1 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
4 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
5 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Add TARGET_DOTPROD.
6 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
7 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75): Enable TARGET_DOTPROD.
8 (cortex-a75.cortex-a55): Likewise.
9 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
11 2017-10-16 Tamar Christina <tamar.christina@arm.com>
13 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
14 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS): New.
15 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane): new.
16 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
17 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
18 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
19 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
20 * config/arm/types.md (neon_dot, neon_dot_q): New.
21 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
23 2017-10-16 Tamar Christina <tamar.christina@arm.com>
25 * config/arm/arm.h (TARGET_DOTPROD): New.
26 * config/arm/arm.c (arm_arch_dotprod): New.
27 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
28 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
29 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
30 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
31 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
32 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
33 * doc/invoke.texi (armv8.2-a): Document dotprod
35 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
37 * i386.c (ix86_vec_cost): New function.
38 (ix86_rtx_costs): Handle vector operations better.
39 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
40 * x86-tune-costs.h: Add new costs to all tables.
42 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
44 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
46 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
47 divsd, sqrtss and sqrtsd
48 * x86-tune-costs.h: Add new entries to all costs.
49 (znver1_cost): Fix to match real instruction latencies.
51 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
52 Michael Collison <michael.collison@arm.com>
54 * compare-elim.c: Include emit-rtl.h.
55 (can_merge_compare_into_arith): New function.
56 (try_validate_parallel): Likewise.
57 (try_merge_compare): Likewise.
58 (try_eliminate_compare): Call the above when no previous clobber
60 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
63 2017-10-14 Jakub Jelinek <jakub@redhat.com>
67 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
68 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
72 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
73 to be any operand_equal_p operands. For & (B - 1) require
74 B to be power of 2. Recognize
75 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
77 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
80 * optabs.c (expand_memory_blockage): Fix call of
81 targetm.have_memory_blockage.
83 2017-10-14 Jakub Jelinek <jakub@redhat.com>
86 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
87 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
88 objects to extra_objs instead of overwriting it.
90 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
92 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
93 Use any_fp_register_operand as operand[3] predicate. Simplify
94 equality test for operands[2] and operands[4] memory location.
95 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
96 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
97 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
98 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
99 any_fp_register_operand as operand[1] predicate. Simplify
100 equality test for operands[0] and operands[3] memory location.
101 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
102 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
103 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
105 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
107 * target-insns.def: Add memory_blockage.
108 * optabs.c (expand_memory_blockage): New function.
109 (expand_asm_memory_barrier): Rename ...
110 (expand_asm_memory_blockage): ... to this.
111 (expand_mem_thread_fence): Call expand_memory_blockage
112 instead of expand_asm_memory_barrier.
113 (expand_mem_singnal_fence): Ditto.
114 (expand_atomic_load): Ditto.
115 (expand_atomic_store): Ditto.
116 * doc/md.texi (Standard Pattern Names For Generation):
117 Document memory_blockage instruction pattern.
119 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
121 * config/rl78/rl78.c (rl78_emit_libcall): New function.
122 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
123 * config/rl78/rl78.md: New define_expand "adddi3".
125 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
127 * cfghooks.c (verify_flow_info): Disable check that all probabilities
130 2017-10-13 Jeff Law <law@redhat.com>
132 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
134 2017-10-13 Jakub Jelinek <jakub@redhat.com>
137 * internal-fn.c (expand_mul_overflow): If both operands have
138 the same highpart of -1 or 0 and the topmost bit of lowpart
139 is different, overflow is if res <= 0 rather than res < 0.
141 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
143 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
144 TARGET_P9_VECTOR code for unaligned_load case.
146 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
148 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
150 2017-10-13 Nathan Sidwell <nathan@acm.org>
152 * tree-core.h (tree_contains_struct): Make bool.
153 * tree.c (tree_contains_struct): Likewise.
154 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
155 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
156 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
157 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
158 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
160 2017-10-13 Richard Biener <rguenther@suse.de>
162 * graphite-isl-ast-to-gimple.c
163 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
164 parameters and dominance check.
165 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
166 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
167 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
168 Do not update SSA form here or do intermediate IL verification.
169 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
170 (graphite_initialize): Remove check on the number of loops in
171 the function and inline into graphite_transform_loops.
172 (graphite_finalize): Inline into graphite_transform_loops.
173 (graphite_transform_loops): Perform SSA update and IL verification
175 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
177 2017-10-13 Richard Biener <rguenther@suse.de>
179 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
180 graphite_expression_type_precision): Avoid global constructor
182 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
183 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
184 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
185 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
186 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
187 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
189 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
192 * config/i386/i386.h (ix86_red_zone_size): New.
193 * config/i386/i386.md (push peephole2s): Replace
194 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
196 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
197 Alan Hayward <alan.hayward@arm.com>
198 David Sherwood <david.sherwood@arm.com>
200 * combine.c (can_change_dest_mode): Reject changes in
201 REGMODE_NATURAL_SIZE.
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 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
208 (expand_debug_source_expr): Likewise.
209 * combine.c (combine_simplify_rtx): Likewise.
210 * cse.c (fold_rtx): Likewise.
211 * fwprop.c (canonicalize_address): Likewise.
212 * targhooks.c (default_shift_truncation_mask): Likewise.
214 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
215 Alan Hayward <alan.hayward@arm.com>
216 David Sherwood <david.sherwood@arm.com>
218 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
219 (widened_mode): Likewise.
220 (expand_unop): Likewise.
221 * ree.c (transform_ifelse): Likewise.
222 (merge_def_and_ext): Likewise.
223 (combine_reaching_defs): Likewise.
224 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
226 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
227 Alan Hayward <alan.hayward@arm.com>
228 David Sherwood <david.sherwood@arm.com>
230 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
231 * combine.c (gen_lowpart_for_combine): Likewise.
232 * dwarf2out.c (rtl_for_decl_location): Likewise.
233 * final.c (alter_subreg): Likewise.
234 * rtlhooks.c (gen_lowpart_general): Likewise.
235 (gen_lowpart_if_possible): Likewise.
237 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
238 Alan Hayward <alan.hayward@arm.com>
239 David Sherwood <david.sherwood@arm.com>
241 * calls.c (expand_call): Use subreg_lowpart_offset.
242 * cse.c (cse_insn): Likewise.
243 * regcprop.c (copy_value): Likewise.
244 (copyprop_hardreg_forward_1): Likewise.
246 2017-10-13 Jakub Jelinek <jakub@redhat.com>
249 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
250 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
251 =Q constraints to +Q and into insn condition add check
252 that operands[0] and operands[1] are equal.
253 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
254 =Q constraints to +Q and into insn condition add check
255 that operands[0] is equal to either operands[1] or operands[2].
258 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
259 instead of handling MINUS_EXPR twice (once for each argument),
260 canonicalize operand order and handle just once, use rtype where
261 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
264 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
265 any values of __C while still being pattern recognizable as a simple
268 2017-10-13 Richard Biener <rguenther@suse.de>
270 PR tree-optimization/82451
272 2017-10-02 Richard Biener <rguenther@suse.de>
274 PR tree-optimization/82355
275 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
276 a mapping for the enclosing loop but avoid generating one for
278 (copy_bb_and_scalar_dependences): Remove premature codegen
279 error on PHIs in blocks duplicated into multiple places.
280 * graphite-scop-detection.c
281 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
282 in the region use it as loop and nest to analyze the DR in.
283 (try_generate_gimple_bb): Likewise.
284 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
285 (add_loop_constraints): For blocks in a loop not in the region
286 create a dimension with a single iteration.
287 * sese.h (gbb_loop_at_index): Remove assert.
289 * cfgloop.c (loop_preheader_edge): For the loop tree root
290 return the single successor of the entry block.
291 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
292 Reset the SCEV hashtable and niters.
293 * graphite-scop-detection.c
294 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
295 assert that we only have POLYNOMIAL_CHREC that vary in loops
296 contained in the region.
297 (scop_detection::graphite_can_represent_expr): Adjust.
298 (scop_detection::stmt_has_simple_data_refs_p): For loops
299 not in the region set loop to NULL. The nest is now the
300 entry edge to the region.
301 (try_generate_gimple_bb): Likewise.
302 * sese.c (scalar_evolution_in_region): Adjust for
303 instantiate_scev change.
304 * tree-data-ref.h (graphite_find_data_references_in_stmt):
305 Make nest parameter the edge into the region.
306 (create_data_ref): Likewise.
307 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
308 entry edge into a region and adjust instantiate_scev calls.
309 (create_data_ref): Likewise.
310 (graphite_find_data_references_in_stmt): Likewise.
311 (find_data_references_in_stmt): Pass the loop preheader edge
312 from the nest argument.
313 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
314 parameter the edge into the region.
315 (instantiate_parameters): Use the loop preheader edge as entry.
316 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
318 (get_instantiated_value_entry): Make instantiate_below parameter
319 the edge into the region.
320 (instantiate_scev_name): Likewise. Adjust dominance checks,
321 when we cannot use loop-based instantiation instantiate by
322 walking use-def chains.
323 (instantiate_scev_poly): Adjust.
324 (instantiate_scev_binary): Likewise.
325 (instantiate_scev_convert): Likewise.
326 (instantiate_scev_not): Likewise.
327 (instantiate_array_ref): Remove.
328 (instantiate_scev_3): Likewise.
329 (instantiate_scev_2): Likewise.
330 (instantiate_scev_1): Likewise.
331 (instantiate_scev_r): Do not blindly handle N-operand trees.
332 Do not instantiate array-refs. Handle all constants and invariants.
333 (instantiate_scev): Make instantiate_below parameter
334 the edge into the region.
335 (resolve_mixers): Use the loop preheader edge for the region
336 parameter to instantiate_scev_r.
337 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
339 2017-10-13 Richard Biener <rguenther@suse.de>
341 PR tree-optimization/82525
342 * graphite-isl-ast-to-gimple.c
343 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
345 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
346 Fail code generation when we cannot represent the isl integer.
347 (binary_op_to_tree): Elide modulo operations that are no-ops
348 in the type we code generate. Remove now superfluous code
351 2017-10-13 Richard Biener <rguenther@suse.de>
353 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
354 (scop_detection::harmful_loop_in_region): Remove premature
356 (scop_detection::graphite_can_represent_scev): We can handle
357 pointer IVs just fine.
359 2017-10-13 Alan Modra <amodra@gmail.com>
361 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
362 "Clobbers and Scratch Registers". Add paragraph on
363 alternative to clobbers for scratch registers and OpenBLAS
366 2017-10-13 Alan Modra <amodra@gmail.com>
368 * doc/extend.texi (Clobbers): Correct vax example. Delete old
369 example of a memory input for a string of known length. Move
370 commentary out of table. Add a number of new examples
371 covering array memory inputs.
373 2017-10-12 Martin Liska <mliska@suse.cz>
375 PR tree-optimization/82493
376 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
377 (test_range_functions): New function.
378 (sbitmap_c_tests): Likewise.
379 * selftest-run-tests.c (selftest::run_tests): Run new tests.
380 * selftest.h (sbitmap_c_tests): New function.
382 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
384 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
386 * config/rs6000/amo.h: Fix spacing issue.
388 2017-10-12 Jakub Jelinek <jakub@redhat.com>
391 * config/i386/i386.md (*ashl<mode>3_mask_1,
392 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
393 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
396 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
398 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
399 (profile_probability): Set max_probability
400 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
402 * profile-count.c (profile_probability::differs_from_p): Do not
403 rely on max_probaiblity == 10000
405 2017-10-12 Jeff Law <law@redhat.com>
407 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
410 2017-10-12 Martin Sebor <msebor@redhat.com>
414 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
415 (handle_alias_pairs): Call it.
416 * common.opt (-Wattribute-alias): New option.
417 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
418 * doc/invoke.texi (-Wattribute-alias): Document.
420 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
423 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
425 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
427 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
428 (make_hard_regno_born, make_hard_regno_dead): Update
429 bb_killed_pseudos and bb_gen_pseudos.
431 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
433 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
435 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
437 * config/alpha/alpha.c (alpha_split_conditional_move):
438 Use std::swap instead of manually swapping.
439 (alpha_stdarg_optimize_hook): Ditto.
440 (alpha_canonicalize_comparison): Ditto.
442 2017-10-12 Bin Cheng <bin.cheng@arm.com>
444 * tree-loop-distribution.c (struct builtin_info): New struct.
445 (struct partition): Refactor fields into struct builtin_info.
446 (partition_free): Free struct builtin_info.
447 (build_size_arg_loc, build_addr_arg_loc): Delete.
448 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
449 information from struct builtin_info.
450 (find_single_drs): New function refactored from classify_partition.
451 Also moved builtin validity checks to this function.
452 (compute_access_range, alloc_builtin): New functions.
453 (classify_builtin_st, classify_builtin_ldst): New functions.
454 (classify_partition): Refactor code into functions find_single_drs,
455 classify_builtin_st and classify_builtin_ldst.
456 (distribute_loop): Don't do runtime alias check when distributing
458 (find_seed_stmts_for_distribution): New function.
459 (pass_loop_distribution::execute): Refactor code finding seed
460 stmts into above function. Support distribution for the innermost
461 two-level loop nest. Adjust dump information.
463 2017-10-12 Bin Cheng <bin.cheng@arm.com>
465 * tree-loop-distribution.c: Adjust the general comment.
466 (NUM_PARTITION_THRESHOLD): New macro.
467 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
468 (classify_partition): Skip builtin pattern of loop nest's inner loop.
469 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
470 in call to build_partition_graph.
471 (finalize_partitions): New parameter. Make loop distribution more
472 conservative by fusing more partitions.
473 (distribute_loop): Don't do runtime alias check in case of loop nest
475 (find_seed_stmts_for_distribution): New function.
476 (prepare_perfect_loop_nest): New function.
477 (pass_loop_distribution::execute): Refactor code finding seed stmts
478 and loop nest into above functions. Support loop nest distribution.
479 Adjust dump information accordingly.
481 2017-10-12 Bin Cheng <bin.cheng@arm.com>
483 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
484 and set PTYPE_SEQUENTIAL for merged partition.
486 2017-10-12 Richard Biener <rguenther@suse.de>
488 PR tree-optimization/69728
490 2017-09-19 Richard Biener <rguenther@suse.de>
492 PR tree-optimization/69728
493 * graphite-sese-to-poly.c (schedule_error): New global.
494 (add_loop_schedule): Handle empty domain by failing the
496 (build_original_schedule): Handle schedule_error.
498 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
499 domain by returning an unchanged schedule.
501 2017-10-12 Jakub Jelinek <jakub@redhat.com>
503 * genrecog.c (validate_pattern): For VEC_SELECT verify that
504 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
506 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
508 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
511 2017-10-12 Jakub Jelinek <jakub@redhat.com>
514 * expr.c (store_field): Don't optimize away bitsize == 0 store
515 from CALL_EXPR with addressable return type.
517 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
519 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
520 * config/rs6000/rs6000.md (sel): Delete mode attribute.
521 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
522 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
523 TARGET_ISEL instead of TARGET_ISEL<sel>.
525 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
527 * config/rs6000/rs6000.c
528 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
530 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
532 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
534 (reg_or_cint_operand): Fix comment.
535 (reg_or_zero_operand): New predicate.
536 * config/rs6000/rs6000-protos.h (output_isel): Delete.
537 * config/rs6000/rs6000.c (output_isel): Delete.
538 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
539 instead of reg_or_cint_operand. Output instruction directly (not via
541 (isel_unsigned_<mode>): Ditto.
542 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
543 gpc_reg_operand. Add an instruction alternative for this. Output
544 instruction directly.
545 (*isel_reversed_unsigned_<mode>): Ditto.
547 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
549 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
550 (TARGET_CANONICALIZE_COMPARISON): Define.
552 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
555 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
556 Check whether the dest is REG before adding REG_EQUIV note.
558 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
561 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
563 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
564 (make_hard_regno_born, make_hard_regno_dead): Update
565 bb_killed_pseudos and bb_gen_pseudos.
567 2017-10-11 Nathan Sidwell <nathan@acm.org>
569 * incpath.h (enum incpath_kind): Name enum, prefix values.
570 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
571 * incpath.c (heads, tails): Use INC_MAX.
572 (add_env_var_paths, add_standard_paths): Use incpath_kind.
573 (merge_include_chains, split_quote_chain,
574 register_include_chains): Update incpath_kind names.
575 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
576 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
578 (add_framework_path, darwin_register_objc_includes): Likewise.
579 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
581 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
583 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
584 Do not use float_operator operator predicate.
585 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
586 * config/i386/predicates.md (float_operator): Remove predicate.
588 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
590 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
591 (*jccxf_i387): Ditto.
592 (*jcc<mode>_i387): Ditto.
593 (*jccu<mode>_i387): Ditto.
594 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
595 (*jcc_*_i387 splitters): Remove.
596 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
597 * config/i386/i386.c (ix86_split_fp_branch): Remove.
598 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
601 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
603 * profile-count.h (slow_safe_scale_64bit): New function.
604 (safe_scale_64bit): New inline.
605 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
606 * profile-count.c: Include wide-int.h
607 (slow_safe_scale_64bit): New.
609 2017-10-11 Nathan Sidwell <nathan@acm.org>
611 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
612 HAS_DECL_ASSEMBLER_NAME_P.
613 * gimple-expr.c (gimple_decl_printable_name: Check
614 HAS_DECL_ASSEMBLER_NAME_P too.
615 * ipa-utils.h (type_in_anonymous_namespace_p): Check
616 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
617 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
618 * passes.c (rest_of_decl_compilation): Check
619 HAS_DECL_ASSEMBLER_NAME_P too.
620 * recog.c (verify_changes): Likewise.
621 * tree-pretty-print.c (dump_decl_name): Likewise.
