cast_before_cand): Likewise.
19042 (replace_one_candidate): Likewise.
19043 * gimplify.c (gimplify_expr): Likewise.
19044 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19045 (set_rename): Likewise.
19046 (rename_uses): Likewise.
19047 (copy_loop_phi_nodes): Likewise.
19048 (add_close_phis_to_merge_points): Likewise.
19049 (copy_loop_close_phi_args): Likewise.
19050 (copy_cond_phi_args): Likewise.
19051 (graphite_copy_stmts_from_block): Likewise.
19052 (translate_pending_phi_nodes): Likewise.
19053 * graphite-poly.c (print_pdr): Likewise.
19054 (dump_gbb_cases): Likewise.
19055 (dump_gbb_conditions): Likewise.
19056 (print_scop_params): Likewise.
19057 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19058 (build_cross_bb_scalars_use): Likewise.
19059 (gather_bbs::before_dom_children): Likewise.
19060 * hsa-dump.c (dump_hsa_immed): Likewise.
19061 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19062 (get_replacement_map): Likewise.
19063 * ipa-inline-analysis.c (dump_condition): Likewise.
19064 (estimate_function_body_sizes): Likewise.
19065 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19066 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19067 * ipa-prop.c (ipa_dump_param): Likewise.
19068 (ipa_print_node_jump_functions_for_edge): Likewise.
19069 (ipa_modify_call_arguments): Likewise.
19070 (ipa_modify_expr): Likewise.
19071 (ipa_dump_param_adjustments): Likewise.
19072 (ipa_dump_agg_replacement_values): Likewise.
19073 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19074 * ipa-pure-const.c (check_stmt): Likewise.
19075 (pass_nothrow::execute): Likewise.
19076 * ipa-split.c (execute_split_functions): Likewise.
19077 * omp-offload.c (dump_oacc_loop_part): Likewise.
19078 (dump_oacc_loop): Likewise.
19079 * trans-mem.c (tm_log_emit): Likewise.
19080 (tm_memopt_accumulate_memops): Likewise.
19081 (dump_tm_memopt_set): Likewise.
19082 (dump_tm_memopt_transform): Likewise.
19083 * tree-cfg.c (gimple_verify_flow_info): Likewise.
19084 (print_loop): Likewise.
19085 * tree-chkp-opt.c (chkp_print_addr): Likewise.
19086 (chkp_gather_checks_info): Likewise.
19087 (chkp_get_check_result): Likewise.
19088 (chkp_remove_check_if_pass): Likewise.
19089 (chkp_use_outer_bounds_if_possible): Likewise.
19090 (chkp_reduce_bounds_lifetime): Likewise.
19091 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
19092 (chkp_mark_completed_bounds): Likewise.
19093 (chkp_register_incomplete_bounds): Likewise.
19094 (chkp_mark_invalid_bounds): Likewise.
19095 (chkp_maybe_copy_and_register_bounds): Likewise.
19096 (chkp_build_returned_bound): Likewise.
19097 (chkp_get_bound_for_parm): Likewise.
19098 (chkp_build_bndldx): Likewise.
19099 (chkp_get_bounds_by_definition): Likewise.
19100 (chkp_generate_extern_var_bounds): Likewise.
19101 (chkp_get_bounds_for_decl_addr): Likewise.
19102 * tree-chrec.c (chrec_apply): Likewise.
19103 * tree-data-ref.c (dump_data_reference): Likewise.
19104 (dump_subscript): Likewise.
19105 (dump_data_dependence_relation): Likewise.
19106 (analyze_overlapping_iterations): Likewise.
19107 * tree-inline.c (expand_call_inline): Likewise.
19108 (tree_function_versioning): Likewise.
19109 * tree-into-ssa.c (dump_defs_stack): Likewise.
19110 (dump_currdefs): Likewise.
19111 (dump_names_replaced_by): Likewise.
19112 (dump_update_ssa): Likewise.
19113 (update_ssa): Likewise.
19114 * tree-object-size.c (pass_object_sizes::execute): Likewise.
19115 * tree-parloops.c (build_new_reduction): Likewise.
19116 (try_create_reduction_list): Likewise.
19117 (ref_conflicts_with_region): Likewise.
19118 (oacc_entry_exit_ok_1): Likewise.
19119 (oacc_entry_exit_single_gang): Likewise.
19120 * tree-pretty-print.h: Likewise.
19121 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
19122 (get_scalar_evolution): Likewise.
19123 (add_to_evolution): Likewise.
19124 (get_loop_exit_condition): Likewise.
19125 (analyze_evolution_in_loop): Likewise.
19126 (analyze_initial_condition): Likewise.
19127 (analyze_scalar_evolution): Likewise.
19128 (instantiate_scev): Likewise.
19129 (number_of_latch_executions): Likewise.
19130 (gather_chrec_stats): Likewise.
19131 (final_value_replacement_loop): Likewise.
19132 (scev_const_prop): Likewise.
19133 * tree-sra.c (dump_access): Likewise.
19134 (disqualify_candidate): Likewise.
19135 (create_access): Likewise.
19136 (reject): Likewise.
19137 (maybe_add_sra_candidate): Likewise.
19138 (create_access_replacement): Likewise.
19139 (analyze_access_subtree): Likewise.
19140 (analyze_all_variable_accesses): Likewise.
19141 (sra_modify_assign): Likewise.
19142 (initialize_constant_pool_replacements): Likewise.
19143 (find_param_candidates): Likewise.
19144 (decide_one_param_reduction): Likewise.
19145 (replace_removed_params_ssa_names): Likewise.
19146 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
19147 * tree-ssa-copy.c (dump_copy_of): Likewise.
19148 (copy_prop_visit_cond_stmt): Likewise.
19149 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
19150 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
19151 (record_equivalences_from_stmt): Likewise.
19152 * tree-ssa-dse.c (compute_trims): Likewise.
19153 (delete_dead_call): Likewise.
19154 (delete_dead_assignment): Likewise.
19155 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
19156 (forward_propagate_into_cond): Likewise.
19157 (pass_forwprop::execute): Likewise.
19158 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
19159 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
19161 (move_computations_worker): Likewise.
19162 (execute_sm): Likewise.
19163 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
19164 (remove_exits_and_undefined_stmts): Likewise.
19165 (remove_redundant_iv_tests): Likewise.
19166 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
19167 (adjust_iv_update_pos): Likewise.
19168 * tree-ssa-math-opts.c (bswap_replace): Likewise.
19169 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
19170 (value_replacement): Likewise.
19171 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
19172 * tree-ssa-pre.c (print_pre_expr): Likewise.
19173 (get_representative_for): Likewise.
19174 (create_expression_by_pieces): Likewise.
19175 (insert_into_preds_of_block): Likewise.
19176 (eliminate_insert): Likewise.
19177 (eliminate_dom_walker::before_dom_children): Likewise.
19178 (eliminate): Likewise.
19179 (remove_dead_inserted_code): Likewise.
19180 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
19181 * tree-ssa-reassoc.c (get_rank): Likewise.
19182 (eliminate_duplicate_pair): Likewise.
19183 (eliminate_plus_minus_pair): Likewise.
19184 (eliminate_not_pairs): Likewise.
19185 (undistribute_ops_list): Likewise.
19186 (eliminate_redundant_comparison): Likewise.
19187 (update_range_test): Likewise.
19188 (optimize_range_tests_var_bound): Likewise.
19189 (optimize_vec_cond_expr): Likewise.
19190 (rewrite_expr_tree): Likewise.
19191 (rewrite_expr_tree_parallel): Likewise.
19192 (linearize_expr): Likewise.
19193 (break_up_subtract): Likewise.
19194 (linearize_expr_tree): Likewise.
19195 (attempt_builtin_powi): Likewise.
19196 (attempt_builtin_copysign): Likewise.
19197 (transform_stmt_to_copy): Likewise.
19198 (transform_stmt_to_multiply): Likewise.
19199 (dump_ops_vector): Likewise.
19200 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
19201 (print_scc): Likewise.
19202 (set_ssa_val_to): Likewise.
19203 (visit_reference_op_store): Likewise.
19204 (visit_use): Likewise.
19205 (sccvn_dom_walker::before_dom_children): Likewise.
19206 (run_scc_vn): Likewise.
19207 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
19209 (expr_hash_elt::print): Likewise.
19210 (const_and_copies::pop_to_marker): Likewise.
19211 (const_and_copies::record_const_or_copy_raw): Likewise.
19212 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
19213 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
19214 (dump_predicates): Likewise.
19215 (find_uninit_use): Likewise.
19216 (warn_uninitialized_phi): Likewise.
19217 (pass_late_warn_uninitialized::execute): Likewise.
19218 * tree-ssa.c (verify_vssa): Likewise.
19219 (verify_ssa): Likewise.
19220 (maybe_optimize_var): Likewise.
19221 * tree-vrp.c (dump_value_range): Likewise.
19222 (dump_all_value_ranges): Likewise.
19223 (dump_asserts_for): Likewise.
19224 (register_edge_assert_for_2): Likewise.
19225 (vrp_visit_cond_stmt): Likewise.
19226 (vrp_visit_switch_stmt): Likewise.
19227 (vrp_visit_stmt): Likewise.
19228 (vrp_visit_phi_node): Likewise.
19229 (simplify_cond_using_ranges_1): Likewise.
19230 (fold_predicate_in): Likewise.
19231 (evrp_dom_walker::before_dom_children): Likewise.
19232 (evrp_dom_walker::push_value_range): Likewise.
19233 (evrp_dom_walker::pop_value_range): Likewise.
19234 (execute_early_vrp): Likewise.
19236 2017-05-16 Richard Biener <rguenther@suse.de>
19238 * dwarf2out.c (loc_list_from_tree_1): Do not create
19239 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
19241 2017-05-16 Richard Biener <rguenther@suse.de>
19243 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
19245 (note_variable_value_in_expr): If we resolved the decl ref
19246 do not push to the stack.
19248 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
19250 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
19251 operations in fast-math mode.
19252 (vaddq_f16): Likewise.
19253 (vmul_f16): Likewise.
19254 (vmulq_f16): Likewise.
19255 (vsub_f16): Likewise.
19256 (vsubq_f16): Likewise.
19257 * config/arm/neon.md (add<mode>3): New.
19259 (fma:<VH:mode>3): New. Also remove outdated comment.
19262 2017-05-16 Martin Liska <mliska@suse.cz>
19266 * ipa-devirt.c (warn_types_mismatch): Fix typo.
19267 (odr_types_equivalent_p): Likewise.
19269 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
19271 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
19273 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
19276 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
19277 non-interunit SSE move alternatives with '?'.
19278 (zero-extendsidi peephole2): New peephole to skip intermediate
19279 general register in SSE zero-extend sequence.
19281 2017-05-15 Jeff Law <law@redhat.com>
19283 * reorg.c (relax_delay_slots): Create a new variable to hold
19284 the temporary target rather than clobbering TARGET_LABEL.
19286 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
19287 missing argument to extract_bit_field call.
19288 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
19290 2017-05-15 Martin Liska <mliska@suse.cz>
19293 * gcc.c (process_command): Do not allow empty argument of -o option.
19295 2017-05-15 Renlin Li <renlin.li@arm.com>
19297 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
19298 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
19299 * config/aarch64/constraints.md (Usf): Add long call check.
19300 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
19301 (call_value): Likewise.
19302 (sibcall): Likewise.
19303 (sibcall_value): Likewise.
19305 (call_value_insn): New.
19306 (sibcall_insn): Update rtx pattern.
19307 (sibcall_value_insn): Likewise.
19308 (call_internal): Remove.
19309 (call_value_internal): Likewise.
19310 (sibcall_internal): Likewise.
19311 (sibcall_value_internal): Likewise.
19312 (call_reg): Likewise.
19313 (call_symbol): Likewise.
19314 (call_value_reg): Likewise.
19315 (call_value_symbol): Likewise.
19317 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
19320 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
19322 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
19324 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
19325 compatible with CCGOCmode and with CCZmode.
19327 2017-05-14 Martin Sebor <msebor@redhat.com>
19329 PR middle-end/77671
19330 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
19331 (gimple_fold_builtin_snprintf): Same.
19332 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
19333 (gimple_fold_builtin_snprintf): Same.
19334 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
19335 of character types.
19336 (is_call_safe): New function.
19337 (try_substitute_return_value): Call it.
19338 (try_simplify_call): New function.
19339 (pass_sprintf_length::handle_gimple_call): Call it.
19341 2017-05-14 Martin Sebor <msebor@redhat.com>
19343 PR middle-end/80669
19344 * builtins.c (expand_builtin_stpncpy): Simplify.
19346 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
19348 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
19349 * config/i386/i386.h
19350 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
19351 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
19352 (struct machine_function): Add new members call_ms2sysv,
19353 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
19354 (struct machine_frame_state): New fields sp_realigned and
19355 sp_realigned_offset.
19356 * config/i386/i386.c
19357 (enum xlogue_stub): New enum.
19358 (enum xlogue_stub_sets): New enum.
19359 (class xlogue_layout): New class.
19360 (struct ix86_frame): New fields stack_realign_allocate_offset,
19361 stack_realign_offset and outlined_save_offset. Modify comments to
19362 detail stack layout when using out-of-line stubs.
19363 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
19364 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
19365 -mcall-ms2sysv-xlogues.
19366 (stub_managed_regs): New static variable.
19367 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
19368 registers managed by out-of-line stub.
19369 (disable_call_ms2sysv_xlogues): New function.
19370 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
19371 m->call_ms2sysv when appropriate and compute frame layout for
19373 (sp_valid_at, fp_valid_at): New inline functions.
19374 (choose_basereg): New function.
19375 (choose_baseaddr): Add align parameter, use choose_basereg and modify
19377 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
19378 Use align parameter of choose_baseaddr to generated aligned SSE movs
19380 (pro_epilogue_adjust_stack): Modify to track
19381 machine_frame_state::sp_realigned.
19382 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
19383 (ix86_nsaved_sseregs): Likewise.
19384 (ix86_emit_save_regs): Likewise.
19385 (ix86_emit_save_regs_using_mov): Likewise.
19386 (ix86_emit_save_sse_regs_using_mov): Likewise.
19387 (get_scratch_register_on_entry): Likewise.
19388 (gen_frame_set): New function.
19389 (gen_frame_load): Likewise.
19390 (gen_frame_store): Likewise.
19391 (emit_outlined_ms2sysv_save): Likewise.
19392 (emit_outlined_ms2sysv_restore): Likewise.
19393 (ix86_expand_prologue): Modify stack re-alignment code and call
19394 emit_outlined_ms2sysv_save when appropriate.
19395 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
19396 parameter rtx_insn *insn, which allows the function to be used to only
19397 generate the notes.
19398 (ix86_expand_epilogue): Modify validity checks of frame and stack
19399 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
19400 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
19401 * config/i386/predicates.md
19402 (save_multiple): New predicate.
19403 (restore_multiple): Likewise.
19404 * config/i386/sse.md
19405 (save_multiple<mode>): New pattern.
19406 (save_multiple_realign<mode>): Likewise.
19407 (restore_multiple<mode>): Likewise.
19408 (restore_multiple_and_return<mode>): Likewise.
19409 (restore_multiple_leave_return<mode>): Likewise.
19410 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
19412 2017-05-14 Julia Koval <julia.koval@intel.com>
19414 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
19415 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
19416 (__builtin_ia32_xsetbv): New builtins.
19417 * config/i386/i386.c (ix86_expand_special_args_builtin):
19419 (ix86_expand_builtin): Special expand for new intrinsics.
19420 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
19421 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
19422 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
19424 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19426 * cfganal.c (inverted_post_order_compute): Change argument type
19428 * cfganal.h (inverted_post_order_compute): Adjust prototype.
19429 * df-core.c (rest_of_handle_df_initialize): Adjust.
19430 (rest_of_handle_df_finish): Likewise.
19431 (df_analyze_1): Likewise.
19432 (df_analyze): Likewise.
19433 (loop_inverted_post_order_compute): Change argument to be a vec *.
19434 (df_analyze_loop): Adjust.
19435 (df_get_n_blocks): Likewise.
19436 (df_get_postorder): Likewise.
19437 * df.h (struct df_d): Change field to be a vec.
19438 * lcm.c (compute_laterin): Adjust.
19439 (compute_available): Likewise.
19440 * lra-lives.c (lra_create_live_ranges_1): Likewise.
19441 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
19442 * tree-ssa-pre.c (compute_antic): Likewise.
19444 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19446 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
19447 (depth_first_search::depth_first_search): Change structure init
19448 function to this constructor.
19449 (depth_first_search::add_bb): Rename function to this member.
19450 (depth_first_search::execute): Likewise.
19451 (flow_dfs_compute_reverse_finish): Adjust.
19453 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19455 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
19456 (longest_simple_path): Likewise.
19457 * shrink-wrap.c (spread_components): Likewise.
19458 (disqualify_problematic_components): Likewise.
19459 (emit_common_heads_for_components): Likewise.
19460 (emit_common_tails_for_components): Likewise.
19461 (insert_prologue_epilogue_for_components): Likewise.
19463 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19465 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
19468 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19470 * df-core.c (df_set_blocks): Start using auto_bitmap.
19471 (df_compact_blocks): Likewise.
19472 * df-problems.c (df_rd_confluence_n): Likewise.
19473 * df-scan.c (df_insn_rescan_all): Likewise.
19474 (df_process_deferred_rescans): Likewise.
19475 (df_update_entry_block_defs): Likewise.
19476 (df_update_exit_block_uses): Likewise.
19477 (df_entry_block_bitmap_verify): Likewise.
19478 (df_exit_block_bitmap_verify): Likewise.
19479 (df_scan_verify): Likewise.
19480 * lra-constraints.c (lra_constraints): Likewise.
19481 (undo_optional_reloads): Likewise.
19482 (lra_undo_inheritance): Likewise.
19483 * lra-remat.c (calculate_gen_cands): Likewise.
19484 (do_remat): Likewise.
19485 * lra-spills.c (assign_spill_hard_regs): Likewise.
19486 (spill_pseudos): Likewise.
19487 * tree-ssa-pre.c (bitmap_set_and): Likewise.
19488 (bitmap_set_subtract_values): Likewise.
19490 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19492 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
19493 management with auto_bitmap.
19494 (fix_inter_tick): Likewise.
19495 (fix_recovery_deps): Likewise.
19496 * ira.c (add_store_equivs): Likewise.
19497 (find_moveable_pseudos): Likewise.
19498 (split_live_ranges_for_shrink_wrap): Likewise.
19499 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
19500 (rtx_reuse_manager::seen_def_p): Likewise.
19501 (rtx_reuse_manager::set_seen_def): Likewise.
19502 * print-rtl.h (class rtx_reuse_manager): Likewise.
19504 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19506 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
19508 (migrate_btr_def): Likewise.
19509 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
19510 * df-core.c (loop_post_order_compute): Likewise.
19511 (loop_inverted_post_order_compute): Likewise.
19512 * hsa-common.h: Likewise.
19513 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
19514 * init-regs.c (initialize_uninitialized_regs): Likewise.
19515 * ipa-inline.c (resolve_noninline_speculation): Likewise.
19516 (inline_small_functions): Likewise.
19517 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
19518 * ira.c (combine_and_move_insns): Likewise.
19519 (build_insn_chain): Likewise.
19520 * loop-invariant.c (find_invariants): Likewise.
19521 * lower-subreg.c (propagate_pseudo_copies): Likewise.
19522 * predict.c (tree_predict_by_opcode): Likewise.
19523 (predict_paths_leading_to): Likewise.
19524 (predict_paths_leading_to_edge): Likewise.
19525 (estimate_loops_at_level): Likewise.
19526 (estimate_loops): Likewise.
19527 * shrink-wrap.c (try_shrink_wrapping): Likewise.
19528 (spread_components): Likewise.
19529 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
19530 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
19531 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
19532 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
19533 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
19534 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
19535 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
19536 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
19537 (mark_threaded_blocks): Likewise.
19538 (thread_through_all_blocks): Likewise.
19539 * tree-ssa.c (verify_ssa): Likewise.
19540 (execute_update_addresses_taken): Likewise.
19541 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
19543 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19545 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
19547 (post_order_compute): Likewise.
19548 (inverted_post_order_compute): Likewise.
19549 (pre_and_rev_post_order_compute_fn): Likewise.
19551 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19553 * genrecog.c (int_set::int_set): Explicitly construct our
19554 auto_vec base class.
19555 * vec.h (auto_vec::auto_vec): New constructor.
19557 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19559 * bitmap.h (class auto_bitmap): New constructor taking
19560 bitmap_obstack * argument.
19562 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19564 * bitmap.h (class auto_bitmap): Change type of m_bits to
19565 bitmap_head, and adjust ctor / dtor and member operators.
19567 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
19569 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
19570 when returned register mode doesn't match original mode.
19572 2017-05-12 Jeff Law <law@redhat.com>
19573 Jakub Jelinek <jakub@redhat.com>
19575 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
19576 we look for cc setter after the compare-elim changes.
19577 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
19578 within the vector to match what compare-elim now expects.
19579 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
19580 (xorsi3_flags, one_cmplsi2_flags): Likewise.
19582 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
19583 after the compare-elim changes.
19584 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
19585 the vector to match what compare-elim now expects.
19586 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
19587 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
19588 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
19589 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
19590 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
19592 * config/visium/visium.c (single_set_and_flags): Fix where
19593 we look for cc setter after the compare-elim changes.
19594 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
19595 with the vector to match what compare-elim now expects.
19596 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
19597 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
19598 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
19599 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
19600 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
19601 (neg<mode>2_insn_set_overflow): Likewise.
19603 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
19605 PR middle-end/79794
19606 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
19607 maybe_expand_insn call, set ops[0].target. If still set after call,
19608 set alt_rtl. Add extra arg to recursive calls.
19609 (extract_bit_field): Add alt_rtl argument. Pass to
19611 * expmed.h (extract_bit_field): Fix prototype.
19612 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
19613 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
19614 to extract_bit_field_calls.
19615 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
19616 Pass alt_rtl to extract_bit_field calls.
19617 * calls.c (store_unaligned_arguments_into_psuedos)
19618 load_register_parameters): Pass extra NULL to extract_bit_field calls.
19619 * optabs.c (maybe_legitimize_operand): Clear op->target when call
19621 * optabs.h (struct expand_operand): Add target bitfield.
19623 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19625 * compare-elim.c (try_eliminate_compare): Canonicalize
19626 operation with embedded compare to
19627 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
19628 (set (reg) (operation)].
19630 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
19632 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19635 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
19636 cost of adding a carry flag for ADC instruction.
19637 [case MINUS]: Ignore the cost of subtracting a carry flag
19638 for SBB instruction.
19640 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
19642 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
19644 * config/rs6000/bmiintrin.h: New file.
19645 * config/rs6000/bmi2intrin.h: New file.
19646 * config/rs6000/x86intrin.h: New file.
19648 2017-05-12 Jeff Law <law@redhat.com>
19650 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
19653 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
19655 PR middle-end/80707
19656 * tree-cfg.c: Remove cfg edges of unreachable case statements.
19658 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19660 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19661 early expansion of vector divide builtins.
19662 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
19663 builtins identified as having unsigned arguments.
19665 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19667 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
19668 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
19669 expansion of vector logical operations (and, andc, or, xor,
19672 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19674 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
19675 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
19677 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19679 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19680 early GIMPLE expansion of vector multiplies.
19682 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19684 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
19685 TARGET_HAVE_MOVT conditional.
19686 (movt splitter): Likewise.
19688 2017-05-12 Richard Biener <rguenther@suse.de>
19690 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
19691 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19692 Fold all stmts not inplace.
19694 2017-05-12 Richard Biener <rguenther@suse.de>
19696 PR tree-optimization/80713
19697 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
19698 inserted_exprs bit for not removed stmts.
19700 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
19702 PR middle-end/69921
19703 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
19704 parallelized" attribute for parallelized OpenACC kernels.
19705 * omp-offload.c (execute_oacc_device_lower): Use it.
19707 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
19708 Set "oacc kernels" attribute.
19709 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
19710 parameter. Adjust all users.
19711 (oacc_fn_attrib_kernels_p): Remove function.
19712 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
19713 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
19714 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
19715 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
19716 assert "oacc kernels" attribute is set.
19718 2017-05-11 Carl Love <cel@us.ibm.com>
19720 * config/rs6000/rs6000-c: Add support for built-in functions
19721 vector unsigned char vec_popcnt (vector signed char)
19722 vector unsigned char vec_popcnt (vector unsigned char)
19723 vector unsigned short vec_popcnt (vector signed short)
19724 vector unsigned short vec_popcnt (vector unsigned short)
19725 vector unsigned int vec_popcnt (vector signed int)
19726 vector unsigned int vec_popcnt (vector unsigned int)
19727 vector unsigned long long vec_popcnt (vector signed long long)
19728 vector unsigned long long vec_popcnt (vector unsigned long long)
19729 vector signed long long vec_slo (vector signed long long,
19730 vector signed char)
19731 vector signed long long vec_slo (vector signed long long,
19732 vector unsigned char)
19733 vector unsigned long long vec_slo (vector unsigned long long,
19734 vector signed char)
19735 vector unsigned long long vec_slo (vector unsigned long long,
19736 vector unsigned char)
19737 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
19738 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
19739 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
19740 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
19741 * doc/extend.texi: Update the built-in documentation file for the
19742 new built-in functions.
19744 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19746 * attribs.h (sorted_attr_string): Move machine independent
19747 functions for target clone support from the i386 port to common
19748 code. Rename ix86_function_versions to common_function_versions.
19749 Rename make_name to make_unique_name.
19750 (common_function_versions): Likewise.
19751 (make_unique_name): Likewise.
19752 (make_dispatcher_decl): Likewise.
19753 (is_function_default_version): Likewise.
19754 * attribs.c (attr_strcmp): Likewise.
19755 (sorted_attr_string): Likewise.
19756 (common_function_versions): Likewise.
19757 (make_unique_name): Likewise.
19758 (make_dispatcher_decl): Likewise.
19759 (is_function_default_version): Likewise.
19760 * config/i386/i386.c (attr_strcmp): Likewise.
19761 (sorted_attr_string): Likewise.
19762 (ix86_function_versions): Likewise.
19763 (make_name): Likewise.
19764 (make_dispatcher_decl): Likewise.
19765 (is_function_default_version): Likewise.
19766 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
19768 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19771 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19772 Account for direct move costs for vec_construct of integer
19775 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
19778 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
19779 (UNSPEC_STX_ATOMIC): Ditto.
19780 (loaddi_via_sse): New insn.
19781 (storedi_via_sse): Ditto.
19782 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
19783 Update corresponding peephole2 patterns.
19784 (atomic_storedi_fpu): Ditto.
19786 2017-05-11 Julia Koval <julia.koval@intel.com>
19788 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
19789 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
19791 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
19792 (__builtin_ia32_rsqrt14ss_mask): New builtins.
19793 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
19795 2017-05-11 Nathan Sidwell <nathan@acm.org>
19797 * graphite-poly.c: Include dumpfile.h.
19799 * dumpfle.h (dump_function): Declare here ...
19800 * tree-dump.h (dump_function): ... not here.
19801 * dumpfile.c: #include tree-cfg.h.
19802 (dump_function): Move here from ...
19803 * tree-dump.c (dump_function): ... here.
19804 * gimplify.c: #include splay-tree.h, not tree-dump.h.
19805 * graphite-poly.c: Don't include tree-dump.h.
19806 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
19807 * print-tree.c: Likewise.
19808 * stor-layout.c: Likewise.
19809 * tree-nested.c: Likewise.
19811 * dumpfile.c (dump_start): Use TDF_FLAGS.
19812 (dump_enable_all): Fix TDF_KIND check thinko.
19814 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19816 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19817 array entries to represent two legal parameterizations of the
19818 overloaded __builtin_cmpb function, as represented by the
19819 P6_OV_BUILTIN_CMPB constant.
19820 (altivec_resolve_overloaded_builtin): Add special case handling
19821 for the __builtin_cmpb function, as represented by the
19822 P6_OV_BUILTIN_CMPB constant.
19823 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
19824 (BU_P6_64BIT_2): New macro.
19825 (BU_P6_OVERLOAD_2): New macro
19826 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
19827 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
19828 (CMPB): Add overload support to represent both 32-bit and 64-bit
19829 compare-bytes function.
19830 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
19831 support for TARGET_CMPB.
19832 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
19833 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19834 documentation of the __builtin_cmpb overloaded built-in function.
19836 2017-05-11 Richard Biener <rguenther@suse.de>
19838 PR tree-optimization/80705
19839 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
19840 bases are not vectorizable.
19842 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19844 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
19845 when counting register pressure.
19847 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19849 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
19850 (iv_ca_more_deps): Renamed to ...
19851 (iv_ca_compare_deps): ... this.
19852 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
19854 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19856 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
19858 (determine_group_iv_costs): ... here.
19859 (find_inv_vars_cb): Record inv var if it's not recorded before.
19861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19863 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
19864 (get_shiftadd_cost): Ditto.
19866 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19868 * tree-ssa-address.c: Include header file.
19869 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
19871 (add_to_parts): Refactor.
19872 (addr_to_parts): New parameter. Update use of move_hint_to_base.
19873 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
19876 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19878 PR tree-optimization/53090
19879 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
19881 (extract_cond_operands): Detect condition with IV on both sides
19882 and return COMP_IV_EXPR_2.
19883 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
19884 (rewrite_use_compare): Simplify by removing call to function
19885 extract_cond_operands.
19887 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19889 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
19890 (extract_cond_operands): Detect condition comparing against non-
19891 invariant bound and return appropriate enum value.
19892 (find_interesting_uses_cond): Update use of extract_cond_operands.
19893 Handle its return value accordingly.
19894 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
19896 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19898 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
19899 nonlinear iv_use computation in loop invariant sensitive way.
19901 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19903 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
19904 (find_iv_candidates): Call relate_compare_use_with_all_cands.
19906 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19908 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
19909 (dump_cand): Support iv_cand.inv_exprs.
19910 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
19912 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
19915 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19917 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
19919 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
19920 as local function. Include necessary header files.
19921 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
19923 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19925 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
19927 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19929 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
19930 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
19931 RSHIFT_EXPR and BIT_NOT_EXPR.
19933 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19935 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
19936 (adjust_setup_cost): New parameter supporting round up adjustment.
19937 (struct address_cost_data): Delete.
19938 (force_expr_to_var_cost): Don't bound cost with spill_cost.
19939 (split_address_cost, ptr_difference_cost): Delete.
19940 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
19941 (struct ainc_cost_data): New struct.
19942 (get_address_cost_ainc): New function.
19943 (get_address_cost, get_computation_cost): Reimplement.
19944 (determine_group_iv_cost_address): Record inv_expr for all uses of
19946 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
19947 (iv_ca_has_deps): Reimplemented to ...
19948 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
19950 (iv_ca_extend): Call iv_ca_more_deps.
19952 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19954 * tree-ssa-address.c (struct mem_address): Move to header file.
19955 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
19956 * tree-ssa-address.h (struct mem_address): Move from C file.
19957 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
19959 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19961 * tree-affine.h (aff_combination_type): New interface.
19962 (aff_combination_zero_p): Remove static.
19963 (aff_combination_const_p): New interface.
19964 (aff_combination_singleton_var_p): New interfaces.
19966 2017-05-11 Richard Biener <rguenther@suse.de>
19968 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19969 Skip unreachable blocks and destinations.
19970 (eliminate): Move stmt removal and fixup ...
19971 (fini_eliminate): ... here. Skip inserted exprs.
19972 (pass_pre::execute): Move fini_pre after fini_eliminate.
19973 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
19974 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
19975 PRE to get rid of dead code that has invalid SSA form and
19976 split critical edges again.
19978 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19980 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
19982 2017-05-11 Richard Biener <rguenther@suse.de>
19984 * passes.c (execute_function_todo): Verify loops if they are
19985 said to be up-to-date.
19986 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
19987 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
19989 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
19992 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
19993 handle calling assemble_external ourself.
19996 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
19997 modes with zero size. Enhance comment.
19999 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20001 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20002 built-ins for vec_xl and vec_xst with short and char pointer
20005 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20007 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20008 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20009 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20010 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20011 (_mm_maskz_min_round_ss): New intrinsics.
20012 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20013 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20014 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20015 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20016 (__builtin_ia32_minss_mask_round): New builtins.
20017 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20018 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20019 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20021 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20022 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20024 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20027 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20029 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20030 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20031 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20032 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20033 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20034 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20035 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20036 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20037 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20038 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20039 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20040 (__builtin_ia32_mulss_mask_round): New builtins.
20041 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20042 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20043 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20045 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20046 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20048 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20051 2017-05-10 Julia Koval <julia.koval@intel.com>
20053 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20054 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20055 (_mm256_setr_m128i): New intrinsics.
20057 2017-05-10 Julia Koval <julia.koval@intel.com>
20059 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20060 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20061 (_mm_maskz_rcp14_ss): New intrinsics.
20062 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20063 (__builtin_ia32_rcp14ss_mask): New builtins.
20064 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20066 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20068 PR tree-optimization/51513
20069 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20070 (assert_unreachable_fallthru_edge_p): Use it.
20071 (group_case_labels_stmt): Likewise.
20072 * tree-cfg.h: Prototype it.
20073 * stmt.c: Include cfghooks.h and tree-cfg.h.
20074 (emit_case_dispatch_table) <gap_label>: New local variable.
20075 Use it to fill dispatch table gaps.
20076 Test for default_label before updating probabilities.
20077 (expand_case) <default_label>: Remove unneeded initialization.
20078 Test for unreachable default case statement and remove its edge.
20079 Set default_label accordingly.
20080 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
20082 2017-05-10 Carl Love <cel@us.ibm.com>
20084 * config/rs6000/rs6000-c: Add support for built-in functions
20085 vector signed char vec_neg (vector signed char)
20086 vector signed short int vec_neg (vector short int)
20087 vector signed int vec_neg (vector signed int)
20088 vector signed long long vec_neg (vector signed long long)
20089 vector float vec_neg (vector float)
20090 vector double vec_neg (vector double)
20091 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
20093 * config/rs6000/altivec.h: Add define for vec_neg
20094 * doc/extend.texi: Update the built-in documentation for the
20095 new built-in functions.
20097 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20099 PR tree-optimization/77644
20100 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
20102 2017-05-10 Nathan Sidwell <nathan@acm.org>
20104 * dumpfile.h (TDI_lang_all): New.
20105 (TDF_KIND): New. Renumber others
20106 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
20108 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
20110 (get_dump_file_name): Adjust suffix generation.
20111 (dump_enable_all): Use TDF_KIND.
20112 * doc/invoke.texi (-fdump-lang-all): Document.
20114 * dumpfile.h: Tabify.
20116 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
20119 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
20120 Move member access before delete.
20122 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
20124 * tree-inline.c (expand_call_inline): Split block at stmt
20127 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
20130 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
20131 are now unused after splitting mov{sf,sd}_hardfloat.
20132 (f32_lr2): Likewise.
20133 (f32_lm): Likewise.
20134 (f32_lm2): Likewise.
20135 (f32_li): Likewise.
20136 (f32_li2): Likewise.
20137 (f32_lv): Likewise.
20138 (f32_sr): Likewise.
20139 (f32_sr2): Likewise.
20140 (f32_sm): Likewise.
20141 (f32_sm2): Likewise.
20142 (f32_si): Likewise.
20143 (f32_si2): Likewise.
20144 (f32_sv): Likewise.
20145 (f32_dm): Likewise.
20146 (f32_vsx): Likewise.
20147 (f32_av): Likewise.
20148 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
20149 For movsf, order stores so the VSX stores occur before the GPR
20150 store which encourages the register allocator to use a traditional
20151 FPR instead of a GPR. For movsd, order the stores so that the GPR
20152 store comes before the VSX stores to allow the power6 to work.
20153 This is due to the power6 not having a 32-bit integer store
20154 instruction from a FPR.
20155 (movsf_hardfloat): Likewise.
20156 (movsd_hardfloat): Likewise.
20158 2017-05-09 Martin Sebor <msebor@redhat.com>
20160 PR translation/80280
20161 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
20164 PR translation/80280
20165 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
20166 data member added in r247778.
20167 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
20169 2017-05-09 Nathan Sidwell <nathan@acm.org>
20171 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
20173 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
20176 2017-05-09 Marek Polacek <polacek@redhat.com>
20178 * doc/invoke.texi: Fix typo.
20180 2017-05-09 Richard Biener <rguenther@suse.de>
20182 * tree-vrp.c (vrp_val_is_max): Adjust comment.
20183 (vrp_val_is_min): Likewise.
20184 (set_value_range_to_value): Likewise.
20185 (set_value_range_to_nonnegative): Likewise.
20186 (gimple_assign_nonzero_p): Likewise.
20187 (gimple_stmt_nonzero_p): Likewise.
20188 (vrp_int_const_binop): Likewise. Remove unreachable case.
20189 (adjust_range_with_scev): Adjust comments.
20190 (compare_range_with_value): Likewise.
20191 (extract_range_from_phi_node): Likewise.
20192 (test_for_singularity): Likewise.
20194 2017-05-09 Richard Biener <rguenther@suse.de>
20196 * tree-vrp.c (get_single_symbol): Add assert that we don't
20197 get overflowed constants as invariant part.
20198 (compare_values_warnv): Add comment before the TREE_NO_WARNING
20199 checks. Use wi::cmp instead of recursing for integer constants.
20200 (compare_values): Just ignore whether we assumed undefined
20201 overflow instead of failing the compare.
20202 (extract_range_for_var_from_comparison_expr): Add comment before the
20203 TREE_NO_WARNING sets.
20204 (test_for_singularity): Likewise.
20205 (extract_range_from_comparison): Do not disable optimization
20206 when we assumed undefined overflow.
20207 (extract_range_basic): Remove init of unused var.
20209 2017-05-09 Richard Biener <rguenther@suse.de>
20211 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
20212 (extract_range_from_multiplicative_op_1): Adjust.
20213 (extract_range_from_binary_expr_1): Use int_const_binop.
20215 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20218 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
20219 rs6000_store_data_bypass_p in seven define_bypass directives and
20220 in several comments.
20221 * config/rs6000/rs6000-protos.h: Add prototype for
20222 rs6000_store_data_bypass_p function.
20223 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
20224 function implements slightly different (rs6000-specific) semantics
20225 than store_data_bypass_p, returning false rather than aborting
20226 with assertion error when arguments do not satisfy the
20227 requirements of store data bypass.
20228 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
20229 rs6000_store_data_bypass_p.
20231 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
20233 * config/xtensa/xtensa-protos.h
20234 (xtensa_initial_elimination_offset): New declaration.
20235 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20236 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
20237 macro definition, add case for FRAME_POINTER_REGNUM when
20238 FRAME_GROWS_DOWNWARD.
20239 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
20240 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
20241 xtensa_initial_elimination_offset.
20243 2017-05-08 Nathan Sidwell <nathan@acm.org>
20245 * doc/invoke.texi: Alphabetize -fdump options.
20247 2017-05-08 Martin Sebor <msebor@redhat.com>
20249 PR translation/80280
20250 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
20252 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
20254 * target.def (compute_frame_layout): New optional target hook.
20255 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
20256 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
20257 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
20259 * reload1.c (verify_initial_elim_offsets): Likewise.
20260 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
20261 (use_simple_return_p): Call arm_compute_frame_layout if needed.
20262 (arm_get_frame_offsets): Split up into this ...
20263 (arm_compute_frame_layout): ... and this function.
20265 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
20267 * config/aarch64/constraints.md (Usa): New constraint.
20268 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
20270 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
20272 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
20273 with_multilib_list after it has been checked.
20275 2017-05-08 Richard Biener <rguenther@suse.de>
20277 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
20278 (bitmap_set_subtract_values): Likewise.
20280 2017-05-08 Richard Biener <rguenther@suse.de>
20282 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
20283 (gimple_assign_nonzero): ... this and remove strict_overflow_p
20285 (gimple_stmt_nonzero_warnv_p): Rename to ...
20286 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
20288 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
20289 (extract_range_basic): Adjust, do not disable propagation on
20290 strict overflow sensitive simplification.
20291 (vrp_visit_cond_stmt): Likewise.
20293 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
20295 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
20296 body size unconditionally.
20298 2017-05-07 Jeff Law <law@redhat.com>
20301 2017-05-06 Jeff Law <law@redhat.com>
20302 PR tree-optimization/78496
20303 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20306 PR tree-optimization/78496
20307 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20308 (simplify_stmt_using_ranges): Call it.
20309 (vrp_dom_walker::before_dom_children): Extract equivalences
20310 from an ASSERT_EXPR with an equality comparison against a
20313 2017-05-06 Jeff Law <law@redhat.com>
20315 PR tree-optimization/78496
20316 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20319 PR tree-optimization/78496
20320 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20321 (simplify_stmt_using_ranges): Call it.
20322 (vrp_dom_walker::before_dom_children): Extract equivalences
20323 from an ASSERT_EXPR with an equality comparison against a
20326 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20328 * lra-constraints.c (lra_copy_reg_equiv): New function.
20329 (split_reg): Use it to copy equivalence information from the
20330 original register to the spill register.
20332 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20334 PR rtl-optimization/75964
20335 * simplify-rtx.c (simplify_const_relational_operation): Remove
20336 invalid handling of comparisons of integer ABS.
20338 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
20340 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
20341 initialize to zero.
20342 (init_regs): Remove declaration.
20343 (function_arg_advance_32): Initialize error_p as boolean variable.
20345 2017-05-05 Nathan Sidwell <nathan@acm.org>
20347 * store-motion.c (remove_reachable_equiv_notes): Reformat long
20348 lines. Use for (;;).
20350 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20352 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
20353 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
20354 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
20355 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
20356 VF=2 that require versioning.
20358 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20360 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
20363 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20365 * diagnostic.h (diagnostic_override_option_index): Convert from
20366 macro to inline function.
20368 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20370 * diagnostic.c (last_module_changed_p): New function.
20371 (set_last_module): New function.
20372 (diagnostic_report_current_module): Convert macro usage to
20373 the above functions.
20374 * diagnostic.h (diagnostic_context::last_module): Strengthen
20375 from const line_map * to const line_map_ordinary *.
20376 (diagnostic_last_module_changed): Delete macro.
20377 (diagnostic_set_last_module): Delete macro.
20379 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20381 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
20382 with diagnostic_report_diagnostic.
20383 (diagnostic_n_impl_richloc): Likewise.
20384 * diagnostic.h (report_diagnostic): Delete macro.
20385 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
20386 with diagnostic_report_diagnostic.
20387 * substring-locations.c (format_warning_va): Likewise.
20389 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20391 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
20392 save/restor of format_spec. Move option-printing code to...
20393 (print_option_information): ...this new function, and
20394 reimplement by simply printing to the pretty_printer,
20395 rather than appending to the format string.
20397 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20399 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
20400 handling logic into...
20401 (update_effective_level_from_pragmas): ...this new function.
20403 2017-05-04 Andrew Waterman <andrew@sifive.com>
20405 * config/riscv/riscv.opt (mstrict-align): New option.
20406 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
20407 (SLOW_UNALIGNED_ACCESS): Define.
20408 (riscv_slow_unaligned_access): Declare.
20409 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
20411 (riscv_slow_unaligned_access): New variable.
20412 (rocket_tune_info): Set slow_unaligned_access to true.
20413 (optimize_size_tune_info): Set slow_unaligned_access to false.
20414 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
20415 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
20416 (riscv_option_override): Set riscv_slow_unaligned_access.
20417 * doc/invoke.texi: Add -mstrict-align to RISC-V.
20419 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
20421 * config/riscv/riscv.md: Unify indentation.
20423 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20428 * config/rs6000/rs6000.md (u code attribute): Add FIX and
20430 (extendsi<mode>2): Add support for doing sign extension via
20431 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
20432 don't have ISA 3.0 instructions.
20433 (extendsi<mode>2 splitter): Likewise.
20434 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
20435 generate the normal insns since SImode can now go in vector
20436 registers. Disallow the special UNSPECs needed for previous
20437 machines to hide SImode being used. Add new insns
20438 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
20439 (fix_trunc<mode>si2_stfiwx): Likewise.
20440 (fix_trunc<mode>si2_internal): Likewise.
20441 (fixuns_trunc<mode>si2): Likewise.
20442 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20443 (fctiw<u>z_<mode>_smallint): Likewise.
20444 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
20445 of floating point to 32-bit integer from doing a direct move to
20446 the GPR registers to do a store.
20447 (fctiwz_<mode>): Break long line.
20449 2017-05-05 Bin Cheng <bin.cheng@arm.com>
20451 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
20452 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
20453 (addr_list, addr_offset_valid_p): New.
20454 (split_address_groups): Check offset validity with above function.
20455 (gt-tree-ssa-loop-ivopts.h): Include header file.
20457 2017-05-05 Nathan Sidwell <nathan@acm.org>
20459 * config.gcc (arm*-*-*): Add missing 'fi'.
20461 2017-05-05 Steve Ellcey <sellcey@cavium.com>
20463 * doc/invoke.texi (-fopt-info): Explicitly say order of options
20464 included in -fopt-info does not matter.
20465 * doc/optinfo.texi (-fopt-info): Fix description of default
20466 behavour. Explicitly say order of options included in -fopt-info
20469 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20471 * config.gcc: Allow combinations of aprofile and rmprofile values for
20472 --with-multilib-list.
20473 * config/arm/t-multilib: New file.
20474 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
20475 variables. Remove setting of ISA and floating-point ABI in
20476 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
20477 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
20478 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
20479 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
20481 * config/arm/t-rmprofile: Likewise except for the matches changes.
20482 * doc/install.texi (--with-multilib-list): Document the combination of
20483 aprofile and rmprofile values and warn about pitfalls in doing that.
20485 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20487 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
20488 (movdi_aarch64): Likewise.
20490 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20492 PR tree-optimization/80632
20493 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
20495 (build_arrays): Initialize it for virtual phis.
20496 (fix_phi_nodes): Use it for virtual phis.
20498 PR tree-optimization/80558
20499 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
20500 [x, y] op z into [x op, y op z] for op & or | if conditions
20503 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
20504 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20507 * config/arm/arm.md (use_literal_pool): Remove.
20508 (64-bit immediate split): No longer takes cost into consideration
20509 if arm_disable_literal_pool is enabled.
20510 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
20511 used when arm_disable_literal_pool is enabled.
20512 (arm_max_const_double_inline_cost): Remove use of
20513 arm_disable_literal_pool.
20514 (push_minipool_fix): Add assert.
20515 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
20516 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
20517 (no_literal_pool_sf_immediate): New.
20519 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20521 PR tree-optimization/80613
20522 * tree-ssa-dce.c (propagate_necessity): Remove cases for
20523 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
20525 2017-05-05 Richard Biener <rguenther@suse.de>
20527 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
20529 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
20531 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
20532 of this flag from insn conditions due to removal from r247495.
20534 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20536 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
20538 (arm_early_store_addr_dep_ptr): Likewise.
20539 * config/arm/aarch-common-protos.h
20540 (arm_early_load_addr_dep_ptr): Add prototype.
20541 (arm_early_store_addr_dep_ptr): Likewise.
20542 * config/arm/cortex-a53.md: Add new bypasses.
20544 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20546 * tree.c (next_type_uid): Change type to unsigned.
20547 (type_hash_canon): Decrement back next_type_uid if
20548 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
20549 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
20552 2017-05-04 Martin Sebor <msebor@redhat.com>
20554 * builtins.c: Fix a trivial typo in a comment.
20556 PR middle-end/79234
20557 * builtins.c (check_sizes): Adjust to handle reading past the end.
20558 Avoid printing excessive upper bound of ranges. Use %E to print
20559 tree nodes instead of converting them to %wu.
20560 (expand_builtin_memchr): New function.
20561 (compute_dest_size): Rename...
20562 (compute_objsize): ...to this.
20563 (expand_builtin_memcpy): Adjust.
20564 (expand_builtin_mempcpy): Adjust.
20565 (expand_builtin_strcat): Adjust.
20566 (expand_builtin_strcpy): Adjust.
20567 (check_strncat_sizes): Adjust.
20568 (expand_builtin_strncat): Adjust.
20569 (expand_builtin_strncpy): Adjust and simplify.
20570 (expand_builtin_memset): Adjust.
20571 (expand_builtin_bzero): Adjust.
20572 (expand_builtin_memcmp): Adjust.
20573 (expand_builtin): Handle memcmp.
20574 (maybe_emit_chk_warning): Check strncat just once.
20576 2017-05-04 Martin Sebor <msebor@redhat.com>
20578 PR preprocessor/79214
20579 PR middle-end/79222
20580 PR middle-end/79223
20581 * builtins.c (check_sizes): Add inlining context and issue
20582 warnings even when -Wno-system-headers is set.
20583 (check_strncat_sizes): Same.
20584 (expand_builtin_strncat): Same.
20585 (expand_builtin_memmove): New function.
20586 (expand_builtin_stpncpy): Same.
20587 (expand_builtin): Handle memmove and stpncpy.
20589 2017-05-04 Bin Cheng <bin.cheng@arm.com>
20591 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
20592 which is not used any more.
20594 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20596 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
20598 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20600 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
20601 (cortexa53_tunings): Likewise.
20602 (cortexa57_tunings): Likewise.
20603 (cortexa72_tunings): Likewise.
20604 (cortexa73_tunings): Likewise.
20606 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20608 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
20609 Set loop alignment to 8.
20611 2017-05-04 Martin Sebor <msebor@redhat.com>
20613 PR translation/80280
20614 * builtins.c (expand_builtin_object_size): Add missing quoting to
20615 %D and like directives.
20616 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
20617 (hsa_type_for_tree_type): Same.
20618 (verify_function_arguments): Same.
20619 * symtab.c (symbol_table::change_decl_assembler_name): Same.
20620 * varasm.c (get_section): Same.
20623 2017-05-04 Martin Sebor <msebor@redhat.com>
20625 PR translation/80280
20626 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
20628 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20630 * config/aarch64/aarch64.c (generic_addrcost_table):
20631 Change HI/TI mode setting.
20633 2017-05-04 Martin Jambor <mjambor@suse.cz>
20635 PR tree-optimization/80622
20636 * tree-sra.c (comes_initialized_p): New function.
20637 (build_accesses_from_assign): Only set write lazily when
20638 comes_initialized_p is false.
20639 (analyze_access_subtree): Use comes_initialized_p.
20640 (propagate_subaccesses_across_link): Assert !comes_initialized_p
20641 instead of testing for PARM_DECL.
20643 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20645 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
20646 constraint on operand 0 to allow more general addressing modes.
20647 Adjust output template.
20648 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
20650 * config/aarch64/aarch64-protos.h
20651 (aarch64_address_valid_for_prefetch_p): Declare prototype.
20652 * config/aarch64/constraints.md (Dp): New address constraint.
20653 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
20656 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
20658 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
20659 update use of estimate_ipcp_clone_size_and_time.
20660 (estimate_local_effects): Update use of
20661 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
20662 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
20663 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
20664 Return nonspecialized time.
20666 2017-05-04 Richard Biener <rguenther@suse.de>
20668 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
20669 for the last VUSE which def dominates the PHI. Directly call
20671 (get_continuation_for_phi_1): Remove.
20673 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
20675 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
20676 to explain the use of truncating division. Cap the number of
20677 iterations to the maximum given by nb_iterations_upper_bound,
20680 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20682 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
20683 * configure: Regenerate.
20684 * config.in: Regenerate.
20685 * config/i386/driver-mingw32.c: new file.
20686 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
20687 * config.host: Link driver-mingw32.o on MinGW host.
20688 * doc/install.texi: Document new --enable-mingw-wildcard configure
20691 2017-05-04 Marek Polacek <polacek@redhat.com>
20693 PR tree-optimization/80612
20694 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
20696 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20697 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
20699 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
20700 (movt splitter): Likewise.
20701 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
20702 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
20703 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
20704 block for Thumb-1 with MOVT.
20705 (thumb2_legitimate_address_p): Move code block ...
20706 (can_avoid_literal_pool_for_label_p): ... into this new function.
20707 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
20709 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
20710 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
20711 "M-profile targets with the MOVT instruction".
20713 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20715 * config/arm/arm-builtins.c (arm_init_builtins): Rename
20716 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
20717 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
20719 2017-05-04 Martin Liska <mliska@suse.cz>
20721 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
20722 variable cond_code.
20724 2017-05-04 Richard Biener <rguenther@suse.de>
20726 * tree.c (array_at_struct_end_p): Handle arrays at struct
20727 end with flexarrays more conservatively. Refactor and treat
20728 arrays of arrays or aggregates more strict. Fix
20729 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
20730 * tree.c (array_at_struct_end_p): Adjust prototype.
20731 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
20732 * gimple-fold.c (get_range_strlen): Likewise.
20733 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
20735 2017-05-04 Richard Biener <rguenther@suse.de>
20737 PR tree-optimization/31130
20738 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
20740 (supports_overflow_infinity): Likewise.
20741 (is_negative_overflow_infinity): Likewise.
20742 (is_positive_overflow_infinity): Likewise.
20743 (is_overflow_infinity): Likewise.
20744 (stmt_overflow_infinity): Likewise.
20745 (overflow_infinity_range_p): Likewise.
20746 (usable_range_p): Remove as always returning true.
20747 (make_overflow_infinity): Remove.
20748 (negative_overflow_infinity): Likewise.
20749 (positive_overflow_infinity): Likewise.
20750 (avoid_overflow_infinity): Likewise.
20751 (set_value_range): Adjust accordingly.
20752 (set_value_range_to_nonnegative): Likewise, remove now unused
20753 overflow_infinity arg.
20754 (vrp_operand_equal_p): Adjust.
20755 (update_value_range): Likewise.
20756 (range_int_cst_singleton_p): Likewise.
20757 (operand_less_p): Likewise.
20758 (compare_values_warnv): Likewise.
20759 (extract_range_for_var_from_comparison_expr): Likewise.
20760 (vrp_int_const_binop): Likewise.
20761 (zero_nonzero_bits_from_vr): Likewise.
20762 (extract_range_from_multiplicative_op_1): Likewise.
20763 (extract_range_from_binary_expr_1): Likewise.
20764 (extract_range_from_unary_expr): Likewise.
20765 (extract_range_from_comparison): Likewise.
20766 (extract_range_basic): Likewise.
20767 (adjust_range_with_scev): Likewise.
20768 (compare_ranges): Likewise.
20769 (compare_range_with_value): Likewise.
20770 (dump_value_range): Likewise.
20771 (test_for_singularity): Likewise, remove strict_overflow_p parameter
20773 (simplify_cond_using_ranges): Adjust.
20775 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20777 * brig-builtins.def: Added a builtin for class_f64.
20778 * builtin-types.def: Added a builtin type needed by class_f64.
20780 2017-05-03 Jason Merrill <jason@redhat.com>
20782 * timevar.def: Add TV_CONSTEXPR.
20784 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20786 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
20788 2017-05-03 Martin Jambor <mjambor@suse.cz>
20790 * ipa-prop.c (ipa_update_after_lto_read): Removed.
20791 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
20792 * ipa-cp.c (ipcp_propagate_stage): Do not call
20793 ipa_update_after_lto_read.
20794 * ipa-inline.c (ipa_inline): Likewise.
20796 2017-05-03 Martin Jambor <mjambor@suse.cz>
20798 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
20799 tag. Added a default constructor and a destructor.
20800 (ipa_edge_args_sum_t): New class;
20801 (ipa_edge_args_sum): Declare.
20802 (ipa_edge_args_vector): Remove declaration.
20803 (IPA_EDGE_REF): Use ipa_edge_args_sum.
20804 (ipa_free_edge_args_substructures): Remove declaration.
20805 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
20806 (ipa_edge_args_info_available_for_edge_p): Likewise.
20807 * ipa-prop.c (ipa_edge_args_vector): Removed.
20808 (edge_removal_hook_holder): Likewise.
20809 (edge_duplication_hook_holder): Likewise.
20810 (ipa_edge_args_sum): New variable.
20811 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
20812 ipa_edge_args_vector.
20813 (ipa_free_edge_args_substructures): Likewise.
20814 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
20815 ipa_edge_args_vector.
20816 (ipa_edge_removal_hook): Turned into method
20817 ipa_edge_args_sum_t::remove.
20818 (ipa_edge_duplication_hook): Turned into method
20819 ipa_edge_args_sum_t::duplicate.
20820 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
20821 registering edge hooks.
20822 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
20823 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
20824 ipa_edge_args_sum instead of ipa_edge_args_vector.
20825 * ipa-profile.c (ipa_profile): Likewise.
20827 2017-05-03 Martin Jambor <mjambor@suse.cz>
20829 * symbol-summary.h (function_summary): New method exists.
20830 (function_summary::symtab_removal): Deallocate through release.
20831 (call_summary): New class.
20832 (gt_ggc_mx): New overload.
20833 (gt_pch_nx): Likewise.
20834 (gt_pch_nx): Likewise.
20836 2017-05-03 Jeff Law <law@redhat.com>
20838 PR tree-optimization/78496
20839 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
20840 from simplify_cond_using_ranges. Split off code to walk
20841 backwards through casts into ...
20842 (simplify_cond_using_ranges_2): New function.
20843 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
20844 (execute_vrp): After identifying jump threads, call
20845 simplify_cond_using_ranges_2.
20847 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
20850 * ipa-inline.h (inline_summary): Add ctor.
20851 (create_ggc): Do not use ggc_cleared_alloc.
20853 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
20854 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20856 * gcc.c (handle_braces): Support escaping in switch matching
20858 * doc/invoke.texi (Spec Files): Document it.
20859 Remove superfluous @code markup in items.
20861 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20863 * diagnostic-show-locus.c (struct column_range): New struct.
20864 (get_affected_columns): New function.
20865 (get_printed_columns): New function.
20866 (struct correction): New struct.
20867 (correction::ensure_capacity): New function.
20868 (correction::ensure_terminated): New function.
20869 (struct line_corrections): New struct.
20870 (line_corrections::~line_corrections): New dtor.
20871 (line_corrections::add_hint): New function.
20872 (layout::print_trailing_fixits): Reimplement in terms of the new
20874 (selftest::test_overlapped_fixit_printing): New function.
20875 (selftest::diagnostic_show_locus_c_tests): Call it.
20877 2017-05-03 Nathan Sidwell <nathan@acm.org>
20879 Canonicalize canonical type hashing
20880 * tree.h (type_hash_canon_hash): Declare.
20881 * tree.c (type_hash_list, attribute_hash_list): Move into
20882 type_hash_canon_hash.
20883 (build_type_attribute_qual_variant): Break out hash code calc into
20884 type_hash_canon_hash.
20885 (type_hash_canon_hash): New. Generic type hash computation.
20886 (build_range_type_1, build_array_type_1, build_function_type,
20887 build_method_type_directly, build_offset_type, build_complex_type,
20888 make_vector_type): Call it.
20890 2017-05-03 Richard Biener <rguenther@suse.de>
20892 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20893 When all DRs have unknown misaligned do not always peel
20894 when there is a store but apply the same costing model as if
20895 there were only loads.
20897 2017-05-03 Richard Biener <rguenther@suse.de>
20900 PR tree-optimization/80492
20901 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20902 compare_base_decls returning dont-know properly.
20904 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
20906 * config/arm/iterators.md (CCSI): New mode iterator.
20907 (arch): New mode attribute.
20908 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
20909 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
20910 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
20911 code iterator for success result mode.
20912 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
20913 the corresponding new insn generators.
20915 2017-05-03 Bin Cheng <bin.cheng@arm.com>
20918 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20919 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20921 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
20923 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
20924 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
20925 (DDR_A): Wrap DDR argument in brackets.
20926 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
20927 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
20928 (DDR_REVERSED_P): Likewise.
20930 2017-05-03 Jakub Jelinek <jakub@redhat.com>
20932 PR tree-optimization/79472
20933 * tree-switch-conversion.c (struct switch_conv_info): Add
20934 contiguous_range and default_case_nonstandard fields.
20935 (collect_switch_conv_info): Compute contiguous_range and
20936 default_case_nonstandard fields, don't clear final_bb if
20937 contiguous_range and only the default case doesn't have the required
20939 (check_all_empty_except_final): Set default_case_nonstandard instead
20940 of failing if contiguous_range and the default case doesn't have empty
20942 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
20943 and only the default case doesn't have the required constants. Skip
20945 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
20946 if default_case_nonstandard.
20947 (build_constructors): Build constant 1 just once. Assert that default
20948 values aren't inserted in between cases if contiguous_range. Skip
20950 (build_arrays): Skip virtual phis.
20951 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
20952 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
20953 Handle virtual phis.
20954 (gen_inbound_check): Handle default_case_nonstandard case.
20955 (process_switch): Adjust check_final_bb caller. Call
20956 gather_default_values with the first non-default case instead of
20957 default case if default_case_nonstandard.
20959 2017-05-02 Nathan Sidwell <nathan@acm.org>
20961 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
20962 check. Fix formatting.
20964 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
20966 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
20967 errors when comparing specialized and unspecialized times.
20969 2017-05-02 David Malcolm <dmalcolm@redhat.com>
20971 * diagnostic-show-locus.c
20972 (layout::should_print_annotation_line_p): Make private.
20973 (layout::print_annotation_line): Make private.
20974 (layout::annotation_line_showed_range_p): Make private.
20975 (layout::show_ruler): Make private.
20976 (layout::print_source_line): Make private. Pass in line and
20977 line_width, rather than calling location_get_source_line. Drop
20979 (layout::print_leading_fixits): New method.
20980 (layout::print_any_fixits): Rename to...
20981 (layout::print_trailing_fixits): ...this, and make private.
20982 Don't print newline fixits.
20983 (diagnostic_show_locus): Move logic for printing one row into...
20984 (layout::print_line): ...this new function. Move the
20985 location_get_source_line call and error-handling from
20986 print_source_line to here. Call print_leading_fixits, and rename
20987 print_any_fixits to print_trailing_fixits.
20988 (selftest::test_fixit_insert_containing_newline): Update now that
20989 newlines are partially supported.
20990 (selftest::test_fixit_insert_containing_newline_2): New test.
20991 (selftest::test_fixit_replace_containing_newline): Update comments.
20992 (selftest::diagnostic_show_locus_c_tests): Call the new test.
20993 * edit-context.c (class added_line): New class.
20994 (class edited_line): Describe newline handling in comment.
20995 (edited_line::actually_edited_p): New method.
20996 (edited_line::print_content): Delete redundant decl.
20997 (edited_line::m_predecessors): New field.
20998 (edited_file::print_content): Call edited_line::print_content.
20999 (edited_file::print_diff): Update to support newlines.
21000 (edited_file::print_diff_hunk): Likewise.
21001 (edited_file::print_run_of_changed_lines): New function.
21002 (edited_file::print_diff_line): Convert to...
21003 (print_diff_line): ...this.
21004 (edited_file::get_effective_line_count): New function.
21005 (edited_line::edited_line): Initialize new field m_predecessors.
21006 (edited_line::~edited_line): Clean up m_predecessors.
21007 (edited_line::apply_fixit): Handle newlines.
21008 (edited_line::get_effective_line_count): New function.
21009 (edited_line::print_content): New function.
21010 (edited_line::print_diff_lines): New function.
21011 (selftest::test_applying_fixits_insert_containing_newline): New
21013 (selftest::test_applying_fixits_replace_containing_newline): New
21015 (selftest::insert_line): New function.
21016 (selftest::test_applying_fixits_multiple_lines): Add example of
21018 (selftest::edit_context_c_tests): Call the new tests.
21020 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21022 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21023 parameter cand. Update dump information.
21024 (get_computation_cost): Update uses.
21026 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21028 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21029 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21030 (get_computation_at, rewrite_use_address): Update use of
21031 get_computation_aff.
21033 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21035 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21036 (get_computation): Delete.
21037 (get_computation_cost): Implement like get_computation_cost_at.
21038 Use get_computation_at.
21039 (get_computation_cost_at): Delete.
21040 (rewrite_use_nonlinear_expr): Use get_computation_at.
21041 (rewrite_use_compare, remove_unused_ivs): Ditto.
21043 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21045 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21047 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21049 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21050 (ivopts_global_cost_for_size): Rename parameter and update uses.
21051 (iv_ca_recount_cost): Update uses.
21052 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21053 candidates seperately in n_invs and n_cands.
21054 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21056 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21058 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21059 (find_inv_vars_cb): New.
21060 (find_depends): Renamed to ...
21061 (find_inv_vars): ... this.
21062 (add_candidate_1, force_var_cost): Call find_inv_vars.
21063 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21065 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21067 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21068 inv_vars. Add inv_exprs.
21069 (struct iv_cand): Rename depends_on to inv_vars.
21070 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21071 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21072 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21073 (dump_cand): Dump inv_vars.
21074 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21075 (record_invariant, find_depends, add_candidate_1): Ditto.
21076 (set_group_iv_cost, force_var_cost): Ditto.
21077 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21078 (get_computation_cost_at, get_computation_cost): Ditto.
21079 (determine_group_iv_cost_generic): Ditto.
21080 (determine_group_iv_cost_address): Ditto.
21081 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
21082 (determine_group_iv_costs): Ditto.
21083 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
21084 (iv_ca_set_remove_invariants): Renamed to ...
21085 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
21086 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
21087 (iv_ca_set_add_invariants): Renamed to ...
21088 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
21089 (iv_ca_set_cp): Use iv_ca_set_add_invs.
21090 (iv_ca_has_deps): Support inv_vars and inv_exprs.
21091 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
21092 (create_new_ivs): Remove useless dump.
21094 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21096 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
21098 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
21099 (iv_ca_set_no_cp, create_new_iv): Ditto.
21101 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21103 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21105 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21107 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
21108 function tree_check2.
21110 2017-05-02 Martin Liska <mliska@suse.cz>
21112 * doc/gcov.texi: Add missing preposition.
21113 * gcov.c (function_info::function_info): Properly fill up
21114 all member variables.
21116 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21118 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
21120 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21122 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
21124 2017-05-02 Martin Liska <mliska@suse.cz>
21127 * lto-streamer-in.c (lto_read_tree_1): Remove
21128 LTO_STREAMER_DEBUG.
21129 * lto-streamer.c (struct tree_hash_entry): Likewise.
21130 (struct tree_entry_hasher): Likewise.
21131 (tree_entry_hasher::hash): Likewise.
21132 (tree_entry_hasher::equal): Likewise.
21133 (lto_streamer_init): Likewise.
21134 (lto_orig_address_map): Likewise.
21135 (lto_orig_address_get): Likewise.
21136 (lto_orig_address_remove): Likewise.
21137 * lto-streamer.h: Likewise.
21138 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
21139 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
21141 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
21143 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
21144 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
21145 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
21146 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
21147 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
21148 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
21149 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
21150 (mm_maskz_sub_ss): New intrinsics.
21151 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21152 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21153 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
21154 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
21155 (__builtin_ia32_subss_mask_round): New builtins.
21156 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21157 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21158 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
21160 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
21161 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21163 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21166 2017-05-02 Martin Jambor <mjambor@suse.cz>
21168 PR tree-optimization/78687
21169 * tree-sra.c (access): New field parent.
21170 (process_subtree_disqualification): New function.
21171 (disqualify_candidate): Call it.
21172 (build_accesses_from_assign): Reset write flag if creating an
21174 (build_access_subtree): Fill in parent field and also prpagate
21175 down grp_write flag.
21176 (create_artificial_child_access): New parameter set_grp_write, set
21177 grp_write to its value.
21178 (propagate_subaccesses_across_link): Also propagate grp_write flag
21180 (propagate_all_subaccesses): Push the closest parent back to work
21181 queue if add_access_to_work_queue returned true.
21183 2017-05-02 Richard Biener <rguenther@suse.de>
21185 * common.opt (fstrict-overflow): Alias negative to fwrapv.
21186 * doc/invoke.texi (fstrict-overflow): Remove all traces of
21187 -fstrict-overflow documentation.
21188 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
21189 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
21190 flag_strict_overflow.
21191 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
21192 * lto-opts.c (lto_write_options): Do not stream it.
21193 * lto-wrapper.c (merge_and_complain): Do not handle it.
21194 * opts.c (default_options_table): Do not set -fstrict-overflow.
21195 (finish_options): Likewise do not clear it when sanitizing.
21196 * simplify-rtx.c (simplify_const_relational_operation): Do not
21197 test flag_strict_overflow.
21199 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21201 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
21202 using enabled attribute.
21203 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
21204 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
21205 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
21206 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
21207 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
21208 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
21209 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
21210 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
21211 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
21212 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
21214 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21216 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
21218 2017-05-02 Richard Biener <rguenther@suse.de>
21220 PR tree-optimization/80591
21222 2017-04-10 Richard Biener <rguenther@suse.de>
21224 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21227 2017-05-02 Richard Biener <rguenther@suse.de>
21229 PR tree-optimization/80549
21230 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
21231 (cleanup_tree_cfg_noloop): Create forwarders to known loop
21232 headers if they do not have a preheader.
21234 2017-05-02 Martin Liska <mliska@suse.cz>
21237 * common.opt: Fix typo.
21238 * doc/invoke.texi: Likewise.
21240 2017-05-01 Jan Beulich <jbeulich@suse.com>
21242 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
21243 swapping, add (x,x,m,x,n) alternative.
21245 2017-05-01 Nathan Sidwell <nathan@acm.org>
21247 * calls.c (combine_pending_stack_adjustment_and_call): Remove
21248 unnecessary unadjusted_alignment check.
21250 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21253 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
21255 * gimplify.c (gimplify_cilk_detach): New function.
21256 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
21257 * tree-core.h: Document EXPR_CILK_SPAWN.
21258 * tree.h (EXPR_CILK_SPAWN): Define.
21260 2017-05-01 David Malcolm <dmalcolm@redhat.com>
21262 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
21263 to use new fixit_hint representation, using the "replace" logic.
21264 (get_line_span_for_fixit_hint): Likewise.
21265 (layout::print_any_fixits): Likewise.
21266 (selftest::test_one_liner_many_fixits): Rename to...
21267 (selftest::test_one_liner_many_fixits_1): ...this, and update
21268 comment and expected output to reflect that the multiple fix-it
21269 hints are now consolidated into one insertion.
21270 (selftest::test_one_liner_many_fixits_2): New test.
21271 (selftest::test_diagnostic_show_locus_one_liner): Update for
21273 (selftest::test_fixit_consolidation): Update for fix-it API
21275 * diagnostic.c (print_parseable_fixits): Likewise.
21276 * edit-context.c (edited_line::m_line_events): Convert from
21277 auto_vec <line_event *> to auto_vec <line_event>.
21278 (class line_event): Convert from abstract base class to a concrete
21279 class, taking over the role of replace_event.
21280 (class insert_event): Delete.
21281 (class replace_event): Rename to class line_event. Convert to
21283 (edit_context::add_fixits): Reimplement.
21284 (edit_context::apply_insert): Delete.
21285 (edit_context::apply_replace): Rename to...
21286 (edit_context::apply_fixit): ...this. Convert to half-open range.
21287 (edited_file::apply_insert): Delete.
21288 (edited_file::apply_replace): Rename to...
21289 (edited_file::apply_fixit): ...this.
21290 (edited_line::~edited_line): Drop deletion of events.
21291 (edited_line::apply_insert): Delete.
21292 (edited_line::apply_replace): Rename to...
21293 (edited_line::apply_fixit): ...this. Convert to half-open range.
21294 Update for change to type of m_line_events.
21295 * edit-context.h (edit_context::apply_insert): Delete.
21296 (edit_context::apply_replace): Rename to...
21297 (edit_context::apply_fixit): ...this.
21299 2017-05-01 Martin Sebor <msebor@redhat.com>
21301 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
21304 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
21307 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
21308 __get_cpuid_max returns 0.
21309 (__get_cpuid_count): Ditto.
21311 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
21313 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
21314 replacement expression is another instance of one of its arguments.
21316 2017-05-01 Jakub Jelinek <jakub@redhat.com>
21319 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
21320 check for stack push/pop autoinc.
21321 * config/i386/i386.c (ix86_agi_dependent): Return false
21322 if the only reason why modified_in_p returned true is that
21323 addr is SP based and set_insn is a push or pop.
21325 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21327 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
21330 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21333 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
21334 (account_size_time): Use two predicates - exec_pred and
21336 (evaluate_conditions_for_known_args): Compute both clause and
21338 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
21339 (inline_summary_t::duplicate): Update.
21340 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
21342 (compute_inline_parameters): Likewise.
21343 (estimate_edge_size_and_time): Update caluclation of time.
21344 (estimate_node_size_and_time): Compute both time and nonspecialized
21346 (estimate_ipcp_clone_size_and_time): Update.
21347 (inline_merge_summary): Update.
21348 (do_estimate_edge_time): Update.
21349 (do_estimate_edge_size): Update.
21350 (do_estimate_edge_hints): Update.
21351 (inline_read_section, inline_write_summary): Stream both new predicates.
21352 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
21354 (compute_inlined_call_time): Cleanup.
21355 (big_speedup_p): Update.
21356 (edge_badness): Update.
21357 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
21358 (size_time_entry): Replace predicate by exec_predicate and
21359 nonconst_predicate.
21360 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
21361 (estimate_edge_time): Return also nonspec_time.
21362 (reset_edge_growth_cache): Update.
21364 2017-04-29 Jakub Jelinek <jakub@redhat.com>
21366 PR rtl-optimization/80491
21367 * ifcvt.c (noce_process_if_block): When looking for x setter
21368 with missing else_bb, don't check only the insn right before
21369 cond_earliest, but look for the last insn that x is modified in
21370 within the same bb.
21372 PR rtl-optimization/80491
21373 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
21375 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
21377 PR tree-optimization/80487
21378 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
21380 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21382 PR tree-optimization/79697
21383 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
21384 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
21385 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
21387 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
21388 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
21390 2017-04-28 Martin Sebor <msebor@redhat.com>
21392 PR tree-optimization/80523
21393 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
21394 (init_target_to_host_charmap, target_to_host, target_strtol10): New
21396 (maybe_warn, format_directive, parse_directive): Use new functions.
21397 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
21399 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
21401 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
21403 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
21405 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
21406 target_header_dir): Set correctly.
21407 * configure: Regenerated.
21408 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21409 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21410 instead of SYSTEM_HEADER_DIR.
21412 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
21414 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
21415 (estimate_local_effects): Likewise.
21416 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
21417 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
21418 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
21419 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
21420 do_estimate_edge_time, estimate_edge_time): Likewise.
21421 * ipa-inline-analysis.c (estimate_node_size_and_time,
21422 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
21423 (estimate_time_after_inlining): Remove.
21425 2017-04-28 Martin Liska <mliska@suse.cz>
21427 * doc/gcov.texi: Enhance documentation of gcov.
21429 2017-04-28 Martin Liska <mliska@suse.cz>
21431 * doc/gcov.texi: Sort options in alphabetic order.
21432 * doc/gcov-dump.texi: Likewise.
21433 * doc/gcov-tool.texi: Likewise.
21434 * gcov.c (print_usage): Likewise.
21435 * gcov-dump.c (print_usage): Likewise.
21436 * gcov-tool.c (print_merge_usage_message): Likewise.
21437 (print_rewrite_usage_message): Likewise.
21438 (print_overlap_usage_message): Likewise.
21440 2017-04-28 Martin Liska <mliska@suse.cz>
21442 PR gcov-profile/53915
21443 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
21445 2017-04-28 Martin Liska <mliska@suse.cz>
21447 PR gcov-profile/79891
21448 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
21449 is marked by compiler as living on a line.
21450 (get_cycles_count): Remove usage of the union.
21451 (output_intermediate_file): Likewise.
21452 (find_source): Fix GNU coding style.
21453 (accumulate_line_counts): Remove old non-all block mode.
21454 (output_lines): Remove usage of the union.
21455 * profile.c (output_location): Include all BBs, even if
21456 belonging to a same line (and file) as a previous BB.
21458 2017-04-28 Martin Liska <mliska@suse.cz>
21460 * gcov.c (process_args): Handle new argument 'w'.
21461 (read_graph_file): Assign ID to BBs.
21462 (output_branch_count): Display BB # if verbose flag is set.
21463 (output_lines): Likewise for arcs.
21464 (print_usage): Add '--verbose' option help.
21465 * doc/gcov.texi: Document --verbose (-w) option.
21467 2017-04-28 Martin Liska <mliska@suse.cz>
21469 * gcov.c (struct block_location_info): New struct.
21470 (process_file): Fill up the new structure.
21471 (read_graph_file): Replace usage of encoding by the newly added
21473 (add_line_counts): Likewise.
21474 (accumulate_line_counts): Remove usage of the union.
21475 (function_info::function_info): New function.
21476 (function_info::~function_info): Likewise.
21477 (process_file): Call delete instead of release_function.
21478 (release_function): Release the function.
21479 (release_structures): Call delete instead of release_function.
21480 (solve_flow_graph): Replace usage of num_blocks.
21481 (find_exception_blocks): Likewise.
21482 (output_lines): Fix GNU coding style.
21484 2017-04-28 Martin Liska <mliska@suse.cz>
21487 * coverage.c (coverage_remove_note_file): New function.
21488 * coverage.h: Declare the function.
21489 * toplev.c (finalize): Clean if an error has been seen.
21491 2017-04-28 Martin Liska <mliska@suse.cz>
21493 PR gcov-profile/80031
21494 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
21495 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
21496 * gcov.c (read_graph_file): Read just number of blocks.
21497 * profile.c (branch_prob): Do not stream 0 flags per a basic
21500 2017-04-28 Martin Liska <mliska@suse.cz>
21502 * gcov-dump.c (tag_*): Add new argument to declarations.
21503 (dump_gcov_file): Likewise.
21504 (tag_blocks): Add and use new argument depth.
21505 (tag_arcs): Likewise.
21506 (tag_lines): Likewise.
21507 (tag_counters): Likewise.
21508 (tag_summary): Likewise.
21509 (dump_working_sets): Use depth to do a proper indentation.
21511 2017-04-28 Jakub Jelinek <jakub@redhat.com>
21514 * cgraph.h (symtab_node::debug_symtab): No longer inline.
21515 * symtab.c (symtab_node::debug_symtab): Move definition here.