622 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
624 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
625 (SET_DECL_ASSEMBLER_NAME): Use it.
626 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
627 (COPY_DECL_ASSEMBLER_NAME): Likewise.
628 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
630 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
632 * config.gcc (i386, x86_64): Add extra objects.
633 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
634 (ix86_min_insn_size): Declare.
635 (ix86_issue_rate): Declare.
636 (ix86_adjust_cost): Declare.
637 (ia32_multipass_dfa_lookahead): Declare.
638 (ix86_macro_fusion_p): Declare.
639 (ix86_macro_fusion_pair_p): Declare.
640 (ix86_bd_has_dispatch): Declare.
641 (ix86_bd_do_dispatch): Declare.
642 (ix86_core2i7_init_hooks): Declare.
643 (ix86_atom_sched_reorder): Declare.
644 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
645 (COSTS_N_BYTES): Move to x86-tune-costs.h.
646 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
647 (rip_relative_addr_p): Rename to ...
648 (ix86_rip_relative_addr_p): ... this one; export.
649 (memory_address_length): Update.
650 (ix86_issue_rate): Move to x86-tune-sched.c.
651 (ix86_flags_dependent): Move to x86-tune-sched.c.
652 (ix86_agi_dependent): Move to x86-tune-sched.c.
653 (exact_dependency_1): Move to x86-tune-sched.c.
654 (exact_store_load_dependency): Move to x86-tune-sched.c.
655 (ix86_adjust_cost): Move to x86-tune-sched.c.
656 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
657 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
658 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
659 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
660 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
661 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
662 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
663 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
664 (min_insn_size): Rename to ...
665 (ix86_min_insn_size): ... this one; export.
666 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
667 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
668 (core2i7_first_cycle_multipass_backtrack): Move to
669 x86-tune-sched-core.c.
670 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
671 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
672 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
673 (ix86_avoid_jump_mispredicts): Update.
674 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
675 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
676 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
677 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
678 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
679 (MAX_INSN): Move to ix86-tune-sched-bd.c.
680 (MAX_IMM): Move to ix86-tune-sched-bd.c.
681 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
682 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
683 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
684 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
685 (MAX_STORE): Move to ix86-tune-sched-bd.c.
686 (BIG): Move to ix86-tune-sched-bd.c.
687 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
688 (enum insn_path): Move to ix86-tune-sched-bd.c.
689 (get_mem_group): Move to ix86-tune-sched-bd.c.
690 (is_cmp): Move to ix86-tune-sched-bd.c.
691 (dispatch_violation): Move to ix86-tune-sched-bd.c.
692 (is_branch): Move to ix86-tune-sched-bd.c.
693 (is_prefetch): Move to ix86-tune-sched-bd.c.
694 (init_window): Move to ix86-tune-sched-bd.c.
695 (allocate_window): Move to ix86-tune-sched-bd.c.
696 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
697 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
698 (process_end_window): Move to ix86-tune-sched-bd.c.
699 (allocate_next_window): Move to ix86-tune-sched-bd.c.
700 (find_constant): Move to ix86-tune-sched-bd.c.
701 (get_num_immediates): Move to ix86-tune-sched-bd.c.
702 (has_immediate): Move to ix86-tune-sched-bd.c.
703 (get_insn_path): Move to ix86-tune-sched-bd.c.
704 (get_insn_group): Move to ix86-tune-sched-bd.c.
705 (count_num_restricted): Move to ix86-tune-sched-bd.c.
706 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
707 (add_insn_window): Move to ix86-tune-sched-bd.c.
708 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
709 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
710 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
711 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
712 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
713 (do_dispatch): Move to ix86-tune-sched-bd.c.
714 (has_dispatch): Move to ix86-tune-sched-bd.c.
715 * config/i386/t-i386: Add new object files.
716 * config/i386/x86-tune-costs.h: New file.
717 * config/i386/x86-tune-sched-atom.c: New file.
718 * config/i386/x86-tune-sched-bd.c: New file.
719 * config/i386/x86-tune-sched-core.c: New file.
720 * config/i386/x86-tune-sched.c: New file.
722 2017-10-11 Liu Hao <lh_mouse@126.com>
724 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
725 the generic version below instead.
726 (should_colorize): Recognize Windows consoles as terminals
728 * pretty-print.c [__MINGW32__] (write_all): New function.
729 [__MINGW32__] (find_esc_head): Likewise.
730 [__MINGW32__] (find_esc_terminator): Likewise.
731 [__MINGW32__] (eat_esc_sequence): Likewise.
732 [__MINGW32__] (mingw_ansi_fputs): New function that handles
734 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
737 2017-10-11 Richard Biener <rguenther@suse.de>
739 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
740 Properly call analyze_scalar_evolution with the loop of the stmt.
742 2017-10-11 Richard Biener <rguenther@suse.de>
744 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
745 * tree-core.h (tree_base): Add chrec_var union member.
746 * tree.h (CHREC_VAR): Remove.
747 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
748 * tree-chrec.h (build_polynomial_chrec): Adjust.
749 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
750 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
752 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
754 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
755 * match.pd: ... here.
756 ((T) X == (T) Y): Relax condition.
758 2017-10-11 Bin Cheng <bin.cheng@arm.com>
760 PR tree-optimization/82472
761 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
763 (break_alias_scc_partitions): Update postorder number.
765 2017-10-11 Martin Liska <mliska@suse.cz>
768 * opts.c (parse_no_sanitize_attribute): Do not use error_value
770 * opts.h (parse_no_sanitize_attribute): Remove last argument.
772 2017-10-11 Martin Liska <mliska@suse.cz>
774 * print-rtl.c (print_insn): Move declaration of idbuf
775 to same scope as name.
777 2017-10-11 Martin Liska <mliska@suse.cz>
782 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
785 2017-10-11 Martin Liska <mliska@suse.cz>
788 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
791 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
793 * config.gcc (default_gnu_indirect_function): Default to yes for
794 arm*-*-linux* with glibc.
796 2017-10-11 Richard Biener <rguenther@suse.de>
798 * tree-scalar-evolution.c (get_scalar_evolution): Handle
799 default-defs and types we do not want to analyze.
800 (interpret_loop_phi): Replace unreachable code with an assert.
801 (compute_scalar_evolution_in_loop): Remove and inline ...
802 (analyze_scalar_evolution_1): ... here, replacing condition with
803 what makes the intent clearer. Remove handling of cases
804 get_scalar_evolution now handles.
806 2017-10-10 Jim Wilson <wilson@tuliptree.org>
808 PR rtl-optimization/81434
809 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
810 comment for main loop. In sched_group_found if, also add checks for
811 pass and min_cost_group.
813 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
815 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
816 (rs6000_insn_cost): New function.
817 * config/rs6000/rs6000.md (cost): New attribute.
819 2017-10-10 Jakub Jelinek <jakub@redhat.com>
820 H.J. Lu <hongjiu.lu@intel.com>
824 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
825 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
826 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
827 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
828 to OPTION_MASK_ISA_AVX512VL - builtins that have both
829 OPTION_MASK_ISA_MMX and some other bit set require both
830 mmx and the ISAs without the mmx bit.
831 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
832 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
833 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
834 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
835 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
836 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
837 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
838 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
839 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
840 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
841 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
842 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
843 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
844 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
845 Add OPTION_MASK_ISA_MMX.
847 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
849 * config.gcc (armv7*-*-freebsd*): New target.
850 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
852 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
854 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
855 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
858 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
860 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
861 a template parameter.
862 (WIDE_INT_REF_FOR): Update accordingly.
863 * tree.h (wi::int_traits <const_tree>): Delete.
864 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
865 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
866 (wi::tree_to_wide_ref): New typedef.
867 (wi::to_wide): New function.
868 * calls.c (get_size_range): Use wi::to_wide when operating on
870 * cgraph.c (cgraph_node::create_thunk): Likewise.
871 * config/i386/i386.c (ix86_data_alignment): Likewise.
872 (ix86_local_alignment): Likewise.
873 * dbxout.c (stabstr_O): Likewise.
874 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
875 * expr.c (const_vector_from_tree): Likewise.
876 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
877 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
878 (fold_negate_expr_1, int_const_binop_1, const_binop)
879 (fold_convert_const_int_from_real, optimize_bit_field_compare)
880 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
881 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
882 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
883 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
884 (fold_not_const, round_up_loc): Likewise.
885 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
886 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
887 (alloca_call_type): Likewise.
888 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
889 * godump.c (go_output_typedef): Likewise.
890 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
891 * internal-fn.c (get_min_precision): Likewise.
892 * ipa-cp.c (ipcp_store_vr_results): Likewise.
893 * ipa-polymorphic-call.c
894 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
895 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
896 (ipa_modify_call_arguments): Likewise.
897 * match.pd: Likewise.
898 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
899 * print-tree.c (print_node_brief, print_node): Likewise.
900 * stmt.c (expand_case): Likewise.
901 * stor-layout.c (layout_type): Likewise.
902 * tree-affine.c (tree_to_aff_combination): Likewise.
903 * tree-cfg.c (group_case_labels_stmt): Likewise.
904 * tree-data-ref.c (dr_analyze_indices): Likewise.
905 (prune_runtime_alias_test_list): Likewise.
906 * tree-dump.c (dequeue_and_dump): Likewise.
907 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
908 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
909 * tree-pretty-print.c (dump_generic_node): Likewise.
910 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
911 (simple_iv_with_niters): Likewise.
912 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
913 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
914 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
915 * tree-ssa-loop-niter.c (split_to_var_and_offset)
916 (refine_value_range_using_guard, number_of_iterations_ne_max)
917 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
918 (get_cst_init_from_scev, record_nonwrapping_iv)
919 (scev_var_range_cant_overflow): Likewise.
920 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
921 * tree-ssa-pre.c (compute_avail): Likewise.
922 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
923 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
924 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
925 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
926 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
927 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
928 (dump_case_nodes, try_switch_expansion): Likewise.
929 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
930 (vect_do_peeling): Likewise.
931 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
932 * tree-vect-stmts.c (vectorizable_load): Likewise.
933 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
934 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
935 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
936 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
937 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
938 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
939 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
940 (evrp_dom_walker::before_dom_children): Likewise.
941 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
942 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
943 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
944 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
945 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
946 (get_range_pos_neg): Likewise.
947 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
948 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
949 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
950 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
951 * config/avr/avr.c (avr_fold_builtin): Likewise.
952 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
953 * config/msp430/msp430.c (msp430_attr): Likewise.
954 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
955 * config/powerpcspe/powerpcspe-c.c
956 (altivec_resolve_overloaded_builtin): Likewise.
957 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
958 (rs6000_expand_ternop_builtin): Likewise.
959 * config/rs6000/rs6000-c.c
960 (altivec_resolve_overloaded_builtin): Likewise.
961 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
962 (rs6000_expand_ternop_builtin): Likewise.
963 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
965 2017-10-10 Bin Cheng <bin.cheng@arm.com>
967 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
968 when copying loop nest with only one inner loop.
970 2017-10-10 Richard Biener <rguenther@suse.de>
972 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
973 blocks if SCEV is active.
974 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
976 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
977 (scev_initialize): Assert we are not yet initialized.
979 2017-10-10 Bin Cheng <bin.cheng@arm.com>
981 * tree-loop-distribution.c (generate_loops_for_partition): Remove
982 inner loop's exit stmt by making it always exit the loop, otherwise
983 we would generate an infinite empty loop.
985 2017-10-10 Bin Cheng <bin.cheng@arm.com>
987 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
988 renaming variables in new preheader if it's deleted.
990 2017-10-10 Bin Cheng <bin.cheng@arm.com>
992 * tree-loop-distribution.c (struct partition): Remove unused field
993 loops of the structure.
994 (partition_alloc, partition_free): Ditto.
995 (build_rdg_partition_for_vertex): Ditto.
997 2017-10-09 Jeff Law <law@redhat.com>
999 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
1000 return type to match prototype and documentation.
1002 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1004 * config/rs6000/rs6000.c (processor_costs): Move to ...
1005 * config/rs6000/rs6000.h: ... here.
1006 (rs6000_cost): Declare.
1008 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
1010 * except.c (setjmp_fn): New global variable.
1011 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
1012 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
1013 if DONT_USE_BUILTIN_SETJMP is defined.
1015 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1017 * target.def (insn_cost): New hook.
1018 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
1019 * doc/tm.texi: Regenerate.
1020 * rtlanal.c (insn_cost): Use the new hook.
1022 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1024 * combine.c (combine_validate_cost): Compute the new insn_cost,
1025 not just pattern_cost.
1026 (try_combine): Adjust comment.
1028 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
1030 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
1032 * combine.c (uid_insn_cost): Adjust comment.
1033 (combine_validate_cost): Adjust comment. Use pattern_cost instead
1035 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
1036 * dse.c (find_shift_sequence): Ditto.
1037 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
1038 (bb_valid_for_noce_process_p): Use pattern_cost.
1039 * rtl.h (insn_rtx_cost): Delete.
1040 (pattern_cost): New prototype.
1041 (insn_cost): New prototype.
1042 * rtlanal.c (insn_rtx_cost): Rename to...
1043 (pattern_cost): ... this.
1046 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
1048 * config/i386/i386.md (*jcc_2): Remove insn pattern.
1049 (*jcc<mode>_0_r_i387): Ditto.
1050 (*jccxf_r_i387): Ditto.
1051 (*jcc<mode>_r_i387): Ditto.
1052 (*jccu<mode>_r_i387): Ditto.
1053 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
1054 (*jcc): Rename from *jcc_1.
1056 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1058 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
1059 deferred rescans after the lvx/stvx recombination pre-pass.
1061 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1063 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
1064 memory operation instruction support.
1065 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
1066 (rs6000-ibm-aix[789]*): Likewise.
1067 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
1068 Document new functions.
1070 2017-10-09 Richard Biener <rguenther@suse.de>
1072 PR tree-optimization/82397
1073 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
1074 equality only for semantically equal trees.
1076 2017-10-09 Richard Biener <rguenther@suse.de>
1078 PR tree-optimization/82449
1079 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
1080 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
1081 allow constant addresses.
1082 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
1085 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1087 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
1090 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1093 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
1096 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1099 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
1101 2017-10-09 Jakub Jelinek <jakub@redhat.com>
1104 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
1105 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
1107 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
1109 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
1110 (WI_BINARY_PREDICATE_RESULT): Likewise.
1111 (wi::binary_traits::operator_result): New type.
1112 (wi::binary_traits::predicate_result): Likewise.
1113 (generic_wide_int::operator~, unary generic_wide_int::operator-)
1114 (generic_wide_int::operator==, generic_wide_int::operator!=)
1115 (generic_wide_int::operator&, generic_wide_int::and_not)
1116 (generic_wide_int::operator|, generic_wide_int::or_not)
1117 (generic_wide_int::operator^, generic_wide_int::operator+
1118 (binary generic_wide_int::operator-, generic_wide_int::operator*):
1120 (operator~, unary operator-, operator==, operator!=, operator&)
1121 (operator|, operator^, operator+, binary operator-, operator*): New
1123 * expr.c (get_inner_reference): Use wi::bit_and_not.
1124 * fold-const.c (fold_binary_loc): Likewise.
1125 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
1126 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
1127 (bit_value_binop): Likewise.
1128 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
1129 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
1130 (extract_range_from_binary_expr_1): Likewise.
1131 (masked_increment): Likewise.
1132 (simplify_bit_ops_using_ranges): Likewise.
1134 2017-10-09 Martin Jambor <mjambor@suse.cz>
1137 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
1138 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
1139 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
1141 (hsa_fixup_mov_insn_type): New function.
1142 (hsa_op_with_type::get_in_type): Use it.
1143 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
1144 immediates in an assert.
1145 (hsa_op_with_type::extend_int_to_32bit): New method.
1146 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
1147 types. Convert to dest type if necessary.
1148 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
1149 (reg_for_gimple_ssa): Pass false as min32int to
1150 hsa_type_for_scalar_tree_type.
1151 (gen_hsa_addr): Fixup type when creating addresable temporary.
1152 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
1153 (gen_hsa_unary_operation): Extend operands and convert to dest type if
1154 necessary. Call hsa_fixup_mov_insn_type.
1155 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
1156 extend operands and convert to dest type if necessary.
1157 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
1158 to dest type if necessary.
1159 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
1160 if conversion nt necessary and size matches.
1161 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
1162 to dest type if necessary.
1163 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
1164 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
1166 (gen_hsa_clrsb): Likewise.
1167 (gen_hsa_ffs): Likewise.
1168 (gen_hsa_divmod): Extend operands and convert to dest type if
1170 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
1172 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1174 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
1175 Remove empty default arguments. Use a brace block as output
1177 (conditional return): Ditto.
1179 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
1180 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
1182 (group_ending_nop): Ditto.
1183 (doloop_end): Ditto.
1184 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
1185 (splitters for those): Ditto.
1187 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
1189 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
1190 a conditional jump (and the compare for it) so that pc_rtx is the
1192 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
1193 for the deleted and renamed ctr<mode>_internal[234] patterns.
1194 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
1195 Delete second conditional return pattern.
1196 (ctr<mode>_internal2): Delete this second bdnz pattern.
1197 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
1198 (ctr<mode>_internal4): Delete this second bdz pattern.
1200 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1202 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
1203 (always_initialized_rtx_for_ssa_name_p): New predicate.
1204 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
1205 (finish_out_of_ssa): Free new field of SA.
1206 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
1207 * tree-ssa-coalesce.c: Include tree-ssa.h.
1208 (get_parm_default_def_partitions): Remove extern keyword.
1209 (get_undefined_value_partitions): New function.
1210 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
1211 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
1213 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
1215 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
1217 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
1219 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1221 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
1222 for targets that preffer 128bit.