21517 2017-04-28 Richard Biener <rguenther@suse.de>
21519 * lto-streamer.h (LTO_major_version): Bump to 7.
21521 2017-04-28 Richard Biener <rguenther@suse.de>
21523 * tree-vrp.c (assert_info): New struct.
21524 (add_assert_info): New helper.
21525 (register_edge_assert_for_2): Refactor to add asserts to a vector
21527 (register_edge_assert_for_1): Likewise.
21528 (register_edge_assert_for): Likewise.
21529 (finish_register_edge_assert_for): New helper actually registering
21530 asserts where live on edge.
21531 (find_conditional_asserts): Adjust.
21532 (find_switch_asserts): Likewise.
21533 (evrp_dom_walker::try_find_new_range): Generalize.
21534 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
21536 2017-04-27 Marek Polacek <polacek@redhat.com>
21539 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
21540 arg10 and arg11 to itype.
21542 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
21544 * doc/extend.texi (Object Size Checking): Improve grammar.
21546 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
21549 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
21550 that the logic for permitting reciprocal estimates matches that
21553 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21556 * tree.c (type_cache_hasher::equal): Only compare
21557 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
21558 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
21559 non-aggregate element types.
21560 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
21561 about the flag on ARRAY_TYPEs in the comment, formatting fix.
21563 2017-04-27 Richard Biener <rguenther@suse.de>
21565 PR middle-end/80533
21566 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
21567 stripping ARRAY_REFs from MEM_EXPR make sure we're not
21568 keeping a reference to a trailing array.
21570 2017-04-27 Richard Biener <rguenther@suse.de>
21572 PR middle-end/80539
21573 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
21574 being in loop-closed SSA form conservatively.
21575 (chrec_fold_multiply_poly_poly): Likewise.
21577 2017-04-27 Tamar Christina <tamar.christina@arm.com>
21579 PR middle-end/79665
21580 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
21581 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
21583 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21586 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
21587 (aarch64_function_arg_alignment): Return unsigned int again, but still
21588 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
21589 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21590 Don't emit -Wpsabi note.
21591 (aarch64_function_arg_boundary): Likewise.
21592 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21595 2017-04-26 Nathan Sidwell <nathan@acm.org>
21597 * tree.h (crc32_unsigned_n): Declare.
21598 (crc32_unsigned, crc32_unsigned): Make inline.
21599 * tree.c (crc32_unsigned_bits): Replace with ...
21600 (crc32_unsigned_n): ... this.
21601 (crc32_unsigned, crc32_byte): Remove.
21602 (crc32_string): Remove unnecessary braces.
21604 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21606 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
21607 * ipa-inline-analysis.c (MAX_TIME): Remove.
21608 (account_size_time): Use sreal for time.
21609 (dump_inline_summary): Update.
21610 (estimate_function_body_sizes): Update.
21611 (estimate_edge_size_and_time): Update.
21612 (estimate_calls_size_and_time): Update.
21613 (estimate_node_size_and_time): Update.
21614 (inline_merge_summary): Update.
21615 (inline_update_overall_summary): Update.
21616 (estimate_time_after_inlining): Update.
21617 (inline_read_section): Update.
21618 (inline_write_summary): Update.
21619 * ipa-inline.c (compute_uninlined_call_time): Update.
21620 (compute_inlined_call_time): Update.
21621 (recursive_inlining): Update.
21622 (inline_small_functions): Update.
21623 (dump_overall_stats): Update.
21624 * ipa-inline.h: Include sreal.h.
21625 (size_time_entry): Turn time to sreal.
21626 (inline_summary): Turn self_time nad time to sreal.
21628 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21630 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
21632 (sreal::stream_out, sreal::stream_in): New.
21633 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
21635 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21637 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
21640 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
21643 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
21645 (dimode_scalar_chain::compute_convert_gain): Ditto.
21646 (dimode_scalar_chain::make_vector_copies): Ditto.
21647 (dimode_scalar_chain::convert_reg): Ditto.
21648 (dimode_scalar_chain::convert_insn): Ditto.
21649 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
21650 (VI248_AVX512BW_1): New mode iterator.
21651 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
21652 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
21655 2017-04-25 Martin Sebor <msebor@redhat.com>
21657 PR tree-optimization/80497
21658 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
21659 constants are representable in HOST_WIDE_INT.
21660 (parse_directive): Ditto.
21662 2017-04-25 Martin Sebor <msebor@redhat.com>
21665 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
21666 (new_zero_array): Adjust signature.
21667 (dom_info::dom_init): Used unsigned rather that size_t.
21668 (dom_info::dom_info): Same.
21670 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21671 Jakub Jelinek <jakub@redhat.com>
21674 * config/arm/arm.c: Include gimple.h.
21675 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21676 returns negative, increment ncrn only if it returned positive.
21677 (arm_needs_doubleword_align): Return int instead of bool,
21678 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
21679 members, but if there is any such non-FIELD_DECL
21680 > PARM_BOUNDARY aligned decl, return -1 instead of false.
21681 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21682 returns negative, increment nregs only if it returned positive.
21683 (arm_setup_incoming_varargs): Likewise.
21684 (arm_function_arg_boundary): Emit -Wpsabi note if
21685 arm_needs_doubleword_align returns negative, return
21686 DOUBLEWORD_ALIGNMENT only if it returned positive.
21688 2017-04-25 Marek Polacek <polacek@redhat.com>
21691 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21692 first argument to type.
21694 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
21697 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
21698 type checks to test for compatibility instead of equality.
21700 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21701 Jakub Jelinek <jakub@redhat.com>
21704 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
21706 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
21707 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
21708 the alignment computation, but return their maximum in warn_alignment.
21709 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21710 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
21712 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
21713 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21716 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21718 * config/arc/simdext.md (dmpyh): Fix typo.
21720 2017-04-25 Richard Biener <rguenther@suse.de>
21722 PR tree-optimization/80492
21723 * alias.c (compare_base_decls): Handle registers with asm
21724 specification conservatively.
21725 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21726 compare_base_decls returning dont-know properly.
21728 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21730 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
21731 (legitimate_offset_address_p): New function.
21732 (arc_legitimate_address_p): Use above function.
21734 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21736 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
21738 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21740 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
21741 ACCH registers whenever they are available.
21743 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21745 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
21746 double regs fix when not used.
21748 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21750 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
21752 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
21753 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
21755 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21757 * config/arc/arc.c (arc_output_addsi): Check for h-register class
21758 when emitting short ADD instructions.
21760 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21762 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
21764 (cmpsi_cc_c_insn): Likewise.
21765 (cbranchsi4_scratch): Compute proper instruction length using
21766 compact_hreg_operand.
21767 * config/arc/predicates.md (compact_hreg_operand): New predicate.
21769 2017-04-25 Richard Biener <rguenther@suse.de>
21771 PR middle-end/80509
21772 * passes.c (pass_manager::pass_manager): Initialize
21773 m_name_to_pass_map.
21775 2017-04-25 Richard Biener <rguenther@suse.de>
21777 PR tree-optimization/79201
21778 * tree-ssa-sink.c (statement_sink_location): Handle calls.
21780 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21783 * config/s390/vector.md: Split MEM->GPR vector moves for
21784 non-s_operand addresses.
21786 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21789 * config/s390/predicates.md (reload_const_wide_int_operand): New
21791 * config/s390/s390.md ("movti"): Remove d/P alternative.
21792 ("movti_bigconst"): New pattern definition.
21794 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21797 * s390-protos.h (s390_expand_cs_hqi): Removed.
21798 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
21799 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
21800 modes as well as CCZ1mode and CCZmode.
21801 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
21802 signature of s390_emit_compare_and_swap.
21803 (s390_expand_cs_hqi): Likewise, make static.
21804 (s390_expand_cs_tdsi): Generate an explicit compare before trying
21805 compare-and-swap, in some cases.
21806 (s390_expand_cs): Wrapper function.
21807 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
21809 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
21810 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
21811 patterns for small and large integers. Forbid symref memory operands.
21812 Move expander to s390.c. Require cc register.
21813 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
21814 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
21815 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
21816 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
21817 symref memory operands. Remove CC mode and call s390_match_ccmode
21819 ("atomic_exchange<mode>"): Allow and implement all integer modes.
21821 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21823 * config/s390/s390.md (define_peephole2): New peephole to help
21824 combining the load-and-test pattern with volatile memory.
21826 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21828 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
21829 with CCZmode for TARGET_Z196.
21831 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21833 PR rtl-optimization/80501
21834 * combine.c (make_compound_operation_int): Set subreg_code to SET
21835 even for AND with mask of the sign bit of mode.
21837 PR rtl-optimization/80500
21838 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
21839 sum's initial value.
21841 2017-04-25 Julian Brown <julian@codesourcery.com>
21842 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21844 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
21846 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
21848 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
21850 2017-04-25 Julian Brown <julian@codesourcery.com>
21851 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21853 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
21854 (thunderx2t99_sha): New Reservation.
21856 2017-04-25 Julian Brown <julian@codesourcery.com>
21857 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21859 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
21860 type for 1-element load.
21862 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
21864 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
21866 2017-04-24 Martin Jambor <mjambor@suse.cz>
21868 PR tree-optimization/80293
21869 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
21870 char arrays not totally scalarizable if it is false.
21871 (analyze_all_variable_accesses): Pass correct value in the new
21872 parameter. Add a statistics counter.
21874 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
21876 PR middle-end/79931
21877 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
21879 2017-04-24 Richard Biener <rguenther@suse.de>
21881 PR tree-optimization/80494
21882 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
21883 out for complex types.
21885 2017-04-24 Richard Biener <rguenther@suse.de>
21887 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
21888 * tree-ssa-sccvn.c (print_scc): Print SCC size.
21889 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
21890 (DFS): Adjust and never fail.
21891 (sccvn_dom_walker::fail): Remove.
21892 (sccvn_dom_walker::before_dom_children): Adjust.
21893 (run_scc_vn): Likewise and never fail.
21894 * tree-ssa-pre.c (pass_pre::execute): Adjust.
21895 (pass_fre::execute): Likewise.
21897 2017-04-24 Richard Biener <rguenther@suse.de>
21899 PR tree-optimization/79725
21900 * tree-ssa-sink.c (statement_sink_location): Return whether
21901 failure reason was zero uses. Move that check later.
21902 (sink_code_in_bb): Deal with zero uses by removing the stmt
21905 2017-04-24 Richard Biener <rguenther@suse.de>
21908 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
21909 pointer-based references.
21911 2017-04-24 Richard Biener <rguenther@suse.de>
21914 * pass_manager.h (pass_manager::operator new): Remove.
21915 (pass_manager::operator delete): Likewise.
21916 * passes.c (pass_manager::operator new): Remove.
21917 (pass_manager::operator delete): Likewise.
21918 (pass_manager::pass_manager): Zero individual pass members.
21920 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
21923 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
21924 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
21925 Check "XEXP (src, 1)" operand here.
21926 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
21927 Check "XEXP (src, 1)" operand here.
21928 (dimode_scalar_chain::make_vector_copies): Detect count register
21929 of a shift instruction. Zero extend count register from QImode
21930 to DImode to satisfy vector shift pattern count operand predicate.
21931 Substitute vector shift count operand with a DImode copy.
21932 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
21935 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
21937 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
21938 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21939 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
21940 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21941 (UNSPEC_NOREX_MEM): Remove definition.
21943 2017-04-21 Richard Biener <rguenther@suse.de>
21945 PR tree-optimization/79547
21946 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21947 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
21948 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
21949 without any constraints.
21951 2017-04-21 Richard Biener <rguenther@suse.de>
21953 PR tree-optimization/78847
21954 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
21956 2017-04-21 Richard Biener <rguenther@suse.de>
21958 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
21959 (build_distinct_type_copy): Likewise.
21960 (build_variant_type_copy): Likewise.
21961 * tree.c (build_qualified_type): Pass down mem-stat info.
21962 (build_distinct_type_copy): Likewise.
21963 (build_variant_type_copy): Likewise.
21965 2017-04-21 Richard Biener <rguenther@suse.de>
21967 PR tree-optimization/80237
21968 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
21970 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
21971 for a simplified result.
21973 2016-04-21 Richard Biener <rguenther@suse.de>
21975 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
21976 sth as strict as a simple_iv but a chrec without symbols and an
21977 operand defined in the loop we are peeling (and not some subloop).
21978 (propagate_constants_for_unrolling): Propagate all constants.
21980 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21983 * config/i386/i386.c (print_reg): Remove assert for disalowed
21984 regno values, call output_operand_lossage instead.
21986 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
21989 * config/i386/constraints.md (Yc): New register constraint.
21990 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
21991 Use Yc constraint for alternative 2 of operand 0. Remove
21992 preferred_for_speed attribute.
21994 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
21996 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
21997 lastprivate clauses in SIMT case.
21999 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22001 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22003 2017-04-20 Richard Biener <rguenther@suse.de>
22005 PR tree-optimization/57796
22006 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22007 as N scalar stores.
22008 (vect_model_load_cost): Cost gathers as N scalar loads.
22010 2017-04-20 Richard Biener <rguenther@suse.de>
22012 * ggc-page.c (ggc_allocated_p): Rename to ...
22013 (safe_lookup_page_table_entry): ... this and return the lookup
22015 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22017 2017-04-20 Richard Biener <rguenther@suse.de>
22019 PR tree-optimization/80453
22020 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22021 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22022 from the conditions.
22023 (vn_phi_eq): Pass them down.
22024 (vn_phi_lookup): Record them.
22025 (vn_phi_insert): Likewise.
22027 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22029 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22030 uninitialized variable warning to avoid buffer overrun.
22032 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22035 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22036 is suppressed for '{ 0 }' in C.
22038 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22040 * BASE-VER: Set to 8.0.0.
22042 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22044 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22045 priority .init_array and .fini_array section with SECTION_NOTYPE
22048 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22050 PR middle-end/80423
22051 * tree.h (build_array_type): Add typeless_storage default argument.
22052 * tree.c (type_cache_hasher::equal): Also compare
22053 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22054 (build_array_type): Add typeless_storage argument, set
22055 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22057 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22058 (build_array_type): Likewise. Add typeless_storage argument.
22060 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22061 Jakub Jelinek <jakub@redhat.com>
22063 PR tree-optimization/80426
22064 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22065 operation on symbolic operands, also compute the overflow for the
22066 invariant part when the operation degenerates into a negation.
22068 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22071 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22072 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22075 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22077 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
22080 * config/avr/avr.c (tree.h): Include it.
22081 (cgraph.h): Include it.
22082 (avr_encode_section_info): Don't warn for uninitialized progmem
22083 variable if it's just an alias.
22085 2017-04-19 Richard Biener <rguenther@suse.de>
22088 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
22089 when needed by AutoPGO.
22091 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
22094 * doc/lto.texi: Remove an extra 'that'.
22096 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
22098 PR rtl-optimization/80429
22099 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
22100 are only used in debug insns.
22102 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22103 Vladimir Makarov <vmakarov@redhat.com>
22105 * config/sparc/predicates.md (input_operand): Add comment. Return
22106 true for any memory operand when LRA is in progress.
22107 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
22109 2017-04-18 Jeff Law <law@redhat.com>
22112 * mips.md ({return,simple_return}_internal): Do not overwrite
22115 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22117 PR tree-optimization/80443
22118 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
22119 instead of adding 1, subtract -1 and similarly instead of subtracting
22122 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
22124 PR rtl-optimization/80357
22125 * haifa-sched.c (tmp_bitmap): New variable.
22126 (model_recompute): Handle duplicate use records.
22127 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
22128 (free_global_sched_pressure_data): Free it.
22130 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
22133 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22134 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22135 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22136 instead of SYSTEM_HEADER_DIR.
22138 2017-04-18 Jeff Law <law@redhat.com>
22140 PR middle-end/80422
22141 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
22142 predecessors after walking up the insn chain.
22144 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22147 * dwarf2out.c (modified_type_die): Try harder not to emit internal
22148 sizetype type into debug info.
22150 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22153 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
22154 unneeded test for TARGET_UPPER_REGS_SF.
22155 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
22157 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22160 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
22161 instead of gsi_after_labels.
22163 2017-04-18 Jeff Law <law@redhat.com>
22165 * regcprop.c (maybe_mode_change): Avoid creating copies of the
22169 2017-04-13 Jeff Law <law@redhat.com>
22170 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22171 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22173 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
22176 * config/avr/avr.c (intl.h): Include it.
22177 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
22179 2017-04-18 Martin Liska <mliska@suse.cz>
22181 PR gcov-profile/78783
22182 * gcov-tool.c (gcov_output_files): Validate that destination
22183 file is either removed by the tool or by a user.
22185 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22186 Guy Benyei <guybe@mellanox.com>
22188 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
22189 block, and do not negate it, the stored id is already negative.
22191 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22193 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
22195 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22198 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
22199 masks of options that should be turned off if the VSX vector
22200 options are turned off.
22201 (OTHER_P8_VECTOR_MASKS): Likewise.
22202 (OTHER_VSX_VECTOR_MASKS): Likewise.
22203 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
22204 rs6000_disable_incompatible_switches to validate no type switches
22206 (rs6000_incompatible_switch): New function to disallow turning on
22207 other vector options if -mno-vsx, -mno-power8-vector, or
22208 -mno-power9-vector are specified.
22210 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22212 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
22214 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22216 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
22217 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
22218 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
22219 (ARG_POINTER_CFA_OFFSET): Likewise.
22221 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22223 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
22224 conditions to take advantage of various optimizations.
22226 2017-04-13 Jeff Law <law@redhat.com>
22228 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22229 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22230 (zero_extendsidi2_dext): Likewise.
22232 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22235 * fold-const.c (fold_ternary_loc): Revert
22236 use op0 instead of fold_convert_loc (loc, type, arg0) part of
22239 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
22241 PR rtl-optimization/80343
22242 * lra-remat.c (update_scratch_ops): Assign original hard reg to
22243 new scratch pseudo.
22245 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
22248 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
22249 to ubsan_encode_value.
22251 2017-04-13 Jeff Law <law@redhat.com>
22253 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
22254 appearing in DEBUG_INSNs.
22256 2017-04-13 Martin Liska <mliska@suse.cz>
22258 PR gcov-profile/80413
22259 * gcov-io.c (gcov_write_string): Copy to buffer just when
22260 allocated size is greater than zero.
22262 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22265 * dwarf2out.c (decls_for_scope): Ignore declarations of
22266 current_function_decl in BLOCK_NONLOCALIZED_VARS.
22268 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
22271 * ipa-visibility.c (non_local_p): Fix typos.
22272 (localize_node): When localizing symbol in same comdat group,
22273 dissolve the group only when we know external symbols are going
22275 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
22277 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22279 PR tree-optimization/79390
22280 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
22281 order does not result in usable sequence, retry with reversed operand
22287 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
22288 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
22289 op0 instead of fold_convert_loc (loc, type, arg0).
22291 2017-04-12 Jeff Law <law@redhat.com>
22293 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
22294 has a delay slot in the generated code.
22296 * config/cris/cris.md (cris_preferred_reload_class): Return
22297 GENNONACR_REGS rather than GENERAL_REGS.
22299 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22302 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
22303 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
22304 signedness of the result type.
22306 2017-04-12 Richard Biener <rguenther@suse.de>
22307 Jeff Law <law@redhat.com>
22309 PR tree-optimization/80359
22310 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
22311 trim stores to TARGET_MEM_REFs.
22313 2017-04-12 Richard Biener <rguenther@suse.de>
22315 PR tree-optimization/79390
22316 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
22317 threading case even more.
22319 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
22322 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
22323 for quad_address_p for TImode, instead of just not indexed_address.
22325 2017-04-12 Richard Biener <rguenther@suse.de>
22326 Bernd Edlinger <bernd.edlinger@hotmail.de>
22328 PR middle-end/79671
22329 * alias.c (component_uses_parent_alias_set_from): Handle
22330 TYPE_TYPELESS_STORAGE.
22331 (get_alias_set): Likewise.
22332 * tree-core.h (tree_type_common): Add typeless_storage flag.
22333 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
22334 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
22335 for types containing members with TYPE_TYPELESS_STORAGE.
22336 (place_field): Likewise.
22337 (layout_type): Likewise for ARRAY_TYPE.
22338 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
22339 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
22340 TYPE_TYPELESS_STORAGE.
22341 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
22343 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22346 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22347 first argument to type.
22349 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22353 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
22354 CONST0_RTX (mode) rather than const0_rtx where appropriate.
22355 (rs6000_expand_binop_builtin): Likewise.
22356 (rs6000_expand_ternop_builtin): Likewise; also add missing
22357 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
22358 vshasigma built-ins.
22359 * doc/extend.texi: Document that vec_xxpermdi's third argument
22360 must be a constant.
22362 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
22364 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
22365 Use shift_const cost parameter when calculating gain of STV shifts.
22367 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
22369 PR rtl-optimization/70478
22370 * lra-constraints.c (process_alt_operands): Check memory for
22371 disfavoring memory insn operand.
22373 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22375 PR middle-end/80100
22376 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
22377 left shift in unsigned HOST_WIDE_INT type.
22379 PR rtl-optimization/80385
22380 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
22381 (not (neg X)) into (plus X -1) for complex or non-integral modes.
22384 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
22385 if they have any depend clauses.
22387 2017-04-11 Martin Liska <mliska@suse.cz>
22390 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
22391 * ipa-split.c (split_function): Create a local comdat symbol
22392 if caller is in a comdat group.
22394 2017-04-11 Martin Liska <mliska@suse.cz>
22397 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
22400 2017-04-11 Martin Sebor <msebor@redhat.com>
22402 PR middle-end/80364
22403 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
22404 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
22406 (directive::set_width, directive::set_precision, format_character):
22408 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
22411 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
22414 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
22415 conflict, set target->arch_name instead of target->cpu_name.
22417 2017-04-11 Richard Biener <rguenther@suse.de>
22419 PR tree-optimization/80374
22420 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
22421 build_zero_cst, remove fold_convertible_p check again.
22423 2017-04-11 Martin Liska <mliska@suse.cz>
22426 * ubsan.c (instrument_object_size): Do not instrument register
22429 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22432 * config/i386/i386-builtin-types.def
22433 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
22434 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
22435 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
22436 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
22437 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
22438 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
22439 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
22440 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
22441 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
22442 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
22443 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
22444 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
22445 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
22446 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
22447 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
22448 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
22449 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
22450 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
22451 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
22452 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
22453 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
22454 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
22455 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
22456 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
22457 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
22458 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
22459 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
22460 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
22461 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
22462 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
22463 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
22464 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
22465 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
22466 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
22467 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
22468 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
22469 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
22470 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
22471 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
22472 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
22473 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
22474 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
22475 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
22476 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
22477 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
22478 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
22480 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
22481 flag to second_arg_count, handle 4 argument function type _COUNT
22482 aliases, handle second_arg_count on second argument rather than last.
22484 2017-04-10 Jeff Law <law@redhat.com>
22486 PR tree-optimization/80374
22487 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
22488 record anything if we can not convert integer_zero_node to the
22491 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22494 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22495 Enhance special handling given to the TARGET_P9_MINMAX option in
22496 relation to certain other options.
22498 2017-04-10 Bin Cheng <bin.cheng@arm.com>
22500 PR tree-optimization/80153
22501 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
22502 remove POINTER_PLUS_EXPR's base part directly, rather than through
22505 2017-04-10 Richard Biener <rguenther@suse.de>
22506 Bin Cheng <bin.cheng@arm.com>
22508 PR tree-optimization/80153
22509 * tree-affine.c (aff_combination_to_tree): Get base pointer from
22510 the first element of pointer type aff_tree. Build result expr in
22512 (add_elt_to_tree): Convert to type unconditionally. Remove other
22513 fold_convert calls.
22514 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
22515 (rewrite_use_nonlinear_expr): Check invariant using iv information.
22517 2017-04-10 Richard Biener <rguenther@suse.de>
22519 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22522 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
22524 PR rtl-optimization/70478
22525 * lra-constraints.c (curr_small_class_check): New.
22526 (update_and_check_small_class_inputs): New.
22527 (process_alt_operands): Update curr_small_class_check. Disfavor
22528 alternative insn memory operands. Check available regs for small
22531 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22534 * config/mips/mips.opt (-mvirt): Update description.
22535 * doc/invoke.texi (-mvirt): Likewise.
22537 2017-04-10 Richard Biener <rguenther@suse.de>
22539 PR middle-end/80362
22540 * fold-const.c (fold_binary_loc): Look at unstripped ops when
22541 looking for NEGATE_EXPR in -A / -B to A / B folding.
22543 2017-04-10 Martin Liska <mliska@suse.cz>
22545 PR gcov-profile/80224
22546 * gcov.c (print_usage): Fix usage string.
22547 (get_gcov_intermediate_filename): Remove.
22548 (output_gcov_file): Use both for normal and intermediate format.
22549 (generate_results): Do not initialize special file for
22550 intermediate format.
22552 2017-04-10 Richard Biener <rguenther@suse.de>
22554 PR tree-optimization/80304
22555 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
22558 2017-04-10 Nathan Sidwell <nathan@acm.org>
22561 * config/rs6000/rs6000.c (rs6000_vector_type): New.
22562 (rs6000_init_builtins): Use it.
22564 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22566 * config/arm/arm.md (<mrc>): Add mode to SET source.
22567 (<mrrc>): Likewise.
22569 2017-04-10 Richard Biener <rguenther@suse.de>
22571 PR middle-end/80344
22572 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
22574 2017-04-10 Jakub Jelinek <jakub@redhat.com>
22577 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
22578 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
22579 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
22580 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
22581 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
22582 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
22583 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
22584 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
22585 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
22586 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
22587 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
22588 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
22589 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
22590 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
22591 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
22592 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
22593 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
22594 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
22595 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
22596 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
22597 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
22598 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
22599 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
22601 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
22603 PR rtl-optimization/70478
22604 * lra-constraints.c: Reverse the last patch.
22606 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
22608 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
22609 Add comment for WCHAR_T.
22611 2017-04-08 Martin Liska <mliska@suse.cz>
22614 2017-04-07 Martin Liska <mliska@suse.cz>
22617 * ipa-split.c (split_function): Add function part to a same comdat
22620 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22623 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
22625 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
22627 * rs6000/rs6000.c (vec_load_pendulum): Rename...
22628 (vec_pairing): ...to this.
22629 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
22630 (rs6000_sched_init): Adjust for name change.
22631 (struct rs6000_sched_context): Likewise.
22632 (rs6000_init_sched_context): Likewise.
22633 (rs6000_set_sched_context): Likewise.
22635 2017-04-07 Jakub Jelinek <jakub@redhat.com>
22641 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
22643 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
22644 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
22645 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
22647 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
22649 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
22651 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22653 PR rtl-optimization/70703
22654 * ira-color.c (update_conflict_hard_regno_costs): Use
22655 int64_t instead of HOST_WIDE_INT.
22657 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22659 PR rtl-optimization/70478
22660 * lra-constraints.c (process_alt_operands): Disfavor alternative
22661 insn memory operands.
22663 2017-04-07 Jeff Law <law@redhat.com>
22665 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
22666 CALL and NOTE_INSN_CALL_ARG_LOCATION.
22668 2017-04-07 Martin Liska <mliska@suse.cz>
22671 * config/aarch64/aarch64.c (aarch64_process_target_attr):
22672 Show error message instead of an ICE.
22674 2017-04-07 Martin Liska <mliska@suse.cz>
22677 * ipa-split.c (split_function): Add function part to a same comdat
22680 2017-04-07 Richard Biener <rguenther@suse.de>
22682 PR middle-end/80341
22683 * tree.c (get_unwidened): Also handle ! for_type case for
22685 * convert.c (do_narrow): Split out from ...
22686 (convert_to_integer_1): ... here. Do not pass final truncation
22687 type to get_unwidened for TRUNC_DIV_EXPR.
22689 2017-04-07 Richard Biener <rguenther@suse.de>
22691 * tree-affine.c (wide_int_ext_for_comb): Take type rather
22693 (aff_combination_const): Adjust.
22694 (aff_combination_scale): Likewise.
22695 (aff_combination_add_elt): Likewise.
22696 (aff_combination_add_cst): Likewise.
22697 (aff_combination_convert): Likewise.
22698 (add_elt_to_tree): Likewise. Remove unused argument.
22699 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
22701 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
22703 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
22705 * config/arm/arm.c (arm_default_short_enums): Use
22706 ARM_DEFAULT_SHORT_ENUMS.
22707 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
22709 2017-04-06 Jakub Jelinek <jakub@redhat.com>
22712 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
22713 members with redundant out-of-class redeclaration.
22715 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22718 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
22719 * config/i386/i386.md (*zero_extendsidi2):
22720 Add (?*x,*x) and (?*v,*v) alternatives.
22722 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22725 * config/i386/i386.c (ix86_expand_builtin)
22726 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
22727 mode from insn data. Convert operands to insn operand mode.
22728 Copy operands that don't satisfy insn predicate to a register.
22730 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
22732 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
22735 2017-04-06 Richard Biener <rguenther@suse.de>
22737 PR tree-optimization/80334
22738 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
22739 preserve alignment of accesses.
22741 2017-04-06 Richard Biener <rguenther@suse.de>
22743 PR tree-optimization/80262
22744 * tree-sra.c (build_ref_for_offset): Preserve address-space
22746 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
22747 Drop useless address-space information on MEM_REF offsets.
22749 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
22751 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
22753 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22755 PR rtl-optimization/70703
22756 * ira-color.c (update_conflict_hard_regno_costs): Use
22757 HOST_WIDE_INT instead of long.
22759 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
22762 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
22763 not defined for x86_64 target. Add -mmmx target option when __SSE2__
22765 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
22766 for x86_64 target. Handle -m3dnowa option.
22768 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22770 PR rtl-optimization/70703
22771 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
22772 (update_conflict_hard_regno_costs): Use long instead of unsigned
22773 arithmetic for cost calculation.
22775 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22776 Bernd Edlinger <bernd.edlinger@hotmail.de>
22779 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
22782 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
22785 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
22786 ptr_mode with Pmode throughout.
22787 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
22788 into probe_stack_range and use DImode.
22790 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
22793 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
22794 call_eh_return is true.
22796 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22798 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
22799 Initialize last_match_fntype_index.
22801 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22804 * tree-nvr.c: Include internal-fn.h.
22805 (pass_return_slot::execute): Ignore internal calls without
22808 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22809 Richard Biener <rguenther@suse.de>
22812 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
22813 captures used multiple times, except for the last use.
22814 * generic-match-head.c: Include gimplify.h.
22816 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22818 PR tree-optimization/79390
22819 * target.h (struct noce_if_info): Declare.
22820 * targhooks.h (default_noce_conversion_profitable_p): Declare.
22821 * target.def (noce_conversion_profitable_p): New target hook.
22822 * ifcvt.h (struct noce_if_info): New type, moved from ...
22823 * ifcvt.c (struct noce_if_info): ... here.
22824 (noce_conversion_profitable_p): Renamed to ...
22825 (default_noce_conversion_profitable_p): ... this. No longer
22827 (noce_try_store_flag_constants, noce_try_addcc,
22828 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
22829 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
22830 instead of noce_conversion_profitable_p.
22831 * config/i386/i386.c: Include ifcvt.h.
22832 (ix86_option_override_internal): Don't override
22833 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
22834 (ix86_noce_conversion_profitable_p): New function.
22835 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
22836 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
22837 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
22838 * doc/tm.texi: Regenerated.
22840 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22842 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
22845 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22848 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
22849 instructions for small multiply cores.
22851 2017-04-04 Jeff Law <law@redhat.com>
22853 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
22855 (mips_expand_vec_perm_const): Initialize elements in orig_perm
22856 that are not set by the loop over the elements.
22858 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22861 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
22862 int mode, convert_modes it to mode as unsigned, otherwise use
22863 lowpart_subreg to mode rather than SImode.
22864 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
22865 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
22866 Use DImode instead of SImode for the shift count operand.
22867 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
22870 2017-04-04 Richard Biener <rguenther@suse.de>
22872 PR middle-end/80281
22873 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
22874 arithmetic done for the negate or the plus. Simplify.
22875 (A - (-B) -> A + B): Likewise.
22876 * fold-const.c (split_tree): Make sure to not negate pointers.
22878 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
22880 PR rtl-optimization/60818
22881 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
22882 a compare of comparisons with the thing compared if this results
22883 in a different machine mode.
22885 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
22887 * alias.c (base_alias_check): Fix typo in comment.
22888 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
22889 * cgraphunit.c (symbol_table::compile): Likewise.
22890 * collect2.c (maybe_run_lto_and_relink): Likewise.
22891 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
22892 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
22893 * config/avr/avr.c (avr_map_op_t): Likewise.
22894 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
22895 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
22896 * config/epiphany/epiphany.md (movcc): Likewise.
22897 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
22898 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
22900 * config/mips/mips.c (mips_save_restore_reg): Likewise.
22901 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
22902 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
22903 * config/sh/sh.c (sh_rtx_costs): Likewise.
22904 * fold-const.c (fold_truth_andor): Likewise.
22905 * genautomata.c (collapse_flag): Likewise.
22906 * gengtype.h (struct type::u::s): Likewise.
22907 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
22908 * input.c (FORMAT_AMOUNT): Likewise.
22909 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
22910 (known_aggs_to_agg_replacement_list): Likewise.
22911 * ipa-inline-analysis.c: Likewise.
22912 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
22913 * ipa-polymorphic-call.c
22914 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
22915 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
22916 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
22918 * modulo-sched.c (apply_reg_moves): Likewise.
22919 * omp-expand.c (build_omp_regions_1): Likewise.
22920 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
22921 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
22922 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
22923 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
22924 * value-prof.c: Likewise.
22925 * var-tracking.c (val_reset): Likewise.
22927 2017-04-03 Richard Biener <rguenther@suse.de>
22929 PR tree-optimization/80275
22930 * fold-const.c (split_address_to_core_and_offset): Handle
22933 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
22935 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
22936 descriptors is at least equal to that of functions.
22938 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22940 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
22942 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22945 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
22946 (mov<IMOD4:mode>): New expander.
22947 (*mov<IMOD4:mode>_internal): New insn and split pattern.
22949 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22951 PR rtl-optimization/79405
22952 * fwprop.c (propagations_left): New variable.
22953 (forward_propagate_into): Decrement it.
22954 (fwprop_init): Initialize it.
22955 (fw_prop): If the variable has reached zero, stop propagating.
22956 (fwprop_addr): Ditto.
22958 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22961 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
22962 a FUNCTION_DECL, pass it as decl instead of origin to
22965 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
22967 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
22970 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
22973 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
22974 TARGET_VSX_SMALL_INTEGER.
22976 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22978 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22979 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
22981 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22983 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
22984 extraction from odd-numbered MSA register.
22986 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22988 PR middle-end/80173
22989 * expmed.c (store_bit_field_1): Don't attempt to create
22990 a word subreg out of hard registers wider than word if they
22991 have HARD_REGNO_NREGS of 1 for their mode.
22993 PR middle-end/80163
22994 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
22995 conversions to integer types wider than word and pointer.
22998 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
22999 (rtx_equal_for_cselib_p): Pass 0 to it.
23000 * cselib.c (cselib_hasher::equal): Likewise.
23001 (rtx_equal_for_cselib_1): Add depth argument. If depth
23002 is 128, don't look up VALUE locs and punt. Increment
23003 depth in recursive calls when walking VALUE locs.
23005 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23007 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23008 (make_gcov_file_name): Use the canonical path name for generating
23010 (read_line): Fix handling of files with ascii null bytes.
23012 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23014 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23015 to initialise a vector register instead
23016 of using a const_int.
23018 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23020 PR translation/80189
23021 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23022 diagnostic messages.
23024 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23027 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23028 (dfp_diex_<mode>): Update mode of operand 1.
23029 * doc/extend.texi (dxex, dxexq): Document change to return type.
23030 (diex, diexq): Document change to argument type.
23032 2017-03-30 Martin Jambor <mjambor@suse.cz>
23035 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23036 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23037 it reflects the signature changes performed at the callee side.
23038 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23039 to cgraph_build_function_type_skip_args.
23040 (build_function_decl_skip_args): Adjust call to the above function.
23042 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23045 * config/i386/sse.md
23046 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23047 register as dest whenever it is a MEM not rtx_equal_p to the
23048 corresponding dup operand, and when forcing into reg move the
23049 reg into the memory afterwards.
23050 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23051 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23052 for the force_reg mode.
23053 (avx512vl_vextractf128<mode>): Use register as dest either
23054 always when a MEM, or when it is a MEM not rtx_equal_p to the
23055 corresponding dup operand, or even not when it is a CONST_VECTOR
23056 depending on the mode and lo vs. hi.
23057 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23059 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23060 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23061 Likewise. Require that operands[2] is even.
23062 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23063 Remove extraneous parens. Require that operands[2] is a multiple
23065 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23066 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23067 disallow memory then.
23069 2017-03-30 Richard Biener <rguenther@suse.de>
23071 PR tree-optimization/77498
23072 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23073 to non-constants over backedges.
23075 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23077 PR rtl-optimization/80233
23078 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23079 as last_combined_insn. Do not test for BARRIER_P separately.
23081 2017-03-29 Andreas Schwab <schwab@suse.de>
23084 * calls.c (prepare_call_address): Convert funexp to Pmode before
23085 copying to temp reg.
23087 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23089 PR tree-optimization/80158
23090 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
23091 Handle possible future case of more than one alternate
23093 (replace_rhs_if_not_dup): Likewise.
23094 (replace_one_candidate): Likewise.
23096 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
23098 PR rtl-optimization/80193
23099 * ira.c (ira): Do not check allocation for LRA.
23101 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
23103 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
23104 (nvptx_output_simt_exit): Declare.
23105 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
23106 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
23107 (init_softstack_frame): Move initialization of crtl->is_leaf to...
23108 (nvptx_declare_function_name): ...here. Emit declaration of local
23109 memory space buffer for omp_simt_enter insn.
23110 (nvptx_output_unisimt_switch): New.
23111 (nvptx_output_softstack_switch): New.
23112 (nvptx_output_simt_enter): New.
23113 (nvptx_output_simt_exit): New.
23114 * config/nvptx/nvptx.h (struct machine_function): New fields
23115 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
23116 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
23117 (UNSPECV_SIMT_EXIT): Ditto.
23118 (omp_simt_enter_insn): New insn.
23119 (omp_simt_enter): New expansion.
23120 (omp_simt_exit): New insn.
23121 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
23123 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
23124 (expand_GOMP_SIMT_ENTER_ALLOC): New.
23125 (expand_GOMP_SIMT_EXIT): New.
23126 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
23127 (GOMP_SIMT_ENTER_ALLOC): Ditto.
23128 (GOMP_SIMT_EXIT): Ditto.
23129 * target-insns.def (omp_simt_enter): New insn.
23130 (omp_simt_exit): Ditto.
23131 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
23133 (lower_rec_simd_input_clauses): Implement SIMT privatization.
23134 (lower_rec_input_clauses): Likewise.
23135 (lower_lastprivate_clauses): Handle SIMT privatization.
23137 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
23138 (ompdevlow_adjust_simt_enter): New.
23139 (find_simtpriv_var_op): New.
23140 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
23141 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
23143 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
23144 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
23145 (copy_decl_for_dup_finish): Ditto.
23147 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
23149 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
23152 * config/i386/i386.c (ix86_option_override_internal): Always
23153 allow -mpreferred-stack-boundary=3 for 64-bit targets.
23155 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23157 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
23159 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23161 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
23162 mark new edge's irreducible flag accordign to it.
23163 (vect_do_peeling): Check loop preheader edge's irreducible flag
23164 and pass it to function slpeel_add_loop_guard.
23166 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
23168 PR tree-optimization/80218
23169 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23170 Update block frequencies and counts.
23172 2017-03-28 Richard Biener <rguenther@suse.de>
23174 PR tree-optimization/78644
23175 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23176 of a simplification result we may not use it at all.
23178 2017-03-28 Richard Biener <rguenther@suse.de>
23181 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
23182 without arguments, generate default definition of a SSA name.
23184 2017-03-28 Richard Biener <rguenther@suse.de>
23186 PR middle-end/80222
23187 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
23188 TYPE_REF_CAN_ALIAS_ALL references.
23189 * fold-const.c (fold_indirect_ref_1): Likewise.
23191 2017-03-28 Martin Liska <mliska@suse.cz>
23194 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
23195 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
23197 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23198 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
23200 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
23201 (EXTRA_SPECS): Define.
23202 (SUBTARGET_EXTRA_SPECS): Likewise.
23203 (SUBTARGET_CPP_SPEC): Likewise.
23204 * config/arc/elf.h (EXTRA_SPECS): Renamed to
23205 SUBTARGET_EXTRA_SPECS.
23206 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
23208 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23210 * config/arc/simdext.md (vst64_insn): Update pattern.
23211 (vld32wh_insn): Likewise.
23212 (vld32wl_insn): Likewise.
23213 (vld64_insn): Likewise.
23214 (vld32_insn): Likewise.
23216 2017-03-28 Marek Polacek <polacek@redhat.com>
23219 * fold-const.c (fold_comparison): Use protected_set_expr_location
23220 instead of SET_EXPR_LOCATION.
23222 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
23224 * tree.c (add_expr): Avoid name lookup warning.
23226 2017-03-27 Jeff Law <law@redhat.com>
23228 PR tree-optimization/80216
23229 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
23230 function name. Limit recursion depth.
23231 (record_temporary_equivalences): Corresponding changes.
23233 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
23235 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
23238 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23241 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
23243 * cfgcleanup.c (reg_note_cfa_p): New array.
23244 (insns_have_identical_cfa_notes): New function.
23245 (old_insns_match_p): Don't cross-jump in between /f
23246 and non-/f instructions. If both i1 and i2 are frame related,
23247 verify all CFA notes, their order and content.
23249 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
23252 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
23253 HImode and SImode with zero extend to DImode to one insn.
23254 (bswap<mode>2_extenddi): Likewise.
23255 (bswapsi2_extenddi): Likewise.
23256 (bswaphi2_extendsi): Likewise.
23257 (bswaphi2): Combine bswap HImode and SImode into one insn.
23258 Separate memory insns from swapping register.
23259 (bswapsi2): Likewise.
23260 (bswap<mode>2): Likewise.
23261 (bswaphi2_internal): Delete, no longer used.
23262 (bswapsi2_internal): Likewise.
23263 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
23264 store, and gpr<-gpr swap insns.
23265 (bswap<mode>2_store): Likewise.
23266 (bswaphi2_reg): Register only splitter, combine with the splitter.
23267 (bswaphi2 splitter): Likewise.
23268 (bswapsi2_reg): Likewise.
23269 (bswapsi2 splitter): Likewise.
23270 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
23271 the insns into load, store, and register/register insns.
23272 (bswapdi2_ldbrx): Likewise.
23273 (bswapdi2_load): Likewise.
23274 (bswapdi2_store): Likewise.
23275 (bswapdi2_reg): Likewise.
23277 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
23279 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
23280 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
23282 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
23285 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
23287 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23288 special handling for target option conflicts between dform
23289 options (-mpower9-dform, -mpower9-dform-vector,
23290 -mpower9-dform-scalar) and -mno-direct-move.
23292 2017-03-27 Richard Biener <rguenther@suse.de>
23294 PR tree-optimization/80181
23295 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
23297 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23299 * config/arc/predicates.md (move_double_src_operand): Replace the
23300 call to move_double_src_operand with a call to address_operand.
23302 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23304 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
23305 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
23306 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
23308 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23310 * config/arc/predicates.md (long_immediate_loadstore_operand):
23311 Consider scaled addresses cases.
23313 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23315 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
23316 restored when in interrupt.
23317 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
23318 doesn't have delay slot.
23320 2017-03-27 Richard Biener <rguenther@suse.de>
23323 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
23324 inlined thunk clones.
23326 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23329 * asan.c (instrument_derefs): Copy over last operand from
23330 original COMPONENT_REF to the new COMPONENT_REF with
23331 DECL_BIT_FIELD_REPRESENTATIVE.
23332 * ubsan.c (instrument_object_size): Likewise.
23334 2017-03-27 Richard Biener <rguenther@suse.de>
23336 PR tree-optimization/80170
23337 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
23338 sure DR/SCEV didnt fold in constants we do not see when looking
23339 at the reference base alignment.
23341 2017-03-27 Richard Biener <rguenther@suse.de>
23343 PR middle-end/80171
23344 * gimple-fold.c (fold_ctor_reference): Properly guard against
23345 NULL return value from canonicalize_constructor_val.
23347 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
23350 * config/i386/i386.c (ix86_expand_builtin)
23351 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
23352 flags reg setting and flags reg using instructions.
23353 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
23354 clobbering instructions to zero extend op2.
23356 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
23358 * doc/install.texi (Configuration) <--with-aix-soname>:
23359 Update link to AIX ld.
23361 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
23363 PR rtl-optimization/80160
23364 PR rtl-optimization/80159
23365 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
23366 reg_alternate_class into account.
23368 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
23371 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
23372 to consider in curr_insn_transform.
23374 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23376 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
23377 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
23378 and emit_mode_inner.
23380 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23382 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
23383 argument to the overloaded builtin variants. Use the new flag to
23384 deprecate certain builtin variants.
23385 * config/s390/s390-builtin-types.def: Add new builtin types.
23386 * config/s390/s390-builtins.h: Support new flags field for
23387 overloaded builtins.
23388 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
23389 (s390_macro_to_expand): Enable vector float data type.
23390 (s390_cpu_cpp_builtins_internal): Indicate support of the new
23391 builtins by incrementing the __VEC__ version number.
23392 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
23394 (s390_resolve_overloaded_builtin): Emit error messages depending
23395 on the builtin flags.
23396 * config/s390/s390.c (s390_expand_builtin): Support additional
23397 flags argument. Change error message to match the messages
23398 emitted in s390-c.c.
23399 * config/s390/s390.md: New UNSPEC_* constants.
23400 (op_type): Add new instruction types.
23401 * config/s390/vecintrin.h: Add new builtins and test data class
23403 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
23404 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
23405 (VEC_INEXACT, VEC_NOINEXACT): New constants.
23406 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
23407 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
23408 ("vec_mergel<mode>"): V_HW -> VEC_HW.
23410 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
23411 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
23412 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
23413 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
23415 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
23416 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
23417 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
23418 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
23420 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
23421 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
23422 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
23423 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
23424 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
23425 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
23426 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
23428 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
23429 ("vec_scatter_element<V_HW_4:mode>_DI")
23430 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
23431 ("vec_fpint<mode>", "vflls")
23432 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
23433 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
23434 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
23435 ("*vec_cmphe<mode>_cc"): ... these.
23437 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
23438 mode constant instead of magic value.
23440 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23442 * config/s390/s390.c (s390_expand_vec_compare): Support other
23443 vector floating point modes than just V2DF.
23444 (s390_expand_vcond): Likewise.
23445 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
23446 (s390_cannot_change_mode_class): Prevent mode changes between TF
23447 and V1TF in vector registers.
23448 * config/s390/s390.md (DF, SF): New mode attributes.
23449 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
23450 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
23451 SFmode support for VRs.
23452 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
23454 (VFT, VF_HW): New mode iterators.
23455 (vw, sdx): New mode attributes.
23456 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
23457 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
23458 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
23459 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
23460 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
23461 also the new vector floating point modes. Renaming to ...
23463 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
23464 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
23465 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
23466 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
23467 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
23468 ("vec_unordered<mode>"): ... these.
23470 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
23471 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
23472 ("*vec_extendv2df"): New insn definitions.
23474 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23476 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
23477 ("mulditi3_2", "*muldi3_sign"): New patterns.
23478 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
23479 rename the pattern definition.
23481 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23483 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
23485 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
23487 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23489 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
23490 instruction if possible.
23491 * config/s390/vector.md (vec_halfnumelts): New mode
23493 ("*vec_vllezlf<mode>"): New pattern.
23495 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23497 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
23498 ("popcountv4si2", "popcountv2di2"): Rename to ...
23499 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
23500 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
23502 ("popcount<mode>2_vxe"): New pattern.
23504 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23506 * common/config/s390/s390-common.c (processor_flags_table): Add
23508 * config.gcc: Add arch12.
23509 * config/s390/driver-native.c (s390_host_detect_local_cpu):
23510 Default to arch12 for unknown CPU model numbers.
23511 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
23512 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
23513 PROCESSOR_max sanity check.
23514 * config/s390/s390-opts.h (enum processor_type): Add
23516 * config/s390/s390.c (processor_table): Add arch12.
23517 (s390_expand_builtin): Add check for B_VXE flag.
23518 (s390_issue_rate): Add PROCESSOR_ARCH12.
23519 (s390_get_sched_attrmask): Likewise.
23520 (s390_get_unit_mask): Likewise.
23521 (s390_sched_score): Enable z13 scheduling for arch12.
23522 (s390_sched_reorder): Likewise.
23523 (s390_sched_variable_issue): Likewise.
23524 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
23526 (s390_tune_attr): Use z13 scheduling also for arch12.
23527 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
23528 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
23529 (TARGET_VXE_P): New macros.
23530 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
23531 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
23532 * config/s390/s390.opt: Add arch12 as processor_type.
23534 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23536 * config/s390/s390.md
23537 ("fixuns_truncdddi2", "fixuns_trunctddi2")
23538 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
23539 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
23541 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
23542 Rename expanders to ...
23544 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
23545 ("fixuns_truncdddi2_emu"): ... these.
23547 ("fixuns_trunc<mode>si2_emu"): New expander.
23549 ("*fixuns_truncdfdi2_z13"): Rename to ...
23550 ("*fixuns_truncdfdi2_vx"): ... this.
23552 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23554 * config/s390/2964.md: Remove the single element vector compare
23555 instructions which are no longer used.
23556 * config/s390/s390.c (s390_select_ccmode): Remove handling of
23558 (s390_canonicalize_comparison): Remove handling of DFmode
23560 (s390_expand_vec_compare_scalar): Remove function.
23561 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
23562 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
23564 ("*cmp<mode>_ccs"): Add wfcdb instruction.
23566 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23568 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
23570 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
23571 will anyway by matched by mov<mode>_64dfp.
23573 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23575 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
23576 vlef/vstef. Add missing operand to vleif.
23578 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23580 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
23581 pair for all vector types with 64 bit elements.
23582 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
23583 * config/s390/vector.md (V_HW_64): ... here.
23584 (V_128_NOSINGLE): New mode iterator.
23585 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
23586 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
23587 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
23588 ("*vec_load_pairv2di"): Change to ...
23589 ("*vec_load_pair<mode>"): ... this one.
23591 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23593 * config/s390/constraints.md: Add comments.
23594 (jKK): Reject element sizes > 8 bytes.
23595 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
23597 * config/s390/s390.md: Add the s_operand checks formerly in
23598 s390_split_ok_p to various splitters where they are still
23600 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
23601 for 128 bit vectors. Plus two splitters.
23603 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23605 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
23608 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23611 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
23612 error if the boundary argument is not constant.
23614 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23616 PR rtl-optimization/80112
23617 * loop-doloop.c (doloop_condition_get): Don't check condition
23618 if cmp isn't SET with IF_THEN_ELSE src.
23620 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23622 PR tree-optimization/80158
23623 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
23624 replacing a candidate statement, also replace it for the
23625 candidate's alternate interpretation.
23626 (replace_rhs_if_not_dup): Likewise.
23627 (replace_one_candidate): Likewise.
23629 2017-03-24 Richard Biener <rguenther@suse.de>
23631 PR tree-optimization/80167
23632 * graphite-isl-ast-to-gimple.c
23633 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
23635 (translate_isl_ast_to_gimple::get_rename): Likewise.
23637 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23639 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
23640 handling of certain combinations of target options, including the
23641 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
23642 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
23644 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23647 * config/arm/arm.md (*load_multiple): Add reload_completed to
23648 matching condition.
23650 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23651 Richard Biener <rguenth@suse.de>
23653 PR tree-optimization/79908
23654 PR tree-optimization/80136
23655 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23656 been cast away, gimplify_and_add suffices.
23658 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
23660 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
23662 2017-03-23 Richard Biener <rguenther@suse.de>
23664 PR tree-optimization/80032
23665 * gimplify.c (gimple_push_cleanup): Forced unconditional
23666 cleanups still have to go to the conditional_cleanups
23669 2017-03-22 Jakub Jelinek <jakub@redhat.com>
23671 PR tree-optimization/80072
23672 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
23674 (next_operand_entry_id): Change type to unsigned int.
23675 (sort_by_operand_rank): Make sure to return the right return value
23676 even if unsigned fields are bigger than INT_MAX.
23677 (struct oecount): Change cnt and id type to unsigned int.
23678 (oecount_hasher::equal): Formatting fix.
23679 (oecount_cmp): Make sure to return the right return value
23680 even if unsigned fields are bigger than INT_MAX.
23681 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
23684 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
23685 TREE_READONLY on result if writing it more than once.
23688 * doc/invoke.texi (-fsanitize=thread): Document that with
23689 -fnon-call-exceptions atomics are not able to throw
23693 * tsan.c: Include tree-eh.h.
23694 (instrument_builtin_call): Call maybe_clean_eh_stmt or
23695 maybe_clean_or_replace_eh_stmt where needed.
23696 (instrument_memory_accesses): Add cfg_changed argument.
23697 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
23698 if it returned true.
23699 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
23701 PR rtl-optimization/63191
23702 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
23703 wrapper function, moved the whole old content into ...
23704 (ix86_delegitimize_address_1): ... this. New inline function.
23705 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
23706 true as last argument instead of ix86_delegitimize_address.
23708 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23710 * config/aarch64/aarch64.c (generic_branch_cost): Copy
23711 cortexa57_branch_cost.
23713 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23715 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
23717 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23720 * doc/md.texi (Constraints): Document wA constraint.
23721 * config/rs6000/constraints.md (wA): New.
23722 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
23723 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
23724 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
23725 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
23727 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
23730 * gimplify.c (is_oacc_declared): New function.
23731 (oacc_default_clause): Use it to set default flags for acc declared
23732 variables inside parallel regions.
23733 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
23734 declared variables.
23735 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
23736 declare attribute to any decl as necessary.
23738 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
23741 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
23742 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
23743 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
23744 (arm_arch_lpae): This.
23745 * config/arm/arm.c (arm_arch7ve): Rename into ...
23746 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
23747 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
23750 2017-03-22 Martin Liska <mliska@suse.cz>
23753 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
23754 error message instead of an ICE.
23756 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23758 * doc/extend.texi (6.11 Additional Floating Types): Revise.
23760 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
23762 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23764 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23767 2017-03-21 Martin Sebor <msebor@redhat.com>
23769 * doc/extend.texi: Use "cannot" instead of "can't."
23770 * doc/hostconfig.texi: Same.
23771 * doc/install.texi: Same.
23772 * doc/invoke.texi: Same.
23773 * doc/loop.texi: Same.
23774 * doc/md.texi: Same.
23775 * doc/objc.texi: Same.
23776 * doc/rtl.texi: Same.
23777 * doc/tm.texi: Same.
23778 * doc/tm.texi.in: Same.
23779 * doc/trouble.texi: Same.
23781 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
23784 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
23785 (collect_checksum_attributes): Set it.
23786 (die_checksum_ordered): Use it.
23788 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23790 PR tree-optimization/79908
23791 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
23792 change: For a VA_ARG whose LHS has been cast away, use
23793 force_gimple_operand to construct the side effects.
23795 2017-03-21 David Malcolm <dmalcolm@redhat.com>
23797 PR translation/80001
23798 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
23799 more amenable to translation.
23800 (oacc_loop_auto_partitions): Likewise.
23802 2017-03-21 Marek Polacek <polacek@redhat.com>
23803 Martin Sebor <msebor@redhat.com>
23805 PR tree-optimization/80109
23806 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
23807 on INTEGRAL_TYPE_P.
23809 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23810 Segher Boessenkool <segher@kernel.crashing.org>
23813 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
23814 check reg_used_between_p between insn and one of succ or succ2
23815 depending on if succ is artificial insn not inserted into insn
23818 2017-03-21 Martin Liska <mliska@suse.cz>
23820 PR gcov-profile/80081
23821 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
23822 * doc/gcc.texi: Include gcov-dump stuff.
23823 * doc/gcov-dump.texi: New file.
23825 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
23827 PR rtl-optimization/79150
23828 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
23829 conditional jump, if the jump is the last insn of the loop.
23831 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23832 Richard Biener <rguenth@suse.de>
23834 PR tree-optimization/79908
23835 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23836 been cast away, use force_gimple_operand to construct the side
23839 2017-03-21 Martin Liska <mliska@suse.cz>
23841 PR libfortran/79956
23842 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
23845 2017-03-21 Brad Spengler <spender@grsecurity.net>
23848 * plugin.c (htab_hash_plugin): New function.
23849 (add_new_plugin): Use it and adjust.
23850 (parse_plugin_arg_opt): Adjust.
23851 (init_one_plugin): Likewise.
23853 2017-03-21 Richard Biener <rguenther@suse.de>
23855 PR tree-optimization/80032
23856 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
23857 if set force the cleanup to happen unconditionally.
23858 (gimplify_target_expr): Push inserted clobbers with force_uncond
23859 to avoid them being removed by control-dependent DCE.
23861 2017-03-21 Richard Biener <rguenther@suse.de>
23863 PR tree-optimization/80122
23864 * tree-inline.c (copy_bb): Do not expans va-arg packs or
23865 va_arg_pack_len when the inlined call stmt requires pack
23867 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
23869 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23872 * tsan.c (instrument_builtin_call): If the memory model argument
23873 is not a constant, assume it is valid.
23876 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
23879 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23881 PR rtl-optimization/79910
23882 * combine.c (can_combine_p): Do not allow combining an I0 or I1
23883 if its dest is used by an insn before I2 (other than the combined
23884 insns themselves, which are properly handled already).
23886 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23889 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23891 * combine.c (record_used_regs): New static function.
23892 (try_combine): Handle situations where there is an additional
23893 instruction between I2 and I3 which needs to have a LOG_LINK
23897 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23899 * combine.c (try_combine): Delete redundant i1 test. Call
23900 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23902 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23905 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
23906 alternatives 13/14.
23908 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23910 PR tree-optimization/80054
23911 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
23912 the optimization if a PHI or any of its arguments is not dominated
23913 by the candidate's basis. Use gphi* rather than gimple* as
23915 (replace_profitable_candidates): Clean up a gimple* variable that
23916 should be a gphi* variable.
23918 2017-03-20 Martin Sebor <msebor@redhat.com>
23921 * doc/extend.texi (attribute constructor): Document present limitation.
23923 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
23926 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
23927 __POWER9_VECTOR__ #ifdef control, change template definition to
23928 use Power9-specific built-in function.
23929 (vec_any_eq): Likewise.
23930 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
23931 to control outcomes from this test.
23932 (vector_ae_<mode>p): For VEC_F modes, likewise.
23934 2017-03-20 Ian Lance Taylor <iant@google.com>
23936 * config/i386/i386.c (ix86_function_regparm): Save an extra
23937 register for -fsplit-stack with DECL_STATIC_CHAIN.
23939 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23942 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
23943 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
23945 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23947 * config/riscv/riscv.c (riscv_print_operand): Use "fence
23949 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
23952 2017-03-20 Marek Polacek <polacek@redhat.com>
23955 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
23957 2017-03-20 Richard Biener <rguenther@suse.de>
23959 PR tree-optimization/80113
23960 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
23961 allocate extra SSA name for PHI def.
23962 (add_close_phis_to_outer_loops): Likewise.
23963 (add_close_phis_to_merge_points): Likewise.
23964 (copy_loop_close_phi_args): Likewise.
23965 (copy_cond_phi_nodes): Likewise.
23967 2017-03-20 Martin Liska <mliska@suse.cz>
23969 PR middle-end/79753
23970 * tree-chkp.c (chkp_build_returned_bound): Do not build
23971 returned bounds for a LHS that's not a BOUNDED_P type.
23973 2017-03-20 Martin Liska <mliska@suse.cz>
23977 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
23978 COMPLEX_CST and VECTOR_CST.
23980 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23983 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
23984 target operand. A new splitter adds the clobber statement in case
23985 the target operand is dead anyway.
23987 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
23989 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
23990 to age-old versions of binutils and glibc.
23992 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
23994 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
23996 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
23998 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24000 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24002 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24003 requirement for binutils 2.13.
24005 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24007 * combine.c (try_combine): Delete redundant i1 test. Call
24008 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24010 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24012 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24013 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24015 <riscv64-*-elf>: Re-arrange section
24016 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24017 <riscv32-*-linux>: Likewise.
24018 <riscv64-*-elf>: Likewise
24019 <riscv64-*-linux>: Likewise.
24021 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24024 * aarch64.opt(verbose-cost-dump): Fix typo.
24026 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24029 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24030 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24032 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24034 * reload.c (find_reloads): When reloading a nonoffsettable address,
24035 use RELOAD_OTHER for it and its address reloads.
24037 PR rtl-optimization/79910
24038 * combine.c (record_used_regs): New static function.
24039 (try_combine): Handle situations where there is an additional
24040 instruction between I2 and I3 which needs to have a LOG_LINK
24043 2017-03-17 Jeff Law <law@redhat.com>
24045 PR tree-optimization/71437
24046 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24047 conditional in the hash table first.
24048 (vrp_dom_walker::before_dom_children): Extract condition from
24049 ASSERT_EXPR. Record condition, its inverion and any implied
24050 conditions as well.
24052 2017-03-17 Marek Polacek <polacek@redhat.com>
24053 Markus Trippelsdorf <markus@trippelsdorf.de>
24055 PR tree-optimization/80079
24056 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24059 2017-03-17 Richard Biener <rguenther@suse.de>
24061 PR middle-end/80075
24062 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24063 Properly verify the LHS before the RHS possibly claims to be
24065 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24068 2017-03-17 Martin Jambor <mjambor@suse.cz>
24070 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24071 (List of -O2 options): Likewise.
24072 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24075 2017-03-17 Tom de Vries <tom@codesourcery.com>
24077 * gcov-dump.c (print_usage): Print bug_report_url.
24079 2017-03-17 Richard Biener <rguenther@suse.de>
24081 PR middle-end/80050
24082 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
24083 (parser::peek): Likewise.
24085 2017-03-17 Richard Biener <rguenther@suse.de>
24087 PR tree-optimization/80048
24088 * sese.c (free_sese_info): Properly release rename_map and
24089 copied_bb_map elements.
24091 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
24093 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
24094 Add linked-list forward and backlinks. Insert on
24095 construction, remove on destruction.
24096 (class pass_store_merging): Add m_stores_head field.
24097 (pass_store_merging::terminate_and_process_all_chains):
24098 Iterate over m_stores_head list.
24099 (pass_store_merging::terminate_all_aliasing_chains):
24101 (pass_store_merging::execute): Check for debug stmts first.
24102 Push new chains onto the m_stores_head stack.
24104 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24107 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
24108 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
24109 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
24111 2017-03-16 Jeff Law <law@redhat.com>
24113 PR tree-optimization/71437
24114 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
24115 member function. Implementation moved into after_dom_children
24116 member function and into the threader's thread_outgoing_edges
24118 (dom_opt_dom_walker::after_dom_children): Simplify by moving
24119 some code into new thread_outgoing_edges.
24120 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
24121 definition. Simplify marker handling (do it here). Assume we always
24122 have the available expression and the const/copies tables.
24123 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
24125 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
24126 * tree-vrp.c (equiv_stack): No longer file scoped.
24127 (vrp_dom_walker): New class.
24128 (vrp_dom_walker::before_dom_children): New member function.
24129 (vrp_dom_walker::after_dom_children): Likewise.
24130 (identify_jump_threads): Setup domwalker. Use it rather than
24131 walking edges in a random order by hand. Simplify setup/finalization.
24132 (finalize_jump_threads): Remove.
24133 (vrp_finalize): Do not call identify_jump_threads here.
24134 (execute_vrp): Do it here instead and call thread_through_all_blocks
24137 PR tree-optimization/71437
24138 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
24140 (simplify_stmt_for_jump_threading): Add basic_block argument. All
24142 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
24143 (dom_opt_dom_walker::thread_across_edge): Remove
24144 handle_dominating_asserts argument. All callers changed.
24145 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
24146 changes. Remove calls to lhs_of_dominating_assert. Other
24147 uses of handle_dominating_asserts turn into unconditional code
24148 (simplify_control_stmt_condition_1): Likewise.
24149 (simplify_control_stmt_condition): Likewise.
24150 (thread_through_normal_block, thread_across_edge): Likewise.
24151 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
24152 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
24153 object if it is not an SSA_NAME.
24154 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
24155 before calling into the VRP specific simplifiers.
24156 (identify_jump_threads): Remove handle_dominating_asserts
24159 2017-03-16 Jakub Jelinek <jakub@redhat.com>
24162 * tree-diagnostic.c (default_tree_printer): No longer static.
24163 * tree-diagnostic.h (default_tree_printer): New prototype.
24165 2017-03-16 Tamar Christina <tamar.christina@arm.com>
24167 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
24168 Change ins into fmov.
24170 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24172 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
24173 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
24174 Use h_con constraint for operand 1.
24175 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
24176 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
24178 2017-03-15 Jeff Law <law@redhat.com>
24180 PR tree-optimization/71437
24181 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24182 (record_temporary_equivalences): Use it.
24184 PR tree-optimization/71437
24185 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
24186 tree-ssa-scopedtables.
24187 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24188 (record_conditions, record_cond, vuse_eq): Likewise.
24189 (record_edge_info): Adjust to API tweak of record_conditions.
24190 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
24191 (record_temporary_equivalences, optimize_stmt): Likewise.
24192 (eliminate_redundant_computations): Likewise.
24193 (record_equivalences_from_stmt): Likewise.
24194 * tree-ssa-scopedtables.c: Include options.h and params.h.
24195 (vuse_eq): New function, moved from tree-ssa-dom.c
24196 (build_and_record_new_cond): Likewise.
24197 (record_conditions): Likewise. Accept vector of conditions rather
24198 than edge_equivalence structure for first argument.
24199 for the first argument.
24200 (avail_exprs_stack::lookup_avail_expr): New member function, moved
24201 from tree-ssa-dom.c.
24202 (avail_exprs_stack::record_cond): Likewise.
24203 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
24204 from tree-ssa-dom.c.
24205 (avail_exprs_stack): Add new member functions lookup_avail_expr
24207 (record_conditions): Declare.
24209 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
24212 * lra-constraints.c (process_alt_operands): Increase reject for
24213 reloading an input/output operand.
24215 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
24218 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
24219 insns to convert from signed/unsigned char/short to IEEE 128-bit
24221 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
24223 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
24226 * config/i386/i386.c (ix86_vector_duplicate_value): Create
24227 subreg of inner mode for values already in registers.
24229 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
24231 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
24232 iteration reg is used after the loop.
24234 2017-03-14 Martin Sebor <msebor@redhat.com>
24236 PR tree-optimization/79800
24237 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
24238 precision in negative-positive range.
24239 (format_floating): Call non-const overload with adjusted precision.
24241 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24244 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
24247 2017-03-14 Martin Sebor <msebor@redhat.com>
24249 PR middle-end/80020
24250 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
24251 * builtins.def (aligned_alloc): Use it.
24254 * Makefile.in (GTFILES): Add calls.c.
24255 * calls.c: Include "gt-calls.h".
24257 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
24259 PR rtl-optimization/79728
24260 * regs.h (struct target_regs): New field
24261 x_contains_allocatable_regs_of_mode.
24262 (contains_allocatable_regs_of_mode): New macro.
24263 * reginfo.c (init_reg_sets_1): Initialize it, and change
24264 contains_reg_of_mode so it includes global regs as well.
24265 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
24266 rather than contains_regs_of_mode.
24268 2017-03-14 Martin Liska <mliska@suse.cz>
24270 * doc/invoke.texi: Document options that can't be combined with
24271 -fcheck-pointer-bounds.
24273 2017-03-14 Martin Liska <mliska@suse.cz>
24275 PR middle-end/79831
24276 * doc/invoke.texi (-Wchkp): Document the option.
24278 2017-03-14 Martin Liska <mliska@suse.cz>
24280 * Makefile.in: Install gcov-dump.
24282 2017-03-14 Martin Liska <mliska@suse.cz>
24284 * multiple_target.c (expand_target_clones): Bail out for
24285 an invalid attribute.
24287 2017-03-14 Richard Biener <rguenther@suse.de>
24289 * alias.c (struct alias_set_entry): Pack properly.
24290 * cfgloop.h (struct loop): Likewise.
24291 * cse.c (struct set): Likewise.
24292 * ipa-utils.c (struct searchc_env): Likewise.
24293 * loop-invariant.c (struct invariant): Likewise.
24294 * lra-remat.c (struct cand): Likewise.
24295 * recog.c (struct change_t): Likewise.
24296 * rtl.h (struct address_info): Likewise.
24297 * symbol-summary.h (function_summary): Likewise.
24298 * tree-loop-distribution.c (struct partition): Likewise.
24299 * tree-object-size.c (struct object_size_info): Likewise.
24300 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
24301 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
24302 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
24303 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
24304 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
24305 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
24306 (struct _stmt_vec_info): Likewise.
24308 2017-03-14 Martin Liska <mliska@suse.cz>
24311 * multiple_target.c (create_dispatcher_calls): Check that
24312 a target can create a function dispatcher.
24314 2017-03-14 Martin Liska <mliska@suse.cz>
24317 * multiple_target.c (expand_target_clones): Drop local.local
24318 flag for default implementation.
24320 2017-03-14 Richard Biener <rguenther@suse.de>
24322 PR tree-optimization/80030
24323 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
24325 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
24327 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
24328 gcc_fallthrough() instead of __attribute__((fallthrough));
24330 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24332 * doc/gcc.texi: Remove "up" link to (DIR).
24333 * doc/gccint.texi: Ditto.
24335 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24337 * doc/install.texi (Specific) <avr>: Remove reference to
24340 2017-03-13 Jeff Law <law@redhat.com>
24342 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
24343 attribute rather than comments.
24345 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
24346 match_scratch operand is highest.
24348 2017-03-13 Martin Liska <mliska@suse.cz>
24350 PR middle-end/78339
24351 * ipa-pure-const.c (warn_function_noreturn): If the declarations
24352 is a CHKP clone, use original declaration.
24354 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24356 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
24357 (arc_conditional_register_usage): Use a different allocation order
24358 when optimizing for size.
24359 * common/config/arc/arc-common.c (arc_option_optimization_table):
24360 Section anchors default on when optimizing for size.
24362 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24364 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
24366 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24368 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
24369 * config/arc/arc.md (cpu_facility): Add cd variant.
24370 (*movqi_insn): Add code density variant.
24371 (*movhi_insn): Likewise.
24372 (*movqi_insn): Likewise.
24373 (*addsi3_mixed): Likewise.
24374 (subsi3_insn): Likewise.
24376 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24378 * config/arc/arc.md (movsi_cond_exec): Update constraint.
24380 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24382 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
24383 expressions with MINUS and UNARY ops.
24385 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24388 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
24390 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
24391 between vec_select and vector argument.
24392 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
24393 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
24394 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
24395 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
24396 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
24397 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
24399 2017-03-13 Richard Biener <rguenther@suse.de>
24402 * params.def (vect-max-peeling-for-alignment): Fix typo.
24404 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24406 * doc/install.texi (Specific) <mips-*-*>: Remove description of
24407 issue that only occurred with binutils below 2.18.
24409 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24411 * doc/install.texi (Specific) <cris-axis-elf>: No longer
24412 refer to binutils 2.11/2.12 minimum.
24414 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24416 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
24417 ftp.kernel.org and simplify binutils requirement.
24419 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
24421 * doc/invoke.texi (Warning Options): Fix spelling of link-time
24423 (Optimize Options): Ditto. Also remove redundancy.
24425 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24427 PR translation/79848
24428 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
24430 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
24431 to G_ to avoid double translation.
24433 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24435 PR translation/79923
24436 * auto-profile.c (get_combined_location): Convert leading
24437 character of diagnostics to lower case and remove trailing period.
24438 (read_profile): Likewise for various diagnostics.
24439 * config/arm/arm.c (arm_option_override): Remove trailing period
24440 from various diagnostics.
24441 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
24442 (msp430_expand_delay_cycles): Likewise.
24444 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24447 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
24448 full command-line argument, rather than just "str".
24449 (aarch64_validate_march): Likewise.
24450 (aarch64_validate_mtune): Likewise.
24452 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
24454 PR rtl-optimization/78911
24455 * lra-assigns.c (must_not_spill_p): New function.
24456 (spill_for): Use it.
24458 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24460 PR tree-optimization/79981
24461 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
24462 ATOMIC_COMPARE_EXCHANGE ifn result.
24463 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
24464 IFN_ATOMIC_COMPARE_EXCHANGE.
24466 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24469 * opts.c (parse_sanitizer_options): Add missing question mark to
24470 "did you mean" message.
24472 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24474 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
24476 (VMULEUH_UNS): Likewise.
24477 (VMULOUB_UNS): Likewise.
24478 (VMULOUH_UNS): Likewise.
24479 * config/rs6000/rs6000.c (builtin_function_type): Remove
24480 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
24482 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24485 * read-rtl-function.c (function_reader::read_rtx_operand): Update
24486 x with result of extra_parsing_for_operand_code_0.
24487 (function_reader::extra_parsing_for_operand_code_0): Convert
24488 return type from void to rtx, returning x. When reading
24489 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
24490 larger size containing struct block_symbol.
24492 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
24494 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
24495 -mfloat128-hardware without -m64.
24497 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
24500 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
24501 entries to the case statement that marks unsigned arguments to
24502 overloaded functions.
24504 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24506 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
24507 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
24509 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
24512 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
24513 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
24515 2017-03-10 Martin Liska <mliska@suse.cz>
24519 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
24520 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
24523 2017-03-10 Olivier Hainque <hainque@adacore.com>
24525 * tree-switch-conversion (array_value_type): Start by resetting
24526 candidate type to it's main variant.
24528 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24530 PR rtl-optimization/79909
24531 * combine.c (try_combine): Use simplify_replace_rtx on individual
24532 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
24533 of the whole CALL_INSN_FUNCTION_USAGE.
24535 PR tree-optimization/79972
24536 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
24537 get_range_info on SSA_NAMEs. Formatting fixes.
24539 2017-03-10 Richard Biener <rguenther@suse.de>
24540 Jakub Jelinek <jakub@redhat.com>
24542 PR tree-optimization/77975
24543 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
24544 edge to be constant.
24545 (get_val_for): For constant x return it. Formatting fix.
24546 (loop_niter_by_eval): Avoid pointless looping if the next iteration
24547 would use the same bases as the current one.
24549 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24551 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
24552 instead of vec_select for V1TImode.
24553 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
24555 (VSX_LE_128): Add V1TI to this mode iterator.
24556 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
24557 (*vsx_le_perm_store_<mode>): Likewise.
24558 (pre-reload splitter for VSX stores): Likewise.
24559 (post-reload splitter for VSX stores): Likewise.
24560 (*vsx_xxpermdi2_le_<mode>): Likewise.
24561 (*vsx_lxvd2x2_le_<mode>): Likewise.
24562 (*vsx_stxvd2x2_le_<mode>): Likewise.
24564 2017-03-09 Michael Eager <eager@eagercon.com>
24566 Correct failures with --enable-checking=yes,rtl.
24568 * config/microblaze/microblaze.c (microblaze_expand_shift):
24569 Replace GET_CODE test with CONST_INT_P and INTVAL test with
24570 test for const0_rtx.
24571 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
24572 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
24574 2017-03-09 Richard Biener <rguenther@suse.de>
24576 PR tree-optimization/79977
24577 * graphite-scop-detection.c (scop_detection::merge_sese):
24578 Handle the case of extra exits to blocks dominating the entry.
24580 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
24582 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24585 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
24587 PR rtl-optimization/79949
24588 * lra-constraints.c (process_alt_operands): Check memory when
24589 trying to predict a cycle. Print about the overall increase.
24591 2017-03-09 Richard Biener <rguenther@suse.de>
24593 PR middle-end/79971
24594 * gimple-expr.c (useless_type_conversion_p): Preserve
24595 TYPE_SATURATING for fixed-point types.
24597 2017-03-09 Richard Biener <rguenther@suse.de>
24600 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
24601 alignment of BLKmode params.
24603 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24606 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
24607 (VALL_NO_V2Q): Likewise.
24609 * config/aarch64/aarch64-simd.md
24610 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
24612 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
24613 VALL_NO_V2Q mode iterator.
24614 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
24616 2017-03-09 Martin Liska <mliska@suse.cz>
24618 PR tree-optimization/79631
24619 * tree-chkp-opt.c (chkp_is_constant_addr): Call
24620 tree_int_cst_sign_bit just for INTEGER constants.
24622 2017-03-09 Martin Liska <mliska@suse.cz>
24626 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
24629 2017-03-09 Marek Polacek <polacek@redhat.com>
24632 * tree.c (inchash::add_expr): Handle TREE_VEC.
24634 2017-03-09 Martin Liska <mliska@suse.cz>
24637 (chkp_narrow_size_and_offset): New function.
24638 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24639 (void chkp_parse_bit_field_ref): New function.
24640 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
24641 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
24643 2017-03-09 Martin Liska <mliska@suse.cz>
24646 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
24647 (chkp_find_bounds_1): Remove gcc_unreachable.
24649 2017-03-09 Jakub Jelinek <jakub@redhat.com>
24652 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
24653 BUILT_IN_SYNC*, determine the access type from the size suffix and
24654 always build a MEM_REF with that type. Handle forgotten
24655 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
24658 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
24659 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
24660 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
24661 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
24662 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
24663 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
24664 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
24665 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
24666 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
24667 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
24668 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
24669 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
24670 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
24671 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
24672 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
24673 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
24674 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
24675 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
24676 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
24677 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
24678 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
24679 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
24680 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
24681 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
24682 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
24683 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
24684 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
24685 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
24686 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
24687 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
24688 definitions outside of __OPTIMIZE__ guarded section.
24691 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
24692 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
24693 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
24694 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
24697 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24699 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
24700 ("vfenez<mode>"): Add missing constraints.