1224 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
1226 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
1228 2017-10-08 Olivier Hainque <hainque@adacore.com>
1230 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
1231 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
1233 (thumb_set_return_address): Likewise.
1235 2017-10-08 Olivier Hainque <hainque@adacore.com>
1237 * common/config/arm/arm-common.c (arm_except_unwind_info):
1238 Handle DWARF2_UNWIND_INFO.
1240 2017-10-07 Michael Collison <michael.collison@arm.com>
1242 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
1245 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
1247 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
1248 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
1249 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
1250 defined, force the creation of a new block for a dispatch label.
1252 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
1254 * invoke.texi (Wsuggest-attribute=cold): Document.
1255 * common.opt (Wsuggest-attribute=cold): New
1256 * ipa-pure-const.c (warn_function_cold): New function.
1257 * predict.c (compute_function_frequency): Use it.
1258 * predict.h (warn_function_cold): Declare.
1260 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
1262 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
1265 2017-10-06 Martin Liska <mliska@suse.cz>
1267 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
1268 keyword for member functions.
1269 (struct sanopt_tree_couple): New struct.
1270 (struct sanopt_tree_couple_hash): New function.
1271 (struct sanopt_ctx): Add new hash_map.
1272 (has_dominating_ubsan_ptr_check): New function.
1273 (record_ubsan_ptr_check_stmt): Likewise.
1274 (maybe_optimize_ubsan_ptr_ifn): Likewise.
1275 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
1276 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
1278 2017-10-06 Sudakshina Das <sudi.das@arm.com>
1281 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
1282 aarch64_simd_valid_immediate on CONST_VECTORs.
1283 (aarch64_reg_or_bic_imm): Likewise.
1285 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
1287 PR rtl-optimization/82396
1288 * haifa-sched.c (ready_sort_real): Disable qsort checking.
1290 2017-10-06 Sebastian Pop <sebpop@gmail.com>
1292 * graphite-dependences.c (scop_get_reads): Move code to...
1293 (scop_get_must_writes): Move code to...
1294 (scop_get_may_writes): Move code to...
1295 (scop_get_reads_and_writes): ... here.
1296 (scop_get_dependences): Call scop_get_reads_and_writes.
1298 2017-10-06 Jakub Jelinek <jakub@redhat.com>
1300 PR tree-optimization/82434
1301 * fold-const.h (can_native_encode_type_p,
1302 can_native_encode_string_p): Remove.
1303 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
1304 don't encode anything, just return what would be otherwise returned.
1305 (native_encode_fixed, native_encode_complex, native_encode_vector):
1307 (native_encode_string): Likewise. Inline by hand
1308 can_native_encode_string_p.
1309 (can_native_encode_type_p): Remove.
1310 (can_native_encode_string_p): Remove.
1311 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
1312 STRING_CSTs using can_native_encode_string_p, test all
1313 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
1314 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
1315 argument from native_encode_expr.
1316 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
1317 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
1320 2017-10-06 Richard Biener <rguenther@suse.de>
1322 PR tree-optimization/82397
1323 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
1324 operand_equal_p but rely on data_ref_compare_tree for detecting
1326 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
1327 to match up with dr_group_sort_cmp.
1329 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1332 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
1334 * config/s390/s390-builtin-types.def: Regenerate.
1336 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1339 * config/s390/s390-builtin-types.def: Regenerate.
1340 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
1341 Change flag from B_VXE to B_VX.
1342 (s390_vec_min_dbl): Remove B_VXE flag.
1344 2017-10-06 Richard Biener <rguenther@suse.de>
1346 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
1347 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
1348 translate_isl_ast_to_gimple::is_valid_rename,
1349 translate_isl_ast_to_gimple::get_rename,
1350 translate_isl_ast_to_gimple::get_def_bb_for_const,
1351 translate_isl_ast_to_gimple::get_new_name,
1352 translate_isl_ast_to_gimple::collect_all_ssa_names,
1353 translate_isl_ast_to_gimple::copy_loop_phi_args,
1354 translate_isl_ast_to_gimple::collect_all_ssa_names,
1355 translate_isl_ast_to_gimple::copy_loop_phi_args,
1356 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
1357 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
1358 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
1359 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
1360 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
1361 translate_isl_ast_to_gimple::copy_cond_phi_args,
1362 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
1363 translate_isl_ast_to_gimple::edge_for_new_close_phis,
1364 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
1365 translate_isl_ast_to_gimple::rename_uses,
1366 translate_isl_ast_to_gimple::rename_all_uses): Remove.
1367 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
1368 (set_rename_for_each_def): Likewise.
1369 (graphite_copy_stmts_from_block): Handle debug stmt resetting
1370 here. Handle rewriting SCEV analyzable uses here.
1371 (copy_bb_and_scalar_dependences): Generate code for PHI
1373 (graphite_regenerate_ast_isl): Adjust.
1374 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
1375 (add_write, add_read): New functions.
1376 (build_cross_bb_scalars_def): Use it and simplify.
1377 (build_cross_bb_scalars_use): Likewise.
1378 (graphite_find_cross_bb_scalar_vars): Inline into...
1379 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
1380 simulating out-of-SSA. Compute liveout and add dependencies.
1381 (build_scops): Force an empty entry block.
1382 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
1384 (sese_build_liveouts): Declare.
1385 (sese_trivially_empty_bb_p): Likewise.
1386 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
1387 compute liveout and debug_liveout.
1388 (sese_bad_liveouts_use): Remove.
1389 (sese_reset_debug_liveouts_bb): Likewise.
1390 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
1391 (sese_build_liveouts): Build liveout and debug_liveout and store
1393 (new_sese_info): Adjust.
1394 (free_sese_info): Likewise.
1395 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
1396 do not build liveout here.
1397 (move_sese_in_condition): Adjust region entry.
1398 (scev_analyzable_p): Match up with chrec_apply requirements.
1399 (sese_trivially_empty_bb_p): New.
1400 * tree-into-ssa.c (get_reaching_def): Properly support generating
1401 default-defs for incremental rewrite of anonymous names.
1403 2017-10-06 Richard Biener <rguenther@suse.de>
1405 * graphite-sese-to-poly.c (extract_affine): For casts increasing
1406 precision do not perform modulo reduction.
1408 2017-10-06 Richard Biener <rguenther@suse.de>
1410 PR tree-optimization/82436
1411 * tree-vect-slp.c (vect_supported_load_permutation_p): More
1412 conservatively choose the vectorization factor when checking
1413 whether we can perform the required load permutation.
1414 (vect_transform_slp_perm_load): Assert when we may not fail.
1416 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
1418 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
1419 message for incompatible -msdata=* and -mcall-* options.
1421 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1423 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
1424 rate for post-reload scheduling.
1426 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1428 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
1430 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1432 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
1433 to improve monte carlo in scimark.
1435 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
1437 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
1438 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
1439 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
1440 pentium4_cost, nocona_cost): Set reassociation width to 1.
1441 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
1442 width to 2 for fp operations and 1 otherwise.
1443 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
1445 (atom_cost): Set reassociation width to 2.
1446 (slm_cost, generic_cost): Set fp reassociation width
1447 to 2 and 1 otherwise.
1448 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
1449 (core_cost): Set fp reassociation width to 4 and vector to 2.
1450 (ix86_reassociation_width): Rewrite using cost table; special case
1451 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
1452 and TARGET_AVX128_OPTIMAL.
1453 * config/i386/i386.h (processor_costs): Add
1454 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
1455 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
1456 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
1457 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
1458 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
1459 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
1461 2017-10-05 Nathan Sidwell <nathan@acm.org>
1463 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
1465 2017-10-05 Tamar Christina <tamar.christina@arm.com>
1467 * config/arm/arm.c (arm_test_fpu_data): New.
1468 (arm_run_selftests): Call arm_test_fpu_data.
1470 2017-10-04 Nathan Sidwell <nathan@acm.org>
1472 * toplev.c (toplev::main): Remove excess parens on pretty_printer
1474 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
1476 2017-10-04 Sudakshina Das <sudi.das@arm.com>
1478 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
1479 check type for aarch64_simd_valid_immediate.
1480 (aarch64_output_simd_mov_immediate): Update prototype.
1481 (aarch64_simd_valid_immediate): Update prototype.
1482 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
1483 support for ORR-immediate.
1484 (and<mode>3): modified pattern to add support for BIC-immediate.
1485 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
1486 now checks for valid immediate for BIC and ORR based on new enum
1488 (aarch64_output_simd_mov_immediate): Function now used to output
1489 BIC/ORR imm as well based on new enum argument.
1490 * config/aarch64/constraints.md (Do): New vector immediate constraint.
1492 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
1493 (aarch64_reg_or_bic_imm): Likewise.
1495 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1497 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
1498 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
1500 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1504 PR rtl-optimization/82396
1505 * haifa-sched.c (autopref_multipass_init): Simplify
1507 (autopref_rank_data): Simplify sort order.
1508 * sched-int.h (autopref_multipass_data_): Remove
1509 multi_mem_insn_p, min_offset and max_offset.
1511 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1513 * doc/sourcebuild.texi: Document vect_peeling_profitable.
1515 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1517 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
1520 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1522 * doc/sourcebuild.texi: Document vect_long_mult.
1524 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
1526 PR tree-optimization/82413
1527 * fold-const.c (build_range_check): Use widest_int when comparing
1528 the maximum ETYPE value with HIGH.
1530 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
1532 PR rtl-optimization/82396
1533 * haifa-sched.c (autopref_multipass_init): Simplify
1535 (autopref_rank_data): Simplify sort order.
1536 * sched-int.h (autopref_multipass_data_): Remove
1537 multi_mem_insn_p, min_offset and max_offset.
1539 2017-10-04 Jakub Jelinek <jakub@redhat.com>
1541 PR tree-optimization/82381
1542 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
1543 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
1546 PR tree-optimization/82374
1547 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
1548 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
1549 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
1550 current_function_decl to the new decl.
1552 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
1554 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
1555 helper macro for IEEE float128 hardware built-in functions.
1556 (SQRTF128_ODD): Add built-in functions with the round-to-odd
1558 (TRUNCF128_ODD): Likewise.
1559 (ADDF128_ODD): Likewise.
1560 (SUBF128_ODD): Likewise.
1561 (MULF128_ODD): Likewise.
1562 (DIVF128_ODD): Likewise.
1563 (FMAF128_ODD): Likewise.
1564 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
1565 UNSPEC_TRUNC_ROUND_TO_ODD.
1566 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
1567 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
1568 floating point round to odd instructions.
1569 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
1570 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
1571 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
1572 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
1573 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
1574 (trunc<mode>sf2_hw): Change the truncate with round to odd
1575 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
1576 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
1577 to odd hardware instructions.
1578 (sub<mode>3_odd): Likewise.
1579 (mul<mode>3_odd): Likewise.
1580 (div<mode>3_odd): Likewise.
1581 (sqrt<mode>2_odd): Likewise.
1582 (fma<mode>4_odd): Likewise.
1583 (fms<mode>4_odd): Likewise.
1584 (nfma<mode>4_odd): Likewise.
1585 (nfms<mode>4_odd): Likewise.
1586 (trunc<mode>df2_odd): Change the truncate with round to odd
1587 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
1589 * doc/extend.texi (PowerPC built-in functions): Update documentation
1590 for existing IEEE float128-bit built-in functions. Add built-in
1591 functions that generate the IEEE 128-bit floating point round to
1594 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
1596 PR rtl-optimization/77729
1597 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
1598 to (X&(C1&~C2))|C2 transformations.
1600 2017-10-03 Martin Jambor <mjambor@suse.cz>
1602 PR tree-optimization/82363
1603 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
1604 mismatch, mark lacc written regardless of racc.
1606 2017-10-03 Jakub Jelinek <jakub@redhat.com>
1608 PR tree-optimization/82381
1609 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
1610 stmt_to_insert nor wheather SSA_NAMEs are default defs.
1611 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
1612 fallthrough into reassoc_stmt_dominates_stmt_p.
1615 * combine.c (combine_instructions): Don't combine in unreachable
1618 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
1621 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
1622 function to not use the have_cpu variable. Do not set cpu_index,
1623 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
1625 (rs6000_valid_attribute_p): Remove duplicate initializations of
1626 old_optimize and func_optimize.
1627 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
1628 (rs6000_activate_target_options): Make global.
1629 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
1632 2017-10-02 Jakub Jelinek <jakub@redhat.com>
1634 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
1635 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
1636 Set *poffset to 0 and *psize and *pmax_size to -1 if
1637 *poffset + *psize overflows in HOST_WIDE_INT.
1639 PR tree-optimization/82387
1640 PR tree-optimization/82388
1641 PR tree-optimization/82389
1642 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
1643 instead of live_bytes non-NULL.
1645 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
1648 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
1651 2017-10-02 Richard Biener <rguenther@suse.de>
1653 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
1654 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
1655 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
1657 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1659 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
1660 requested precision matches the type's.
1661 * calls.c (alloc_max_size): Calculate the new candidate size as
1662 a widest_int and use wi::to_widest when comparing it with the
1663 current candidate size.
1664 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
1665 zero rather than integer_zero_node.
1666 * match.pd: Check for a no-op conversion before using wi::add
1667 rather than after. Use tree_to_uhwi when summing small shift
1668 counts into an unsigned int.
1670 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
1671 Alan Hayward <alan.hayward@arm.com>
1672 David Sherwood <david.sherwood@arm.com>
1675 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
1676 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
1677 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
1678 POINTER_AND_FP_REGS.
1680 2017-10-02 Richard Biener <rguenther@suse.de>
1682 PR tree-optimization/82355
1683 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
1684 a mapping for the enclosing loop but avoid generating one for
1686 (copy_bb_and_scalar_dependences): Remove premature codegen
1687 error on PHIs in blocks duplicated into multiple places.
1688 * graphite-scop-detection.c
1689 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
1690 in the region use it as loop and nest to analyze the DR in.
1691 (try_generate_gimple_bb): Likewise.
1692 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
1693 (add_loop_constraints): For blocks in a loop not in the region
1694 create a dimension with a single iteration.
1695 * sese.h (gbb_loop_at_index): Remove assert.
1697 2017-10-01 Kevin Buettner <kevinb@redhat.com>
1699 * omp-expand.c (adjust_context_scope): New function.
1700 (expand_parallel_call): Call adjust_context_scope.
1702 2017-10-01 Jeff Law <law@redhat.com>
1704 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
1705 dom_opt_dom_walker class with direct access to private members.
1706 Add comments. Call test_for_singularity.
1707 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
1708 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
1709 m_dummy_cond anymore.
1710 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
1712 (pass_dominator:execute): Build the dummy_cond here and pass it
1713 to the dom_opt_dom_walker ctor.
1714 (test_for_singularity): New function.
1716 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
1717 Maya Rashish <coypu@sdf.org>
1719 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
1720 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
1721 (alpha*-*-netbsd*) Use nbsd_tm_file.
1722 (arm*-*-netbsdelf*) Likewise.
1723 (i[34567]86-*-netbsdelf*) Likewise.
1724 (x86_64-*-netbsd*) Likewise.
1725 (mips*-*-netbsd*) Likewise.
1726 (powerpc-*-netbsd*) Likewise.
1727 (sh*-*-netbsd*) Likewise.
1728 (sparc-*-netbsdelf*) Likewise.
1729 (sparc64-*-netbsd*) Likewise.
1730 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
1732 (vax-*-netbsdelf*) Likewise.
1733 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
1734 (UINT_FAST8_TYPE) Likewise.
1735 (INT_FAST16_TYPE) Check CHAR_FAST16.
1736 (UINT_FAST16_TYPE) Likewise.
1738 2017-09-30 Jakub Jelinek <jakub@redhat.com>
1741 * config/i386/i386.md
1742 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
1743 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
1744 *divmodsi4_zext_2): New define_insn_and_split.
1745 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
1746 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
1747 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
1748 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
1749 New define_insn_and_split.
1750 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
1751 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
1752 operands[1] having DImode when mode is SImode.
1754 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
1755 always SImode for DIV and MOD in REG_EQUAL notes.
1757 2017-09-29 Yury Gribov <tetra2005@gmail.com>
1760 * match.pd: Fix handling of NaNs in pattern.
1762 2017-09-29 Jeff Law <law@redhat.com>
1764 * sbitmap.c (bitmap_bit_in_range_p): New function.
1765 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
1766 * tree-ssa-dse.c (live_bytes_read): New function.
1767 (dse_classify_store): Ignore reads of dead bytes.
1769 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
1770 typos and whitespace errors.
1771 * config/i386/predicates.md (address_no_seg_operand): Likewise.
1772 * config/s390/s390.c (s390_emit_prologue): Likewise.
1774 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1777 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
1778 with a symbol for LRA.
1780 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
1782 PR rtl-optimization/82338
1783 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
1785 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
1787 * genmodes.c (calc_wider_mode): Suppress qsort macro.
1788 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
1789 (qsort_chk): Declare.
1790 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
1791 (qsort_chk): New function.
1793 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1795 PR tree-optimization/82337
1796 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
1797 phi definition if the PHI result appears in an abnormal PHI.
1798 (find_basis_for_base_expr): Don't record a basis if the LHS of the
1799 basis appears in an abnormal PHI.
1801 2017-09-29 Richard Biener <rguenther@suse.de>
1803 * graphite-isl-ast-to-gimple.c
1804 (translate_isl_ast_to_gimple::set_codegen_error): New function.
1805 (binary_op_to_tree): Use it.
1806 (get_rename_from_scev): Likewise.
1807 (copy_loop_phi_nodes): Likewise.
1808 (copy_bb_and_scalar_dependences): Likewise.
1809 (translate_pending_phi_nodes): Likewise.