24702 2017-03-08 Martin Sebor <msebor@redhat.com>
24705 * config/nds32/nds32.c (nds32_option_override):
24706 Fix misspelled diagnostic.
24708 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24711 * gimplify.c (gimplify_omp_for): Replace index var in outer
24712 taskloop statement with an artificial variable and add
24713 OMP_CLAUSE_PRIVATE clause for it.
24715 2017-03-08 Richard Biener <rguenther@suse.de>
24717 PR tree-optimization/79955
24718 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
24719 for accesses that are completely outside of the variable.
24721 2017-03-08 Andrew Haley <aph@redhat.com>
24723 PR tree-optimization/79943
24724 * tree-ssa-loop-split.c (compute_new_first_bound): When
24725 calculating the new upper bound, (END-BEG) should be added, not
24728 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24730 * config/avr/avr.md (setmemhi): Make sure match_dup
24731 operand number comes before match_scratch.
24733 2017-03-08 Richard Biener <rguenther@suse.de>
24735 PR tree-optimization/79920
24736 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
24737 with ncopies == 1 to ...
24738 (vect_transform_slp_perm_load): ... here. Properly compute
24739 all element loads by iterating VF times over the group. Do
24740 not handle ncopies (computed in a broken way) in
24741 vect_create_mask_and_perm.
24743 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24746 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
24747 is a uniform vector, use uniform_vector_p return value instead of
24748 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
24750 2017-03-07 Marek Polacek <polacek@redhat.com>
24752 PR middle-end/79809
24753 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
24754 (alloca_call_type): Likewise.
24756 2017-03-07 Martin Liska <mliska@suse.cz>
24758 * gcov.c (process_args): Put comment to correct location.
24760 2017-03-07 Martin Liska <mliska@suse.cz>
24762 PR middle-end/68270
24763 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
24764 Use array_at_struct_end_p instead of DECL_CHAIN (field).
24765 (chkp_narrow_bounds_for_field): Likewise.
24766 (chkp_parse_array_and_component_ref): Pass one more argument to
24769 2017-03-07 Richard Biener <rguenther@suse.de>
24771 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
24774 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24776 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
24777 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
24779 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24782 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
24783 to end of description.
24784 (PARAM_MAX_STORES_TO_MERGE): Likewise.
24786 2017-03-07 Jakub Jelinek <jakub@redhat.com>
24788 PR rtl-optimization/79901
24789 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
24791 (*avx512f_<code><mode>3<mask_name>): ... this.
24792 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
24793 iterator instead of VI8_AVX2_AVX512BW.
24795 PR rtl-optimization/79901
24796 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
24797 min/max expander, expand it using expand_vec_cond_expr.
24800 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
24803 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24806 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
24807 to void * for PCH reasons.
24808 * dwarf2out.c (output_loc_operands, output_die): Cast
24809 v.val_vec.array to unsigned char *.
24811 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
24814 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
24817 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
24819 PR rtl-optimization/79571
24820 * lra-constraints.c (process_alt_operands): Calculate static
24821 reject and subtract it from overall when only addresses will be
24824 2017-03-06 Julia Koval <julia.koval@intel.com>
24827 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
24828 incoming stack boundary to 128 for 64-bit targets.
24830 2017-03-06 Richard Biener <rguenther@suse.de>
24832 PR tree-optimization/79894
24833 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
24834 to NULL after folding it.
24836 2017-03-06 Richard Biener <rguenther@suse.de>
24838 PR tree-optimization/79824
24839 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
24840 check disabling peeling for gaps.
24842 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
24844 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24845 attributes): Document gettimeofday.
24847 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24849 * config/s390/s390.c (s390_option_override_internal): Set
24850 PARAM_MIN_VECT_LOOP_BOUND
24852 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24854 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
24855 * config/s390/s390.md: Likewise.
24857 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24860 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
24861 (<avx2_avx512>_perm<mode>): Rename to ...
24862 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
24864 (<avx512>_perm<mode>_mask): Rename to ...
24865 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
24867 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
24868 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
24869 instead of VI8F_256_512.
24870 (avx512f_perm<mode>): New define_expand.
24871 (avx512f_perm<mode>_mask): Likewise.
24872 (avx512f_perm<mode>_1<mask_name>): New define_insn.
24873 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
24875 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24877 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
24878 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
24880 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
24882 2017-03-06 Martin Liska <mliska@suse.cz>
24885 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
24886 when having a SSA NAME w/o VAR_DECL assigned to it.
24888 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24890 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
24891 msa_dpsub_<su>_d): Fix MODE for vec_select.
24893 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24895 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
24897 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
24899 2017-03-06 Richard Biener <rguenther@suse.de>
24901 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
24902 * plugin.c (register_plugin_info): Likewise.
24903 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
24905 2017-03-05 Jakub Jelinek <jakub@redhat.com>
24907 * config/i386/sse.md (sse_storehps, sse_storelps,
24908 avx_<castmode><avxsizesuffix>_<castmode>,
24909 avx512f_<castmode><avxsizesuffix>_<castmode>,
24910 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
24911 in condition that at least one operand is not a MEM.
24913 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24915 PR middle-end/79805
24916 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
24917 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
24919 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
24920 gimple_call_nothrow_p flag based on whether original builtin can throw.
24921 If it can, emit following stmts on the fallthrough edge.
24922 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
24923 don't create new bb if inserting just debug stmts on the edge, try to
24924 insert them on the fallthru bb or just reset debug stmts.
24926 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
24929 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
24930 restore recog_data (including the operand rtxes inside it) around
24931 the call to get_insn_template.
24933 2017-03-03 Martin Sebor <msebor@redhat.com>
24935 PR tree-optimization/79699
24936 * context.c (context::~context): Free MPFR caches to avoid
24937 a memory leak on program exit.
24939 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24941 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
24942 Use wide_int::ulow () instead of .elt (0).
24944 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24946 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
24947 (*pushxf): Limit oF constraint to 32bit targets and add oC
24948 constraint for 64bit targets.
24949 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
24950 (*pushdf): Change rmF constraint to rmC.
24952 2017-03-03 Martin Liska <mliska@suse.cz>
24954 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
24955 Remove unused variable.
24957 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24960 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
24961 is a memory operand, increase num_memory.
24962 (ix86_expand_args_builtin): Likewise.
24964 2017-03-03 Jan Hubicka <jh@suse.cz>
24967 * ipa-devirt.c (maybe_record_node): Properly handle
24968 __cxa_pure_virtual visibility.
24970 2017-03-03 Martin Liska <mliska@suse.cz>
24972 PR tree-optimization/79803
24973 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
24975 (pass_loop_prefetch::execute): Disabled optimization if an
24976 assumption about L1 cache size is not met.
24978 2017-03-03 Martin Liska <mliska@suse.cz>
24980 PR rtl-optimization/79574
24981 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
24982 (hash_scan_set): Likewise.
24983 (dump_hash_table): Likewise.
24984 (hoist_code): Likewise.
24986 2017-03-03 Richard Biener <rguenther@suse.de>
24988 * fixed-value.c (fixed_from_string): Restore use of elt (1)
24989 in place of uhigh ().
24990 (fixed_convert_from_real): Likewise.
24992 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24995 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
24997 2017-03-03 Richard Biener <rguenther@suse.de>
24999 PR middle-end/79818
25000 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25001 TYPE_OVERFLOW_UNDEFINED check.
25003 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25005 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25007 (vector_ae_<mode>_p): Likewise.
25008 (vector_nez_<mode>_p): Likewise.
25009 (vector_ne_v2di_p): Likewise.
25010 (vector_ae_v2di_p): Likewise.
25011 (vector_ne_<mode>_p): Likewise.
25012 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25014 (vsx_tsqrt<mode>2_fe): Likewise.
25016 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25019 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25021 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25023 PR rtl-optimization/79780
25024 * cprop.c (one_cprop_pass): When second and further conditional trap
25025 in a single basic block is turned into an unconditional trap, turn it
25026 into a deleted note to avoid RTL verification failures.
25028 2017-03-02 Richard Biener <rguenther@suse.de>
25030 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25032 2017-03-02 Richard Biener <rguenther@suse.de>
25034 PR tree-optimization/79345
25036 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25037 param and abort the walk, returning -1 if it is hit.
25038 (walk_aliased_vdefs): Take a limit param and pass it on.
25039 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25040 defaulting to 0 and return a signed int.
25041 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25042 (check_defs): New helper.
25043 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25044 about uninitialized memory.
25045 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25046 bogus uninitialized warning.
25047 (fixed_convert_from_real): Likewise.
25049 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25051 PR tree-optimization/66768
25052 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25053 iv_use if base object can't be determined.
25055 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25057 PR tree-optimization/79345
25058 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25059 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25060 (get_pattern_stats): Initialize it.
25061 * genemit.c (gen_expand): Verify match_scratch numbers come after
25062 match_operand/match_dup numbers.
25063 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25064 match_scratch numbers.
25065 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25067 * config/s390/s390.md (trunctdsd2): Likewise.
25069 2017-03-02 Richard Biener <rguenther@suse.de>
25071 * wide-int.h (wide_int_storage::operator=): Implement in terms
25074 2017-03-02 Richard Biener <rguenther@suse.de>
25076 PR tree-optimization/79777
25077 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25078 the to insert expression to sth existing.
25080 2017-03-01 Martin Sebor <msebor@redhat.com>
25082 PR middle-end/79692
25083 * gimple-ssa-sprintf.c
25084 (directive::known_width_and_precision): New function.
25085 (format_integer): Use it.
25086 (get_mpfr_format_length): Consider the full range of precision
25087 when computing %g output with the # flag. Set the likely byte
25088 count to 3 rather than 1 when precision is indeterminate.
25089 (format_floating): Correct the lower bound of precision.
25091 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25093 * doc/invoke.texi: Document default code model for 64-bit Linux.
25095 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25098 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
25099 udiv rather than div since input pattern is unsigned.
25101 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
25103 * config/i386/i386.c (print_reg): Warn for values of
25104 unsupported size in integer register.
25106 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
25109 * config/rs6000/predicates.md (current_file_function_operand): Do
25110 not allow self calls to be local if the function is replaceable.
25112 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
25115 * config/rs6000/altivec.h (vec_ctz and others): Change the
25116 preprocessor macro that controls conditional compilation from
25117 _ARCH_PWR9 to __POWER9_VECTOR__.
25118 (vec_all_ne): Change parameterization of __altivec_scalar_pred
25119 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
25120 control (instead of _ARCH_PWR9 control) so that template
25121 definition uses power9-specific function.
25122 (vec_any_eq): Likewise.
25123 (vec_all_ne): Change macro definition to use a power9-specific
25124 expansion under #ifdef __POWER9_VECTOR__ control (instead of
25125 _ARCH_PWR9 control).
25126 (vec_any_eq) Likewise.
25127 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
25128 expansion for CMPNEF to remove support for xvcmpnesp instruction.
25129 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
25130 support for xvcmpnedp instruction.
25131 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
25132 macro expansion so that Power9 implementation of vec_all_ne does
25133 not use the AltiVec predicate framework.
25134 (VCMPNEH_P): Likewise.
25135 (VCMPNEW_P): Likewise.
25136 (VCMPNED_P): Likewise.
25137 (VCMPNEFP_P): Likewise.
25138 (VCMPNEDP_P): Likewise.
25139 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
25140 implementation of vec_any_eq to not use AltiVec predicate
25142 (VCMPAEH_P): Likewise.
25143 (VCMPAEW_P): Likewise.
25144 (VCMPAED_P): Likewise.
25145 (VCMPAEFP_P): Likewise.
25146 (VCMPAEDP_P): Likewise.
25147 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
25148 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
25149 not use the AltiVec predicate framework.
25150 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
25151 of vec_any_eq to not use AltiVec predicate framework.
25152 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25153 support for predefined __POWER9_VECTOR__ macro to indicate that
25154 Power9 instruction selection is enabled.
25155 (altivec_overloaded_builtins): Remove extraneous
25156 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
25157 function argument types RS6000_BTI_bool_V16QI and
25158 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
25159 entry for overloaded function argument types RS6000_BTI_bool_V4SI
25160 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
25161 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
25162 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
25163 Power9 for implementations of vec_cmpne. Change the signature for
25164 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
25165 (representing vec_all_ne) to remove the previously described first
25166 argument of type RS6000_BTI_INTSI, as this was an artifact of
25167 reliance on the AltiVec predicate framework, which is no longer
25168 used in the implementation of these functions. Add
25169 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
25170 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
25171 since, unlike the AltiVec predicate framework implementation, we
25172 do not share function descriptors between vec_alle and vec_anyeq.
25173 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
25174 set of modes that receive special treatment even when
25175 TARGET_P9_VECTOR is true. The special treatment emits code that
25176 does not depend on Power9 instructions.
25177 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
25178 define_expand to not rely on AltiVec predicate framework.
25179 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25181 (vector_ne_v2di_p): Change this define_expand to not rely on
25182 AltiVec predicate framework.
25183 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
25185 (vector_ne_<mode>_p): Change this define_expand to not rely on
25186 AltiVec predicate framework.
25187 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25189 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
25190 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
25191 define_insn pattern.
25192 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
25193 define_insn pattern because the xvcmpne<VSs>. instruction is not
25195 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
25196 instruction is not supported.
25198 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25200 * config/nvptx/nvptx.c: Include intl.h.
25202 2017-03-01 Martin Jambor <mjambor@suse.cz>
25205 * ipa-prop.h (ipa_bits): Removed field known.
25206 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
25207 to pointers. Adjusted their comments to warn about their sharing.
25208 (ipcp_transformation_summary): Change bits to a vector of pointers.
25209 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
25210 (ipa_get_ipa_bits_for_value): Declare.
25211 * tree-vrp.h (value_range): Mark as GTY((for_user)).
25212 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
25213 (ipa_bits_hash_table): Likewise.
25214 (ipa_vr_ggc_hash_traits): Likewise.
25215 (ipa_vr_hash_table): Likewise.
25216 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
25217 being pointers and vr_known being removed.
25218 (ipa_set_jf_unknown): Likewise.
25219 (ipa_get_ipa_bits_for_value): New function.
25220 (ipa_set_jfunc_bits): Likewise.
25221 (ipa_get_value_range): New overloaded functions.
25222 (ipa_set_jfunc_vr): Likewise.
25223 (ipa_compute_jump_functions_for_edge): Use the above functions to
25224 construct bits and vr parts of jump functions.
25225 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
25226 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25228 (ipcp_grow_transformations_if_necessary): Also allocate
25229 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25231 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
25232 them. Fix too long lines.
25233 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
25234 vr_known being removed.
25235 (ipa_read_jump_function): Use new setter functions to construct bits
25236 and vr parts of jump functions or set them to NULL.
25237 (write_ipcp_transformation_info): Adjust for bits being pointers.
25238 (read_ipcp_transformation_info): Likewise.
25239 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
25241 Include gt-ipa-prop.h.
25242 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
25244 (ipcp_store_bits_results): Likewise.
25245 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
25246 Do not write to existing jump functions but use a temporary instead.
25248 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25251 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
25252 attempt to use its first operand as BIT_FIELD_REF base.
25254 2017-03-01 Richard Biener <rguenther@suse.de>
25256 PR middle-end/79721
25257 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
25258 interpolating formula in wrapping arithmetic.
25259 (chrec_apply): Convert chrec_evaluate return value to wanted type.
25261 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25263 PR tree-optimization/79734
25264 * tree-vect-generic.c (expand_vector_condition): Optimize
25265 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
25266 Handle VEC_COND_EXPR where comparison has different inner width from
25267 type's inner width.
25269 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
25271 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
25272 markup, and similar issues. Remove @opindex entries for things
25273 that aren't options. Add missing -mmpy-option entries.
25275 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25277 PR tree-optimization/79737
25278 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
25279 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
25280 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
25281 instead of byte_size. Formatting fix.
25282 (shift_bytes_in_array_right): Formatting fix.
25284 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
25287 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
25288 condition on optimize for the leaf function test.
25290 2017-02-28 Martin Liska <mliska@suse.cz>
25293 * read-rtl-function.c (function_reader::handle_unknown_directive):
25294 Bail out when one uses -flto.
25296 2017-02-28 Martin Liska <mliska@suse.cz>
25298 * common.opt: Replace space with tabular for options of <number>
25300 * config/i386/i386.opt: Show <number> value for
25301 -mlarge-data-threshold.
25302 * opts.c (print_filtered_help): Do not display number in hexadecimal
25305 2017-02-28 Martin Liska <mliska@suse.cz>
25307 * common.opt: Fix --help=option -Q for options which are of
25310 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
25312 * config/i386/i386.c (print_reg): Error out for values
25313 of 8-bit size in invalid integer register.
25315 2017-02-28 Martin Sebor <msebor@redhat.com>
25317 PR tree-optimization/79691
25318 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
25320 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25323 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
25324 gcc_unreachable with output_operand_lossage.
25326 2017-02-28 Richard Biener <rguenther@suse.de>
25328 PR tree-optimization/79740
25329 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
25331 (visit_nary_op): Insert the nary into the hashtable if we
25332 pattern-matched sth.
25333 * tree-ssa-pre.c (eliminate_insert): Robustify.
25335 2017-02-28 Richard Biener <rguenther@suse.de>
25337 PR middle-end/79731
25338 * fold-const.c (decode_field_reference): Reject out-of-bound
25341 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25343 * config/i386/i386.c: Include intl.h.
25344 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
25345 instead of just cond ? "..." : "...".
25346 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
25347 * coverage.c (read_counts_file): Likewise.
25348 * omp-offload.c: Include intl.h.
25349 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
25350 of just cond ? "..." : "...".
25351 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
25352 of just cond ? "..." : "...".
25354 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
25357 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
25359 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
25360 'tune for' CPU name.
25361 * config/arm/arm-cpu-data.h: Regenerated.
25363 2017-02-28 Richard Biener <rguenther@suse.de>
25365 PR tree-optimization/79732
25366 * tree-inline.c (expand_call_inline): Do not shadow var.
25368 2017-02-28 Richard Biener <rguenther@suse.de>
25370 PR tree-optimization/79723
25371 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
25372 address-space properly.
25374 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
25376 * doc/optinfo.texi (Optimization groups): Fix option used for
25378 * doc/invoke.texi (-fopt-info): Document "omp".
25379 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
25380 (OPTGROUP_ALL): Add OPTGROUP_OMP.
25381 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
25382 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
25383 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
25385 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
25387 * dumpfile.c (optgroup_options): Instead of "openmp", associate
25388 OPTGROUP_OMP with "omp".
25390 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
25393 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
25394 for arithmetic shift of unsigned V2DI.
25396 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
25398 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
25399 arc/linux.h headers.
25400 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
25401 (LINK_SPEC): Likewise.
25402 (ARC_TLS_EXTRA_START_SPEC): Likewise.
25403 (EXTRA_SPECS): Likewise.
25404 (STARTFILE_SPEC): Likewise.
25405 (ENDFILE_SPEC): Likewise.
25406 (LIB_SPEC): Likewise.
25407 (TARGET_SDATA_DEFAULT): Likewise.
25408 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
25409 (MULTILIB_DEFAULTS): Likewise.
25410 (DWARF2_UNWIND_INFO): Likewise.
25411 * config/arc/big.h: New file.
25412 * config/arc/elf.h: Likewise.
25413 * config/arc/linux.h: Likewise.
25414 * config/arc/t-uClibc: Remove.
25416 2017-02-27 Bin Cheng <bin.cheng@arm.com>
25418 PR tree-optimization/77536
25419 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
25420 (tree_transform_and_unroll_loop): Use above function to compute the
25421 estimated niter of unrolled loop and use it when scaling profile.
25422 Also use count info rather than frequency if it's non-zero.
25423 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
25424 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
25425 (vect_transform_loop): Call above function.
25427 2017-02-27 Richard Biener <rguenther@suse.de>
25429 PR tree-optimization/45397
25430 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
25431 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
25432 (visit_nary_op): Add pattern matching for CSEing sign-changed
25433 or truncated operations with wider ones.
25435 2017-02-27 Richard Biener <rguenther@suse.de>
25437 PR tree-optimization/79690
25438 * tree-vect-stmts.c (vectorizable_store): Use vector type
25439 built from the DR with address-space.
25441 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
25443 * doc/invoke.texi (Optimize Options): Refine the description
25444 of asan-use-after-return.
25446 2017-02-25 Alan Modra <amodra@gmail.com>
25448 PR rtl-optimization/79584
25449 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
25450 base, not ad->base_term, the reg within base. Remove assertion
25451 that ad->base == ad->base_term. Replace gen_int_mode using
25452 bogus mode with const0_rtx.
25454 2017-02-25 Jakub Jelinek <jakub@redhat.com>
25456 PR middle-end/79396
25457 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
25458 FMA_EXPR like tcc_binary or tcc_unary.
25460 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
25463 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
25465 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
25466 (output_loc_operands): Handle DW_OP_call_ref and
25467 DW_OP_GNU_variable_value.
25468 (struct variable_value_struct): New type.
25469 (struct variable_value_hasher): Likewise.
25470 (variable_value_hash): New variable.
25471 (string_types): Remove.
25472 (copy_loc_descr): New function.
25473 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
25474 (prepend_loc_descr_to_each): New function.
25475 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
25476 instead of add_loc_descr_to_each if the first argument is single
25477 location list and the second has multiple.
25478 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
25479 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
25480 when looking for variable value which doesn't have other location info.
25481 (loc_list_from_tree): Formatting fix.
25482 (gen_array_type_die): Simplify DW_AT_string_length handling.
25483 (adjust_string_types): Remove.
25484 (gen_subprogram_die): Don't call adjust_string_types nor test/set
25485 string_types. Call resolve_variable_values.
25486 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
25487 (resolve_addr_in_expr): Likewise. Add A argument.
25488 (copy_deref_exprloc): Remove deref argument. Adjust for the
25489 original expression being DW_OP_GNU_variable_value with optionally
25490 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
25491 optionally after it.
25492 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
25493 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
25494 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
25495 (variable_value_hasher::hash, variable_value_hasher::equal): New
25497 (resolve_variable_value_in_expr, resolve_variable_value,
25498 resolve_variable_values, note_variable_value_in_expr,
25499 note_variable_value): New functions.
25500 (dwarf2out_early_finish): Call note_variable_value on all toplevel
25503 2017-02-24 Jakub Jelinek <jakub@redhat.com>
25506 * opts.h (handle_generated_option): Add GENERATED_P argument.
25507 * opts-common.c (handle_option): Adjust function comment.
25508 (handle_generated_option): Add GENERATED_P argument, pass it to
25510 (control_warning_option): Pass false to handle_generated_option
25512 * opts.c (maybe_default_option): Pass true to handle_generated_option
25514 * optc-gen.awk: Likewise.
25516 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25518 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
25519 a REG, look at the REG it is a SUBREG of.
25520 (splitter for cmpeqsi_t): Ditto.
25522 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25524 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
25525 the special USEs with the pattern of the insn, not the insn itself.
25527 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
25530 * doc/invoke.texi: Document -mload-store-pairs.
25532 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25533 Sandra Loosemore <sandra@codesourcery.com>
25535 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
25536 argument isn't a CONST_INT.
25537 (nios2_alternate_compare_const): Assert op is a CONST_INT.
25538 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
25539 (nios2_validate_compare): Bypass alternate compare logic if *op2
25540 is not a CONST_INT.
25541 (ldstwm_operation_p): Return false if first_base is not a REG or
25542 if first_offset is not a CONST_INT.
25544 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25546 * config/cris/cris.md: Use correct operand in a define_peephole2.
25548 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25550 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
25552 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25554 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
25555 this_insn if it is an INSN or JUMP_INSN.
25556 (force_offsettable): Look at base, not at addr.
25557 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
25558 on things that aren't necessarily CONST_INTs.
25560 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
25562 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
25563 -mfpmath=sse is the default also for x86-32 targets with SSE2
25564 instruction set when @option{-ffast-math} is enabled
25566 2017-02-24 Jeff Law <law@redhat.com>
25568 PR rtl-optimizatoin/79286
25569 * ira.c (update_equiv_regs): Drop may_trap_p exception to
25572 2017-02-24 Richard Biener <rguenther@suse.de>
25574 PR tree-optimization/79389
25575 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
25578 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
25580 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
25581 function comment to reflect reality.
25582 (loop_exits_before_overflow): Fix typo in function description.
25584 2017-02-24 Richard Biener <rguenther@suse.de>
25586 PR tree-optimization/79389
25587 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
25588 properly that a threading opportunity exists. Detect conditional
25589 copy/constant propagation opportunities.
25591 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
25593 * config/visium/visium.md (type): Add trap.
25594 (b): New mode attribute.
25595 (*btst): Rename into...
25596 (*btst<mode>): ...this and adjust.
25597 (*cbranchsi4_btst_insn): Rename into...
25598 (*cbranch<mode>4_btst_insn): ...this and adjust.
25599 (trap): New define_insn.
25601 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25603 PR tree-optimization/79389
25604 * ifcvt.c (struct noce_if_info): Add rev_cond field.
25605 (noce_reversed_cond_code): New function.
25606 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
25607 reversed_comparison_code. Formatting fix.
25608 (noce_try_store_flag): Test rev_cond != NULL in addition to
25609 reversed_comparison_code.
25610 (noce_try_store_flag_constants): Likewise.
25611 (noce_try_store_flag_mask): Likewise.
25612 (noce_try_addcc): Use rev_cond if non-NULL instead of
25613 reversed_comparison_code.
25614 (noce_try_cmove_arith): Likewise. Formatting fixes.
25615 (noce_try_minmax, noce_try_abs): Clear rev_cond.
25616 (noce_find_if_block): Initialize rev_cond.
25617 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
25618 instead of false as last argument never attempt to reverse it
25621 2017-02-23 Bin Cheng <bin.cheng@arm.com>
25623 PR tree-optimization/79663
25624 * tree-predcom.c (combine_chains): Process refs in reverse order
25625 only for ZERO length chains, and add explaining comment.
25627 2017-02-23 Jeff Law <law@redhat.com>
25629 PR tree-optimization/79578
25630 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
25631 in call to operand_equal_p.
25633 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
25636 * config/i386/cpuid.h: Fix another undefined behavior.
25638 2017-02-23 Richard Biener <rguenther@suse.de>
25640 PR tree-optimization/79683
25641 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
25642 vector types for data-refs.
25644 2017-02-23 Martin Liska <mliska@suse.cz>
25646 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
25648 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25650 PR middle-end/79665
25651 * internal-fn.c (get_range_pos_neg): Moved to ...
25652 * tree.c (get_range_pos_neg): ... here. No longer static.
25653 * tree.h (get_range_pos_neg): New prototype.
25654 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
25655 are known to be in between 0 and signed maximum inclusive, try to
25656 expand both unsigned and signed divmod and use the cheaper one from
25659 2017-02-22 Jeff Law <law@redhat.com>
25661 PR tree-optimization/79578
25662 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
25663 to compare base operands.
25665 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
25668 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
25669 gpc_reg_operand instead of fpr_reg_operand.
25671 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
25673 * config/mips/mips.c (mips_return_in_memory): Force FP
25674 vector types to be returned in memory for o32 ABI.
25676 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25678 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
25679 instead of DW_TAG_member for static data member declarations and don't
25680 set no_linkage_name for static inline data members.
25681 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
25684 2017-02-22 Martin Liska <mliska@suse.cz>
25686 * doc/invoke.texi: Replace inequality signs with square brackets
25689 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25691 PR tree-optimization/68644
25692 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
25694 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
25697 * lra-constraints.c (simplify_operand_subreg): Handle
25698 WORD_REGISTER_OPERATIONS targets.
25700 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25703 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
25704 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
25705 elimination by swapping fld*.
25707 2017-02-22 Richard Biener <rguenther@suse.de>
25709 PR tree-optimization/79673
25710 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
25711 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
25712 irrelevant address-space qualifiers and avoiding a
25713 ADDR_SPACE_CONVERT_EXPR from fold_convert.
25715 2017-02-22 Richard Biener <rguenther@suse.de>
25717 PR tree-optimization/79666
25718 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
25719 to not symbolically negate if that may introduce undefined
25722 2017-02-22 Martin Liska <mliska@suse.cz>
25725 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
25726 * data-streamer-out.c (streamer_write_gcov_count_stream):
25728 * value-prof.c (stream_out_histogram_value): Make assert more
25729 precise based on type of counter.
25731 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
25734 * config/i386/i386.md (standard_x87sse_constant_load splitter):
25735 Use nonimmediate_operand instead of memory_operand for operand 1.
25736 (float-extend standard_x87sse_constant_load splitter): Ditto.
25738 2017-02-21 Jeff Law <law@redhat.com>
25740 PR tree-optimization/79621
25741 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
25742 blocks with edges to themselves.
25744 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25747 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
25748 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
25749 Use gimple_call_builtin_p.
25752 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
25753 on temporarily removed DEBUG_INSNs.
25755 PR tree-optimization/79649
25756 * tree-loop-distribution.c (classify_partition): Give up on
25757 non-generic address space loads/stores.
25759 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
25761 * doc/loop.texi (Loop manipulation): Remove nonexistent
25762 tree_ssa_loop_version from the documentation.
25763 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
25765 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25768 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
25769 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
25770 * config/rs6000/rs6000.c: Include except.h.
25771 (rs6000_expand_split_stack_prologue): Call
25772 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
25774 2017-02-21 Martin Jambor <mjambor@suse.cz>
25777 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
25778 have been analyzed.
25780 2017-02-21 Martin Jambor <mjambor@suse.cz>
25782 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
25783 for backward compatibility only.
25784 * doc/invoke.texi (Option Summary): Remove all references to
25785 -fipa-cp-alignment.
25787 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
25791 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25793 * lra-constraints.c (curr_insn_transform): Handle
25794 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25796 2017-02-21 Martin Liska <mliska@suse.cz>
25798 * config/i386/i386.opt: Replace -masm-dialect with -masm.
25800 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
25802 PR translation/79638
25803 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
25805 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
25808 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
25809 (arm_function_ok_for_sibcall): Return false for an indirect call by
25810 descriptor if all the argument registers are used.
25811 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
25812 alignment of the function.
25814 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25816 PR tree-optimization/61441
25817 * simplify-rtx.c (simplify_const_unary_operation): For
25818 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
25819 the sNaN unmodified.
25821 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25823 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25824 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25825 instead of SYSTEM_HEADER_DIR.
25827 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
25828 Martin Liška <mliska@suse.cz>
25830 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
25831 Fix typos and grammar, use active voice, and clarify.
25833 2017-02-20 Marek Polacek <polacek@redhat.com>
25835 PR middle-end/79537
25836 * gimplify.c (gimplify_expr): Handle unused *&&L;.
25839 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
25841 2017-02-20 Jakub Jelinek <jakub@redhat.com>
25844 * config/i386/i386.c (ix86_expand_builtin): Handle
25845 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
25846 ix86_builtins_isa[fcode].isa as a requirement of those
25847 flags and any other flag in the bitmask.
25848 (ix86_init_mmx_sse_builtins): Use 0 instead of
25849 ~OPTION_MASK_ISA_64BIT as mask.
25850 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
25851 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
25852 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
25853 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
25855 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25858 * lra-constraints.c (split_reg): Check requested split mode
25859 is supported by the register.
25861 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25863 * lra-constraints.c (simplify_operand_subreg): Remove early
25866 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25869 * lra-constraints.c (curr_insn_transform): Tighten condition
25870 for converting SUBREG reloads from OP_OUT to OP_INOUT.
25872 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25875 * lra-constraints.c (curr_insn_transform): Handle
25876 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25878 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
25881 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
25883 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
25885 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
25888 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
25891 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25893 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
25896 2017-02-18 Jakub Jelinek <jakub@redhat.com>
25898 * final.c (last_columnnum, override_columnnum): New variables.
25899 (final_start_function): Set last_columnnum, pass it to begin_prologue
25900 hook and pass 0 to dwarf2out_begin_prologue.
25901 (final_scan_insn): Update override_columnnum. Pass last_columnnum
25902 to source_line debug hook.
25903 (notice_source_line): Compute last_columnnum and for debug_column_info
25904 return true on column changes.
25905 * debug.h (struct gcc_debug_hooks): Add column argument to
25906 source_line and begin_prologue hooks.
25907 (debug_nothing_int_charstar_int_bool): Remove prototype.
25908 (debug_nothing_int_int_charstar,
25909 debug_nothing_int_int_charstar_int_bool): New prototypes.
25910 (dwarf2out_begin_prologue): Add column argument.
25911 * debug.c (do_nothing_debug_hooks): Adjust source_line and
25912 begin_prologue hooks.
25913 (debug_nothing_int_charstar_int_bool): Remove.
25914 (debug_nothing_int_int_charstar,
25915 debug_nothing_int_int_charstar_int_bool): New functions.
25916 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
25917 through to dwarf2out_source_line.
25918 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
25919 (dwarf2out_source_line): Add column argument, emit it if requested.
25920 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
25922 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25923 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25924 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
25925 through to dwarf2out_begin_prologue.
25926 (vmsdbgout_source_line): Add column argument, pass it through to
25927 dwarf2out_source_line.
25928 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
25929 dbxout_source_line caller.
25930 (dbxout_source_line): Add column argument.
25932 * common.opt (gno-column-info, gcolumn-info): New options.
25933 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
25934 (check_die): Also test for multiple DW_AT_decl_column attributes.
25935 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
25936 DW_AT_decl_column if requested.
25937 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
25939 (gen_variable_die): Likewise.
25940 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
25941 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
25944 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
25945 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
25946 (ix86_handle_option): Handle OPT_m3dnowa.
25947 * doc/invoke.texi (-m3dnowa): Document.
25948 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
25949 -m3dnowa instead of -m3dnow -march=athlon.
25952 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
25953 instead of gcc_assert for K, r and R code checks. Formatting fixes.
25955 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25958 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
25959 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
25960 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
25961 generator for vsx_xxpermdi_<mode>_be.
25962 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
25963 force big-endian semantics.
25964 (vsx_xxpermdi_<mode>_be): New define_expand with same
25965 implementation as previous version of vsx_xxpermdi_<mode>.
25967 2017-02-17 Jakub Jelinek <jakub@redhat.com>
25969 PR tree-optimization/79327
25970 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
25971 variable, its initialization and use.
25973 2017-02-17 Julia Koval <julia.koval@intel.com>
25975 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
25976 (OPTION_MASK_ISA_PKU_UNSET): New.
25977 (ix86_handle_option): Handle -mrdpid.
25978 * config/i386/cpuid.h (bit_RDPID): New.
25979 * config/i386/driver-i386.c (host_detect_local_cpu):
25980 Detect RDPID feature.
25981 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
25982 * config/i386/i386-c.c (ix86_target_macros_internal):
25984 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
25985 (ix86_valid_target_attribute_inner_p): Add "rdpid".
25986 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
25987 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
25988 * config/i386/i386.md (define_insn "rdpid"): New.
25989 * config/i386/i386.opt Add -mrdpid.
25990 * config/i386/immintrin.h (_rdpid_u32): New.
25992 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
25994 PR rtl-optimization/79541
25995 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
25996 instead of transforming it into USE.
25998 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26000 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26001 If HONOR_SNANS (SFmode) force the input to a register.
26002 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26003 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26004 an frsp or similar insn.
26006 2017-02-17 Martin Liska <mliska@suse.cz>
26008 PR rtl-optimization/79577
26009 * params.def (selsched-max-sched-times): Increase minimum to 1.
26011 2017-02-17 Martin Liska <mliska@suse.cz>
26013 PR rtl-optimization/79574
26014 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26016 2017-02-17 Martin Liska <mliska@suse.cz>
26018 PR tree-optimization/79529
26019 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26020 ssa_defined_default_def_p to handle cases which are implicitly
26022 * tree-ssa.c (ssa_defined_default_def_p): New function.
26023 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26024 which are implicitly defined.
26025 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26027 2017-02-17 Richard Biener <rguenther@suse.de>
26029 PR middle-end/79576
26030 * params.def (max-ssa-name-query-depth): Limit to 10.
26032 2017-02-17 Richard Biener <rguenther@suse.de>
26034 PR tree-optimization/79552
26035 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26038 2017-02-17 Richard Biener <rguenther@suse.de>
26041 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26043 2017-02-17 Marek Polacek <polacek@redhat.com>
26045 PR middle-end/79536
26046 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26047 (fold_negate_expr): New wrapper.
26049 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26051 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26052 Correct terminology and de-emphasize pre-standard behavior.
26054 2017-02-16 Alan Modra <amodra@gmail.com>
26056 PR rtl-optimization/79286
26057 * ira.c (def_dominates_uses): New function.
26058 (update_equiv_regs): Don't create an equivalence for insns that
26059 may trap where the register def does not dominate the use.
26061 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26063 PR rtl-optimization/78127
26064 * lra.c (lra): Call lra_eliminate before finish the loop after
26067 2017-02-16 Richard Biener <rguenther@suse.de>
26069 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26071 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26072 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26073 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26074 (isl_val_int_from_wi): New function.
26075 (extract_affine_gmp): Rename to ...
26076 (extract_affine_wi): ... this, take a widest_int.
26077 (extract_affine_int): Just wrap extract_affine_wi.
26078 (add_param_constraints): Use isl_val_int_from_wi.
26079 (add_loop_constraints): Likewise, and extract_affine_wi.
26081 2017-02-15 Jeff Law <law@redhat.com>
26083 PR middle-end/79521
26084 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
26085 ira_init_register_move_cost_if_necessary.
26087 2017-02-15 Martin Sebor <msebor@redhat.com>
26089 PR middle-end/32003
26090 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
26091 removed in a prior commit.
26093 2017-02-15 Bin Cheng <bin.cheng@arm.com>
26095 PR tree-optimization/79347
26096 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
26097 counters during peeling.
26099 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
26101 * Makefile.in (site.exp): Remove "set ISLVER".
26103 2017-02-15 Jakub Jelinek <jakub@redhat.com>
26106 * real.c (real_from_integer): Call real_convert even for decimal.
26108 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
26111 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
26113 2017-02-14 Andrew Pinski <apinski@cavium.com>
26115 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
26116 cores and change the partno/implementer to be correct.
26117 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
26118 the 'B" as the implementer.
26119 * config/aarch64/aarch64-tune.md: Regenerate.
26121 2017-02-14 Carl Love <cel@us.ibm.com>
26123 * config/rs6000/rs6000.c: Add case statement entry to make the
26124 xvcvuxdsp built-in argument unsigned.
26125 * config/rs6000/vsx.md: Fix the source and return operand types so they
26126 match the instruction definitions from the ISA document. Fix typo
26127 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
26130 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
26133 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
26134 member early_clobber_alts.
26135 * lra-lives.c (reg_early_clobber_p): New.
26136 (process_bb_lives): Use it.
26137 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
26138 (debug_operand_data): Initialize early_clobber_alts.
26139 (setup_operand_alternative): Set up early_clobber_alts.
26140 (collect_non_operand_hard_regs): Ditto. Pass early clobber
26141 alternatives to new_insn_reg.
26142 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
26144 (lra_update_insn_regno_info): Pass the new arg.
26146 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26148 PR middle-end/79505
26149 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
26150 (new_oacc_loop_raw): Don't clear already cleared fields.
26153 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
26154 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
26155 _mm512_prefetch_i64gather_ps): New inline functions and macros.