1811 2017-09-29 Jakub Jelinek <jakub@redhat.com>
1814 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
1815 for movabsq $(i32 << shift), r64.
1817 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
1819 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
1820 index when encoding %esp as %rsp to avoid 0x67 prefix.
1822 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
1824 * config/i386/i386.md (*movsf_internal, *movdf_internal):
1825 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
1827 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
1829 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
1830 Extensions with more than 16 double VFP registers.
1831 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
1832 to_clear_mask and all code related to it. Replace the remaining
1833 entry by a sbitmap and adapt code accordingly.
1835 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
1837 * brig-builtins.def: Change pure attributes to const.
1839 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1841 * config.gcc (default_gnu_indirect_function): Default to yes for
1842 sparc*-*-linux* with glibc.
1844 2017-09-28 Joseph Myers <joseph@codesourcery.com>
1846 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
1847 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
1848 when creating .init_array and .fini_array sections with priority
1851 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
1854 * config/aarch64/aarch64.c
1855 (aarch64_builtin_support_vector_misalignment): Always return false
1856 when misalignment is unknown.
1858 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
1860 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
1861 this function to return false if the definition used by the swap
1862 instruction is artificial, or if the memory address from which the
1863 constant value is loaded is not represented by a base address held
1864 in a register or if the base address register is a frame or stack
1865 pointer. Additionally, return false if the base address of the
1866 loaded constant is a SYMBOL_REF but is not considered to be a
1868 (replace_swapped_load_constant): New function.
1869 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
1870 loaded constant vector with a load of a swapped constant vector.
1872 2017-09-27 Carl Love <cel@us.ibm.com>
1874 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
1875 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
1876 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
1879 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
1881 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
1882 first, always call autopref_rank_data otherwise.
1884 2017-09-27 Richard Biener <rguenther@suse.de>
1886 * graphite-scop-detection.c (find_scop_parameters): Move
1887 loop bound handling ...
1888 (gather_bbs::before_dom_children): ... here, avoiding the need
1889 to build scop_info->loop_nest.
1890 (record_loop_in_sese): Remove.
1891 * sese.h (sese_info_t::loop_nest): Remove.
1892 * sese.c (new_sese_info): Do not allocate loop_nest.
1893 (free_sese_info): Do not free loop_nest.
1895 2017-09-27 Jakub Jelinek <jakub@redhat.com>
1898 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
1899 lhs from calls if the lhs has addressable type.
1901 2017-09-27 Richard Biener <rguenther@suse.de>
1903 * graphite.h (scop::max_alias_set): New member.
1904 * graphite-scop-detection.c: Remove references to non-existing
1905 --param in comments.
1906 (build_alias_sets): Record the maximum alias set used for drs.
1907 (build_scops): Support zero as unlimited for
1908 --param graphite-max-arrays-per-scop.
1909 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
1911 (build_poly_sr_1): ... here. Compute alias set based on the
1912 maximum alias set used for drs rather than
1913 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
1915 2017-09-27 Richard Biener <rguenther@suse.de>
1917 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
1918 --param loop-block-tile-size=0 to disable tiling.
1920 2017-09-27 Richard Biener <rguenther@suse.de>
1922 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
1923 (graphite-max-nb-scop-params): Document special value zero.
1924 * domwalk.h (dom_walker::STOP): New symbolical constant.
1925 (dom_walker::dom_walker): Add optional parameter for bb to
1927 (dom_walker::~dom_walker): Declare.
1928 (dom_walker::before_dom_children): Document STOP return value.
1929 (dom_walker::m_user_bb_to_rpo): New member.
1930 (dom_walker::m_bb_to_rpo): Likewise.
1931 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
1932 mapping here if not provided by the user.
1933 (dom_walker::~dom_walker): Free bb to RPO mapping if not
1934 provided by the user.
1935 (dom_walker::STOP): Define.
1936 (dom_walker::walk): Do not compute bb to RPO mapping here.
1937 Support STOP return value from before_dom_children to stop
1939 * graphite-optimize-isl.c (optimize_isl): If the schedule
1940 is the same still generate code if -fgraphite-identity
1941 or -floop-parallelize-all are given.
1942 * graphite-scop-detection.c: Include cfganal.h.
1943 (gather_bbs::gather_bbs): Get and pass through bb to RPO
1945 (gather_bbs::before_dom_children): Return STOP for BBs
1947 (build_scops): Compute bb to RPO mapping and pass it to
1948 the domwalk. Treat --param graphite-max-nb-scop-params=0
1949 as not limiting the number of params.
1950 * graphite.c (graphite_initialize): Remove limit on the
1951 number of basic-blocks in a function.
1952 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
1953 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
1954 default value of 10.
1956 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1958 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
1959 Adjust code to eliminate needing to do the shift right 32-bits
1960 operation after XSCVDPSPN.
1962 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1964 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
1965 ((X / Y) != 0 -> X >= Y): Likewise.
1967 2017-09-26 Carl Love <cel@us.ibm.com>
1969 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
1970 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
1971 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
1972 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
1973 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
1974 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
1975 definitions and overloading.
1976 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
1977 statement for P9V_BUILTIN_XST_LEN_R.
1978 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
1979 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
1980 define_expand and define_insn for the instructions and builtins.
1981 * doc/extend.texi: Update the built-in documentation file for the new
1983 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
1984 define_insn for the instructions
1986 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
1989 * gcc/config/netbsd-protos.h: New file.
1990 * gcc/config/netbsd.c: New file.
1991 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
1992 * gcc/config/t-netbsd: New file.
1993 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
1994 (tmake_file) Add t-netbsd.
1995 (extra_objs) Add netbsd.o.
1997 2017-09-26 Janus Weil <janus@gcc.gnu.org>
2001 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
2003 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2005 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
2006 sign extension from a vector register to a GPR by doing a 32-bit
2007 direct move and then an EXTSW.
2008 (extendsi<mode>2 splitter): Likewise.
2009 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
2010 right or vector extract after doing XSCVDPSPN. Use
2011 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
2013 (movdi_from_sf_zero_ext): Likewise.
2014 (reload_gpr_from_vsxsf): Likewise.
2015 (p8_mfvsrd_4_disf): Delete, no longer used.
2016 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
2017 then needing to move the SFmode to a GPR to use the XSCVDPSP
2018 instruction instead of FRSP and XSCVDPSPN.
2019 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
2020 it is adjacent to the other XSCVSPDP insns.
2021 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
2022 SFmode to be in traditional Altivec registers.
2023 (vsx_xscvdpspn): Eliminate useless alternative constraint.
2024 (vsx_xscvspdpn): Likewise.
2025 (vsx_xscvspdpn_scalar): Likewise.
2027 2017-09-26 Martin Jambor <mjambor@suse.cz>
2029 * tree-sra.c (compare_access_positions): Put integral types first,
2030 stabilize sorting of integral types, remove conditions putting
2031 non-full-precision integers last.
2032 (sort_and_splice_var_accesses): Disable scalarization if a
2033 non-integert would be represented by a non-full-precision integer.
2035 2017-09-26 Joseph Myers <joseph@codesourcery.com>
2037 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
2038 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
2039 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
2040 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
2041 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
2042 conditionals inside the function instead of around it. Call
2043 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
2044 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
2046 2017-09-26 Richard Biener <rguenther@suse.de>
2048 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
2050 (scop_detection::build_scop_breadth): ... this. Removed.
2051 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
2052 (scop_detection::harmful_stmt_in_bb): Likewise.
2053 (scop_detection::graphite_can_represent_stmt): Likewise.
2054 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
2055 (scop_detection::can_represent_loop): Remove recursion, fold in ...
2056 (scop_detection::can_represent_loop_1): ... this. Removed.
2057 (scop_detection::harmful_loop_in_region): Simplify after inlining
2058 the above and remove more quadraticness.
2059 (build_scops): Adjust.
2060 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
2063 2017-09-26 Jakub Jelinek <jakub@redhat.com>
2066 * config/i386/i386.c (ix86_print_operand_address_as): Only test
2067 REGNO (base) == SP_REG if base is a REG.
2070 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
2071 if it is different SSA_NAME.
2072 (optimize_range_tests_cmp_bitwise): New function.
2073 (optimize_range_tests): Call it.
2075 2017-09-26 Richard Biener <rguenther@suse.de>
2077 PR tree-optimization/82321
2078 * graphite.c (canonicalize_loop_closed_ssa): Properly check
2079 for the def being inside the loop.
2081 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2083 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
2085 * config/s390/s390-builtins.def: Fix constraint on op4.
2087 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2089 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
2090 independent expanders.
2091 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
2092 ("vec_ordered", "vec_unordered"): New expanders.
2094 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2096 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
2099 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2101 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
2102 vec_unpacks_lo_v16qi.
2103 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
2105 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2107 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
2108 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
2109 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
2111 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2113 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
2115 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
2117 ("*vec_slb<mode>"): New insn pattern.
2118 ("vec_shr_<mode>"): New expander.
2119 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
2120 and force the shift count operand to V16QImode.
2121 ("vec_srb<mode>"): Set shift count mode to V16QI.
2123 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2125 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
2126 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
2127 ("vec_widen_smult_hi_<mode>"): New expander definitions.
2129 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
2132 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
2134 2017-09-26 Richard Biener <rguenther@suse.de>
2136 PR tree-optimization/82320
2137 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
2140 2017-09-25 Jeff Law <law@redhat.com>
2142 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
2143 prototype for new argument.
2144 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
2145 mostly extracted from rs6000_emit_allocate_stack.
2146 (rs6000_emit_probe_stack_range_stack_clash): New function.
2147 (rs6000_emit_allocate_stack): Call
2148 rs6000_emit_probe_stack_range_stack_clash as needed.
2149 (rs6000_emit_probe_stack_range): Add additional argument
2150 to call to gen_probe_stack_range{si,di}.
2151 (output_probe_stack_range): New.
2152 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
2153 (output_probe_stack_range_stack_clash): New.
2154 (rs6000_emit_prologue): Emit notes into dump file as requested.
2155 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
2156 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
2157 Add additional operand and pass it to output_probe_stack_range.
2159 2017-09-25 Bin Cheng <bin.cheng@arm.com>
2161 PR tree-optimization/82163
2162 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
2163 (checking_verify_loop_closed_ssa): New parameter.
2164 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
2165 (check_loop_closed_ssa_stmt): Delete.
2166 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
2167 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
2168 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
2171 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
2173 * brig-builtins.def: Treat HSAIL barrier builtins as
2174 setjmp/longjump style functions.
2176 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2178 * target.def (constant_alignment): New hook.
2179 * defaults.h (CONSTANT_ALIGNMENT): Delete.
2180 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
2181 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
2182 * doc/tm.texi: Regenerate.
2183 * targhooks.h (default_constant_alignment): Declare.
2184 (constant_alignment_word_strings): Likewise.
2185 * targhooks.c (default_constant_alignment): New function.
2186 (constant_alignment_word_strings): Likewise.
2187 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
2188 instead of CONSTANT_ALIGNMENT.
2189 * varasm.c (align_variable, get_variable_align, build_constant_desc)
2190 (force_const_mem): Likewise.
2191 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
2192 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
2193 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
2194 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2195 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
2197 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
2198 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2199 constant_alignment_word_strings.
2200 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
2201 (CONSTANT_ALIGNMENT): Likewise.
2202 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2203 (arm_constant_alignment): New function.
2204 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
2205 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2206 constant_alignment_word_strings.
2207 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
2208 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2209 constant_alignment_word_strings.
2210 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
2211 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2212 (cris_constant_alignment): New function.
2213 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
2214 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2215 (epiphany_constant_alignment): New function.
2216 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
2217 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2218 constant_alignment_word_strings.
2219 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
2220 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2221 constant_alignment_word_strings.
2222 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
2223 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2224 constant_alignment_word_strings.
2225 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
2226 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
2227 * config/i386/i386.c (ix86_constant_alignment): Make static.
2228 Use the same interface as the target hook.
2229 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2230 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
2231 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2232 constant_alignment_word_strings.
2233 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
2234 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
2235 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2236 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
2237 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2238 constant_alignment_word_strings.
2239 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
2240 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2241 constant_alignment_word_strings.
2242 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
2243 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2244 constant_alignment_word_strings.
2245 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
2246 * config/microblaze/microblaze.c (microblaze_constant_alignment):
2248 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2249 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
2250 * config/mips/mips.c (mips_constant_alignment): New function.
2251 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2252 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
2253 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
2254 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2255 (mmix_constant_alignment): Make static. Use the same interface
2257 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
2258 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2259 constant_alignment_word_strings.
2260 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
2261 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2262 constant_alignment_word_strings.
2263 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
2264 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2265 constant_alignment_word_strings.
2266 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
2267 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2268 (rs6000_constant_alignment): New function.
2269 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
2270 * config/riscv/riscv.c (riscv_constant_alignment): New function.
2271 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2272 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
2273 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2274 (rs6000_constant_alignment): New function.
2275 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
2276 * config/s390/s390.c (s390_constant_alignment): New function.
2277 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2278 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
2279 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2280 constant_alignment_word_strings.
2281 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
2282 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2283 (sparc_constant_alignment): New function.
2284 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
2285 * config/spu/spu.c (spu_constant_alignment): New function.
2286 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2287 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
2288 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2289 constant_alignment_word_strings.
2290 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
2291 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2292 constant_alignment_word_strings.
2293 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
2294 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
2295 constant_alignment_word_strings.
2296 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
2297 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2298 (visium_constant_alignment): New function.
2299 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
2300 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
2301 (xtensa_constant_alignment): New function.
2302 * system.h (CONSTANT_ALIGNMENT): Poison.
2304 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
2306 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2307 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
2308 (rs6000_builtin_valid_without_lhs): New helper function.
2309 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2310 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
2312 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
2314 * target.h (vec_perm_indices): Use unsigned short rather than
2316 (auto_vec_perm_indices): Likewise.
2317 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
2318 Use unsigned int rather than unsigned char.
2319 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
2321 2017-09-25 Richard Biener <rguenther@suse.de>
2323 * cfgloop.h (sort_sibling_loops): Declare.
2324 * cfgloop.c (sort_sibling_loops_cmp): New helper.
2325 (sort_sibling_loops): New function sorting the sibling loop list
2327 * graphite.c (graphite_transform_loops): Sort sibling loops.
2329 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
2331 * target.def (vec_perm_const_ok): Change sel parameter to
2333 * optabs-query.c (can_vec_perm_p): Update accordingly.
2334 * doc/tm.texi: Regenerate.
2335 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
2336 auto_vec_perm_indices and remove separate nelt field.
2337 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
2338 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
2339 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
2340 (aarch64_expand_vec_perm_const): Update accordingly.
2341 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
2342 to vec_perm_indices.
2343 * config/arm/arm.c (expand_vec_perm_d): Change perm to
2344 auto_vec_perm_indices and remove separate nelt field.
2345 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
2346 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
2347 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
2349 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
2350 to vec_perm_indices.
2351 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
2352 sel to vec_perm_indices.
2353 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
2354 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
2355 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
2357 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
2360 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
2363 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
2364 on the FUNCTION_DECL function context if it has a DIE that is a
2367 2017-09-25 Richard Biener <rguenther@suse.de>
2369 PR tree-optimization/82285
2370 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
2373 2017-09-25 Tom de Vries <tom@codesourcery.com>
2377 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
2380 2017-09-25 Richard Biener <rguenther@suse.de>
2382 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
2383 ISL errors other than isl_error_quota happen. Dump if the
2384 schedule is the same.
2385 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
2386 errors instead of aborting inside ISL.
2388 2017-09-25 Iain Sandoe <iain@codesourcery.com>
2391 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
2392 of libgcc_eh for m64.
2393 * config/i386/darwin64.h: Likewise.
2395 2017-09-25 Richard Biener <rguenther@suse.de>
2398 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
2399 attribute for incomplete types nor twice for complete ones.
2401 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
2404 * config/i386/i386.c (ix86_print_operand_address_as): Encode
2405 %esp as %rsp to avoid 0x67 prefix if there is no index or base
2408 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
2411 * config/i386/i386.opt (mprefer-avx256): Use
2412 ix86_target_flags variable.
2413 * config/i386/i386.c (ix86_target_string): Move
2414 -mprefer-avx256 to flag2_opts.
2416 2017-09-22 Jakub Jelinek <jakub@redhat.com>
2419 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
2420 and x != -1 | y != -1 into (x & y) != -1.
2422 2017-09-22 Steve Ellcey <sellcey@cavium.com>
2424 * config.gcc: Add new case statement to set
2425 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
2426 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
2427 s390x-*-linux* case statements. Added aarch64 to the list of
2428 supported architectures.
2430 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2432 PR tree-optimization/82289
2433 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
2434 STMT_VINFO_RELEVANT_P.
2436 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2437 Alan Hayward <alan.hayward@arm.com>
2438 David Sherwood <david.sherwood@arm.com>
2440 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
2441 for VR_RANGE only; don't allow VR_ANTI_RANGE.
2442 (extract_range_from_binary_expr_1): Don't call
2443 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
2445 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2446 Alan Hayward <alan.hayward@arm.com>
2447 David Sherwood <david.sherwood@arm.com>
2449 * target.def (preferred_vector_alignment): New hook.
2450 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
2452 * doc/tm.texi: Regenerate.
2453 * targhooks.h (default_preferred_vector_alignment): Declare.
2454 * targhooks.c (default_preferred_vector_alignment): New function.
2455 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
2457 (DR_TARGET_ALIGNMENT): New macro.
2458 (aligned_access_p): Update commentary.
2459 (vect_known_alignment_in_bytes): New function.
2460 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
2462 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
2463 Calculate the misalignment based on the target alignment rather than
2465 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
2466 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
2467 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
2468 the target alignment, rather than masking the element misalignment
2469 with the number of elements in a vector. Also use the target
2470 alignment when calculating the maximum number of peels.
2471 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
2472 instead of TYPE_ALIGN_UNIT.
2473 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
2474 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
2475 (vect_create_addr_base_for_vector_ref): Update call accordingly.
2476 (vect_create_data_ref_ptr): Likewise.
2477 (vect_setup_realignment): Realign by ANDing with
2478 -DR_TARGET_MISALIGNMENT.
2479 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
2480 the number of peels based on DR_TARGET_ALIGNMENT.
2481 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
2482 with the guaranteed alignment boundary when deciding whether
2484 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
2485 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
2486 (ensure_base_align): Remove stmt_info parameter. Get the
2487 target base alignment from DR_TARGET_ALIGNMENT.
2488 (vectorizable_store): Update call accordingly. Interpret
2489 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
2491 (vectorizable_load): Likewise.
2493 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
2494 Alan Hayward <alan.hayward@arm.com>
2495 David Sherwood <david.sherwood@arm.com>
2497 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
2498 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
2499 (vect_enhance_data_refs_alignment): Likewise.
2501 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2503 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
2504 error. Only quit immediately if parsing is complete.
2505 (BEGIN): Initialize fatal_err and parse_done.
2506 (begin fpu, end fpu): Check number of arguments.
2507 (begin arch, end arch): Likewise.
2508 (begin cpu, end cpu): Likewise.
2509 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
2510 (optalias): Likewise.
2512 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
2514 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
2515 * config/arm/arm-isa.h: Delete. Move definitions to ...
2516 * arm-cpus.in: ... here. Use new feature and fgroup values.
2517 * config/arm/arm.c (arm_option_override): Use lower case for feature
2519 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
2520 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
2521 * config/arm/parsecpu.awk (END): Add new command 'isa'.
2523 (print_isa_bits_for): New function.
2524 (gen_isa): New function.
2525 (gen_comm_data): Use print_isa_bits_for.
2526 (define feature): New keyword.
2527 (define fgroup): New keyword.
2528 * config/arm/t-arm (TM_H): Remove.
2529 (GTM_H): Add arm-isa.h.
2530 (arm-isa.h): Add rule to generate file.
2531 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
2532 case for feature bit names.
2534 2017-09-22 Richard Biener <rguenther@suse.de>
2536 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
2538 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
2539 print of no dependency loops ...
2540 * graphite.c (graphite_transform_loops): ... here.
2541 (canonicalize_loop_closed_ssa_form): Work from inner to outer
2543 (same_close_phi_node, remove_duplicate_close_phi,
2544 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
2545 (canonicalize_loop_closed_ssa): ... here and simplify.
2546 * graphite-optimize-isl.c: Include tree-vectorizer.h.
2547 (optimize_isl): Use dump_printf_loc to tell when we stopped
2548 optimizing because of an ISL timeout.
2550 2017-09-22 Richard Biener <rguenther@suse.de>
2552 PR tree-optimization/82291
2553 * tree-if-conv.c (predicate_mem_writes): Make sure to
2554 remove writes in blocks predicated with false.
2556 2017-09-22 Richard Biener <rguenther@suse.de>
2558 * sese.c: Include cfganal.h.
2559 (if_region_set_false_region): Remove.
2560 (create_if_region_on_edge): Likewise.
2561 (move_sese_in_condition): Re-implement without destroying
2564 2017-09-22 Richard Biener <rguenther@suse.de>
2566 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
2567 Verify both BBs contain loop PHI nodes before dispatching to
2569 (graphite_regenerate_ast_isl): Do not recompute dominators,
2570 do not verify three times. Restructure for clarity.
2571 * graphite-scop-detection.c (same_close_phi_node,
2572 remove_duplicate_close_phi, make_close_phi_nodes_unique,
2573 defined_in_loop_p, canonicalize_loop_closed_ssa,
2574 canonicalize_loop_closed_ssa_form): Simplify, remove excess
2575 checking and SSA rewrite, move to ...
2576 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
2577 (graphite_initialize): Do not pass in ctx, do not reset the
2578 SCEV cache, compute only dominators.
2579 (graphite_transform_loops): Allocate ISL ctx after
2580 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
2581 Maintain post-dominators only around build_scops.
2582 * sese.c (if_region_set_false_region): Make static. Free
2583 and recompute dominators.
2584 (move_sese_in_condition): Assert we don't get called with
2585 post-dominators computed.
2586 * sese.h (if_region_set_false_region): Remove.
2588 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
2590 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
2591 mode attribute for TARGET_AVX512VL.
2593 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
2595 * config/i386/i386.opt (mprefer-avx256): New option.
2596 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
2598 (ix86_preferred_simd_mode): Return 256-bit AVX modes
2599 for TARGET_PREFER_AVX256.
2600 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
2602 2017-09-21 Jeff Law <law@redhat.com>
2604 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
2605 Fix dump output if the only stack space is for pushed registers.
2607 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2609 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
2612 2017-09-21 Martin Sebor <msebor@redhat.com>
2615 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
2616 code (in C++) or code that triggers warnings.
2618 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
2620 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
2622 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
2624 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
2625 * sched-rgn.c: Ditto.
2626 * sel-sched-ir.c: Ditto.
2628 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
2630 * toplev.h (set_random_seed): Adjust return type.
2631 * toplev.c (init_local_tick): Move eager initialization of random_seed
2632 to get_random_seed. Adjust comment.
2633 (init_random_seed): Inline to get_random_seed, delete.
2634 (get_random_seed): Initialize random_seed lazily.
2635 (set_random_seed): Do not return previous value.
2636 (print_switch_value): Do not call get_random_seed.
2638 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
2640 * cgraph.c (delete_function_version): New, broken out from...
2641 (cgraph_node::delete_function_version): ...here. Rename to
2642 cgraph_node::delete_function_version_by_decl. Update all uses.
2643 (cgraph_node::remove): Call delete_function_version.
2645 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2648 * tree-inline.c (expand_call_inline): Emit clobber stmts for
2649 VAR_DECLs to which addressable non-volatile parameters are mapped
2650 and for id->retvar after the return value assignment. Clear
2651 id->retval and id->retbnd after inlining.
2653 2017-09-21 Richard Biener <rguenther@suse.de>
2655 PR tree-optimization/82276
2656 PR tree-optimization/82244
2657 * tree-vrp.c (build_assert_expr_for): Set
2658 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
2660 (remove_range_assertions): Revert earlier change.
2662 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
2665 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
2667 2017-09-21 Richard Biener <rguenther@suse.de>
2669 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2670 Restore valid IL after code generation errors.
2671 * graphite.c (graphite_transform_loops): Diagnose code
2672 generation issues as MSG_MISSED_OPTIMIZATION and continue
2673 with processing SCOPs.
2675 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2676 Alan Hayward <alan.hayward@arm.com>
2677 David Sherwood <david.sherwood@arm.com>
2679 * calls.c (compute_argument_addresses): Use simplify_gen_binary
2680 rather than choosing between plus_constant and gen_rtx_<CODE>.
2681 * expr.c (emit_push_insn): Likewise.
2682 (expand_expr_real_2): Likewise.
2684 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2685 Alan Hayward <alan.hayward@arm.com>
2686 David Sherwood <david.sherwood@arm.com>
2688 * loop-unroll.c (split_iv): Call copy_rtx on the step.
2690 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2691 Alan Hayward <alan.hayward@arm.com>
2692 David Sherwood <david.sherwood@arm.com>
2694 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
2695 calling tree_to_uhwi.
2697 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2698 Alan Hayward <alan.hayward@arm.com>
2699 David Sherwood <david.sherwood@arm.com>
2701 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
2702 INTEGER_CST rather than a negative test for ADDR_EXPR.
2704 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
2705 Alan Hayward <alan.hayward@arm.com>
2706 David Sherwood <david.sherwood@arm.com>
2708 * tree-vrp.c (extract_range_from_binary_expr_1): Check
2709 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
2711 2017-09-21 Richard Biener <rguenther@suse.de>
2713 PR tree-optimization/71351
2714 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
2715 graphite_create_new_loop_guard): Remove, fold remaining parts
2717 (translate_isl_ast_node_for): ... here and simplify.
2719 2017-09-21 Jakub Jelinek <jakub@redhat.com>
2722 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
2723 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
2724 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
2725 alternative always use QI mode, for -Os imov (=R,R) alternative
2726 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
2729 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2730 Jeff Law <law@redhat.com>
2732 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
2733 (allocate_stack_space): New function, partially extracted from
2735 (s390_emit_prologue): Track offset to most recent stack probe.
2736 Code to allocate space moved into allocate_stack_space.
2737 Dump actions when no stack is allocated.
2738 (s390_prologue_plus_offset): New function.
2739 (s390_emit_stack_probe): Likewise.
2741 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
2743 * common.opt (Wa, Wl, Wp, g, gz=): Add
2745 (gno-column-info): Remove.
2746 (gcolumn-info): Drop RejectNegative.
2748 (gno-record-gcc-switches): Remove.
2749 (grecord-gcc-switches): Drop RejectNegative.
2750 (gno-split-dwarf): Remove.
2751 (gsplit-dwarf): Drop RejectNegative.
2752 (gno-strict-dwarf): Remove.
2753 (gstrict-dwarf): Drop RejectNegative.
2754 * config/darwin.opt (gfull, gused): Add RejectNegative.
2755 * dwarf2out.c (gen_producer_string): Drop
2756 gno-record-gcc-switches handler.
2757 * optc-gen.awk: Add g to prefixes with negative forms.
2758 * opts-common.c (remapping_prefix_p): New.
2759 (find_opt): Check it.
2760 (generate_canonical_option): Test g prefix.
2761 (option_map): Add -gno- mapping.
2762 (add_misspelling_candidates): Check remapping_prefix_p.
2764 2017-09-20 Jeff Law <law@redhat.com>
2766 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
2767 thinko in stack clash protection support.
2769 * explow.c (compute_stack_clash_protection_loop_data): Use
2770 CONST_INT_P instead of explicit test. Verify object is a
2771 CONST_INT_P before looking at INTVAL.
2772 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
2773 instead of explicit test.
2775 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
2778 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
2779 address instead of to r1 and r11.
2781 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
2783 * config.gcc: Support "knm".
2784 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
2785 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
2787 * config/i386/i386.c (m_KNM): Define.
2788 (processor_target_table): Add "knm".
2790 (ix86_option_override_internal): Add "knm".
2791 (ix86_issue_rate): Add PROCESSOR_KNM.
2792 (ix86_adjust_cost): Ditto.
2793 (ia32_multipass_dfa_lookahead): Ditto.
2794 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
2795 (fold_builtin_cpu): Add M_INTEL_KNM.
2796 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
2797 (processor_type): Add PROCESSOR_KNM.
2798 * config/i386/x86-tune.def: Add m_KNM.
2799 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
2801 2017-09-20 Richard Biener <rguenther@suse.de>
2803 PR tree-optimization/80213
2804 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
2805 are allowed in empty BBs as well.
2806 (canonicalize_loop_closed_ssa): Also look for other complex
2808 (scop_detection::get_sese): Include the loop-closed PHI block
2810 (scop_detection::merge_sese): Remove code adding extra blocks.
2811 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
2812 (build_scops): Assert the final returned scop is invalid.
2814 2017-09-20 Richard Biener <rguenther@suse.de>
2816 PR tree-optimization/82264
2817 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
2819 (vn_phi_lookup): Likewise.
2820 (vn_phi_insert): Likewise.
2822 2017-09-20 Jakub Jelinek <jakub@redhat.com>
2824 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
2825 that fits into uhwi or shwi, add DW_AT_const_value regardless
2826 of early_dwarf without going through RTL, using add_AT_unsigned
2829 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
2830 (DEBUG_LTO_ABBREV_SECTION): Likewise.
2831 (DEBUG_LTO_MACINFO_SECTION): Likewise.
2832 (DEBUG_MACRO_SECTION): Likewise.
2833 (DEBUG_LTO_MACRO_SECTION): Likewise.
2834 (DEBUG_STR_DWO_SECTION): Likewise.
2835 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
2836 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
2837 (DEBUG_LTO_DWO_LINE_SECTION): Define.
2838 (DEBUG_LTO_LINE_STR_SECTION): Define.
2839 (init_sections_and_labels): Initialize debug_line_str_section
2840 variable. Initialize debug_loc_section for -gdwarf-5 to
2841 DEBUG_LOCLISTS_SECTION. Formatting fixes.
2843 2017-09-20 Richard Biener <rguenther@suse.de>
2845 * graphite-sese-to-poly.c (extract_affine): Properly handle
2846 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
2848 2017-09-20 Richard Biener <rguenther@suse.de>
2850 PR tree-optimization/81373
2851 * graphite-scop-detection.c (build_cross_bb_scalars_def):
2852 Force SESE live-out defs to be handled even if they are
2855 2017-09-19 Jeff Law <law@redhat.com>
2857 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
2858 nothing for stack adjustments with REG_STACK_CHECK.
2859 * sched-deps.c (parse_add_or_inc): Reject insns with
2860 REG_STACK_CHECK from dependency breaking.
2861 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
2862 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
2863 * reg-notes.def (STACK_CHECK): New note.
2865 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
2866 (ix86_expand_prologue): Dump stack clash info as needed.
2867 Call ix86_adjust_stack_and_probe_stack_clash as needed.
2869 * function.c (dump_stack_clash_frame_info): New function.
2870 * function.h (dump_stack_clash_frame_info): Prototype.
2871 (enum stack_clash_probes): New enum.
2873 * config/alpha/alpha.c (alpha_expand_prologue): Also check
2874 flag_stack_clash_protection.
2875 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
2876 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
2877 (arm_frame_pointer_required): Likewise.
2878 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
2879 (ia64_expand_prologue): Likewise.
2880 * config/mips/mips.c (mips_expand_prologue): Likewise.
2881 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
2882 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2883 (sparc_flat_expand_prologue): Likewise.
2884 * config/spu/spu.c (spu_expand_prologue): Likewise.
2886 * explow.c: Include "params.h".
2887 (anti_adjust_stack_and_probe_stack_clash): New function.
2888 (get_stack_check_protect): Likewise.
2889 (compute_stack_clash_protection_loop_data): Likewise.
2890 (emit_stack_clash_protection_loop_start): Likewise.
2891 (emit_stack_clash_protection_loop_end): Likewise.
2892 (allocate_dynamic_stack_space): Use get_stack_check_protect.
2893 Use anti_adjust_stack_and_probe_stack_clash.
2894 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
2895 (emit_stack_clash_protection_loop_start): Likewise.
2896 (emit_stack_clash_protection_loop_end): Likewise.
2897 * rtl.h (get_stack_check_protect): Prototype.
2898 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
2899 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
2900 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
2902 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
2904 * doc/tm.texi: Rebuilt.
2905 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
2906 get_stack_check_protect.
2907 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
2908 * config/arm/arm.c (arm_expand_prologue): Likewise.
2909 (arm_frame_pointer_required): Likewise.
2910 * config/i386/i386.c (ix86_expand_prologue): Likewise.
2911 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
2912 * config/mips/mips.c (mips_expand_prologue): Likewise.
2913 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
2914 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
2915 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
2916 (sparc_flat_expand_prologue): Likewise.
2918 * common.opt (-fstack-clash-protection): New option.
2919 * flag-types.h (enum stack_check_type): Note difference between
2920 -fstack-check= and -fstack-clash-protection.
2921 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
2922 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
2923 * toplev.c (process_options): Issue warnings/errors for cases
2924 not handled with -fstack-clash-protection.
2925 * doc/invoke.texi (-fstack-clash-protection): Document new option.
2926 (-fstack-check): Note additional problem with -fstack-check=generic.
2927 Note that -fstack-check is primarily for Ada and refer users
2928 to -fstack-clash-protection for stack-clash-protection.
2929 Document new params for stack clash protection.
2931 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
2933 * config/i386/i386.c (ix86_split_long_move): Do not handle
2934 address used for LEA in a special way.
2936 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
2938 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
2940 2017-09-19 Martin Sebor <msebor@redhat.com>
2943 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
2944 of incompatible types.
2946 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
2948 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
2949 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
2950 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
2951 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
2953 2017-09-19 Richard Biener <rguenther@suse.de>
2955 PR tree-optimization/82244
2956 * tree-vrp.c (remove_range_assertions): Do not propagate
2957 a constant to abnormals but replace the assert with a copy.
2959 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
2961 PR rtl-optimization/57878
2962 PR rtl-optimization/68988
2963 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
2964 avoidance test involving non_reload_pseudos. Move frequency test
2965 below the general fragmentation avoidance test.
2967 2017-09-19 Richard Biener <rguenther@suse.de>
2969 PR tree-optimization/69728
2970 * graphite-sese-to-poly.c (schedule_error): New global.
2971 (add_loop_schedule): Handle empty domain by failing the
2973 (build_original_schedule): Handle schedule_error.
2975 2017-09-19 Richard Biener <rguenther@suse.de>
2977 * graphite-scop-detection.c (scop_detection::can_represent_loop):
2978 Do not iterate to sibling loops but only to siblings of inner
2981 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
2984 * config/m68k/m68k.md (moveq feeding equality comparison): Check
2985 that the registers are different.
2987 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
2989 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
2990 to processor_model and "amdfam17h" to arch_names_table.
2991 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
2993 2017-09-18 Jakub Jelinek <jakub@redhat.com>
2996 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
2998 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
2999 Alan Hayward <alan.hayward@arm.com>
3000 David Sherwood <david.sherwood@arm.com>
3002 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
3004 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
3006 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
3007 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
3008 vect_schedule_slp_instance.