26157 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
26160 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
26162 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
26165 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
26166 the extra instruction to the right place to store 128-bit constant
26169 2017-02-14 Martin Sebor <msebor@redhat.com>
26171 PR middle-end/79448
26172 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
26173 warning for strings of unknown length.
26175 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26177 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
26179 2017-02-14 Jeff Law <law@redhat.com>
26182 * ira-costs.c (scan_one_insn): Initialize register move costs
26183 for pseudos seen in USE/CLOBBER insns.
26185 PR tree-optimization/79095
26186 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
26187 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
26188 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
26189 if the operands are known to be not equal, then the resulting range
26191 (intersect_ranges): If the new range is ~[0,0] and the old range is
26192 wide, then prefer ~[0,0].
26193 * tree-vrp.c (overflow_comparison_p_1): New function.
26194 (overflow_comparison_p): New function.
26195 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
26196 if NAME is used in an overflow test.
26197 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
26198 overflow check that can be expressed as an equality test, then adjust
26199 ops to be that equality test.
26201 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26203 * config/s390/s390-builtin-types.def: Remove flags argument.
26204 * config/s390/s390.c (s390_init_builtins): Likewise.
26206 2017-02-14 Martin Liska <mliska@suse.cz>
26208 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
26209 vector. Fix trailing white spaces.
26211 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
26213 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
26216 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26218 PR rtl-optimization/68664
26219 * config/arm/arm.c (arm_sched_can_speculate_insn):
26220 New function. Declare prototype.
26221 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26223 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26225 PR rtl-optimization/68664
26226 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
26228 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26230 2017-02-14 Amit Pawar <amit.pawar@amd.com>
26232 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
26233 max skip bytes for function, loop and jump.
26235 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26237 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
26238 ABS_EXPR for gimple dump.
26240 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26243 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
26245 PR tree-optimization/79408
26246 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
26247 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
26248 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
26249 also if rhs1 is INTEGER_CST.
26251 2017-02-14 Richard Biener <rguenther@suse.de>
26253 PR middle-end/79432
26254 * tree-into-ssa.c (insert_phi_nodes): When the function can
26255 have abnormal edges rewrite SSA names with broken use-def
26256 dominance out of SSA and register them for PHI insertion.
26258 2017-02-13 Martin Sebor <msebor@redhat.com>
26260 PR middle-end/79496
26261 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
26262 clearing info.nowrite flag when snprintf size argument is a range.
26264 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26266 * cprop.c (cprop_jump): Add missing space in string literal.
26267 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
26268 (get_constraint_for_component_ref): Likewise.
26269 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
26270 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
26271 * lra-constraints.c (process_alt_operands): Likewise.
26272 * ipa-inline.c (inline_small_functions): Likewise.
26273 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
26274 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26275 * trans-mem.c (diagnose_tm_1_op): Likewise.
26276 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
26277 (grid_parallel_clauses_gridifiable): Likewise.
26279 * config/nvptx/mkoffload.c (process): Add space in between
26282 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
26283 "MOD4_SSE_REGS" and "ALL_REGS".
26285 * spellcheck.c (test_data): Add , in between "foo" and "food".
26287 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26290 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
26291 boundary crossing check and subsequent code generation agree.
26293 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26295 * config/aarch64/aarch64.c (has_memory_op): Delete.
26296 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
26299 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26301 PR rtl-optimization/79388
26302 PR rtl-optimization/79450
26303 * combine.c (distribute_notes): When removing TEM_INSN for which
26304 corresponding dest has last value recorded, invalidate that last
26307 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26309 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
26310 of explicit '@'. Add missing assembly comment marker on branch costs
26313 2017-02-13 Nathan Sidwell <nathan@acm.org>
26315 * gengtype-lex.l (<in_struct>): Add '/'.
26317 2017-02-13 Martin Liska <mliska@suse.cz>
26320 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
26322 2017-02-13 Richard Biener <rguenther@suse.de>
26324 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
26326 * configure: Re-generate.
26327 * config.in: Likewise.
26328 * graphite-dependences.c: Simplify as if
26329 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
26330 * graphite-isl-ast-to-gimple.c: Likewise.
26331 * graphite-optimize-isl.c: Likewise.
26332 * graphite-poly.c: Likewise.
26333 * graphite-sese-to-poly.c: Likewise.
26334 * graphite.h: Likewise.
26335 * toplev.c: Include isl/version.h and use isl_version () for
26336 printing the ISL version.
26337 * doc/install.texi: Update ISL requirement.
26339 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26341 * doc/standards.texi (Standards): Update reference to
26344 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26346 * doc/extend.texi (Named Address Spaces): sourceware.org now
26348 * doc/install.texi (Binaries): Ditto.
26351 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26353 * doc/cpp.texi: Replace "stringify"/"stringification" with C
26354 standard terminology "stringize"/"stringizing" throughout.
26355 * doc/cppinternals.texi: Likewise.
26357 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26359 * doc/extend.texi: Fix some spelling mistakes and typos.
26360 * doc/invoke.texi: Likewise.
26362 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26365 * params.def (inline-min-speedup) Change from 10 to 8.
26367 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26369 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
26372 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26375 * ipa-inline-analysis.c (get_minimal_bb): New function.
26376 (record_modified): Use it.
26377 (remap_edge_change_prob): Handle also ancestor functions.
26379 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
26381 * doc/contrib.texi (Contributors): Remove broken link into
26382 the Mauve CVS repository.
26384 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26386 PR middle-end/79454
26387 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
26388 result computation whenever lhs doesn't have vector mode, not
26389 just when it has BLKmode.
26391 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26393 * doc/makefile.texi (profiledbootstrap): Refer to the
26394 installation instructions only in textual form.
26396 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26399 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
26401 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26403 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
26404 (Specific): Update mingw-w64 reference.
26406 (Specific): Remove broken link to Renesas RX processor.
26408 2017-02-10 Richard Biener <rguenther@suse.de>
26410 * toplev.c (process_options): Do not mention obsolete graphite
26411 options when printing sorry message about missing graphite support.
26412 Mention -floop-nest-optimize.
26414 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
26416 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
26417 (vtst_p16): Likewise.
26418 (vtstq_p8): Likewise.
26419 (vtstq_p16): Likewise.
26421 (vtstq_p64): Likewise.
26422 * config/arm/arm_neon.h (vgetq_lane_p64): New.
26423 (vset_lane_p64): New.
26424 (vsetq_lane_p64): New.
26426 2017-02-10 Jakub Jelinek <jakub@redhat.com>
26428 PR tree-optimization/79411
26429 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
26430 stmt operands are SSA_NAMEs used in abnormal phis.
26431 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
26434 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26437 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
26440 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26442 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
26444 2017-02-09 Jakub Jelinek <jakub@redhat.com>
26446 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
26450 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
26451 not arbitrary TREE_CONSTANT.
26454 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
26455 "omp declare target link" attribute unless is_global_var.
26456 * omp-offload.c (find_link_var_op): Likewise.
26458 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
26459 Chung-Lin Tang <cltang@codesourcery.com>
26461 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
26463 (gimplify_adjust_omp_clauses): Don't delete TILE.
26464 (gimplify_omp_for): Deal with TILE.
26465 * internal-fn.c (expand_GOACC_TILE): New function.
26466 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
26468 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
26469 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
26471 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
26472 avoid DIV for outermost collapse var.
26473 (expand_oacc_for): Insert tile element loop as needed. Adjust.
26474 Remove out of date comments, fix whitespace.
26475 * omp-general.c (omp_extract_for_data): Deal with tiling.
26476 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
26477 adjust OLF_DIM_BASE value.
26478 (struct omp_for_data): Add tiling field.
26479 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
26480 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
26481 for auto loops. Remove default auto determining, moved to
26482 oacc_loop_fixed_partitions.
26483 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
26484 stmts, add e_mask field.
26485 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
26486 (oacc_thread_numbers): Use oacc_dim_call.
26487 (oacc_xform_tile): New.
26488 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
26489 (finish_oacc_loop): Adjust for ifns vector.
26490 (oacc_loop_discover_walk): Append loop abstraction sites to list,
26491 add case for GOACC_TILE fns.
26492 (oacc_loop_xform_loop): Delete.
26493 (oacc_loop_process): Iterate over call list directly, and add
26494 handling for GOACC_TILE fns.
26495 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
26497 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
26498 vector partitioning to outer loops. Assign 2 partitions to loops
26499 when available. Add TILE handling.
26500 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
26501 (execite_oacc_device_lower): Process GOACC_TILE fns,
26502 ignore unknown specs.
26503 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
26504 * tree.c (omp_clause_num_ops): Adjust TILE ops.
26505 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
26507 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
26509 * configure.ac (ACX_BUGURL): Update.
26510 * configure: Regenerate.
26512 2017-02-09 Richard Biener <rguenther@suse.de>
26514 PR tree-optimization/69823
26515 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26516 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
26518 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26520 * config/arc/arc-c.def: Add __NPS400__ definition.
26521 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
26522 (TARGET_NPS400): Define.
26524 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26526 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
26528 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
26529 pointer, arch_info.
26530 (arc_cpu_types): Fill the arch_info field with a pointer into the
26531 arc_arch_types table.
26532 (arc_selected_cpu): Declare.
26533 * config/arc/arc.c (arc_selected_cpu): Make global.
26534 (arc_selected_arch): Delete.
26535 (arc_base_cpu): Delete.
26536 (arc_override_options): Remove references to deleted variables,
26537 update access to arch information.
26538 (ARC_OPT): Update access to arch information.
26539 (ARC_OPTX): Likewise.
26540 * config/arc/arc.h (arc_base_cpu): Remove declaration.
26541 (TARGET_ARC600): Update access to arch information.
26542 (TARGET_ARC601): Likewise.
26543 (TARGET_ARC700): Likewise.
26544 (TARGET_EM): Likewise.
26545 (TARGET_HS): Likewise.
26546 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
26549 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
26552 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
26553 condition/operands for integer GE/LE/GEU/LEU operations.
26555 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
26557 PR translation/79397
26558 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
26561 2017-02-08 Martin Jambor <mjambor@suse.cz>
26564 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
26565 whether allocation happened.
26566 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
26567 nothing was allocated.
26569 2017-02-08 Jakub Jelinek <jakub@redhat.com>
26571 PR tree-optimization/79408
26572 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
26573 constant, but SSA_NAME with a known integer range, use the minimum
26574 of that range instead of op1 to determine if modulo can be replaced
26575 with its first operand.
26577 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26579 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
26581 2017-02-08 Richard Biener <rguenther@suse.de>
26583 PR tree-optimization/71824
26584 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26585 Check all loops contained in the merged region.
26587 2017-02-07 Andrew Pinski <apinski@cavium.com>
26589 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
26591 2017-02-07 Andrew Pinski <apinski@cavium.com>
26593 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
26594 (thunderxt88): Likewise.
26595 (thunderxt81): Disable LSE and change v8.1 to v8.
26596 (thunderxt83): Likewise.
26598 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26599 Richard Biener <rguenther@suse.de>
26601 PR middle-end/79399
26602 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
26603 type from int to size_t.
26604 * ira-costs.c (struct_costs_size): Change type from int to size_t.
26606 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26608 PR rtl-optimization/79386
26609 * cprop.c (bypass_conditional_jumps): Initialize
26610 bypass_last_basic_block already before splitting bbs after
26611 unconditional traps...
26612 (bypass_conditional_jumps): ... rather than here.
26615 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
26616 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
26617 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
26618 fix -masm=intel patterns.
26620 2017-02-07 Richard Biener <rguenther@suse.de>
26622 PR tree-optimization/79256
26623 PR middle-end/79278
26624 * builtins.c (get_object_alignment_2): Use min_align_of_type
26625 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
26626 and ADJUST_FIELD_ALIGN.
26628 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
26630 * doc/tm.texi: Regenerate.
26631 * stor-layout.c (layout_decl): Adjust.
26632 (update_alignment_for_field): Likewise.
26633 (place_field): Likewise.
26634 (min_align_of_type): Likewise.
26635 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
26636 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
26637 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
26638 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
26639 * config/frv/frv.c (frv_adjust_field_align): Likewise.
26640 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
26641 * config/i386/i386.c (x86_field_alignment): Likewise.
26642 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
26643 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
26644 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
26645 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
26646 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
26647 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
26651 2017-01-30 Richard Biener <rguenther@suse.de>
26653 PR tree-optimization/79256
26654 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26655 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26658 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
26660 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
26661 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
26662 builtins to SImode and emit a zero-extend, if necessary.
26664 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26666 * docs/invoke.texi (RISC-V Options): Alphabetize.
26668 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26670 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
26673 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26675 * config/riscv/riscv.c: New file.
26676 * common/config/riscv/riscv-common.c: Likewise.
26677 * config.gcc: Likewise.
26678 * config/riscv/constraints.md: Likewise.
26679 * config/riscv/elf.h: Likewise.
26680 * config/riscv/generic.md: Likewise.
26681 * config/riscv/linux.h: Likewise.
26682 * config/riscv/multilib-generator: Likewise.
26683 * config/riscv/peephole.md: Likewise.
26684 * config/riscv/pic.md: Likewise.
26685 * config/riscv/predicates.md: Likewise.
26686 * config/riscv/riscv-builtins.c: Likewise.
26687 * config/riscv/riscv-c.c: Likewise.
26688 * config/riscv/riscv-ftypes.def: Likewise.
26689 * config/riscv/riscv-modes.def: Likewise.
26690 * config/riscv/riscv-opts.h: Likewise.
26691 * config/riscv/riscv-protos.h: Likewise.
26692 * config/riscv/riscv.h: Likewise.
26693 * config/riscv/riscv.md: Likewise.
26694 * config/riscv/riscv.opt: Likewise.
26695 * config/riscv/sync.md: Likewise.
26696 * config/riscv/t-elf-multilib: Likewise.
26697 * config/riscv/t-linux: Likewise.
26698 * config/riscv/t-linux-multilib: Likewise.
26699 * config/riscv/t-riscv: Likewise.
26700 * configure.ac: Likewise.
26701 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
26702 Waterman as RISC-V maintainers.
26703 * doc/install.texi: Add RISC-V entries.
26704 * doc/invoke.texi: Add RISC-V options section.
26705 * doc/md.texi: Add RISC-V constraints section.
26706 * configure: Regenerated.
26708 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26711 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
26712 false values to be constant vectors with all 0 or all 1 bits set.
26713 (vcondu<mode><mode>): Likewise.
26714 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
26716 (fpmask_comparison_operator): Update comment.
26717 (vecint_comparison_operator): New predicate.
26718 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
26719 vector conditionals when the true and false values are constant
26720 vectors with all 0 bits or all 1 bits set.
26722 2017-02-06 Martin Sebor <msebor@redhat.com>
26724 PR tree-optimization/79376
26725 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
26727 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
26729 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
26730 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
26731 to simplify split condition.
26733 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26735 * omp-expand.c (oxpand_omp_atomic_fetch_op,
26736 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
26739 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
26741 PR rtl-optimization/68664
26742 * target.def (can_speculate_insn): New hook.
26743 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
26744 * doc/tm.texi: Regenerate.
26745 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
26746 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
26747 (rs6000_sched_can_speculate_insn): New function.
26749 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26751 PR tree-optimization/79284
26752 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
26753 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
26754 vectorizable_mask_load_store, vectorizable_operation,
26755 vect_is_simple_cond, get_same_sized_vectype): Use it instead
26756 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
26757 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
26758 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
26759 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
26760 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26761 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
26762 is_gimple_assign (stmt). Replace another such test with
26763 is_gimple_assign (stmt).
26765 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26768 * config/avr/avr.c (rtl-iter.h): Include it.
26769 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
26770 (avr_legitimate_combined_insn): ...and implementation.
26772 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
26774 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
26775 * config/s390/s390.c (s390_const_operand_ok)
26776 (s390_canonicalize_comparison, s390_extract_part)
26777 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
26778 (s390_contiguous_bitmask_p, s390_rtx_costs)
26779 (legitimize_pic_address): Likewise.
26780 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
26781 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
26782 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
26783 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
26784 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
26786 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26788 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
26789 REGNO($0) == REGNO($1).
26791 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26793 * config/s390/linux.h(SIZE_TYPE): Add comment.
26795 2017-02-06 Julian Brown <julian@codesourcery.com>
26796 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26797 Virendra Pathak <virendra.pathak@broadcom.com>
26799 * config/aarch64/aarch64-cores.def: Change the scheduler
26801 * config/aarch64/aarch64.md: Include thunderx2t99.md.
26802 * config/aarch64/thunderx2t99.md: New file.
26804 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26806 * doc/standards.texi (Go Language): Update link to language
26809 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26811 * tree-eh.c (lower_resx): Sanitize profile.
26812 (cleanup_empty_eh_move_lp): Likewise.
26814 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26817 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
26819 * cfgloopmanip.h (loop_version): Update prototype.
26820 * modulo-sched.c (sms_schedule): Update call of loop_version.
26821 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
26822 * tree-parloops.c (gen_parallel_loop): Likewise.
26823 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
26824 * tree-ssa-loop-split.c (split_loop): Likewise.
26825 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
26826 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26828 2017-02-05 Martin Liska <mliska@suse.cz>
26831 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
26833 (print_operand_address): Initialize a struct to zero.
26835 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26837 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
26838 garbage collector only in textual form.
26840 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26842 * doc/extend.texi (x86 specific memory model extensions for
26843 transactional memory): Simplify a phrase.
26845 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
26848 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
26849 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
26850 (atomic_storedi_1): Likewise.
26852 2017-02-04 Jakub Jelinek <jakub@redhat.com>
26854 PR tree-optimization/79338
26855 * tree-parloops.c (gather_scalar_reductions): Don't call
26856 vect_analyze_loop_form for loop->inner before destroying loop's
26859 2017-02-03 Martin Sebor <msebor@redhat.com>
26861 PR tree-optimization/79327
26862 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
26863 when precision has resulted in leading zeros.
26864 (format_integer): Adjust the likely counter to assume an unknown
26865 argument that may be zero is non-zero.
26867 2017-02-03 Jason Merrill <jason@redhat.com>
26870 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
26871 avoid copying non-taken branch.
26873 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26875 PR tree-optimization/79340
26876 * tree-vect-loop.c (vectorizable_reduction): Release
26877 vec_defs elements after safe_splicing them into other vectors.
26880 PR tree-optimization/79327
26881 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
26882 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
26884 (format_integer): Use wide_int_to_tree instead of build_int_cst
26885 + to_?hwi. If argmin is NULL, just set argmin and argmax to
26886 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
26887 of shortest and longest sequence.
26889 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
26891 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
26892 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
26894 2017-02-03 Walter Lee <walt@tilera.com>
26897 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
26898 after initial stackframe link reg save.
26899 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
26901 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26904 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
26905 wu for stxssp alternative.
26907 2017-02-03 Martin Sebor <msebor@redhat.com>
26909 PR tree-optimization/79352
26910 * gimple-fold.c (get_range_strlen): Add argument.
26911 (get_range_strlen): Change return type to bool.
26912 (get_maxval_strlen): Pass in a dummy argument.
26913 * gimple-fold.h (get_range_strlen): Change return type to bool.
26914 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
26915 * tree.h (array_at_struct_end_p): Add argument.
26916 * tree.c (array_at_struct_end_p): Handle it.
26918 2017-02-03 Martin Liska <mliska@suse.cz>
26921 * multiple_target.c (create_dispatcher_calls): Redirect edge
26922 from a caller of a dispatcher.
26923 (expand_target_clones): Make the clones local.
26924 (ipa_target_clone): Do both target clones and resolvers.
26925 (ipa_dispatcher_calls): Remove the pass.
26926 (pass_dispatcher_calls::gate): Likewise.
26927 (make_pass_dispatcher_calls): Likewise.
26928 * passes.def (pass_target_clone): Put as very first IPA early
26931 2017-02-03 Martin Liska <mliska@suse.cz>
26933 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
26934 in case of a function with ifunc attribute.
26936 2017-02-03 Martin Liska <mliska@suse.cz>
26938 * cgraph.c (cgraph_node::dump): Dump function version info.
26939 * symtab.c (symtab_node::dump_base): Add missing new line.
26941 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26943 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
26944 (ifcombine_ifandif): Use it.
26946 2017-02-03 Martin Liska <mliska@suse.cz>
26948 * doc/invoke.texi: Document default value for
26949 use-after-scope-direct-emission-threshold.
26951 2017-02-03 Martin Liska <mliska@suse.cz>
26953 PR tree-optimization/79339
26954 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
26955 (format_floating): Likewise.
26957 2017-02-03 Martin Liska <mliska@suse.cz>
26960 * ipa-prop.c (ipa_node_params_t::insert): Remove current
26962 (ipa_node_params_t::remove): Likewise.
26963 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
26964 initialization from removed ipa_node_params_t::insert.
26965 (ipa_node_params::~ipa_node_params): Move from removed
26966 ipa_node_params_t::release.
26967 * symbol-summary.h (symbol_summary::m_released): New member.
26968 Do not release a summary twice. Do not allow to call finalizer
26969 for types of a summary that live in GGC memory.
26971 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26973 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
26976 2017-02-02 Martin Sebor <msebor@redhat.com>
26978 PR middle-end/79275
26979 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
26980 (format_string): Tighten up the range of output for non-constant
26981 strings and correct the expected range for wide non-constant strings.
26983 2017-02-02 Martin Sebor <msebor@redhat.com>
26985 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
26987 PR middle-end/32003
26988 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
26990 (-fdump-tree-@var): Add to index and document how to come up
26991 with pass-specific option and dump file names.
26992 (-fdump-passes): Clarify where to look for output.
26994 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26996 PR middle-end/77445
26997 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
26998 statistics of the analyzed path; allow threading for speed when
26999 any of BBs along the path are optimized for speed.
27001 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27003 PR middle-end/78468
27004 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27005 settings of the virtual registers.
27008 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27010 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27011 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27014 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27016 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27017 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27019 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27021 * config/s390/s390.md: Add missing comments with the expanded
27023 * config/s390/vector.md: Likewise.
27024 * config/s390/vx-builtins.md: Likewise.
27026 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27029 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27030 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27031 conditions on a single line.
27033 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27035 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27036 __S390_VX__ to __VX__.
27038 2017-02-01 Andrew Pinski <apinski@cavium.com>
27040 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27041 stmt_info to record_stmt_cost.
27042 (vect_get_known_peeling_cost): Pass stmt_info if known to
27044 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27045 cpu_vector_cost field into
27046 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27047 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27048 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27049 splitting of scalar_stmt_cost and vec_stmt_cost.
27050 (thunderx_vector_cost): Likewise.
27051 (cortexa57_vector_cost): LIkewise.
27052 (exynosm1_vector_cost): Likewise.
27053 (xgene1_vector_cost): Likewise.
27054 (thunderx2t99_vector_cost): Improve after the splitting of the two
27056 (aarch64_builtin_vectorization_cost): Update for the splitting of
27057 scalar_stmt_cost and vec_stmt_cost.
27059 2017-02-01 Torvald Riegel <triegel@redhat.com>
27060 Richard Henderson <rth@redhat.com>
27062 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27063 conditional on existance of a fast atomic load.
27064 * optabs-query.c (can_atomic_load_p): New function.
27065 * optabs-query.h (can_atomic_load_p): Declare it.
27066 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27067 no fast atomic load is available for the particular size of access.
27068 (expand_atomic_compare_and_swap): Likewise.
27069 (expand_atomic_load): Likewise.
27070 (expand_atomic_store): Likewise.
27071 (expand_atomic_fetch_op): Likewise.
27072 * testsuite/lib/target-supports.exp
27073 (check_effective_target_sync_int_128): Remove x86 because it provides
27074 no fast atomic load.
27075 (check_effective_target_sync_int_128_runtime): Likewise.
27077 2017-02-01 Richard Biener <rguenther@suse.de>
27079 * graphite.c: Include tree-vectorizer.h for find_loop_location.
27080 (graphite_transform_loops): Provide opt-info for optimized nests.
27081 * tree-parloop.c (parallelize_loops): Provide opt-info for
27082 parallelized loops.
27084 2017-02-01 Richard Biener <rguenther@suse.de>
27086 PR middle-end/79315
27087 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
27088 was not set before.
27090 2017-02-01 Richard Biener <rguenther@suse.de>
27092 PR tree-optimization/71824
27093 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27094 Verify the loops are valid in the merged SESE region.
27095 (scop_detection::can_represent_loop_1): Check analyzing the
27096 evolution of the number of iterations in the region succeeds.
27098 2017-01-31 Ian Lance Taylor <iant@golang.org>
27100 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
27101 REG_ARGS_SIZE note to 32-bit push insns and call insn.
27103 2017-01-31 David Malcolm <dmalcolm@redhat.com>
27105 PR preprocessor/79210
27106 * input.c (get_substring_ranges_for_loc): Replace line_width
27107 assertion with error-handling.
27109 2017-01-31 Richard Biener <rguenther@suse.de>
27111 PR tree-optimization/77318
27112 * graphite-sese-to-poly.c (extract_affine): Fix assert.
27113 (create_pw_aff_from_tree): Take loop parameter.
27114 (add_condition_to_pbb): Pass loop of the condition to
27115 create_pw_aff_from_tree.
27117 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27119 * config/s390/s390.c (s390_asan_shadow_offset): New function.
27120 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
27122 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
27126 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
27128 (convert_int_to_float128): Likewise.
27129 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
27130 (convert_int_to_float128): Likewise.
27131 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
27132 (UNSPEC_IEEE128_CONVERT): Likewise.
27133 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
27134 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
27135 Use local variables for IBM extended format.
27136 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
27137 (fix_trunc<mode>si2_fprs): Likewise.
27138 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
27139 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
27140 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
27141 to know that we can now have integers of all sizes in vector
27143 (fix<uns>_<mode>di2_hw): Likewise.
27144 (float<uns>_<mode>si2_hw): Likewise.
27145 (fix_<mode>si2_hw): Likewise.
27146 (fixuns_<mode>si2_hw): Likewise.
27147 (float<uns>_<mode>di2_hw): Likewise.
27148 (float_<mode>di2_hw): Likewise.
27149 (float_<mode>si2_hw): Likewise.
27150 (floatuns_<mode>di2_hw): Likewise.
27151 (floatuns_<mode>si2_hw): Likewise.
27152 (xscvqp<su>wz_<mode>): Delete, no longer used.
27153 (xscvqp<su>dz_<mode>): Likewise.
27154 (xscv<su>dqp_<mode>): Likewise.
27155 (ieee128_mfvsrd_64bit): Likewise.
27156 (ieee128_mfvsrd_32bit): Likewise.
27157 (ieee128_mfvsrwz): Likewise.
27158 (ieee128_mtvsrw): Likewise.
27159 (ieee128_mtvsrd_64bit): Likewise.
27160 (ieee128_mtvsrd_32bit): Likewise.
27162 2017-01-31 Martin Liska <mliska@suse.cz>
27165 * ipa-prop.c (ipa_free_all_node_params): Call release method
27166 instead of ~sumbol_summary to not to trigger double times
27169 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
27171 PR tree-optimization/71691
27172 * bitmap.h (class auto_bitmap): New.
27173 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
27174 is_maybe_undefined instead of ssa_undefined_value_p.
27176 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27178 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27179 __S390_ARCH_LEVEL__ to __ARCH__.
27181 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27183 PR tree-optimization/79267
27184 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
27185 if should_remove_lhs_p is true.
27187 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
27190 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
27191 (add_alignment_attribute): New.
27192 (base_type_die): Add alignment attribute.
27193 (subrange_type_die): Likewise.
27194 (modified_type_die): Likewise.
27195 (gen_array_type_die): Likewise.
27196 (gen_descr_array_type_die: Likewise.
27197 (gen_enumeration_type_die): Likewise.
27198 (gen_subprogram_die): Likewise.
27199 (gen_variable_die): Likewise.
27200 (gen_field_die): Likewise.
27201 (gen_ptr_to_mbr_type_die): Likewise.
27202 (gen_struct_or_union_type_die): Likewise.
27203 (gen_subroutine_type_die): Likewise.
27204 (gen_typedef_die): Likewise.
27205 (base_type_cmp): Compare alignment attribute.
27207 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27210 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
27211 (setb_unsigned) New pattern for setb with CCUNS.
27212 * config/rs6000/rs6000.c (expand_block_compare): Use a different
27213 subfc./subfe sequence to avoid overflow problems. Generate a
27214 shorter sequence with cmpld/setb for power9.
27215 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
27216 for generating subfc. instruction.
27217 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
27218 now uses this instruction.
27220 2017-01-30 Ian Lance Taylor <iant@google.com>
27223 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
27224 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
27226 2017-01-30 Martin Sebor <msebor@redhat.com>
27228 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
27229 Move constant to the right of a relational operator.
27230 (get_mpfr_format_length, format_character, format_string): Ditto.
27231 (should_warn_p, maybe_warn): Same.
27233 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
27235 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
27238 * asan.c (get_translation_unit_decl): Remove function.
27239 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
27241 2017-01-30 Martin Liska <mliska@suse.cz>
27243 PR gcov-profile/79259
27244 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
27245 -fprofile-generate.
27247 2017-01-30 Martin Liska <mliska@suse.cz>
27250 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
27251 Initialize variables with NULL value.
27253 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
27256 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
27258 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
27260 2017-01-30 Richard Biener <rguenther@suse.de>
27262 PR tree-optimization/79276
27263 * tree-vrp.c (process_assert_insertions): Properly adjust common
27264 when removing a duplicate.
27266 * gcc.dg/torture/pr79276.c: New testcase.
27268 2017-01-30 Richard Biener <rguenther@suse.de>
27270 PR tree-optimization/79256
27271 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27272 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27274 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
27276 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
27279 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
27280 ("*r<noxa>sbg_<mode>_sll_bitmask")
27281 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
27282 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
27283 Use contiguous_bitmask_nowrap_operand.
27285 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27288 * config/rs6000/altivec.h (vec_xl): Revise #define.
27289 (vec_xst): Likewise.
27291 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
27293 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
27295 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
27297 PR rtl-optimization/79194
27298 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
27299 traps before call to bypass_conditional_jumps.
27301 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27303 PR tree-optimization/71374
27304 * lra-constraints.c (check_conflict_input_operands): New.
27305 (match_reload): Use it.
27307 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27310 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
27311 account to calculate conflict_set.
27313 2017-01-27 Bin Cheng <bin.cheng@arm.com>
27315 PR rtl-optimization/78559
27316 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
27317 other_insn in combine.
27319 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27321 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
27322 uint16_type_node for BT_UINT16.
27324 2017-01-27 David Malcolm <dmalcolm@redhat.com>
27326 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
27327 "RTL Tests" to menu.
27328 (GIMPLE Tests): New node.
27329 (RTL Tests): New node.
27331 2017-01-27 Richard Biener <rguenther@suse.de>
27333 PR tree-optimization/79245
27334 * tree-loop-distribution.c (distribute_loop): Apply cost
27335 modeling also to detected patterns.
27337 2017-01-27 Richard Biener <rguenther@suse.de>
27339 PR tree-optimization/71433
27340 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
27341 (compare_assert_loc): New function.
27342 (process_assert_insertions): Sort and optimize assert locations
27343 to remove duplicates and push down identical assertions on
27344 edges to their destination block.
27346 2017-01-27 Richard Biener <rguenther@suse.de>
27348 PR tree-optimization/79244
27349 * tree-vrp.c (remove_range_assertions): Forcefully propagate
27350 out SSA names even if abnormal.
27352 2017-01-27 Jakub Jelinek <jakub@redhat.com>
27354 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
27355 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
27356 instead of MPFR_RNDN.
27358 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
27361 * arm.c (arm_option_override): Don't call build_target_option_node
27362 until after doing all option overrides.
27363 (arm_valid_target_attribute_tree): Likewise.
27365 2017-01-27 Martin Liska <mliska@suse.cz>
27367 * doc/invoke.texi (-fprofile-arcs): Document profiling support
27368 for {cd}tors and C++ {cd}tors.
27370 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
27372 * config/s390/s390.md ("*setmem_long_and")
27373 ("*setmem_long_and_31z"): Use zero_extend instead of and.
27375 2017-01-26 Martin Sebor <msebor@redhat.com>
27377 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
27380 2017-01-26 Martin Sebor <msebor@redhat.com>
27382 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
27383 HAVE_DFmode before using XFmode or DFmode.
27384 (parse_directive): Avoid using the z length modifier to avoid
27385 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
27387 PR middle-end/78703
27388 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
27389 to accept adjustment as an array.
27390 (get_int_range): New function.
27391 (struct directive): Make width and prec arrays.
27392 (directive::set_width, directive::set_precision): Call get_int_range.
27393 (format_integer, format_floating): Handle width and precision ranges.
27394 (format_string, parse_directive): Same.
27396 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27399 * dwarf2out.c (generate_skeleton_bottom_up): For children with
27400 comdat_type_p set, just clone them, but keep the children in the
27404 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
27405 which have direct callers with -fvar-tracking-assignments enabled
27407 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
27408 inside of type units.
27410 2017-01-26 Martin Sebor <msebor@redhat.com>
27412 PR middle-end/78703
27413 * gimple-ssa-sprintf.c (struct result_range): Add likely and
27415 (struct format_result): Replace number_chars, number_chars_min,
27416 and number_chars_max with a single member of struct result_range.
27418 (format_result::operator+=): Adjust.
27419 (struct fmtresult): Remove bounded. Handle likely and unlikely
27421 (fmtresult::adjust_for_width_or_precision): New function.
27422 (fmtresult:type_max_digits): New function.
27423 (bytes_remaining): Handle likely and unlikely counters.
27424 (min_bytes_remaining): Remove.
27425 (format_percent): Simplify.
27426 (format_integer, format_floating): Set likely and unlikely counters.
27427 (get_string_length, format_character, format_string): Same.
27428 (format_plain, should_warn_p): New function.
27429 (maybe_warn): Call should_warn_p. Update diagnostic messages
27430 and handle those for all directives, including plain strings.
27431 (format_directive): Handle likely and unlikely counters.
27432 Remove unnecessary quoting from diagnostics. Add an informational
27434 (add_bytes): Remove.
27435 (pass_sprintf_length::compute_format_length): Simplify.
27436 (try_substitute_return_value): Handle likely and unlikely counters.
27438 2017-01-26 Carl Love <cel@us.ibm.com>
27440 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
27441 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
27443 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
27446 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
27447 endianess for subregs into account.
27448 * lra-constraints.c (lra_constraints): Do risky transformations
27449 always on the first iteration.
27450 * lra-lives.c (check_pseudos_live_through_calls): Add arg
27451 last_call_used_reg_set.
27452 (process_bb_lives): Define and use last_call_used_reg_set.
27453 * lra.c (lra): Always continue after lra_constraints on the first
27456 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
27458 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
27460 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
27462 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27464 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
27465 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
27466 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
27467 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
27468 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
27469 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
27470 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
27471 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
27472 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
27474 2017-01-26 Marek Polacek <polacek@redhat.com>
27477 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
27478 for the third operand.
27480 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27482 PR middle-end/79236
27483 * omp-low.c (struct omp_context): Add simt_stmt field.
27484 (scan_omp_for): Return omp_context *.
27485 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
27486 context to the _simt_ SIMD stmt.
27487 (lower_omp_for): For combined SIMD with sibling _simt_
27488 SIMD, make sure to use the same decls in _looptemp_
27489 clauses as in the sibling.
27491 2017-01-26 David Sherwood <david.sherwood@arm.com>
27493 PR middle-end/79212
27494 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
27497 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27500 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
27501 emit fld b; fld a; if possible.
27503 * brig-builtins.def: Update copyright years.
27504 * config/arm/arm_acle_builtins.def: Update copyright years.
27506 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
27509 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
27510 constraint instead of o for the stxsd instruction.
27512 2017-01-25 Carl Love <cel@us.ibm.com>
27514 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
27515 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
27517 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
27519 * doc/invoke.texi (C++ Dialect Options): Fix typo.
27521 2017-01-25 Richard Biener <rguenther@suse.de>
27523 PR tree-optimization/69264
27524 * target.def (vector_alignment_reachable): Improve documentation.
27525 * doc/tm.texi: Regenerate.
27526 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
27528 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
27529 earlier changes with respect to TYPE_USER_ALIGN.
27530 (vector_alignment_reachable_p): Likewise. Improve dumping.
27532 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27535 * config/arm/arm.md (xordi3): Force constant operand into a register
27538 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27540 * doc/invoke.texi (-fstore-merging): Correct default optimization
27541 levels at which it is enabled.
27542 (-O): Move -fstore-merging from list to...
27545 2017-01-25 Richard Biener <rguenther@suse.de>
27548 * omp-expand.c: Include debug.h.
27549 (expand_omp_taskreg): Make sure to generate early debug before
27550 outlining anything from a function.
27551 (expand_omp_target): Likewise.
27552 (grid_expand_target_grid_body): Likewise.
27554 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
27557 * asan.c (get_translation_unit_decl): New function.
27558 (asan_add_global): Extract modules file name from globals
27559 TRANSLATION_UNIT_DECL name.
27561 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
27564 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
27565 for long calls with APCS frame and VFP.
27567 2017-01-24 David Malcolm <dmalcolm@redhat.com>
27569 * cfg.c (original_copy_tables_initialized_p): New function.
27570 * cfg.h (original_copy_tables_initialized_p): New decl.
27571 * cfgrtl.c (relink_block_chain): Guard the call to
27572 free_original_copy_tables with a call to
27573 original_copy_tables_initialized_p.
27574 * cgraph.h (symtab_node::native_rtl_p): New decl.
27575 * cgraphunit.c (symtab_node::native_rtl_p): New function.
27576 (symtab_node::needed_p): Don't assert for early assembly output
27577 for __RTL functions.
27578 (cgraph_node::finalize_function): Set "force_output" for __RTL
27580 (cgraph_node::analyze): Bail out early for __RTL functions.
27581 (analyze_functions): Update assertion to support __RTL functions.
27582 (cgraph_node::expand): Bail out early for __RTL functions.
27583 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
27585 * function.h (struct function): Update comment for field
27587 * gimple-expr.c: Include "tree-pass.h".
27588 (gimple_has_body_p): Return false for __RTL functions.
27589 * Makefile.in (OBJS): Add run-rtl-passes.o.
27590 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
27592 (gcc::pass_manager::get_clean_slate): New accessor.
27593 * passes.c: Include "insn-addr.h".
27594 (should_skip_pass_p): Add logging. Update logic for running
27595 "expand" to be compatible with both __GIMPLE and __RTL. Guard
27596 property-provider override so it is only done for gimple passes.
27598 (skip_pass): New function.
27599 (execute_one_pass): Call skip_pass when skipping passes.
27600 * read-md.c (md_reader::read_char): Support filtering
27601 the input to a subset of line numbers.
27602 (md_reader::md_reader): Initialize fields
27603 m_first_line and m_last_line.
27604 (md_reader::read_file_fragment): New function.
27605 * read-md.h (md_reader::read_file_fragment): New decl.
27606 (md_reader::m_first_line): New field.
27607 (md_reader::m_last_line): New field.
27608 * read-rtl-function.c (function_reader::create_function): Only
27609 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
27610 curr_properties. Set DECL_INITIAL to a dummy block.