3009 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
3010 Update call to vect_slp_analyze_node_operations. Simplify return
3012 (vect_slp_analyze_bb_1): Update call accordingly.
3013 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
3014 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
3015 (vect_schedule_slp): Update call accordingly.
3017 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3018 Alan Hayward <alan.hayward@arm.com>
3019 David Sherwood <david.sherwood@arm.com>
3021 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
3022 with types that aren't in fact scalar.
3024 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3026 * tree-vect-slp.c (vect_record_max_nunits): New function,
3028 (vect_build_slp_tree_1): ...here.
3029 (vect_build_slp_tree_2): Call it for phis too.
3031 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3033 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
3034 to vect_get_vec_def_for_operand when getting the mask operand.
3036 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3037 Alan Hayward <alan.hayward@arm.com>
3038 David Sherwood <david.sherwood@arm.com>
3040 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
3043 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3044 Alan Hayward <alan.hayward@arm.com>
3045 David Sherwood <david.sherwood@arm.com>
3047 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
3048 calculation for vector booleans.
3050 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
3051 Alan Hayward <alan.hayward@arm.com>
3052 David Sherwood <david.sherwood@arm.com>
3054 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
3056 (vect_transform_stmt): ...here.
3057 (vect_analyze_stmt): Use it instead of calling
3058 vectorizable_live_operation directly.
3060 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
3062 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
3063 non-SIMT targets in acc vector loops.
3065 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
3067 * configure.ac: Add arc and check if assembler supports gdwarf2.
3068 * configure: Regenerate.
3070 2017-09-18 Richard Biener <rguenther@suse.de>
3072 PR tree-optimization/82220
3073 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
3074 epilogue niters from the min_profitable_iters compute.
3076 2017-09-18 Jakub Jelinek <jakub@redhat.com>
3079 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
3080 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
3081 (ix86_init_pic_reg): Revert 2017-09-01 changes.
3083 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
3086 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
3087 switching to a new text section.
3089 2017-09-18 Richard Biener <rguenther@suse.de>
3091 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
3093 (build_alias_set): Reject aliases with no access function.
3095 2017-09-18 Richard Biener <rguenther@suse.de>
3097 PR tree-optimization/79622
3098 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
3100 (build_cross_bb_scalars_use): Likewise.
3102 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
3104 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
3106 2017-09-18 Alan Modra <amodra@gmail.com>
3109 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
3110 stack_pointer_rtx for count 0. Update comments. Break up
3111 large rtl expression.
3113 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
3115 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
3116 Increase to 20 bytes.
3117 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
3118 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
3119 or avx version of the stub.
3121 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
3124 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
3125 compute the minimum stack alignment. Also update preferred stack
3126 boundary for leaf functions.
3128 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
3130 PR tree-optimization/82228
3131 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
3134 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3136 * common/config/nds32/nds32-common.c
3137 (nds32_option_optimization_table): Refine formatting.
3138 (nds32_option_optimization_table): Use -fsched-pressure and
3139 -fomit-frame-pointer for specific optimization level.
3141 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
3143 * config/nds32/nds32.c: Refine formatting and comments.
3144 * config/nds32/nds32.h: Likewise.
3145 * config/nds32/nds32.md: Likewise.
3146 * config/nds32/nds32-cost.c: Likewise.
3147 * config/nds32/nds32-isr.c: Likewise.
3148 * config/nds32/nds32-md-auxiliary.c: Likewise.
3149 * config/nds32/nds32-multiple.md: Likewise.
3150 * config/nds32/nds32-predicates.c: Likewise.
3152 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
3153 Jakub Jelinek <jakub@redhat.com>
3155 Add support for -std=c++2a.
3156 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
3158 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
3160 2017-09-15 Steve Ellcey <sellcey@cavium.com>
3163 * doc/extend.texi (Common Function Attributes): Add
3164 references to ARM, AArch64, and S/390 specific attributes.
3165 (Function Specific Option Pragmas): Add AArch64 and S/390
3166 to list of back ends that support the target pragma.
3168 2017-09-15 Nathan Sidwell <nathan@acm.org>
3170 * doc/standards.texi: Fix C++17 description. Update URLs for
3173 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
3175 * common.opt (Wcast-align=strict): New warning option.
3176 * doc/invoke.texi: Document -Wcast-align=strict.
3178 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
3180 * cgraph.h (cgraph_thunk_info): Add comments.
3181 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
3182 assert for VIRTUAL_* arguments stricter.
3184 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
3186 PR tree-optimization/71026
3187 * match.pd: Move RDIV patterns from fold-const.c
3188 * fold-const.c (distribute_real_division): Removed.
3189 (fold_binary_loc): Remove calls to distribute_real_divison.
3191 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3193 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
3194 c++1z and gnu++1z as deprecated. Change other references to
3195 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
3196 Change -Wc++1z-compat to -Wc++17-compat.
3197 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
3198 * dwarf2out.c (highest_c_language): Handle C++17.
3199 (gen_compile_unit_die): Likewise.
3201 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3203 PR rtl-optimization/82192
3204 * combine.c (make_extraction): Don't look through non-paradoxical
3205 SUBREGs or TRUNCATE if pos + len is or might be bigger than
3208 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3209 Alan Hayward <alan.hayward@arm.com>
3210 David Sherwood <david.sherwood@arm.com>
3212 * target.def (function_arg_offset): New hook.
3213 * targhooks.h (default_function_arg_offset): Declare.
3214 * targhooks.c (default_function_arg_offset): New function.
3215 * function.c (locate_and_pad_parm): Use
3216 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
3217 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
3218 (TARGET_FUNCTION_ARG_OFFSET): ...this.
3219 * doc/tm.texi: Regenerate.
3220 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
3221 * config/spu/spu.c (spu_function_arg_offset): New function.
3222 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
3223 * system.h (FUNCTION_ARG_OFFSET): Poison.
3225 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3226 Alan Hayard <alan.hayward@arm.com>
3227 David Sherwood <david.sherwood@arm.com>
3229 * target.def (truly_noop_truncation): New hook.
3230 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
3231 than TRULY_NOOP_TRUNCATION.
3232 * hooks.h (hook_bool_uint_uint_true): Declare.
3233 * hooks.c (hook_bool_uint_uint_true): New function.
3234 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
3235 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
3236 * doc/tm.texi: Regenerate.
3237 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
3238 rather than TRULY_NOOP_TRUNCATION in comments.
3239 (simplify_comparison): Likewise.
3240 (record_truncated_value): Likewise.
3241 * expmed.c (extract_bit_field_1): Likewise.
3242 (extract_split_bit_field): Likewise.
3243 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
3244 instead of TRULY_NOOP_TRUNCATION.
3245 * function.c (assign_parm_setup_block): Likewise.
3246 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
3247 * rtlhooks.c: Include target.h.
3248 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
3249 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
3250 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
3251 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
3252 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
3253 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
3254 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
3255 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
3256 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
3257 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
3258 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
3259 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
3260 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
3261 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
3262 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
3263 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
3264 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
3265 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
3266 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
3267 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
3268 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
3269 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
3270 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
3271 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
3272 * config/mips/mips.c (mips_truly_noop_truncation): New function.
3273 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3274 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3275 rather than TRULY_NOOP_TRUNCATION in comments.
3276 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
3277 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
3278 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
3279 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
3280 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
3281 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
3282 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
3283 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
3284 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
3285 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
3286 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
3287 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3288 rather than TRULY_NOOP_TRUNCATION in comments.
3289 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
3290 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
3291 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
3292 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
3293 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
3294 TRULY_NOOP_TRUNCATION condition.
3295 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
3296 (TRULY_NOOP_TRUNCATION): Delete.
3297 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
3298 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
3299 * config/spu/spu.c (spu_truly_noop_truncation): New function.
3300 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3301 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
3302 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
3303 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
3304 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
3305 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
3306 rather than TRULY_NOOP_TRUNCATION in comments.
3307 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
3308 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
3309 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
3310 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
3311 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
3312 * system.h (TRULY_NOOP_TRUNCATION): Poison.
3314 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
3317 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
3318 (*cmp_ior): Likewise.
3319 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
3320 (*ior_scc_scc_cmp): Likewise.
3321 (*and_scc_scc): Likewise.
3322 (*and_scc_scc_cmp): Likewise.
3324 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
3325 Alan Hayard <alan.hayward@arm.com>
3326 David Sherwood <david.sherwood@arm.com>
3328 * target.def (can_change_mode_class): New hook.
3329 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
3330 (hard_regno_nregs): Likewise.
3331 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
3332 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
3333 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
3334 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
3335 (LOAD_EXTEND_OP): Update accordingly.
3336 * doc/tm.texi: Regenerate.
3337 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
3338 CANNOT_CHANGE_MODE_CLASS.
3339 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
3340 (REG_CAN_CHANGE_MODE_P): ...this new macro.
3341 * combine.c (simplify_set): Update accordingly.
3342 * emit-rtl.c (validate_subreg): Likewise.
3343 * recog.c (general_operand): Likewise.
3344 * regcprop.c (mode_change_ok): Likewise.
3345 * reload1.c (choose_reload_regs): Likewise.
3346 (inherit_piecemeal_p): Likewise.
3347 * rtlanal.c (simplify_subreg_regno): Likewise.
3348 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
3349 instead of CANNOT_CHANGE_MODE_CLASS.
3350 (reload_cse_simplify_operands): Likewise.
3351 * reload.c (push_reload): Use targetm.can_change_mode_class
3352 instead of CANNOT_CHANGE_MODE_CLASS.
3353 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
3354 REG_CANNOT_CHANGE_MODE_P.
3355 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3356 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
3357 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3358 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3359 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3360 (arm_can_change_mode_class): New function.
3361 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3362 than CANNOT_CHANGE_MODE_CLASS in comments.
3363 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3364 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
3365 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
3366 (ix86_can_change_mode_class): ...this new function, inverting the
3367 sense of the return value.
3368 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3369 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3370 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3371 (ia64_can_change_mode_class): New function.
3372 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3373 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
3374 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
3375 (m32c_can_change_mode_class): ...this new function, inverting the
3376 sense of the return value.
3377 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3378 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3379 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
3380 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
3381 (mips_can_change_mode_class): ...this new function, inverting the
3382 sense of the return value.
3383 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3384 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3385 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3386 (msp430_can_change_mode_class): New function.
3387 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3388 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
3389 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3390 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3391 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3392 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
3393 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3394 (pa_cannot_change_mode_class): Replace with...
3395 (pa_can_change_mode_class): ...this new function, inverting the
3396 sense of the return value.
3397 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
3398 than CANNOT_CHANGE_MODE_CLASS in comments.
3399 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3400 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
3401 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3402 (pdp11_cannot_change_mode_class): Replace with...
3403 (pdp11_can_change_mode_class): ...this new function, inverting the
3404 sense of the return value.
3405 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3406 * config/powerpcspe/powerpcspe-protos.h
3407 (rs6000_cannot_change_mode_class_ptr): Delete.
3408 * config/powerpcspe/powerpcspe.c
3409 (rs6000_cannot_change_mode_class_ptr): Delete.
3410 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3411 (rs6000_option_override_internal): Assign to
3412 targetm.can_change_mode_class instead of
3413 rs6000_cannot_change_mode_class_ptr.
3414 (rs6000_cannot_change_mode_class): Replace with...
3415 (rs6000_can_change_mode_class): ...this new function, inverting the
3416 sense of the return value.
3417 (rs6000_debug_cannot_change_mode_class): Replace with...
3418 (rs6000_debug_can_change_mode_class): ...this new function.
3419 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3420 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
3421 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3422 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3423 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
3425 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
3426 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3427 (rs6000_option_override_internal): Assign to
3428 targetm.can_change_mode_class instead of
3429 rs6000_cannot_change_mode_class_ptr.
3430 (rs6000_cannot_change_mode_class): Replace with...
3431 (rs6000_can_change_mode_class): ...this new function, inverting the
3432 sense of the return value.
3433 (rs6000_debug_cannot_change_mode_class): Replace with...
3434 (rs6000_debug_can_change_mode_class): ...this new function.
3435 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3436 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
3437 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
3438 (s390_can_change_mode_class): ...this new function, inverting the
3439 sense of the return value.
3440 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3441 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3442 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
3443 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3444 (sh_cannot_change_mode_class): Replace with...
3445 (sh_can_change_mode_class): ...this new function, inverting the
3446 sense of the return value.
3447 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3448 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3449 (sparc_can_change_mode_class): New function.
3450 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3451 * config/spu/spu.c (spu_can_change_mode_class): New function.
3452 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3453 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
3454 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
3455 (visium_can_change_mode_class): New function.
3456 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
3458 2017-09-15 Richard Biener <rguenther@suse.de>
3460 PR tree-optimization/82217
3461 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
3462 but not undefined case.
3464 2017-09-15 Jakub Jelinek <jakub@redhat.com>
3467 * postreload.c (reload_cse_simplify_operands): Skip
3468 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
3470 2017-09-15 Richard Biener <rguenther@suse.de>
3472 PR tree-optimization/68823
3473 * graphite-scop-detection.c (build_alias_set): If we have a
3474 possible dependence check whether we can handle them by just
3475 looking at the DRs DR_ACCESS_FNs.
3476 (build_scops): If build_alias_set fails, fail the SCOP.
3478 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
3480 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
3481 to support float128 built-in functions that require the ISA 3.0
3483 (BU_FLOAT128_3_HW): Likewise.
3484 (SQRTF128): Add support for the IEEE 128-bit square root and fma
3486 (FMAF128): Likewise.
3488 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
3489 support for built-in functions that need the ISA 3.0 IEEE 128-bit
3490 floating point instructions.
3491 (rs6000_invalid_builtin): Likewise.
3492 (rs6000_builtin_mask_names): Likewise.
3493 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
3494 (RS6000_BTM_FLOAT128_HW): Likewise.
3495 (RS6000_BTM_COMMON): Likewise.
3496 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
3498 * doc/extend.texi (RS/6000 built-in functions): Document the
3499 IEEE 128-bit floating point square root and fused multiply-add
3502 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
3504 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
3505 reg (r2) isn't in the set of registers defined in the prologue.
3507 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3508 Alan Hayward <alan.hayward@arm.com>
3509 David Sherwood <david.sherwood@arm.com>
3511 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
3512 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
3513 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
3514 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
3515 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
3517 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
3518 max_vectorization_factor.
3519 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
3521 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3522 Alan Hayward <alan.hayward@arm.com>
3523 David Sherwood <david.sherwood@arm.com>
3525 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
3526 (vect_worthwhile_without_simd_p): Declare.
3527 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
3528 (vectorizable_reduction): Use it.
3529 * tree-vect-stmts.c (vectorizable_shift): Likewise.
3530 (vectorizable_operation): Likewise.
3532 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3533 Alan Hayward <alan.hayward@arm.com>
3534 David Sherwood <david.sherwood@arm.com>
3536 * tree-vectorizer.h (vect_get_num_copies): New function.
3537 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
3538 * tree-vect-loop.c (vectorizable_reduction): Likewise.
3539 (vectorizable_induction): Likewise.
3540 (vectorizable_live_operation): Likewise.
3541 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
3542 (vectorizable_bswap): Likewise.
3543 (vectorizable_call): Likewise.
3544 (vectorizable_conversion): Likewise.
3545 (vectorizable_assignment): Likewise.
3546 (vectorizable_shift): Likewise.
3547 (vectorizable_operation): Likewise.
3548 (vectorizable_store): Likewise.
3549 (vectorizable_load): Likewise.
3550 (vectorizable_condition): Likewise.
3551 (vectorizable_comparison): Likewise.
3552 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
3554 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3555 Alan Hayward <alan.hayward@arm.com>
3556 David Sherwood <david.sherwood@arm.com>
3558 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
3559 of vect_init_vector.
3561 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3562 Alan Hayward <alan.hayward@arm.com>
3563 David Sherwood <david.sherwood@arm.com>
3565 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
3566 an inline wrapper that provides a location.
3567 (gimple_build_vector): Likewise.
3568 * gimple-fold.c (gimple_build_vector_from_val): New function.
3569 (gimple_build_vector): Likewise.
3570 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
3571 functions to build the initial value. Always return a gimple value.
3572 (get_initial_defs_for_reduction): Likewise. Only compute
3574 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
3575 vect_init_vector on the results from get_initial_def(s)_for_reduction.
3576 (vectorizable_induction): Use gimple_build_vector rather than
3579 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3580 Alan Hayward <alan.hayward@arm.com>
3581 David Sherwood <david.sherwood@arm.com>
3583 * target.h (vec_perm_indices): New typedef.
3584 (auto_vec_perm_indices): Likewise.
3585 * optabs-query.h: Include target.h
3586 (can_vec_perm_p): Take a vec_perm_indices *.
3587 * optabs-query.c (can_vec_perm_p): Likewise.
3588 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
3589 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3590 * tree-vect-generic.c (lower_vec_perm): Likewise.
3591 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
3592 (vect_grouped_load_supported): Likewise.
3593 (vect_shift_permute_load_chain): Likewise.
3594 (vect_permute_store_chain): Use auto_vec_perm_indices.
3595 (vect_permute_load_chain): Likewise.
3596 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
3597 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
3598 Update uses of can_vec_perm_p.
3599 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
3600 mode with a number of elements. Take a vec_perm_indices *.
3601 (vect_create_epilog_for_reduction): Update accordingly.
3602 Use auto_vec_perm_indices.
3603 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
3604 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
3605 (vect_transform_slp_perm_load): Likewise.
3606 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
3607 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
3608 (vect_gen_perm_mask_checked): Likewise.
3609 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
3610 (vect_gen_perm_mask_checked): Likewise.
3611 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
3612 (vectorizable_store): Likewise.
3613 (vectorizable_load): Likewise.
3614 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
3615 (vectorizable_bswap): Likewise.
3617 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3618 Alan Hayward <alan.hayward@arm.com>
3619 David Sherwood <david.sherwood@arm.com>
3621 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
3622 * tree.c (build_vector): Likewise.
3623 (build_vector_from_ctor): Update accordingly.
3624 (build_vector_from_val): Likewise.
3625 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
3626 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
3627 * tree-vect-generic.c (add_rshift): Likewise.
3628 (expand_vector_divmod): Likewise.
3629 (optimize_vector_constructor): Likewise.
3630 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
3631 (vect_transform_slp_perm_load): Likewise.
3632 (vect_schedule_slp_instance): Likewise.
3633 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
3634 (vectorizable_call): Likewise.
3635 (vect_gen_perm_mask_any): Likewise. Add elements in order.
3636 * expmed.c (make_tree): Likewise.
3637 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
3638 a vector passed to build_vector.
3639 (fold_convert_const): Likewise.
3640 (exact_inverse): Likewise.
3641 (fold_ternary_loc): Likewise.
3642 (fold_relational_const): Likewise.
3643 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
3644 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
3645 (const_unop): Likewise. Store the reduction accumulator in a
3646 variable rather than an array.
3647 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
3648 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
3649 the new vector, rather than constructing it after the input arrays.
3650 (native_interpret_vector): Use auto_vec<tree> when building
3651 a vector passed to build_vector. Add elements in order.
3652 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
3653 auto_vec<tree> when building a vector passed to build_vector.
3654 (vect_create_epilog_for_reduction): Likewise.
3655 (vectorizable_induction): Likewise.
3656 (get_initial_def_for_reduction): Likewise. Fix indentation of
3658 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
3660 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
3661 passed to build_vector.
3663 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
3664 Alan Hayward <alan.hayward@arm.com>
3665 David Sherwood <david.sherwood@arm.com>
3667 * tree-core.h (tree_base::u): Add an "nelts" field.
3668 (tree_vector): Use VECTOR_CST_NELTS as the length.
3669 * tree.c (tree_size): Likewise.
3670 (make_vector): Initialize VECTOR_CST_NELTS.
3671 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
3672 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
3673 TYPE_VECTOR_SUBPARTS.
3674 * expr.c (const_vector_mask_from_tree): Consistently use "units"
3675 as the number of units, setting it from VECTOR_CST_NELTS.
3676 (const_vector_from_tree): Likewise.
3677 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
3678 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
3679 (fold_negate_expr_1): Likewise.
3680 (fold_convert_const): Likewise.
3681 (const_binop): Likewise. Differentiate the number of output and
3683 (const_unop): Likewise.
3684 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
3685 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
3686 in cases that did the opposite.
3688 2017-09-14 Richard Biener <rguenther@suse.de>
3690 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
3693 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
3695 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
3697 2017-09-14 Jakub Jelinek <jakub@redhat.com>
3700 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
3701 if and where to split a bb, except for splitting before debug insn
3702 sequences followed by non-label real insn. Delete debug insns
3703 in between basic blocks.
3705 * combine.c (make_compound_operation_int): Formatting fixes.
3707 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
3708 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3709 * config/netbsd.h (LINK_EH_SPEC): Likewise.
3710 * config/sol2.h (LINK_EH_SPEC): Likewise.
3711 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3712 * config/s390/linux.h (LINK_SPEC): Likewise.
3713 * config/freebsd.h (LINK_EH_SPEC): Likewise.
3714 * config/openbsd.h (LINK_EH_SPEC): Likewise.
3715 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3716 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
3717 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
3718 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3719 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
3720 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
3722 2017-09-13 Jakub Jelinek <jakub@redhat.com>
3724 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
3726 (ENDFILE_LINUX_SPEC): Likewise.
3727 (LINK_EH_SPEC): Likewise.
3728 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
3729 (LINK_OS_LINUX_SPEC32): Likewise.
3730 (LINK_OS_LINUX_SPEC64): Likewise.
3731 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
3732 (LINK_OS_LINUX_SPEC): Likewise.
3734 2017-09-13 Martin Liska <mliska@suse.cz>
3737 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
3738 CASE_HIGH is NULL_TREE.
3740 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3741 Alan Hayward <alan.hayward@arm.com>
3742 David Sherwood <david.sherwood@arm.com>
3744 * target.def (secondary_memory_needed): New hook.
3745 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
3746 instead of SECONDARY_MEMORY_NEEDED.
3747 (secondary_memory_needed_mode): Likewise.
3748 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
3749 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
3750 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
3751 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
3752 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
3753 * doc/tm.texi: Regenerate.
3754 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
3755 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
3756 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3757 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
3758 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
3759 * config/i386/i386.c (inline_secondary_memory_needed): Put the
3760 mode argument first and change the reg_class arguments to reg_class_t.
3761 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
3762 Make static. Update the call to inline_secondary_memory_needed.
3763 (ix86_register_move_cost): Update the call to
3764 inline_secondary_memory_needed.
3765 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3766 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
3768 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
3769 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3771 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
3772 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
3773 * config/mips/mips.c (mips_secondary_memory_needed): Make static
3774 and match hook interface. Add comment from mips.h.
3775 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3776 * config/mmix/mmix.md (truncdfsf2): Refer to
3777 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3779 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
3780 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
3781 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3782 (pa_secondary_memory_needed): New function.
3783 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
3784 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
3785 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3786 (pdp11_secondary_memory_needed): Make static and match hook interface.
3787 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
3788 * config/powerpcspe/powerpcspe-protos.h
3789 (rs6000_secondary_memory_needed_ptr): Delete.
3790 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
3792 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3793 (rs6000_option_override_internal): Assign to
3794 targetm.secondary_memory_needed rather than
3795 rs6000_secondary_memory_needed_ptr.
3796 (rs6000_secondary_memory_needed): Match hook interface.
3797 (rs6000_debug_secondary_memory_needed): Likewise.
3798 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
3799 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
3800 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
3801 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3802 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
3803 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
3805 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
3806 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3807 (rs6000_option_override_internal): Assign to
3808 targetm.secondary_memory_needed rather than
3809 rs6000_secondary_memory_needed_ptr.
3810 (rs6000_secondary_memory_needed): Match hook interface.
3811 (rs6000_debug_secondary_memory_needed): Likewise.
3812 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
3813 * config/s390/s390.c (s390_secondary_memory_needed): New function.
3814 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3815 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
3816 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
3817 (sparc_secondary_memory_needed): New function.
3818 * lra-constraints.c (check_and_process_move): Refer to
3819 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
3821 (curr_insn_transform): Likewise.
3822 (process_alt_operands): Use targetm.secondary_memory_needed
3823 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3824 (check_secondary_memory_needed_p): Likewise.
3825 (choose_split_class): Likewise.
3826 * reload.c: Unconditionally include code that was previously
3827 conditional on SECONDARY_MEMORY_NEEDED.
3828 (push_secondary_reload): Use targetm.secondary_memory_needed
3829 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3830 (push_reload): Likewise.
3831 * reload1.c: Unconditionally include code that was previously
3832 conditional on SECONDARY_MEMORY_NEEDED.
3833 (choose_reload_regs): Use targetm.secondary_memory_needed
3834 instead of TARGET_SECONDARY_MEMORY_NEEDED.
3835 (gen_reload): Likewise.
3836 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
3838 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
3839 Alan Hayward <alan.hayward@arm.com>
3840 David Sherwood <david.sherwood@arm.com>
3842 * target.def (secondary_memory_needed_mode): New hook:
3843 * targhooks.c (default_secondary_memory_needed_mode): Declare.
3844 * targhooks.h (default_secondary_memory_needed_mode): New function.
3845 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
3846 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
3847 * doc/tm.texi: Regenerate.
3848 * lra-constraints.c (check_and_process_move): Use
3849 targetm.secondary_memory_needed_mode instead of
3850 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
3851 (curr_insn_transform): Likewise.
3852 * reload.c (get_secondary_mem): Likewise.
3853 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3854 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
3856 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3857 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3858 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
3859 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3860 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
3862 * config/powerpcspe/powerpcspe-protos.h
3863 (rs6000_secondary_memory_needed_mode): Delete.
3864 * config/powerpcspe/powerpcspe.c
3865 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3866 (rs6000_secondary_memory_needed_mode): Make static.
3867 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3868 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
3870 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3872 (rs6000_secondary_memory_needed_mode): Make static.
3873 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3874 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
3875 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
3876 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
3877 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
3879 (sparc_secondary_memory_needed_mode): New function.
3880 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
3882 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
3884 * config/aarch64/constraints.md (Umq): New constraint.
3885 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
3887 (mov<mode>): Update condition.
3889 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3891 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
3892 when bitposition is the same.
3894 2017-09-13 Richard Biener <rguenther@suse.de>
3896 * dwarf2out.c (output_die_symbol): Remove.
3897 (output_die): Do not output a DIEs symbol.
3899 2017-09-13 Richard Biener <rguenther@suse.de>
3902 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
3903 default-def to avoid breaking iterator update with the weird
3904 interaction with cgraph_update_edges_for_call_stmt_node.
3906 2017-09-13 Richard Biener <rguenther@suse.de>
3908 * tree-cfg.c (verify_gimple_assign_binary): Add verification
3909 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
3910 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
3911 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
3913 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
3915 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
3916 Disable pc relative literal load irrespective of
3917 TARGET_FIX_ERR_A53_84341 for default.
3919 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
3921 * config/sparc/sparc.c (output_return): Output the source location of
3922 the insn in the delay slot, if any.
3923 (output_sibcall): Likewise.
3925 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
3928 * common.opt (-static-pie): New alias.
3929 (shared): Negate static-pie.
3930 (-no-pie): Update help text.
3932 (static-pie): New option.
3933 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
3934 -static-pie support.
3935 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
3936 (LINK_EH_SPEC): Likewise.
3937 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
3938 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3939 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
3940 * gcc.c (LINK_COMMAND_SPEC): Likewise.
3941 (init_gcc_specs): Likewise.
3942 (init_spec): Likewise.
3943 (display_help): Update help message for -pie.
3944 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
3947 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
3949 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
3950 (movdi_aarch64): Likewise.
3951 (movti_aarch64): Likewise.
3953 2017-09-12 Simon Wright <simon@pushface.org>
3956 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
3957 calculation of the minor version, always output as 0.
3959 2017-09-12 Jakub Jelinek <jakub@redhat.com>
3962 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
3963 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
3964 on it early, rather than manual conversion late. For
3965 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
3966 instead of performing manual conversion.
3968 2017-09-12 Carl Love <cel@us.ibm.com>
3970 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
3971 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
3973 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
3974 VMULOSW): Add definitions.
3975 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3976 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
3977 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
3978 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
3979 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
3981 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
3983 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
3985 (movti_aarch64): Likewise.
3986 (movdf_aarch64): Likewise.
3987 (movtf_aarch64): Likewise.
3988 (load_pairdi): Likewise.
3989 (store_pairdi): Likewise.
3990 (load_pairdf): Likewise.
3991 (store_pairdf): Likewise.
3992 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
3993 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
3994 (ldr_got_small_<mode>): Likewise.
3995 (ldr_got_small_28k_<mode>): Likewise.
3996 (ldr_got_tiny): Likewise.
3997 * config/aarch64/iterators.md (ldst_sz): New.
3998 (ldpstp_sz): Likewise.
3999 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
4001 (thunderx_load): Split load_8 to load_16.
4002 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
4004 (thunderx2t99_storepair_basic): Split store_8 to store_16.
4005 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
4006 (xgene1_store_pair): Split store_8 to store_16.
4007 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
4008 (falkor_st_0_st_sd): Split store_8 to store_16.
4010 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
4012 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
4013 and store1/2/3/4 to store_4/8/12/16.
4014 * config/aarch64/aarch64.md: Update for rename.
4015 * config/arm/arm.md: Likewise.: Likewise.
4016 * config/arm/arm.c: Likewise.
4017 * config/arm/thumb1.md: Likewise.
4018 * config/arm/thumb2.md: Likewise.
4019 * config/arm/vfp.md: Likewise.
4020 * config/arm/arm-generic.md: Likewise.
4021 * config/arm/arm1020e.md: Likewise.
4022 * config/arm/arm1026ejs.md: Likewise.
4023 * config/arm/arm1136jfs.md: Likewise.
4024 * config/arm/arm926ejs.md: Likewise.
4025 * config/arm/cortex-a15.md: Likewise.
4026 * config/arm/cortex-a17.md: Likewise.
4027 * config/arm/cortex-a5.md: Likewise.
4028 * config/arm/cortex-a53.md: Likewise.
4029 * config/arm/cortex-a57.md: Likewise.
4030 * config/arm/cortex-a7.md: Likewise.
4031 * config/arm/cortex-a8.md: Likewise.
4032 * config/arm/cortex-a9.md: Likewise.
4033 * config/arm/cortex-m4.md: Likewise.
4034 * config/arm/cortex-m7.md: Likewise.
4035 * config/arm/cortex-r4.md: Likewise.
4036 * config/arm/exynos-m1.md: Likewise.
4037 * config/arm/fa526.md: Likewise.
4038 * config/arm/fa606te.md: Likewise.
4039 * config/arm/fa626te.md: Likewise.
4040 * config/arm/fa726te.md: Likewise.
4041 * config/arm/fmp626.md: Likewise.
4042 * config/arm/iwmmxt.md: Likewise.
4043 * config/arm/ldmstm.md: Likewise.
4044 * config/arm/marvell-pj4.md: Likewise.
4045 * config/arm/xgene1.md: Likewise.
4046 * config/aarch64/thunderx.md: Likewise.
4047 * config/aarch64/thunderx2t99.md: Likewise.
4048 * config/aarch64/falkor.md: Likewise.
4050 2017-09-12 Martin Liska <mliska@suse.cz>
4052 * attribs.c (private_lookup_attribute): New function.
4053 * attribs.h (private_lookup_attribute): Declared here.
4054 (lookup_attribute): Called from this place.
4056 2017-09-12 Richard Biener <rguenther@suse.de>
4058 PR tree-optimization/82157
4059 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
4060 stmts with side-effects.
4062 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4063 Alan Hayward <alan.hayward@arm.com>
4064 David Sherwood <david.sherwood@arm.com>
4066 * target.def (hard_regno_nregs): New hook.
4067 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
4068 * targhooks.h (default_hard_regno_nregs): Declare.
4069 * targhooks.c (default_hard_regno_nregs): New function.
4070 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
4071 (TARGET_HARD_REGNO_NREGS): ...this hook.
4072 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
4073 (CLASS_MAX_NREGS): Likewise.
4074 * doc/tm.texi: Regenerate.
4075 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
4076 instead of HARD_REGNO_NREGS.
4077 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
4078 HARD_REGNO_NREGS in the comment.
4079 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
4080 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
4081 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
4082 Return an unsigned int.
4083 (TARGET_HARD_REGNO_NREGS): Redefine.
4084 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
4085 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
4086 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4087 (arc_hard_regno_nregs): New function.
4088 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
4089 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
4090 (arm_hard_regno_nregs): New function.
4091 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
4092 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
4093 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
4094 (TARGET_HARD_REGNO_NREGS): Redefine.
4095 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
4096 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
4097 (HARD_REGNO_NREGS): Delete.
4098 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
4099 (cr16_hard_regno_nregs): New function.
4100 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
4101 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
4102 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
4103 (cris_hard_regno_nregs): New function.
4104 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
4105 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
4106 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4107 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
4108 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
4109 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
4110 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
4111 (frv_hard_regno_nregs): Make static. Take and return an
4113 (frv_class_max_nregs): Remove outdated copy of documentation.
4114 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
4115 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
4116 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
4117 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
4118 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
4119 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
4120 (TARGET_HARD_REGNO_NREGS): Redefine.
4121 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
4122 (CLASS_MAX_NREGS): Update comment.
4123 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
4124 (ia64_hard_regno_nregs): New function.
4125 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
4126 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
4127 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
4128 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
4129 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
4131 (m32c_hard_regno_nregs): Likewise. Make static.
4132 (TARGET_HARD_REGNO_NREGS): Redefine.
4133 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
4134 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
4135 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
4136 (m68k_hard_regno_nregs): New function.
4137 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
4138 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
4139 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
4140 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
4141 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
4142 Take and return an unsigned int.
4143 (TARGET_HARD_REGNO_NREGS): Redefine.
4144 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
4145 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
4146 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
4147 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
4148 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
4149 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
4150 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
4151 (msp430_hard_regno_nregs): Make static. Take and return an
4153 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
4154 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
4155 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
4156 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
4157 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
4158 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
4159 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
4160 (TARGET_HARD_REGNO_NREGS): Redefine.
4161 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
4162 (PA_HARD_REGNO_NREGS): ...this.
4163 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
4164 (PA_HARD_REGNO_NREGS): ...this.
4165 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4166 (pa_hard_regno_nregs): New function.
4167 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
4168 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
4169 (pdp11_hard_regno_nregs): New function.
4170 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
4171 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
4172 (rs6000_hard_regno_nregs_hook): New function.
4173 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
4174 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
4175 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
4176 Take and return an unsigned int. Move earlier in file.
4177 (TARGET_HARD_REGNO_NREGS): Redefine.
4178 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
4179 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
4180 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
4181 (rl78_hard_regno_nregs): Make static. Take and return an
4183 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
4184 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
4185 (rs6000_hard_regno_nregs_hook): New function.
4186 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
4187 * config/rx/rx.c (rx_hard_regno_nregs): New function.
4188 (TARGET_HARD_REGNO_NREGS): Redefine.
4189 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
4190 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
4191 instead of HARD_REGNO_NREGS.
4192 (s390_hard_regno_nregs): New function.
4193 (s390_hard_regno_mode_ok): Add comment from s390.h.