27611 (read_rtl_function_body_from_file_range): New function.
27612 * read-rtl-function.h (read_rtl_function_body_from_file_range):
27614 * run-rtl-passes.c: New file.
27615 * run-rtl-passes.h: New file.
27617 2017-01-24 Jeff Law <law@redhat.com>
27619 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
27622 2017-01-24 Bin Cheng <bin.cheng@arm.com>
27624 PR tree-optimization/79159
27625 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
27626 (record_nonwrapping_iv): Improve boundary using above function if no
27627 value range information.
27629 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
27630 Martin Jambor <mjambor@suse.cz>
27632 * brig-builtins.def: New file.
27633 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
27634 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
27635 (DEF_HSAIL_SAT_BUILTIN): Likewise.
27636 (DEF_HSAIL_INTR_BUILTIN): Likewise.
27637 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
27638 * builtin-types.def (BT_INT8): New.
27639 (BT_INT16): Likewise.
27640 (BT_UINT8): Likewise.
27641 (BT_UINT16): Likewise.
27642 (BT_FN_ULONG): Likewise.
27643 (BT_FN_UINT_INT): Likewise.
27644 (BT_FN_UINT_ULONG): Likewise.
27645 (BT_FN_UINT_LONG): Likewise.
27646 (BT_FN_UINT_PTR): Likewise.
27647 (BT_FN_ULONG_PTR): Likewise.
27648 (BT_FN_INT8_FLOAT): Likewise.
27649 (BT_FN_INT16_FLOAT): Likewise.
27650 (BT_FN_UINT32_FLOAT): Likewise.
27651 (BT_FN_UINT16_FLOAT): Likewise.
27652 (BT_FN_UINT8_FLOAT): Likewise.
27653 (BT_FN_UINT64_FLOAT): Likewise.
27654 (BT_FN_UINT16_UINT32): Likewise.
27655 (BT_FN_UINT32_UINT16): Likewise.
27656 (BT_FN_UINT16_UINT16_UINT16): Likewise.
27657 (BT_FN_INT_PTR_INT): Likewise.
27658 (BT_FN_UINT_PTR_UINT): Likewise.
27659 (BT_FN_LONG_PTR_LONG): Likewise.
27660 (BT_FN_ULONG_PTR_ULONG): Likewise.
27661 (BT_FN_VOID_UINT64_UINT64): Likewise.
27662 (BT_FN_UINT8_UINT8_UINT8): Likewise.
27663 (BT_FN_INT8_INT8_INT8): Likewise.
27664 (BT_FN_INT16_INT16_INT16): Likewise.
27665 (BT_FN_INT_INT_INT): Likewise.
27666 (BT_FN_UINT_FLOAT_UINT): Likewise.
27667 (BT_FN_FLOAT_UINT_UINT): Likewise.
27668 (BT_FN_ULONG_UINT_UINT): Likewise.
27669 (BT_FN_ULONG_UINT_PTR): Likewise.
27670 (BT_FN_ULONG_ULONG_ULONG): Likewise.
27671 (BT_FN_UINT_UINT_UINT): Likewise.
27672 (BT_FN_VOID_UINT_PTR): Likewise.
27673 (BT_FN_UINT_UINT_PTR: Likewise.
27674 (BT_FN_UINT32_UINT64_PTR): Likewise.
27675 (BT_FN_INT_INT_UINT_UINT): Likewise.
27676 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
27677 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
27678 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
27679 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
27680 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
27681 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
27682 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
27683 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
27684 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
27685 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
27686 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
27687 * doc/frontends.texi: List BRIG FE.
27688 * doc/install.texi (Testing): Add BRIG tesring requirements.
27689 * doc/invoke.texi (Overall Options): Mention BRIG.
27690 * doc/standards.texi (Standards): Doucment BRIG HSA version.
27692 2017-01-24 Richard Biener <rguenther@suse.de>
27694 PR translation/79208
27695 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
27697 2017-01-24 Martin Jambor <mjambor@suse.cz>
27700 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
27701 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
27702 and known_contexts.
27704 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
27706 PR middle-end/79123
27707 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
27708 casts from signed to unsigned really don't have a range.
27710 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
27712 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
27713 GMP_RNDx for compatiblity.
27715 2017-01-24 Martin Liska <mliska@suse.cz>
27718 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
27719 that would prevent us to call alloca with -1 as argument.
27721 2017-01-24 Jakub Jelinek <jakub@redhat.com>
27723 * dwarf2out.c (output_compilation_unit_header, output_file_names):
27724 Avoid -Wformat-security warning.
27726 2017-01-23 Andrew Pinski <apinski@cavium.com>
27728 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
27731 2017-01-23 Martin Sebor <msebor@redhat.com>
27733 PR middle-end/78703
27734 * gimple-ssa-sprintf.c (warn_level): New global.
27735 (format_integer): Use it here and throughout the rest of the file.
27736 Use the same switch to compute sign as base.
27737 (maybe_warn): New function.
27738 (format_directive): Factor out warnings into maybe_warn.
27739 Add debugging output. Use warn_level.
27740 (add_bytes): Use warn_level.
27741 (pass_sprintf_length::compute_format_length): Add debugging output.
27742 (try_substitute_return_value): Same.
27743 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
27745 PR middle-end/78703
27746 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
27747 (struct fmtresult, format_integer, format_floating): Adjust.
27748 (fmtresult::fmtresult): Set max correctly in two argument ctor.
27749 (get_string_length, format_string,format_directive): Same.
27750 (pass_sprintf_length::compute_format_length): Same.
27751 (try_substitute_return_value): Simplify slightly.
27753 PR middle-end/78703
27754 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
27755 (fmtresult::operator+=): Outlined.
27756 (struct fmtresult): Add ctors.
27757 (struct conversion_spec): Rename...
27758 (struct directive): ...to this. Add and remove data members.
27759 (directive::set_width, directive::set_precision): New functions.
27760 (format_percent): Use fmtresult ctor.
27761 (get_width_and_precision): Remove.
27762 (format_integer): Make naming changes. Avoid computing width and
27764 (format_floating): Same. Adjust indentation.
27765 (format_character, format_none): New functions.
27766 (format_string): Moved character handling to format_character.
27767 (format_directive): Remove arguments, change return type.
27768 (parse_directive): New function.
27769 (pass_sprintf_length::compute_format_length): Move directive
27770 parsing to parse_directive.
27772 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27774 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
27775 (assign_assembler_name_if_needed): ... this.
27776 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
27777 (assign_assembler_name_if_needed): ... this.
27778 (free_lang_data_in_cgraph): Adjust callers.
27779 * cgraphunit.c (cgraph_node::analyze): Likewise.
27780 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
27783 2017-01-23 Richard Biener <rguenther@suse.de>
27785 PR tree-optimization/79088
27786 PR tree-optimization/79188
27787 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
27788 resetting loop bounds after last path deletion. Reset loop
27789 bounds of the target loop, make code match the comments.
27790 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
27791 Make sure loops need no fixups.
27793 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27795 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
27796 exponent support with double type for first argument.
27797 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
27798 type returned by __builtin_vec_extract_sig,
27799 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
27800 functions from "vector int" to "vector unsigned int" or from
27801 "vector long long int" to "vector unsigned long long int".
27802 Changed type returned by __builtin_vec_extract_exp,
27803 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
27804 functions from "vector int" to "vector unsigned int" or from
27805 "vector long long int" to "vector unsigned long long int".
27806 Changed return type of __builtin_vec_test_data_class,
27807 __builtin_vec_test_data_class_sp, and
27808 __builtin_vec_test_data_class_dp from "vector int" to
27809 "vector bool int" or from "vector long long int" to "vector bool
27810 long long int" and changed second argument type from "unsigned
27811 int" to "int". Added new overloaded function forms "vector float
27812 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
27813 "vector float __builtin_vec_insert_exp_sp (vector float, vector
27814 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
27815 double, vector unsigned long long int)" and "vector double
27816 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
27817 long int)". Changed return type of
27818 __builtin_scalar_test_data_class and
27819 __builtin_scalar_test_data_class_sp and
27820 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
27821 int" and changed second argument from "unsigned int" to "int".
27822 Changed type returned by __builtin_scalar_test_neg,
27823 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
27824 from "int" to "bool int". Added new overloaded function form
27825 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
27826 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
27827 exponent double-precision with floating point first argument.
27828 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
27829 documentation of scalar_test_data_class, scalar_test_neg,
27830 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
27831 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
27832 vec_test_data_class built-in functions to reflect refinements in
27833 their type signatures.
27835 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
27837 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
27839 (aarch64_elf_asm_destructor): Likewise.
27841 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
27843 PR rtl-optimization/78634
27844 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
27845 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
27846 * ifcvt.c (noce_try_cmove): Add missing cost check.
27848 PR rtl-optimization/71724
27849 * combine.c (if_then_else_cond): Look for situations where it is
27850 beneficial to undo the work of one of the recursive calls.
27852 2017-01-23 Bin Cheng <bin.cheng@arm.com>
27854 PR tree-optimization/70754
27855 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
27856 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
27857 combined stmt before it if not NULL.
27858 (combine_chains): Process refs reversely and compute dominance point
27861 2017-01-23 Martin Liska <mliska@suse.cz>
27863 PR tree-optimization/79196
27864 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
27865 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
27867 (strlen_optimize_stmt): Call the renamed function.
27869 2017-01-23 Michael Matz <matz@suse.de>
27871 PR tree-optimization/78384
27872 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
27874 2017-01-23 Richard Biener <rguenther@suse.de>
27876 PR tree-optimization/79186
27877 * tree-vrp.c (register_new_assert_for): Make sure we've seen
27878 both incoming edges before moving an assert.
27880 2017-01-23 Martin Jambor <mjambor@suse.cz>
27882 * ipa-prop.c (load_from_param_1): Removed.
27883 (load_from_unmodified_param): Bits from load_from_param_1 put back
27885 (load_from_param): Removed.
27886 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
27887 with stmt. Reverted back to use of load_from_unmodified_param.
27889 2017-01-23 Martin Jambor <mjambor@suse.cz>
27892 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
27893 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
27894 field a pointer to garbage collected vector, mark lattices and
27895 ipcp_orig_node with GTY((skip)).
27896 (ipa_get_param_count): Adjust to descriptors being a pointer.
27897 (ipa_get_param): Likewise.
27898 (ipa_get_type): Likewise.
27899 (ipa_get_param_move_cost): Likewise.
27900 (ipa_set_param_used): Likewise.
27901 (ipa_get_controlled_uses): Likewise.
27902 (ipa_set_controlled_uses): Likewise.
27903 (ipa_is_param_used): Likewise.
27904 (ipa_node_params_t): Move into garbage collector. New methods insert
27906 (ipa_node_params_sum): Annotate wth GTY(()).
27907 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
27909 (ipa_load_from_parm_agg): Adjust declaration.
27910 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
27911 * ipa-profile.c (ipa_profile): Likewise.
27912 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
27913 (ipa_populate_param_decls): Make descriptors parameter garbage
27915 (ipa_dump_param): Adjust to descriptors being a pointer.
27916 (ipa_alloc_node_params): Likewise.
27917 (ipa_initialize_node_params): Likewise.
27918 (load_from_param_1): Make descriptors parameter garbage collected.
27919 (load_from_unmodified_param): Likewise.
27920 (load_from_param): Likewise.
27921 (ipa_load_from_parm_agg): Likewise.
27922 (ipa_node_params::~ipa_node_params): Removed.
27923 (ipa_free_all_node_params): Remove call to delete operator.
27924 (ipa_node_params_t::insert): New.
27925 (ipa_node_params_t::remove): Likewise.
27926 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
27927 copy known_csts and known_contexts vectors.
27928 (ipa_read_node_info): Adjust to descriptors being a pointer.
27929 (ipcp_modif_dom_walker): Make m_descriptors field garbage
27931 (ipcp_transform_function): Make descriptors variable garbage
27934 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
27936 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
27937 * config/i386/avx512dqintrin.h: Ditto.
27938 * config/i386/avx512fintrin.h: Ditto.
27939 * config/i386/i386.c: Handle new builtins.
27940 * config/i386/i386-builtin.def: Add new builtins.
27941 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
27942 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
27944 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27945 Martin Liska <mliska@suse.cz>
27947 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
27948 * asan.c (asan_expand_poison_ifn): Support stores and use
27949 appropriate ASAN report function.
27950 * internal-fn.c (expand_ASAN_POISON_USE): New function.
27951 * internal-fn.def (ASAN_POISON_USE): Declare.
27952 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
27953 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
27954 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
27955 ASAN_POISON calls w/o LHS.
27956 * tree-ssa.c (execute_update_addresses_taken): Create clobber
27957 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
27958 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
27959 * gimplify.c (asan_poison_variables): Add attribute
27960 use_after_scope_memory to variables that really needs to live
27962 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
27963 having the attribute.
27965 2017-01-23 Martin Liska <mliska@suse.cz>
27967 * asan.c (create_asan_shadow_var): New function.
27968 (asan_expand_poison_ifn): Likewise.
27969 * asan.h (asan_expand_poison_ifn): New declaration.
27970 * internal-fn.c (expand_ASAN_POISON): Likewise.
27971 * internal-fn.def (ASAN_POISON): New builtin.
27972 * sanopt.c (pass_sanopt::execute): Expand
27973 asan_expand_poison_ifn.
27974 * tree-inline.c (copy_decl_for_dup_finish): Make function
27976 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
27977 * tree-ssa.c (is_asan_mark_p): New function.
27978 (execute_update_addresses_taken): Rewrite local variables
27979 (identified just by use-after-scope as addressable) into SSA.
27981 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27983 * doc/install.texi (Specific): opensource.apple.com uses https
27984 now. Remove trailing slash.
27986 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
27988 * README.Portability: Remove note on an Irix compatibility issue.
27990 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
27992 * gcov.c (INCLUDE_ALGORITHM): Define.
27993 (INCLUDE_VECTOR): Define.
27994 No longer include <vector> and <algorithm> directly.
27996 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
27998 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28000 * doc/invoke.texi (Code Gen Options): Ditto.
28002 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28005 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28007 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28009 rtl-optimization/79125
28010 * cprop.c (local_cprop_pass): Handle cases where we make an
28011 unconditional trap.
28013 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28017 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28018 read from, for big endian.
28020 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28022 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28023 register pauth builtins for LP64 only.
28025 2017-01-20 Marek Polacek <polacek@redhat.com>
28028 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28031 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28033 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28035 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28036 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28037 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28039 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28042 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28043 in big-endian mode when they are not a single duplicated value.
28045 2017-01-20 Richard Biener <rguenther@suse.de>
28047 * BASE-VER: Bump to 7.0.1.
28049 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28051 * omp-low.c (omplow_simd_context): New struct. Use it...
28052 (lower_rec_simd_input_clauses): ...here and...
28053 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28054 references to idx, lane, max_vf, is_simt.
28056 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28058 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28061 2017-01-20 Martin Jambor <mjambor@suse.cz>
28063 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28064 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28066 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28067 (GTFILES): Rename hsa.c to hsa-common.c.
28068 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28069 * hsa-dump.c: Likewise.
28070 * hsa-gen.c: Likewise.
28071 * hsa-regalloc.c: Likewise.
28072 * ipa-hsa.c: Likewise.
28073 * omp-expand.c: Likewise.
28074 * omp-low.c: Likewise.
28075 * toplev.c: Likewise.
28077 2017-01-20 Marek Polacek <polacek@redhat.com>
28080 * doc/invoke.texi: Document -Wduplicated-branches.
28081 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
28082 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
28083 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
28084 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
28085 return 0 only when not OEP_LEXICOGRAPHIC.
28086 (fold_build_cleanup_point_expr): Use the expression
28087 location when building CLEANUP_POINT_EXPR.
28088 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
28089 * tree.c (add_expr): Handle error_mark_node.
28091 2017-01-20 Martin Liska <mliska@suse.cz>
28094 * tree-profile.c (init_ic_make_global_vars): Do not call
28096 (gimple_init_gcov_profiler): Likewise.
28098 2017-01-20 Martin Liska <mliska@suse.cz>
28101 * cgraph.h (maybe_create_reference): Remove argument and
28103 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
28105 * ipa-cp.c (create_specialized_node): Likewise.
28106 * symtab.c (symtab_node::maybe_create_reference): Handle
28107 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
28109 2017-01-20 Martin Liska <mliska@suse.cz>
28111 * read-rtl-function.c (function_reader::create_function): Use
28112 build_decl instread of build_decl_stat.
28114 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
28116 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
28117 * config/i386/avx512dqintrin.h: Ditto.
28118 * config/i386/avx512fintrin.h: Ditto.
28119 * config/i386/i386-builtin-types.def: Add new types.
28120 * config/i386/i386.c: Handle new types.
28121 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
28122 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
28123 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
28124 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
28125 (__builtin_ia32_kshiftridi): New.
28126 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
28128 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
28132 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
28133 define to rs6000_init_stack_protect_guard.
28134 (rs6000_init_stack_protect_guard): New function.
28136 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28137 Yunqiang Su <yunqiang.su@imgtec.com>
28139 * config.gcc (supported_defaults): Add madd4.
28140 (with_madd4): Add validation.
28141 (all_defaults): Add madd4.
28142 * config/mips/mips.opt (mmadd4): New option.
28143 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28145 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28147 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
28148 (ISA_HAS_FUSED_MADD4): Likewise.
28149 * doc/invoke.texi (-mmadd4): Document the new option.
28150 * doc/install.texi (--with-madd4): Document the new option.
28152 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28154 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
28155 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
28156 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
28157 (aarch64_init_pauth_hint_builtins): New.
28158 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
28159 (aarch64_expand_builtin): Expand new builtins.
28161 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28163 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
28164 * combine-stack-adj.c (no_unhandled_cfa): Handle
28165 REG_CFA_TOGGLE_RA_MANGLE.
28166 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
28167 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
28168 info for return address signing.
28169 (aarch64_expand_epilogue): Likewise.
28171 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28173 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
28174 * config/aarch64/aarch64-protos.h
28175 (aarch64_return_address_signing_enabled): New declaration.
28176 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
28178 (aarch64_expand_prologue): Sign return address before it's pushed onto
28180 (aarch64_expand_epilogue): Authenticate return address fetched from
28182 (aarch64_override_options): Sanity check for ILP32 and ISA level.
28183 (aarch64_attributes): New function attributes for "sign-return-address".
28184 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
28185 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
28186 ("*do_return"): Generate combined instructions according to key index.
28187 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
28188 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
28190 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
28191 * config/aarch64/aarch64.opt (msign-return-address=): New.
28192 * doc/extend.texi (AArch64 Function Attributes): Documents
28193 "sign-return-address=".
28194 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
28196 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28198 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
28199 overall option summary.
28201 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28203 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
28204 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
28205 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
28206 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
28208 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
28210 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
28211 -mpower9-minmax by default for -mcpu=power9.
28212 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
28213 128-bit floating point.
28215 2017-01-20 Alan Modra <amodra@gmail.com>
28217 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
28218 optimizing for size.
28220 2017-01-20 Alan Modra <amodra@gmail.com>
28223 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
28224 for strcmp and strncmp from corresponding builtin decl.
28226 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28228 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
28229 instead of i386/rtems-64.h.
28230 * config/i386/rtems-64.h: Remove.
28232 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28236 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
28238 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
28240 2017-01-19 Tamar Christina <tamar.christina@arm.com>
28242 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
28243 Change int to HOST_WIDE_INT.
28244 * config/aarch64/aarch64-protos.h
28245 (aarch64_simd_gen_const_vector_dup): Likewise.
28246 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
28248 2017-01-19 David Malcolm <dmalcolm@redhat.com>
28250 * langhooks-def.h (lhd_type_for_size): New decl.
28251 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
28252 * langhooks.c (lhd_type_for_size): New function, taken from
28255 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
28257 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
28258 define_bypass for CR latency.
28259 (power9-cracked-alu): Update bypass latency and remove power9-branch.
28260 (power9-alu2): Add define_bypass for CR latency.
28262 (power9-mul): Update insn latency.
28263 (power9-mul-compare): Update insn latency, bypass latency and remove
28266 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28268 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
28270 * config/aarch64/aarch64.md
28271 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
28272 aarch64_nopcrelative_literal_loads.
28273 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
28275 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
28277 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
28278 TARGET_LOONGSON_3A.
28279 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
28281 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
28284 * config.gcc (supported_defaults): Add lxc1-sxc1.
28285 (with_lxc1_sxc1): Add validation.
28286 (all_defaults): Add lxc1-sxc1.
28287 * config/mips/mips.opt (mlxc1-sxc1): New option.
28288 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28290 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28291 __mips_no_lxc1_sxc1.
28292 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
28293 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
28294 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
28296 2017-01-19 Richard Biener <rguenther@suse.de>
28298 PR tree-optimization/72488
28299 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
28300 sure to restore SSA info.
28301 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
28303 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
28305 PR rtl-optimization/79121
28306 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
28307 of the inner type when shifting an extended value.
28309 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28312 * symtab.c (symtab_node::equal_address_to): Fix comparing of
28313 interposable aliases.
28315 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
28318 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
28319 Use the evmergelohi instruction.
28320 (mov_si<mode>_e500_subreg4_2_le): Likewise.
28321 (mov_sitf_e500_subreg8_2_be): Likewise.
28322 (mov_sitf_e500_subreg12_2_le): Likewise.
28323 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
28324 (mov_si<mode>_e500_subreg4_2_be): Likewise.
28325 (mov_sitf_e500_subreg8_2_le): Likewise.
28326 (mov_sitf_e500_subreg12_2_be): Likewise.
28328 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28330 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
28331 attribute from vecsimple to vecperm.
28332 (altivec_vbpermq2): Likewise.
28334 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28337 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
28339 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28340 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
28341 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
28342 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
28343 case where N arg is SIZE_MAX.
28344 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
28345 (cmpstrsi): Add pattern.
28347 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28349 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28350 __builtin_vec_revb builtins.
28351 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
28352 built-in functions to support generation of the ISA 3.0 XXBR<x>
28353 vector byte reverse instructions.
28354 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
28355 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
28356 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
28357 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
28358 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
28359 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
28360 (P9V_BUILTIN_VEC_REVB): Likewise.
28361 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
28362 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
28363 (p9_xxbrq_v16qi): Likewise.
28364 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
28365 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
28366 (p9_xxbrh_v8hi): Likewise.
28367 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
28368 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
28369 vec_revb built-in functions.
28371 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
28373 PR rtl-optimization/78952
28374 * config/i386/i386.md (any_extract): New code iterator.
28375 (*insvqi_2): Use any_extract for source operand.
28376 (*insvqi_3): Use any_shiftrt for source operand.
28378 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
28380 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
28382 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
28384 2017-01-18 Matthias Klose <doko@ubuntu.com>
28386 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
28388 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28390 * config/rs6000/altivec.h (vec_bperm): Change #define.
28391 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
28392 (altivec_vbpermq2): New define_insn.
28393 (altivec_vbpermd): Likewise.
28394 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
28395 function interface.
28396 (VBPERMD): Likewise.
28397 (VBPERM): New polymorphic function interface.
28398 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
28399 Add entries for P9V_BUILTIN_VEC_VBPERM.
28400 * doc/extend.texi: Add interfaces for vec_bperm.
28402 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28404 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
28405 first letter of error messages.
28406 (s390_resolve_overloaded_builtin): Likewise.
28407 * config/s390/s390.c (s390_expand_builtin): Likewise.
28408 (s390_invalid_arg_for_unprototyped_fn): Likewise.
28409 (s390_valid_target_attribute_inner_p): Likewise.
28410 * config/s390/s390.md ("tabort"): Likewise.
28412 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
28414 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
28415 (ISA_AVOID_DIV_HILO): New macro.
28416 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
28417 (ISA_HAS_DDIV): Likewise.
28419 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28421 * doc/invoke.texi (fabi-version): Correct number of occurrences.
28423 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28425 * doc/invoke.texi (fabi-version): Spelling fix.
28427 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28430 * doc/invoke.texi (fabi-version): Mention mangling fix for
28433 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28436 * doc/invoke.texi (fabi-version): Document discriminator mangling.
28438 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
28441 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
28442 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
28444 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
28446 (stack_protect_test): Ditto.
28447 * config/rs6000/rs6000.opt (mstack-protector-guard=,
28448 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
28450 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
28451 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
28452 -mstack-protector-guard-offset=.
28453 (RS/6000 and PowerPC Options): Ditto.
28455 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28457 * config/i386/i386.h (MASK_CLASS_P): New define.
28458 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
28459 there are no registers from different register sets also when
28460 mask registers are used. Update function comment.
28461 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
28462 to (*k/*r) and (*k/*km) alternatives.
28464 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28466 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
28467 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
28468 (EH_RETURN_HANDLER_RTX): New define.
28469 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
28470 Force frame pointer in EH return functions.
28471 (aarch64_expand_epilogue): Add barrier for eh_return.
28472 (aarch64_final_eh_return_addr): Remove.
28473 (aarch64_eh_return_handler_rtx): New function.
28474 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
28476 (aarch64_eh_return_handler_rtx): New prototype.
28478 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28480 * config/rs6000/altivec.h (vec_rlmi): New #define.
28481 (vec_vrlnm): Likewise.
28482 (vec_rlnm): Likewise.
28483 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
28484 (UNSPEC_VRLNM): Likewise.
28485 (VIlong): New mode iterator.
28486 (altivec_vrl<VI_char>mi): New define_insn.
28487 (altivec_vrl<VI_char>nm): Likewise.
28488 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
28490 (VRLDNM): Likewise.
28491 (RLNM): New polymorphic function entry.
28492 (VRLWMI): New monomorphic function entry.
28493 (VRLDMI): Likewise.
28494 (RLMI): New polymorphic function entry.
28495 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
28496 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
28497 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
28500 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28503 * dwarf2out.c (field_byte_offset): Restore the
28504 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
28505 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
28506 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
28509 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28512 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
28514 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28517 * dwarf2out.c (add_data_member_location_attribute): For constant
28518 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
28519 instead of DW_AT_data_member_location, DW_AT_bit_offset and
28520 DW_AT_byte_size attributes.
28522 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28524 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
28525 after forcing to constant memory when the code model is medium.
28527 2017-01-17 Julia Koval <julia.koval@intel.com>
28530 * config/i386/avx512fintrin.h
28531 (_mm512_i32gather_ps): Change __addr type to void const*.
28532 (_mm512_mask_i32gather_ps): Ditto.
28533 (_mm512_i32gather_pd): Ditto.
28534 (_mm512_mask_i32gather_pd): Ditto.
28535 (_mm512_i64gather_ps): Ditto.
28536 (_mm512_mask_i64gather_ps): Ditto.
28537 (_mm512_i64gather_pd): Ditto.
28538 (_mm512_mask_i64gather_pd): Ditto.
28539 (_mm512_i32gather_epi32): Ditto.
28540 (_mm512_mask_i32gather_epi32): Ditto.
28541 (_mm512_i32gather_epi64): Ditto.
28542 (_mm512_mask_i32gather_epi64): Ditto.
28543 (_mm512_i64gather_epi32): Ditto.
28544 (_mm512_mask_i64gather_epi32): Ditto.
28545 (_mm512_i64gather_epi64): Ditto.
28546 (_mm512_mask_i64gather_epi64): Ditto.
28547 (_mm512_i32scatter_ps): Change __addr type to void*.
28548 (_mm512_mask_i32scatter_ps): Ditto.
28549 (_mm512_i32scatter_pd): Ditto.
28550 (_mm512_mask_i32scatter_pd): Ditto.
28551 (_mm512_i64scatter_ps): Ditto.
28552 (_mm512_mask_i64scatter_ps): Ditto.
28553 (_mm512_i64scatter_pd): Ditto.
28554 (_mm512_mask_i64scatter_pd): Ditto.
28555 (_mm512_i32scatter_epi32): Ditto.
28556 (_mm512_mask_i32scatter_epi32): Ditto.
28557 (_mm512_i32scatter_epi64): Ditto.
28558 (_mm512_mask_i32scatter_epi64): Ditto.
28559 (_mm512_i64scatter_epi32): Ditto.
28560 (_mm512_mask_i64scatter_epi32): Ditto.
28561 (_mm512_i64scatter_epi64): Ditto.
28562 (_mm512_mask_i64scatter_epi64): Ditto.
28563 * config/i386/avx512pfintrin.h
28564 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
28565 (_mm512_mask_prefetch_i32gather_ps): Ditto.
28566 (_mm512_mask_prefetch_i64gather_pd): Ditto.
28567 (_mm512_mask_prefetch_i64gather_ps): Ditto.
28568 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
28569 (_mm512_prefetch_i32scatter_ps): Ditto.
28570 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
28571 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
28572 (_mm512_prefetch_i64scatter_pd): Ditto.
28573 (_mm512_prefetch_i64scatter_ps): Ditto.
28574 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
28575 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
28576 * config/i386/avx512vlintrin.h
28577 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
28578 (_mm_mmask_i32gather_ps): Ditto.
28579 (_mm256_mmask_i32gather_pd): Ditto.
28580 (_mm_mmask_i32gather_pd): Ditto.
28581 (_mm256_mmask_i64gather_ps): Ditto.
28582 (_mm_mmask_i64gather_ps): Ditto.
28583 (_mm256_mmask_i64gather_pd): Ditto.
28584 (_mm_mmask_i64gather_pd): Ditto.
28585 (_mm256_mmask_i32gather_epi32): Ditto.
28586 (_mm_mmask_i32gather_epi32): Ditto.
28587 (_mm256_mmask_i32gather_epi64): Ditto.
28588 (_mm_mmask_i32gather_epi64): Ditto.
28589 (_mm256_mmask_i64gather_epi32): Ditto.
28590 (_mm_mmask_i64gather_epi32): Ditto.
28591 (_mm256_mmask_i64gather_epi64): Ditto.
28592 (_mm_mmask_i64gather_epi64): Ditto.
28593 (_mm256_i32scatter_ps): Change __addr type to void*.
28594 (_mm256_mask_i32scatter_ps): Ditto.
28595 (_mm_i32scatter_ps): Ditto.
28596 (_mm_mask_i32scatter_ps): Ditto.
28597 (_mm256_i32scatter_pd): Ditto.
28598 (_mm256_mask_i32scatter_pd): Ditto.
28599 (_mm_i32scatter_pd): Ditto.
28600 (_mm_mask_i32scatter_pd): Ditto.
28601 (_mm256_i64scatter_ps): Ditto.
28602 (_mm256_mask_i64scatter_ps): Ditto.
28603 (_mm_i64scatter_ps): Ditto.
28604 (_mm_mask_i64scatter_ps): Ditto.
28605 (_mm256_i64scatter_pd): Ditto.
28606 (_mm256_mask_i64scatter_pd): Ditto.
28607 (_mm_i64scatter_pd): Ditto.
28608 (_mm_mask_i64scatter_pd): Ditto.
28609 (_mm256_i32scatter_epi32): Ditto.
28610 (_mm256_mask_i32scatter_epi32): Ditto.
28611 (_mm_i32scatter_epi32): Ditto.
28612 (_mm_mask_i32scatter_epi32): Ditto.
28613 (_mm256_i32scatter_epi64): Ditto.
28614 (_mm256_mask_i32scatter_epi64): Ditto.
28615 (_mm_i32scatter_epi64): Ditto.
28616 (_mm_mask_i32scatter_epi64): Ditto.
28617 (_mm256_i64scatter_epi32): Ditto.
28618 (_mm256_mask_i64scatter_epi32): Ditto.
28619 (_mm_i64scatter_epi32): Ditto.
28620 (_mm_mask_i64scatter_epi32): Ditto.
28621 (_mm256_i64scatter_epi64): Ditto.
28622 (_mm256_mask_i64scatter_epi64): Ditto.
28623 (_mm_i64scatter_epi64): Ditto.
28624 (_mm_mask_i64scatter_epi64): Ditto.
28625 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
28626 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
28627 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
28628 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
28629 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
28630 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
28631 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
28632 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
28633 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
28634 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
28635 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
28636 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
28637 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
28638 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
28639 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
28640 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
28641 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
28642 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
28643 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
28644 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
28645 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
28646 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
28647 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
28648 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
28649 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
28650 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
28651 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
28652 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
28653 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
28654 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
28655 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
28656 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
28657 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
28658 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
28659 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
28660 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
28661 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
28662 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
28663 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
28664 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
28665 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
28666 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
28667 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
28668 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
28669 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
28670 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
28671 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
28672 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
28673 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
28674 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
28675 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
28676 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
28677 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
28678 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
28679 definitions accordingly.
28681 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
28682 Kuan-Lin Chen <kuanlinchentw@gmail.com>
28685 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
28688 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
28691 * ira-conflicts.c (ira_build_conflicts): Update total conflict
28692 hard regs for inner regno.
28694 2017-01-17 Martin Liska <mliska@suse.cz>
28697 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
28698 assumption and add comment.
28700 2017-01-17 Nathan Sidwell <nathan@acm.org>
28702 * ipa-visibility.c (localize_node): New function, broken out of ...
28703 (function_and_variable_visibility): ... here. Call it.
28705 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28707 PR middle-end/77445
28708 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
28709 correctly set frequency of oudgoing edge.
28710 (duplicate_thread_path): Fix profile updating.
28712 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28715 * configure.ac: Add GCC_BASE_VER.
28716 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
28717 version from BASE-VER file.
28718 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
28719 (gcc.o): Depend on $(BASEVER).
28720 * common.opt (dumpfullversion): New option.
28721 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
28722 * doc/invoke.texi: Document -dumpfullversion.
28723 * doc/install.texi: Document --with-gcc-major-version-only.
28724 * configure: Regenerated.
28726 2017-01-17 Richard Biener <rguenther@suse.de>
28728 PR tree-optimization/71433
28729 * tree-vrp.c (register_new_assert_for): Merge same asserts
28730 on all incoming edges.
28731 (process_assert_insertions_for): Handle insertions at the
28734 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
28736 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
28737 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
28739 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
28742 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
28745 2017-01-17 Alan Modra <amodra@gmail.com>
28748 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
28749 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
28750 symbolic stack limit when pic.
28752 2017-01-16 Martin Sebor <msebor@redhat.com>
28754 PR tree-optimization/78608
28755 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
28757 2017-01-16 Jeff Law <law@redhat.com>
28760 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
28761 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
28762 for several include directories that may be relative to sysroot.
28763 * config/i386/x-mingw32 (gplus_includedir): Define.
28764 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
28765 (native_system_includedir): Likewise.
28766 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
28767 override if TARGET_SYSTEM_ROOT is defined.
28768 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
28770 PR tree-optimization/79090
28771 PR tree-optimization/33562
28772 PR tree-optimization/61912
28773 PR tree-optimization/77485
28774 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
28775 and computed trims into the dump file.
28777 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28779 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
28781 2017-01-16 Jakub Jelinek <jakub@redhat.com>
28784 * gimplify.c (gimplify_init_constructor): If want_value and
28785 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
28789 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
28790 sequence. Formatting fixes.
28791 (doloop_optimize): Formatting fixes.
28794 * gcc.c (debug_level_greater_than_spec_func): New function.
28795 (static_spec_functions): Add debug-level-gt spec function.
28796 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
28798 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28799 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28800 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
28801 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
28802 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
28803 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
28805 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
28807 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
28808 QImode fixups to general and mask registers only.
28810 2017-01-16 Carl Love <cel@us.ibm.com>
28812 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
28813 for built-in functions
28814 vector signed char vec_nabs (vector signed char)
28815 vector signed short vec_nabs (vector signed short)
28816 vector signed int vec_nabs (vector signed int)
28817 vector signed long long vec_nabs (vector signed long long)
28818 vector float vec_nabs (vector float)
28819 vector double vec_nabs (vector double)
28820 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
28822 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
28823 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
28824 * doc/extend.texi: Update the documentation file for the new built-in
28827 2017-01-16 Martin Sebor <msebor@redhat.com>
28829 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
28832 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28834 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
28835 UNSPEC_VSX__XXSPLTD to require special splat handling.
28837 2017-01-16 David Malcolm <dmalcolm@redhat.com>
28840 * system.h: Poison strndup.
28842 2017-01-16 Alan Modra <amodra@gmail.com>
28845 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
28848 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
28850 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
28852 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
28854 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
28855 call recog here. Assert that INSN_CODE (insn) is non-negative.
28857 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28860 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
28862 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
28863 in the currently scheduled RTL fragment.
28865 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28867 PR rtl-optimization/78751
28868 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
28871 2017-01-14 Jeff Law <law@redhat.com>
28873 PR tree-optimization/79090
28874 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
28875 variable length stores.
28876 (compute_trims): Delete dead assignment to *trim_tail.
28877 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
28880 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
28882 PR rtl-optimization/78626
28883 PR rtl-optimization/78727
28884 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
28885 of a block, and split such blocks after everything else is finished.
28887 2017-01-14 Alan Modra <amodra@gmail.com>
28890 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
28891 target legitimate_combined_insn.
28892 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
28893 (rs6000_legitimate_combined_insn): New function.
28894 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
28896 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
28897 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
28898 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
28900 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
28902 * doc/frontends.texi (G++ and GCC): Remove references to Java.
28904 2017-01-13 Jeff Law <law@redhat.com>
28906 PR tree-optimization/33562
28907 PR tree-optimization/61912
28908 PR tree-optimization/77485
28909 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
28911 (delete_dead_assignment): Likewise.
28912 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
28913 statement to delete_dead_call and delete_dead_assignment.
28915 2017-01-13 David Malcolm <dmalcolm@redhat.com>
28918 * substring-locations.c (format_warning_va): Strengthen case 1 so
28919 that both endpoints of the substring must be within the format
28920 range for just the substring to be printed.
28922 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
28924 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
28925 * config/i386/i386.c (ix86_target_string): Add missing options
28926 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
28927 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
28928 flags_other and ix86_target_other to flags2_other. Display unknown
28930 (ix86_valid_target_attribute_inner_p): Add missing options and
28931 reorder options by implied ISAs, as in ix86_target_string.
28933 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28935 * hash-table.h (hash_table::too_empty_p): New function.
28936 (hash_table::expand): Use it.
28937 (hash_table::traverse): Likewise.
28938 (hash_table::empty_slot): Use sizeof (value_type) instead of
28939 sizeof (PTR) to convert bytes to elements. Shrink the table
28940 if the current size is excessive for the current number of
28943 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28945 * ira-costs.c (record_reg_classes): Break from the inner loop
28946 early once alt_fail is known to be true. Update outer loop
28947 handling accordingly.
28949 2017-01-13 Jeff Law <law@redhat.com>
28951 * tree-ssa-dse.c (decrement_count): New function.
28952 (increment_start_addr, maybe_trim_memstar_call): Likewise.
28953 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
28954 when we know the partially dead statement is a mem* function.
28956 PR tree-optimization/61912
28957 PR tree-optimization/77485
28958 * tree-ssa-dse.c: Include expr.h.
28959 (maybe_trim_constructor_store): New function.
28960 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
28962 PR tree-optimization/33562
28963 PR tree-optimization/61912
28964 PR tree-optimization/77485
28965 * doc/invoke.texi: Document new dse-max-object-size param.
28966 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
28967 * tree-ssa-dse.c: Include params.h.
28968 (dse_store_status): New enum.
28969 (initialize_ao_ref_for_dse): New, partially extracted from
28971 (valid_ao_ref_for_dse, normalize_ref): New.
28972 (setup_live_bytes_from_ref, compute_trims): Likewise.
28973 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
28974 (maybe_trim_partially_dead_store): Likewise.
28975 (maybe_trim_complex_store): Likewise.
28976 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
28977 Track what bytes live from the original store. Return tri-state
28978 for dead, partially dead or live.
28979 (dse_dom_walker): Add constructor, destructor and new private members.
28980 (delete_dead_call, delete_dead_assignment): New extracted from
28982 (dse_optimize_stmt): Make a member of dse_dom_walker.
28983 Use initialize_ao_ref_for_dse.
28985 PR tree-optimization/33562
28986 PR tree-optimization/61912
28987 PR tree-optimization/77485
28988 * sbitmap.h (bitmap_count_bits): Prototype.
28989 (bitmap_clear_range, bitmap_set_range): Likewise.
28990 * sbitmap.c (bitmap_clear_range): New function.
28991 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
28993 2017-01-13 Martin Liska <mliska@suse.cz>
28996 * function.c (set_cfun): Add new argument force.
28997 * function.h (set_cfun): Likewise.
28998 * ipa-inline-transform.c (inline_call): Use the function when
28999 strict alising from is dropped for function we inline to.
29001 2017-01-13 Richard Biener <rguenther@suse.de>
29003 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29004 for dumping GIMPLE INTEGER_CSTs.
29006 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29008 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29009 to 201112L since C++17.
29011 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29014 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29015 if -fsanitize=kernel-address is present.
29017 2017-01-13 Richard Biener <rguenther@suse.de>
29019 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29020 as _Literal ( type ) number in case usual suffixes do not
29021 preserve all information.
29023 2017-01-13 Richard Biener <rguenther@suse.de>
29025 PR tree-optimization/77283
29026 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29027 and ssa-iterators.h.
29028 (is_feasible_trace): Implement a cost model based on joiner
29031 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29034 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29035 char or short to __float128/_Float128 directly.
29037 2017-01-12 Martin Sebor <msebor@redhat.com>
29039 to -Wformat-overflow.
29040 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29041 (min_bytes_remaining): Same.
29042 (get_string_length): Same.
29043 (format_string): Same.
29044 (format_directive): Same.
29046 (pass_sprintf_length::handle_gimple_call): Same.
29048 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29050 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29051 info.nowrite calls with no lhs that can't throw. Return bool
29052 whether gsi_remove has been called or not.
29053 (pass_sprintf_length::handle_gimple_call): Return bool whether
29054 try_substitute_return_value called gsi_remove. Formatting fix.
29055 (pass_sprintf_length::execute): Don't use gsi_remove if
29056 handle_gimple_call returned true.
29059 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29060 be removed due to side-effects, don't remove following barrier nor
29061 turn the successor edge into fallthru edge.
29063 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29066 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29067 element-reversing loads and stores as not swappable.
29069 2017-01-12 Nathan Sidwell <nathan@acm.org>
29070 Nicolai Stange <nicstange@gmail.com>
29072 * combine.c (try_combine): Don't ignore result of overlap checking
29073 loop. Combine overlap & asm check into single loop.
29075 2017-01-12 Richard Biener <rguenther@suse.de>
29077 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29078 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
29080 2017-01-12 Richard Biener <rguenther@suse.de>
29082 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
29083 and TS_TARGET_OPTION directly derive from TS_BASE.
29084 * tree-core.h (tree_optimization_option): Derive from tree_base.
29085 (tree_target_option): Likewise.
29087 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
29089 * config/i386/i386.c (memory_address_length): Increase len
29090 only when rip_relative_addr_p returns false.
29092 2017-01-11 Julia Koval <julia.koval@intel.com>
29094 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
29095 (OPTION_MASK_ISA_SGX_SET): New.
29096 (ix86_handle_option): Handle OPT_msgx.
29097 * config.gcc: Added sgxintrin.h.
29098 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
29099 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
29100 * config/i386/i386.c (ix86_target_string): Add -msgx.
29102 (ix86_option_override_internal): Handle new options.
29103 (ix86_valid_target_attribute_inner_p): Add sgx.
29104 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
29105 * config/i386/i386.opt: Add msgx.
29106 * config/i386/sgxintrin.h: New file.
29107 * config/i386/x86intrin.h: Add sgxintrin.h.
29109 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29112 * fold-const.c (maybe_nonzero_address): Return 1 for function
29114 (tree_single_nonzero_warnv_p): Don't handle function local objects
29118 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
29121 2017-01-11 David Malcolm <dmalcolm@redhat.com>
29124 * opts.c: Include "spellcheck.h"
29125 (struct string_fragment): New struct.
29126 (struct edit_distance_traits<const string_fragment &>): New
29128 (get_closest_sanitizer_option): New function.
29129 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
29131 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29133 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
29135 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
29136 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
29137 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
29138 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
29139 for initial die_offset if dwarf_split_debug_info.
29140 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
29141 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
29143 (output_skeleton_debug_sections): Formatting fix. Use
29144 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
29145 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
29147 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
29149 * config/arm/cortex-a53.md: Add bypasses for
29150 cortex_a53_r2f_cvt.
29151 (cortex_a53_r2f): Only use for transfers.
29152 (cortex_a53_f2r): Likewise.
29153 (cortex_a53_r2f_cvt): Add reservation for conversions.
29154 (cortex_a53_f2r_cvt): Likewise.
29156 2017-01-11 Tamar Christina <tamar.christina@arm.com>
29158 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
29159 to all inlined functions, change static to extern.
29161 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
29164 * config/arm/arm.c (legitimize_pic_address): Handle reference to
29166 (arm_assemble_integer): Likewise.
29168 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
29170 * config.gcc: Use new awk script to check CPU, FPU and architecture
29171 parameters for --with-... options.
29172 * config/arm/parsecpu.awk: New file
29173 * config/arm/arm-cpus.in: New file.
29174 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
29176 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
29178 * config/arm/t-arm: Update dependency rules.
29179 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
29180 of processing .def files.
29181 * config/arm/genopt.sh: Deleted.
29182 * config/arm/gentune.sh: Deleted.
29183 * config/arm/arm-cores.def: Deleted.
29184 * config/arm/arm-arches.def: Deleted.
29185 * config/arm/arm-fpus.def: Deleted.
29186 * config/arm/arm-tune.md: Regenerated.
29187 * config/arm/arm-tables.opt: Regenerated.
29188 * config/arm/arm-cpu.h: New generated file.
29189 * config/arm/arm-cpu-data.h: New generated file.
29190 * config/arm/arm-cpu-cdata.h: New generated file.
29192 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
29195 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
29197 (input_varpool_node): Unpack dynamically_initialized bit.
29199 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
29201 PR rtl-optimization/79032
29202 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
29203 the alignment of the adjusted memory reference against that of MODE,
29204 instead of the alignment of the original memory reference.
29206 2017-01-11 Martin Jambor <mjambor@suse.cz>
29208 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
29210 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
29211 decorated functions.
29213 2017-01-11 Richard Biener <rguenther@suse.de>
29215 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
29216 set range/nonnull info for PHI results. Do not set it on
29217 stmts marked for removal.
29219 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
29221 * expr.c (store_field): In the bitfield case, fetch the return value
29222 from the registers before applying a single big-endian adjustment.
29223 Always do a final load for a BLKmode value not larger than a word.
29225 2017-01-10 David Malcolm <dmalcolm@redhat.com>
29228 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29229 that we correctly handle column numbers greater than
29230 LINE_MAP_MAX_COLUMN_NUMBER.
29232 2017-01-10 Martin Sebor <msebor@redhat.com>
29234 PR middle-end/78245
29235 * gimple-ssa-sprintf.c (get_destination_size): Call
29236 {init,fini}object_sizes.
29237 * tree-object-size.c (addr_object_size): Adjust.
29238 (pass_through_call): Adjust.
29239 (pass_object_sizes::execute): Adjust.
29240 * tree-object-size.h (fini_object_sizes): Declare.
29242 2017-01-10 Martin Sebor <msebor@redhat.com>
29244 PR tree-optimization/78775
29245 * builtins.c (get_size_range): Move...
29246 * calls.c: ...to here.
29247 (alloc_max_size): Accept zero argument.
29248 (operand_signed_p): Remove.
29249 (maybe_warn_alloc_args_overflow): Call get_size_range.
29250 * calls.h (get_size_range): Declare.
29252 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
29254 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
29255 from TI's devices.csv file as of September 2016.
29256 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
29258 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
29260 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
29261 * doc/invoke.texi: Likewise.
29262 * doc/md.texi: Likewise.
29263 * doc/objc.texi: Likewise.
29265 2017-01-10 Joshua Conner <joshconner@google.com>
29267 * config/arm/fuchsia-elf.h: New file.
29268 * config/fuchsia.h: New file.
29269 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
29270 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
29272 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
29274 2016-01-10 Richard Biener <rguenther@suse.de>
29276 PR tree-optimization/79034
29277 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29278 Propagate out degenerate PHIs in the joiner.
29280 2017-01-10 Martin Liska <mliska@suse.cz>
29282 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
29283 (sort_congruence_classes_by_decl_uid): Likewise.
29284 (sort_congruence_class_groups_by_decl_uid): Likewise.
29285 (sem_item_optimizer::merge_classes): Sort class, groups in these
29286 classes and members in the groups by DECL_UID of declarations.
29287 This would make merge operations stable.
29289 2017-01-10 Martin Liska <mliska@suse.cz>
29291 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
29292 usage of m_classes_vec.
29293 (sem_item_optimizer::~sem_item_optimizer): Likewise.
29294 (sem_item_optimizer::get_group_by_hash): Likewise.
29295 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
29296 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
29297 (sem_item_optimizer::verify_classes): Likewise.
29298 (sem_item_optimizer::process_cong_reduction): Likewise.
29299 (sem_item_optimizer::dump_cong_classes): Likewise.
29300 (sem_item_optimizer::merge_classes): Likewise.
29301 * ipa-icf.h (congruence_class_hash): Rename from
29302 congruence_class_group_hash. Remove declaration of m_classes_vec.
29304 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
29306 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
29307 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
29308 * config.gcc: Add avx512vpopcntdqintrin.h.
29309 * config/i386/avx512vpopcntdqintrin.h: New.
29310 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
29311 * config/i386/i386-builtin-types.def: Add new types.
29312 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
29313 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
29314 __builtin_ia32_vpopcountq_v8di_mask): New.
29315 * config/i386/i386-c.c (ix86_target_macros_internal): Define
29316 __AVX512VPOPCNTDQ__.
29317 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
29318 (PTA_AVX512VPOPCNTDQ): Define.
29319 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
29320 TARGET_AVX512VPOPCNTDQ_P): Define.
29321 * config/i386/i386.opt: Add mavx512vpopcntdq.
29322 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
29323 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
29325 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29327 PR middle-end/77484
29328 * predict.def (PRED_CALL): Set to 67.
29330 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
29332 * expr.c (store_field): In the bitfield case, if the value comes from
29333 a function call and is of an aggregate type returned in registers, do
29334 not modify the field mode; extract the value in all cases if the mode
29335 is BLKmode and the size is not larger than a word.
29337 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
29340 * config/i386/cpuid.h: Fix undefined behavior.
29342 2017-01-04 Jeff Law <law@redhat.com>
29344 PR tree-optimization/79007
29345 PR tree-optimization/67955
29346 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
29347 conservative for pt.null when flag_non_call_exceptions is on.
29349 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29351 PR translation/79019
29352 PR translation/79020
29353 * params.def (PARAM_INLINE_MIN_SPEEDUP,
29354 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
29355 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
29357 * config/avr/avr.opt (maccumulate-args): Likewise.
29358 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
29359 * common.opt (freport-bug): Likewise.
29360 * cif-code.def (CIF_FINAL_ERROR): Likewise.
29361 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
29362 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
29363 translatable string.
29364 * config/i386/i386.c (function_value_32): Likewise.
29365 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
29366 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
29368 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
29369 * common/config/msp430/msp430-common.c (msp430_handle_option):
29371 * symtab.c (symtab_node::verify_base): Likewise.
29372 * opts.c (set_debug_level): Likewise.
29373 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
29374 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
29375 missing whitespace to translatable strings.
29376 * config/avr/avr.md (bswapsi2): Fix typo in comment.
29377 * config/sh/superh.h: Likewise.
29378 * config/i386/xopintrin.h: Likewise.
29379 * config/i386/znver1.md: Likewise.
29380 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
29381 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
29382 * double-int.h (struct double_int): Likewise.
29383 * double-int.c (div_and_round_double): Likewise.
29384 * wide-int.cc: Likewise.
29385 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
29386 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
29387 * cfgcleanup.c (crossjumps_occured): Renamed to ...
29388 (crossjumps_occurred): ... this.
29389 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
29392 PR tree-optimization/78899
29393 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
29394 returning bool return struct loop *, NULL for failure and the new
29396 (versionable_outer_loop_p): Don't version outer loop if it has
29397 dont_vectorized bit set.
29398 (tree_if_conversion): When versioning outer loop, ensure
29399 tree_if_conversion is performed also on the inner loop of the
29400 non-vectorizable outer loop copy.
29401 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
29402 LOOP_VECTORIZED in inner loop of the scalar outer loop and
29403 prevent vectorization of it.
29404 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
29405 the outer loop vectorization of the non-scalar version is attempted
29406 before vectorization of the inner loop in scalar version. If
29407 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
29408 vectorization of its inner loop.
29409 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
29410 has 2 inner loops, rename also on edges from bb whose single pred
29411 is outer_loop->header. Fix typo in function comment.
29413 2017-01-09 Martin Sebor <msebor@redhat.com>
29416 * asan.c (asan_emit_stack_protection): Increase local buffer size
29417 to avoid snprintf truncation warning.
29419 2017-01-09 Andrew Pinski <apinski@cavium.com>
29421 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
29422 to reference thunderx2t99 for the tuning structure
29423 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
29425 (thunderx2t99_extra_costs): This.
29426 * config/aarch64/aarch64-tune.md: Regenerate.
29427 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
29428 (vulcan_addrcost_table): This.
29429 (vulcan_regmove_cost): Rename to ...
29430 (thunderx2t99_regmove_cost): This.
29431 (vulcan_vector_cost): Rename to ...
29432 (thunderx2t99_vector_cost): this.
29433 (vulcan_branch_cost): Rename to ...
29434 (thunderx2t99_branch_cost): This.
29435 (vulcan_tunings): Rename to ...
29436 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
29437 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
29439 2017-01-09 Martin Jambor <mjambor@suse.cz>
29443 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
29444 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
29445 (propagate_vr_accross_jump_function): Use the above function for all
29446 value range computations for pass-through jump functions and type
29447 converasion from explicit value range values.
29448 (ipcp_propagate_stage): Do not attempt to deduce types of formal
29449 parameters from TYPE_ARG_TYPES.
29450 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
29451 (ipa_write_node_info): Stream type of the actual argument.
29452 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
29454 2017-01-09 Martin Liska <mliska@suse.cz>
29457 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
29458 (lookup_compiler): Do not show error message with have_E.
29460 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29462 PR tree-optimization/78938
29463 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
29464 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
29465 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
29466 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
29469 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29471 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
29474 2017-01-09 Richard Biener <rguenther@suse.de>
29476 PR tree-optimization/78997
29477 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
29478 name condition properly.
29480 2017-01-09 Richard Biener <rguenther@suse.de>
29483 * dwarf2out.c (is_cxx): New overload with context.
29484 (is_naming_typedef_decl): Use it.
29486 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29488 * invoke.texi (Option Summary): Correct spacing in option lists
29489 and add line breaks to fix over-long lines.
29491 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29493 PR middle-end/17660
29495 * extend.texi (Common Variable Attributes): Add xref to GCC
29496 Internals manual to explain mode attribute keywords.
29498 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29501 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
29502 and Preprocessor Options.
29503 (Options for Linking): Document -pthread here....
29504 (RS/6000 and PowerPC Options): ...not here.
29505 (Solaris 2 Options): ...or here.
29506 * doc/cppopts.texi: Document -pthread.
29508 2017-01-08 Martin Sebor <msebor@redhat.com>
29510 PR middle-end/77708
29511 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
29512 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
29513 New member functions.
29514 (format_directive): Used them.
29516 (pass_sprintf_length::handle_gimple_call): Same.
29517 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
29518 to avoid truncation for any argument.
29519 (extract_affine_mul): Same.
29520 * tree.c (get_file_function_name): Same.
29522 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29524 PR middle-end/77484
29525 * predict.def (PRED_INDIR_CALL): Set to 86.
29527 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29529 PR preprocessor/54124
29530 * doc/cppopts.texi: Reformat -d subtable to list the full name
29531 of the options. Add cross-reference to the docs for the general
29532 compiler -d options.
29533 * doc/invoke.texi (Developer Options): Add cross-reference to the
29534 preprocessor-specific -d option documentation.
29536 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29538 PR preprocessor/13498
29539 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
29540 redudant material, and reflect new command-line options.
29541 (System Headers): Likewise.
29543 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29545 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
29546 -isystem, and -idirafter. Copy-edit.
29547 * doc/cppopts.texi: Copy-edit. Remove contradiction about
29548 default for -ftrack-macro-expansion. Delete obsolete and
29549 badly-formatted implementation details about -fdebug-cpp output.
29550 * doc/cppwarnopts.texi: Copy-edit.
29552 2017-01-07 David Malcolm <dmalcolm@redhat.com>
29555 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29556 that the transition from a max line width >= 1<<10 to narrower
29557 lines works correctly.
29559 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
29561 * doc/options.texi (PerFunction): New.
29562 * opt-functions.awk (switch_flags): Map both Optimization and
29563 PerFunction to CL_OPTIMIZATION.
29564 * opth-gen.awk: Test for PerFunction flag along with
29566 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
29567 it only when the latter is present. Skip those that don't in
29568 the hash function generator.
29569 * common.opt (fvar-tracking): Mark as PerFunction instead of
29571 (fvar-tracking-assignments): Likewise.
29572 (fvar-tracking-assignments-toggle): Likewise.
29573 (fvar-tracking-uninit): Likewise.
29575 2017-01-07 Jakub Jelinek <jakub@redhat.com>
29577 PR translation/79018
29578 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
29581 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
29584 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
29585 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
29586 TARGET_LONG_JUMP_TABLE_OFFSETS.
29587 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
29588 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
29589 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
29590 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
29591 * config/m68k/m68k.md (tablejump expander): Likewise.
29592 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
29593 TARGET_LONG_JUMP_TABLE_OFFSETS.
29594 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
29595 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
29597 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
29598 David Holsgrove <david.holsgrove@xilinx.com>
29600 * common/config/microblaze/microblaze-common.c
29601 (TARGET_EXCEPT_UNWIND_INFO): Remove.
29602 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
29604 * config/microblaze/microblaze.c (microblaze_must_save_register)
29605 (microblaze_expand_epilogue, microblaze_return_addr): Handle
29607 (microblaze_eh_return): New function.
29608 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
29609 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
29610 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
29611 * config/microblaze/microblaze.md (eh_return): New pattern.
29613 2017-01-06 Jakub Jelinek <jakub@redhat.com>
29615 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
29616 GCC_DIAGNOSTIC_STRINGIFY): Define.
29618 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
29620 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29622 * config/arm/arm.md (<mcrr>): New.
29624 * config/arm/arm.c (arm_arch5te): New.
29625 (arm_option_override): Set arm_arch5te.
29626 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
29628 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
29629 (arm_mcrr_qualifiers): ... this. New.
29630 (MRRC_QUALIFIERS): Define to...
29631 (arm_mrrc_qualifiers): ... this. New.
29632 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
29634 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
29635 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
29636 (MRRCI, mrrc, MRRC): New.
29637 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
29638 VUNSPEC_MRRC2): New.
29640 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29642 * config/arm/arm.md (<mcr>): New.
29644 * config/arm/arm.c (arm_coproc_builtin_available): Add
29645 support for mcr, mrc, mcr2 and mrc2.
29646 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
29647 (arm_mcr_qualifiers): ... this. New.
29648 (MRC_QUALIFIERS): Define to ...
29649 (arm_mrc_qualifiers): ... this. New.
29650 (MCR_QUALIFIERS): Define to ...
29651 (arm_mcr_qualifiers): ... this. New.
29652 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
29654 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
29655 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
29656 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
29657 VUNSPEC_MRC2): New.
29659 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29661 * config/arm/arm.md (*ldc): New.
29665 * config/arm/arm.c (arm_coproc_builtin_available): Add
29666 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
29667 (arm_coproc_ldc_stc_legitimate_address): New.
29668 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
29669 'qualifier_const_pointer'.
29670 (LDC_QUALIFIERS): Define to...
29671 (arm_ldc_qualifiers): ... this. New.
29672 (STC_QUALIFIERS): Define to...
29673 (arm_stc_qualifiers): ... this. New.
29674 * config/arm/arm-protos.h
29675 (arm_coproc_ldc_stc_legitimate_address): New.
29676 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
29677 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
29678 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
29679 stc2, stcl, stc2l): New.
29680 * config/arm/constraints.md (Uz): New.
29681 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
29682 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
29683 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
29684 VUNSPEC_STC2L): New.
29686 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29688 * config/arm/arm.md (<cdp>): New.
29689 * config/arm/arm.c (neon_const_bounds): Rename this ...
29690 (arm_const_bounds): ... this.
29691 (arm_coproc_builtin_available): New.
29692 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
29693 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
29694 (CDP_QUALIFIERS): Define to...
29695 (arm_cdp_qualifiers): ... this. New.
29697 (arm_expand_builtin_args): Add case for 6 arguments.
29698 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
29699 (arm_const_bounds): ... this.
29700 (arm_coproc_builtin_available): New.
29701 * config/arm/arm_acle.h (__arm_cdp): New.
29703 * config/arm/arm_acle_builtins.def (cdp): New.
29705 * config/arm/iterators.md (CDPI,CDP,cdp): New.
29706 * config/arm/neon.md: Rename all 'neon_const_bounds' to
29707 'arm_const_bounds'.
29708 * config/arm/types.md (coproc): New.
29709 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
29710 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
29711 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
29712 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
29714 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29716 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
29717 (UBINOP_QUALIFIERS): New.
29719 (acle_builtin_data): New. Change comment.
29720 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
29721 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
29722 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
29723 arm_acle_builtins.def.
29724 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
29725 (arm_init_acle_builtins): New.
29726 (CRC32_BUILTIN): Remove.
29727 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
29728 crc32cb, crc32ch and crc32cw.
29729 (arm_init_crc32_builtins): Remove.
29730 (arm_init_builtins): Use arm_init_acle_builtins rather
29731 than arm_init_crc32_builtins.
29732 (arm_expand_acle_builtin): New.
29733 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
29734 * config/arm/arm_acle_builtins.def: New.
29736 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29738 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
29739 (arm_builtin_datum): ... this.
29740 (arm_init_neon_builtin): Rename to ...
29741 (arm_init_builtin): ... this. Add a new parameters PREFIX
29742 and USE_SIG_IN_NAME.
29743 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
29744 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
29745 'arm_builtin_datum'.
29746 (arm_init_vfp_builtins): Likewise.
29747 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
29748 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
29749 (arm_expand_neon_args): Rename to ...
29750 (arm_expand_builtin_args): ... this. Rename builtin_arg
29751 enum values and differentiate between ARG_BUILTIN_MEMORY
29752 and ARG_BUILTIN_NEON_MEMORY.
29753 (arm_expand_neon_builtin_1): Rename to ...
29754 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
29755 values, arm_expand_builtin_args and add bool parameter NEON.
29756 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
29757 (arm_expand_vfp_builtin): Likewise.
29758 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
29760 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29762 PR middle-end/77484
29763 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
29764 * predict.c (tree_estimate_probability_bb): Reverse direction of
29765 polymorphic call predictor.
29767 2017-01-06 David Malcolm <dmalcolm@redhat.com>
29769 * passes.c (execute_one_pass): Split out pass-skipping logic into...
29770 (determine_pass_name_match): ...this new function and...
29771 (should_skip_pass_p): ...this new function.
29773 2017-01-06 Nathan Sidwell <nathan@acm.org>
29775 * ipa-visibility.c (function_and_variable_visibility): Reformat
29776 comments and long lines. Remove extrneous if.
29777 * symtab.c (symtab_node::make_decl_local): Fix code format.
29778 (symtab_node::set_section_for_node): Fix comment typo.
29780 2017-01-06 Martin Liska <mliska@suse.cz>
29783 * lra-constraints.c: Rename invariant to lra_invariant.
29784 * predict.c (set_even_probabilities): Initialize e to NULL.
29786 2017-01-05 Martin Sebor <msebor@redhat.com>
29788 PR tree-optimization/78910
29789 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
29790 (format_integer): Correct off-by-one error in the handling
29791 of precision with negative numbers in signed conversions..
29793 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
29795 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
29797 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29799 PR tree-optimization/71016
29800 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
29801 factor_out_conditional_conversion. Formatting fix.
29802 (factor_out_conditional_conversion): Add cond_stmt argument.
29803 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
29804 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
29807 2017-01-05 David Malcolm <dmalcolm@redhat.com>
29809 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
29810 read-rtl-function.o, and selftest-rtl.o.
29811 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
29812 (selftest::aarch64_test_loading_full_dump): New function.
29813 (selftest::aarch64_run_selftests): New function.
29814 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
29815 selftest::aarch64_run_selftests.
29816 * config/i386/i386.c
29817 (selftest::ix86_test_loading_dump_fragment_1): New function.
29818 (selftest::ix86_test_loading_call_insn): New function.
29819 (selftest::ix86_test_loading_full_dump): New function.
29820 (selftest::ix86_test_loading_unspec): New function.
29821 (selftest::ix86_run_selftests): Call the new functions.
29822 * emit-rtl.c (maybe_set_max_label_num): New function.
29823 * emit-rtl.h (maybe_set_max_label_num): New decl.
29824 * function.c (instantiate_decls): Guard call to
29825 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
29826 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
29828 * gensupport.c (gen_reader::gen_reader): Pass "false"
29829 for new "compact" param of rtx_reader.
29830 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
29831 rather than an empty string for NULL strings.
29832 * read-md.c: Potentially include config.h rather than bconfig.h.
29833 Wrap include of errors.h with #ifdef GENERATOR_FILE.
29834 (have_error): New global, copied from errors.c.
29835 (md_reader::read_name): Rename to...
29836 (md_reader::read_name_1): ...this, adding "out_loc" param,
29837 and converting "missing name or number" to returning false, rather
29839 (md_reader::read_name): Reimplement in terms of read_name_1.
29840 (md_reader::read_name_or_nil): New function.
29841 (md_reader::read_string): Handle "(nil)" by returning NULL.
29842 (md_reader::md_reader): Add new param "compact".
29843 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
29844 (md_reader::read_file): New method.
29845 * read-md.h (md_reader::md_reader): Add new param "compact".
29846 (md_reader::read_file): New method.
29847 (md_reader::is_compact): New accessor.
29848 (md_reader::read_name): Convert return type from void to file_location.
29849 (md_reader::read_name_or_nil): New decl.
29850 (md_reader::read_name_1): New decl.
29851 (md_reader::m_compact): New field.
29852 (noop_reader::noop_reader): Pass "false" for new "compact" param
29854 (rtx_reader::rtx_reader): Add new "compact" param.
29855 (rtx_reader::read_rtx_operand): Make virtual and convert return
29856 type from void to rtx.
29857 (rtx_reader::read_until): New decl.
29858 (rtx_reader::handle_any_trailing_information): New virtual function.
29859 (rtx_reader::postprocess): New virtual function.
29860 (rtx_reader::finalize_string): New virtual function.
29861 (rtx_reader::m_in_call_function_usage): New field.
29862 (rtx_reader::m_reuse_rtx_by_id): New field.
29863 * read-rtl-function.c: New file.
29864 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
29865 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
29866 (selftest::verify_three_block_rtl_cfg): New decl.
29867 * read-rtl-function.h: New file.
29868 * read-rtl.c: Potentially include config.h rather than bconfig.h.
29869 For host, include function.h, memmodel.h, and emit-rtl.h.
29870 (one_time_initialization): New function.
29871 (struct compact_insn_name): New struct.
29872 (compact_insn_names): New array.
29873 (find_code): Handle insn codes in compact dumps.
29874 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
29875 (bind_subst_iter_and_attr): Likewise.
29876 (add_condition_to_string): Likewise.
29877 (add_condition_to_rtx): Likewise.
29878 (apply_attribute_uses): Likewise.
29879 (add_current_iterators): Likewise.
29880 (apply_iterators): Likewise.
29881 (initialize_iterators): Guard usage of apply_subst_iterator with
29882 #ifdef GENERATOR_FILE.
29883 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
29884 (md_reader::read_mapping): Likewise.
29885 (add_define_attr_for_define_subst): Likewise.
29886 (add_define_subst_attr): Likewise.
29887 (read_subst_mapping): Likewise.
29888 (check_code_iterator): Likewise.
29889 (rtx_reader::read_rtx): Likewise. Move one-time initialization
29891 (one_time_initialization): New function.
29892 (rtx_reader::read_until): New method.
29893 (read_flags): New function.
29894 (parse_reg_note_name): New function.
29895 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
29896 Handle reuse_rtx ids.
29897 Wrap iterator lookup within #ifdef GENERATOR_FILE.
29898 Add parsing support for RTL dumps, mirroring the special-cases in
29899 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
29900 values, and calling handle_any_trailing_information.
29901 (rtx_reader::read_rtx_operand): Convert return type from void
29902 to rtx, returning return_rtx. Handle case 'e'. Call
29903 finalize_string on XSTR and XTMPL fields.
29904 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
29905 "(nil)" values were omitted. Call the postprocess vfunc on the
29907 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
29908 class ctor. Initialize m_in_call_function_usage. Call
29909 one_time_initialization.
29910 * rtl-tests.c (selftest::test_uncond_jump): Call
29911 set_new_first_and_last_insn.
29912 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
29913 * selftest-rtl.c: New file.
29914 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
29915 (selftest::get_insn_by_uid): New decl.
29916 * selftest-run-tests.c (selftest::run_tests): Call
29917 read_rtl_function_c_tests.
29918 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
29919 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
29922 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
29924 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
29925 operands in a special way. Assert that pos+len <= mode precision.
29927 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29929 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
29930 3 argument Alias with unlimited for the negative form.
29931 (fno-vect-cost-model): Removed.
29933 2017-01-05 Martin Liska <mliska@suse.cz>
29935 * hsa-gen.c (gen_hsa_divmod): New function.
29936 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
29938 2017-01-05 Martin Liska <mliska@suse.cz>
29941 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
29944 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29946 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
29947 small constant length operands.
29949 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29951 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
29952 between loop iterations.
29954 2017-01-05 Martin Liska <mliska@suse.cz>
29957 * gimplify.c (gimplify_decl_expr): Compare to
29958 asan_poisoned_variables instread of checking flags.
29959 (gimplify_target_expr): Likewise.
29960 (gimplify_expr): Likewise.
29961 (gimplify_function_tree): Conditionally initialize
29962 asan_poisoned_variables.
29964 2017-01-04 Jeff Law <law@redhat.com>
29966 PR tree-optimizatin/78812
29967 * rtl.h (contains_mem_rtx_p): Prototype.
29968 * ifcvt.c (containts_mem_rtx_p): Move from here to...
29969 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
29970 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
29971 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
29972 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
29974 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29976 * input.c (assert_char_at_range): Default-initialize actual_range.
29978 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29980 * df-scan.c (df_ref_create_structure): Make regno unsigned,
29981 to match the caller.
29983 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29985 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
29986 insns after final jump in test to emit dummy move.
29988 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29990 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
29991 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
29993 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
29995 * multiple_target.c (create_dispatcher_calls): Init e_next.
29996 * tree-ssa-loop-split.c (split_loop): Init border.
29997 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30000 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30005 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30006 (altivec_register_operand): Do not return true if the operand
30007 contains a SUBREG mixing SImode and SFmode.
30008 (vsx_register_operand): Likewise.
30009 (vsx_reg_sfsubreg_ok): New predicate.
30010 (vfloat_operand): Do not return true if the operand contains a
30011 SUBREG mixing SImode and SFmode.
30012 (vint_operand): Likewise.
30013 (vlogical_operand): Likewise.
30014 (gpc_reg_operand): Likewise.
30015 (int_reg_operand): Likewise.
30016 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30017 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30018 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30020 (rs6000_emit_move_si_sf_subreg): New helper function.
30021 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30022 fixup SUBREGs involving SImode and SFmode.
30023 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30024 numbers for the new peephole2 optimization.
30025 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30026 the GLIBC math library that do AND/IOR/XOR operations on single
30027 precision floating point.
30028 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30029 target macros to say whether we need to avoid SUBREGs mixing
30031 (TARGET_ALLOW_SF_SUBREG): Likewise.
30032 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30033 (UNSPEC_SI_FROM_SF): Likewise.
30034 (iorxor): Change spacing.
30035 (and_ior_xor): New iterator for AND, IOR, and XOR.
30036 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30037 (movdi_from_sf_zero_ext): Likewise.
30038 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30039 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30040 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30041 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30042 (fms<mode>4): Likewise.
30043 (fnma<mode>4): Likewise.
30044 (fnms<mode>4): Likewise.
30045 (nfma<mode>4): Likewise.
30046 (nfms<mode>4): Likewise.
30048 2017-01-04 Marek Polacek <polacek@redhat.com>
30051 * doc/invoke.texi: Document -Wpointer-compare.
30053 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30055 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30058 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30059 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30062 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30065 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30066 documentation of the powerpc_popcntb_ok attribute.
30067 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30068 code to issue warning messages if a requested CPU configuration is
30069 not supported by the binary (assembler and loader) toolchain.
30070 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30071 made to define a built-in function that has been disabled.
30072 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30073 made to define a built-in function that has been disabled.
30074 (altivec_init_builtins): Add comment explaining why definition
30075 of the DST built-in functions is not preceded by an assertion
30076 check. Add assertions to prevent ICE if attempts are made to
30077 define an altivec predicate or an abs* built-in function that has
30079 (htm_init_builtins): Add comment explaining why definition of the
30080 htm built-in functions is not preceded by an assertion check.
30082 2017-01-04 Jeff Law <law@redhat.com>
30084 PR tree-optimizatin/67955
30085 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
30086 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
30087 the points-to solution does not include pt_null. Use DECL_PT_UID
30090 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
30092 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
30093 Use gen_int_mode instead of gen_lopwart for const_int operands.
30095 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30097 PR tree-optimization/71563
30098 * match.pd: Simplify X << Y into X if Y is known to be 0 or
30099 out of range value - has low bits known to be zero.
30101 2017-01-04 Alan Modra <amodra@gmail.com>
30103 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
30104 * configure: Regenerate.
30105 * config.in: Regenerate.
30107 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30110 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
30111 a substring of the message, but strcmp with the whole message. Ifdef
30112 ENABLE_NLS, translate the message first using dgettext.
30114 2017-01-03 Jeff Law <law@redhat.com>
30116 PR tree-optimizatin/78856
30117 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
30118 (mark_threaded_blocks): Remove code to truncate thread paths that
30119 cross multiple loop headers. Instead invalidate the cached loop
30120 iteration information and handle case of a thread path walking
30121 into an irreducible region.
30123 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
30126 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
30127 assertions. Add support for doing the signbit if the IEEE 128-bit
30128 floating point value is in a GPR.
30129 * config/rs6000/rs6000.md (Fsignbit): Delete.
30130 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
30131 Update the length attribute if the value is in a GPR.
30132 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
30133 the sign or zero extension instruction, since the value is always 0/1.
30134 (signbit<mode>2_dm2): Delete using <Fsignbit>.
30137 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
30138 extracting SImode to a GPR register so that we can generate a
30139 store, limit the vector to be in a traditional Altivec register
30140 for the vextuwrx instruction.
30142 2017-01-03 Ian Lance Taylor <iant@google.com>
30144 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
30146 2017-01-03 Martin Sebor <msebor@redhat.com>
30148 PR tree-optimization/78696
30149 * gimple-ssa-sprintf.c (format_floating): Correct handling of
30150 precision. Use MPFR for %f for greater fidelity. Correct handling
30152 (pass_sprintf_length::compute_format_length): Set width and precision
30153 specified by asrerisk to void_node for vararg functions.
30154 (try_substitute_return_value): Adjust dump output.
30156 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
30158 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
30160 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
30162 * doc/invoke.texi (SPARC options): Document -mlra as the default.
30163 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
30164 -mlra/-mno-lra was passed to the compiler.
30166 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
30168 PR rtl-optimization/65618
30169 * emit-rtl.c (try_split): Move initialization of "before" and
30170 "after" to just before the call to emit_insn_after_setloc.
30172 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
30174 * doc/md.texi (Standard Names): Remove reference to Java frontend.
30176 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
30178 * dwarf2out.c (gen_enumeration_type_die): When
30179 -gno-strict-dwarf, add a DW_AT_encoding attribute.
30181 2017-01-03 Jakub Jelinek <jakub@redhat.com>
30183 PR tree-optimization/78965
30184 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
30185 Change first argument from const call_info & to call_info &. For %n
30186 set info.nowrite to false.
30188 PR middle-end/78901
30189 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
30190 possibly throwing calls.
30192 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
30193 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
30194 and fns handling, rather than in a separate case SSA_NAME.
30196 2017-01-02 Jeff Law <law@redhat.com>
30198 * config/darwin-driver.c (darwin_driver_init): Const-correctness
30199 fixes for first_period and second_period variables.
30201 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
30204 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
30205 (*insvqi_1): New insn pattern.
30206 (*insvqi_1_mem_rex64): Ditto.
30207 (*insvqi_2): Ditto.
30208 (*insvqi_3): Rename from *insvqi.
30210 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
30212 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
30214 * doc/cfg.texi (Edges): Remove reference to Java.
30215 (Maintaining the CFG): Ditto.
30217 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30219 PR middle-end/77674
30220 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
30221 transparent aliases.
30223 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30225 PR middle-end/77484
30226 * predict.def (PRED_CALL): Update hitrate.
30227 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
30228 * predict.c (tree_estimate_probability_bb): Split CALL predictor
30229 into direct/indirect/polymorphic variants.
30231 2017-01-01 Jakub Jelinek <jakub@redhat.com>
30233 Update copyright years.
30235 * gcc.c (process_command): Update copyright notice dates.
30236 * gcov-dump.c (print_version): Ditto.
30237 * gcov.c (print_version): Ditto.
30238 * gcov-tool.c (print_version): Ditto.
30239 * gengtype.c (create_file): Ditto.
30240 * doc/cpp.texi: Bump @copying's copyright year.
30241 * doc/cppinternals.texi: Ditto.
30242 * doc/gcc.texi: Ditto.
30243 * doc/gccint.texi: Ditto.
30244 * doc/gcov.texi: Ditto.
30245 * doc/install.texi: Ditto.
30246 * doc/invoke.texi: Ditto.
30248 Copyright (C) 2017 Free Software Foundation, Inc.
30250 Copying and distribution of this file, with or without modification,
30251 are permitted in any medium without royalty provided the copyright
30252 notice and this notice are preserved.