4194 (TARGET_HARD_REGNO_NREGS): Redefine.
4195 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
4196 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
4197 (sh_hard_regno_nregs): New function.
4198 (sh_pass_in_reg_p): Use it.
4199 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
4200 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
4201 (sparc_hard_regno_nregs): New function.
4202 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
4203 * config/spu/spu.c (spu_hard_regno_nregs): New function.
4204 (spu_function_arg_advance): Use it, supplying a valid register number.
4205 (TARGET_HARD_REGNO_NREGS): Redefine.
4206 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
4207 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
4208 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
4209 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
4210 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
4211 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
4212 (CLASS_MAX_NREGS): Remove copy of old documentation.
4213 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
4214 (visium_hard_regno_nregs): New function.
4215 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
4216 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
4217 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
4218 xtensa_hard_regno_nregs): New function.
4219 * system.h (HARD_REGNO_NREGS): Poison.
4221 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4223 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
4224 hard_regno_nregs instead of HARD_REGNO_NREGS.
4225 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
4226 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
4227 (c6x_expand_epilogue): Likewise.
4228 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
4229 (frv_read_iacc_argument): Likewise.
4230 * config/sh/sh.c: Include regs.h.
4231 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4232 (regs_used): Likewise.
4233 (output_stack_adjust): Likewise.
4234 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
4235 * expmed.c: Include regs.h.
4236 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
4237 * ree.c: Include regs.h.
4238 (combine_reaching_defs): Use hard_regno_nregs instead of
4240 (add_removable_extension): Likewise.
4242 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4244 * regs.h (hard_regno_nregs): Turn into a function.
4245 (end_hard_regno): Update accordingly.
4246 * caller-save.c (setup_save_areas): Likewise.
4247 (save_call_clobbered_regs): Likewise.
4248 (replace_reg_with_saved_mem): Likewise.
4249 (insert_restore): Likewise.
4250 (insert_save): Likewise.
4251 * combine.c (can_change_dest_mode): Likewise.
4252 (move_deaths): Likewise.
4253 (distribute_notes): Likewise.
4254 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
4255 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
4256 (rs6000_split_multireg_move): Likewise.
4257 (rs6000_register_move_cost): Likewise.
4258 (rs6000_memory_move_cost): Likewise.
4259 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
4260 (rs6000_split_multireg_move): Likewise.
4261 (rs6000_register_move_cost): Likewise.
4262 (rs6000_memory_move_cost): Likewise.
4263 * cselib.c (cselib_reset_table): Likewise.
4264 (cselib_lookup_1): Likewise.
4265 * emit-rtl.c (set_mode_and_regno): Likewise.
4266 * function.c (aggregate_value_p): Likewise.
4267 * ira-color.c (setup_profitable_hard_regs): Likewise.
4268 (check_hard_reg_p): Likewise.
4269 (calculate_saved_nregs): Likewise.
4270 (assign_hard_reg): Likewise.
4271 (improve_allocation): Likewise.
4272 (calculate_spill_cost): Likewise.
4273 * ira-emit.c (modify_move_list): Likewise.
4274 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
4275 (ira_hard_reg_in_set_p): Likewise.
4276 * ira.c (setup_reg_mode_hard_regset): Likewise.
4277 (clarify_prohibited_class_mode_regs): Likewise.
4278 (check_allocation): Likewise.
4279 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4280 (lra_setup_reg_renumber): Likewise.
4281 (setup_try_hard_regno_pseudos): Likewise.
4282 (spill_for): Likewise.
4283 (assign_hard_regno): Likewise.
4284 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
4285 * lra-constraints.c (in_class_p): Likewise.
4286 (lra_constraint_offset): Likewise.
4287 (simplify_operand_subreg): Likewise.
4288 (lra_constraints): Likewise.
4289 (split_reg): Likewise.
4290 (split_if_necessary): Likewise.
4291 (invariant_p): Likewise.
4292 (inherit_in_ebb): Likewise.
4293 * lra-lives.c (process_bb_lives): Likewise.
4294 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
4295 (get_hard_regs): Likewise.
4296 (do_remat): Likewise.
4297 * lra-spills.c (assign_spill_hard_regs): Likewise.
4298 * mode-switching.c (create_pre_exit): Likewise.
4299 * postreload.c (reload_combine_recognize_pattern): Likewise.
4300 * recog.c (peep2_find_free_register): Likewise.
4301 * regcprop.c (kill_value_regno): Likewise.
4302 (set_value_regno): Likewise.
4303 (copy_value): Likewise.
4304 (maybe_mode_change): Likewise.
4305 (find_oldest_value_reg): Likewise.
4306 (copyprop_hardreg_forward_1): Likewise.
4307 * regrename.c (check_new_reg_p): Likewise.
4308 (regrename_do_replace): Likewise.
4309 * reload.c (push_reload): Likewise.
4310 (combine_reloads): Likewise.
4311 (find_dummy_reload): Likewise.
4312 (operands_match_p): Likewise.
4313 (find_reloads): Likewise.
4314 (find_equiv_reg): Likewise.
4315 (reload_adjust_reg_for_mode): Likewise.
4316 * reload1.c (count_pseudo): Likewise.
4317 (count_spilled_pseudo): Likewise.
4318 (find_reg): Likewise.
4319 (clear_reload_reg_in_use): Likewise.
4320 (free_for_value_p): Likewise.
4321 (allocate_reload_reg): Likewise.
4322 (choose_reload_regs): Likewise.
4323 (reload_adjust_reg_for_temp): Likewise.
4324 (emit_reload_insns): Likewise.
4325 (delete_output_reload): Likewise.
4326 * rtlanal.c (subreg_get_info): Likewise.
4327 * sched-deps.c (sched_analyze_reg): Likewise.
4328 * sel-sched.c (init_regs_for_mode): Likewise.
4329 (mark_unavailable_hard_regs): Likewise.
4330 (choose_best_reg_1): Likewise.
4331 (verify_target_availability): Likewise.
4332 * valtrack.c (dead_debug_insert_temp): Likewise.
4333 * var-tracking.c (track_loc_p): Likewise.
4334 (emit_note_insn_var_location): Likewise.
4335 * varasm.c (make_decl_rtl): Likewise.
4336 * reginfo.c (choose_hard_reg_mode): Likewise.
4337 (init_reg_modes_target): Refer directly to
4338 this_target_regs->x_hard_regno_nregs.
4340 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4342 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
4343 instead of hard_regno_nregs.
4345 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4347 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
4348 end_hard_regno instead of hard_regno_nregs.
4349 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
4350 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
4351 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
4352 * ira-color.c (improve_allocation): Likewise.
4353 * lra-assigns.c (find_hard_regno_for_1): Likewise.
4354 * lra-lives.c (mark_regno_live): Likewise.
4355 (mark_regno_dead): Likewise.
4356 * lra-remat.c (operand_to_remat): Likewise.
4357 * lra.c (collect_non_operand_hard_regs): Likewise.
4358 * postreload.c (reload_combine_note_store): Likewise.
4359 (move2add_valid_value_p): Likewise.
4360 * reload.c (regno_clobbered_p): Likewise.
4362 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4364 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
4366 * config/v850/v850.c (v850_reorg): Likewise.
4367 * reload.c (refers_to_regno_for_reload_p): Likewise.
4368 (find_equiv_reg): Likewise.
4369 * reload1.c (reload_reg_reaches_end_p): Likewise.
4371 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4373 * caller-save.c (add_used_regs): Use REG_NREGS instead of
4375 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
4376 * config/arm/arm.c (output_move_neon): Likewise.
4377 (arm_attr_length_move_neon): Likewise.
4378 (neon_split_vcombine): Likewise.
4379 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
4380 (c6x_mark_reg_written): Likewise.
4381 (c6x_dwarf_register_span): Likewise.
4382 * config/i386/i386.c (ix86_save_reg): Likewise.
4383 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
4384 (rws_access_reg): Likewise.
4385 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
4386 * mode-switching.c (create_pre_exit): Likewise.
4387 * ree.c (combine_reaching_defs): Likewise.
4388 (add_removable_extension): Likewise.
4389 * regcprop.c (find_oldest_value_reg): Likewise.
4390 (copyprop_hardreg_forward_1): Likewise.
4391 * reload.c (reload_inner_reg_of_subreg): Likewise.
4392 (push_reload): Likewise.
4393 (combine_reloads): Likewise.
4394 (find_dummy_reload): Likewise.
4395 (reload_adjust_reg_for_mode): Likewise.
4396 * reload1.c (find_reload_regs): Likewise.
4397 (forget_old_reloads_1): Likewise.
4398 (reload_reg_free_for_value_p): Likewise.
4399 (reload_adjust_reg_for_temp): Likewise.
4400 (emit_reload_insns): Likewise.
4401 (delete_output_reload): Likewise.
4402 * sel-sched.c (choose_best_reg_1): Likewise.
4403 (choose_best_pseudo_reg): Likewise.
4405 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4406 Alan Hayward <alan.hayward@arm.com>
4407 David Sherwood <david.sherwood@arm.com>
4409 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
4410 * target.def (slow_unaligned_access): New hook.
4411 * targhooks.h (default_slow_unaligned_access): Declare.
4412 * targhooks.c (default_slow_unaligned_access): New function.
4413 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
4414 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
4415 * doc/tm.texi: Regenerate.
4416 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
4417 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
4418 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
4420 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
4421 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
4423 (rs6000_slow_unaligned_access): New function.
4424 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4425 (expand_block_compare): Likewise.
4426 (expand_strn_compare): Likewise.
4427 (rs6000_rtx_costs): Likewise.
4428 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
4429 (riscv_slow_unaligned_access): Likewise.
4430 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
4431 (riscv_slow_unaligned_access_p): ...this and make static.
4432 (riscv_option_override): Update accordingly.
4433 (riscv_slow_unaligned_access): New function.
4434 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4435 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
4436 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
4437 (rs6000_slow_unaligned_access): New function.
4438 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
4439 (rs6000_rtx_costs): Likewise.
4440 * config/rs6000/rs6000-string.c (expand_block_compare)
4441 (expand_strn_compare): Use targetm.slow_unaligned_access instead
4442 of SLOW_UNALIGNED_ACCESS.
4443 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
4444 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
4445 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
4446 of SLOW_UNALIGNED_ACCESS.
4447 * expmed.c (simple_mem_bitfield_p): Likewise.
4448 * expr.c (alignment_for_piecewise_move): Likewise.
4449 (emit_group_load_1): Likewise.
4450 (emit_group_store): Likewise.
4451 (copy_blkmode_from_reg): Likewise.
4452 (emit_push_insn): Likewise.
4453 (expand_assignment): Likewise.
4454 (store_field): Likewise.
4455 (expand_expr_real_1): Likewise.
4456 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
4457 * lra-constraints.c (simplify_operand_subreg): Likewise.
4458 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
4459 * gimple-ssa-store-merging.c: Likewise in block comment at start
4461 * tree-ssa-strlen.c: Include target.h.
4462 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
4463 of SLOW_UNALIGNED_ACCESS.
4464 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
4466 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
4468 PR rtl-optimization/82185
4469 * expmed.c (emit_store_flag_int): Only test tem if it has been
4472 2017-09-12 Richard Biener <rguenther@suse.de>
4475 * match.pd ((FTYPE) N CMP CST): Fix typo.
4477 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
4479 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
4481 (mips_near_type_p): Add 'short_call' attribute as a synonym
4483 * doc/extend.texi (short_call): Document new function attribute.
4485 2017-09-12 Jakub Jelinek <jakub@redhat.com>
4488 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
4489 assertion check that in the condition.
4490 (get_atomic_generic_size): Likewise. Before testing if parameter
4491 has pointer type, if it has array type, call for C++
4492 default_conversion to perform array-to-pointer conversion.
4494 2017-09-12 Richard Biener <rguenther@suse.de>
4496 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
4497 for operations we cannot scalarize.
4499 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
4501 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
4502 vectors heap vectors. Clean up comments.
4503 Make visited_bbs a reference.
4504 (profitable_jump_thread_path): Make GC
4505 vectors heap vectors. Clean up comments.
4507 (convert_and_register_jump_thread_path): Make GC vectors heap
4509 (check_subpath_and_update_thread_path): Same. Clean up comments.
4510 Make visited_bbs a reference.
4511 (handle_phi): Abstract common code to to
4512 register_jump_thread_path_if_profitable.
4513 Rename VAR_BB to DEF_BB.
4515 Make GC vectors heap vectors.
4516 Make visited_bbs a reference.
4517 (handle_assignment): Same.
4518 (register_jump_thread_path_if_profitable): New.
4519 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
4521 Make GC vectors heap vectors. Clean up comments.
4522 Make visited_bbs a reference.
4523 (find_jump_threads_backwards): Make visited_bbs live in the stack.
4524 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
4527 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
4530 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
4531 words of E_DImode object are reachable by xtensa_uimm8x4 access.
4533 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
4535 Revert r251800 and r251799.
4537 2017-09-11 Martin Jambor <mjambor@suse.cz>
4540 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
4541 arguments in advance.
4542 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
4543 use it to find predecessor edges.
4544 (naive_outof_ssa): Collect vector of predecessors.
4546 2017-09-08 Jason Merrill <jason@redhat.com>
4548 PR c++/70029 - ICE with ref-qualifier and -flto
4549 * langhooks.h (struct lang_hooks_for_types): Add
4550 copy_lang_qualifiers.
4551 * attribs.c (build_type_attribute_qual_variant): Use it.
4552 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
4554 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
4555 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
4557 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
4560 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
4561 (*mulsi3_sp64): New instruction.
4562 (mulsi3): New expander.
4564 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
4566 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
4568 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4570 * sancov.c: Include memmodel.h.
4572 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
4575 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
4578 2017-09-07 Carl Love <cel@us.ibm.com>
4580 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
4581 the sldi instruction.
4583 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
4585 * sancov.c: Include tm_p.h.
4587 2017-09-07 Jakub Jelinek <jakub@redhat.com>
4590 * output.h (switch_to_other_text_partition): New declaration.
4591 * varasm.c (switch_to_other_text_partition): New function.
4592 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
4593 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
4594 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
4595 to the other text partition before emitting LCL label and switch back
4596 after emitting the word after it.
4598 2017-09-07 Richard Biener <rguenther@suse.de>
4600 * passes.def (pass_split_crit_edges): Remove instance before PRE.
4601 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
4602 critical edges here, after loop init.
4603 (pass_data_pre): Remove PROP_no_crit_edges flags.
4604 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
4605 for valueization of call args to avoid leaking VN_TOP.
4606 (visit_use): Assert we do not visit default defs.
4607 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
4608 Use error_mark_node to more easily detect leaking VN_TOP.
4609 All default-defs are varying, not VN_TOP. Mark them visited.
4610 (run_scc_vn): Make code match comment.
4612 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
4614 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
4615 OPTION_MASK_FLOAT128_KEYWORD.
4616 (POWERPC_MASKS): Likewise.
4617 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
4618 support for the -mfloat128-type option, and make -mfloat128
4619 default on PowerPC Linux systems. Define or undefine
4620 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
4621 Define __float128 to be __ieee128 if IEEE 128-bit support is
4622 enabled, or undefine it.
4623 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
4624 Delete defining __FLOAT128_TYPE__.
4625 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
4626 -mfloat128-type option and make -mfloat128 default on PowerPC
4628 (TARGET_FLOAT128_TYPE): Likewise.
4629 (-mfloat128-type): Likewise.
4630 * config/rs6000/rs6000.c (rs6000_option_override_internal):
4631 Delete the -mfloat128-type option and make -mfloat128 default on
4632 PowerPC Linux systems. Always use __ieee128 to be the keyword for
4633 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
4634 128-bit floating point is enabled. Change tests from using
4635 -mfloat128-type to -mfloat128.
4636 (rs6000_mangle_type): Use the correct mangling for the __float128
4637 type even if normal long double is restricted to 64-bits.
4638 (floatn_mode): Enable the _Float128 type by default on VSX Linux
4640 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
4641 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
4642 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
4644 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
4645 documentation for -mfloat128.
4647 2017-09-06 Olivier Hainque <hainque@adacore.com>
4649 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
4651 2017-09-06 Wish Wu <wishwu007@gmail.com>
4652 Jakub Jelinek <jakub@redhat.com>
4654 * asan.c (initialize_sanitizer_builtins): Add
4655 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
4656 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
4657 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
4658 BT_FN_VOID_UINT64_PTR variables.
4659 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
4660 (BT_FN_VOID_UINT16_UINT16): Likewise.
4661 (BT_FN_VOID_UINT32_UINT32): Likewise.
4662 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
4663 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
4664 (BT_FN_VOID_UINT64_PTR): Likewise.
4665 * common.opt (flag_sanitize_coverage): New variable.
4666 (fsanitize-coverage=trace-pc): Remove.
4667 (fsanitize-coverage=): Add.
4668 * flag-types.h (enum sanitize_coverage_code): New enum.
4669 * fold-const.c (fold_range_test): Disable non-short-circuit
4670 optimization if flag_sanitize_coverage.
4671 (fold_truth_andor): Likewise.
4672 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
4673 * opts.c (COVERAGE_SANITIZER_OPT): Define.
4674 (coverage_sanitizer_opts): New array.
4675 (get_closest_sanitizer_option): Add OPTS argument, handle also
4676 OPT_fsanitize_coverage_.
4677 (parse_sanitizer_options): Adjusted to also handle
4678 OPT_fsanitize_coverage_.
4679 (common_handle_option): Add OPT_fsanitize_coverage_.
4680 * sancov.c (instrument_comparison, instrument_switch): New function.
4681 (sancov_pass): Add trace-cmp support.
4682 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
4683 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
4684 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
4685 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
4686 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
4687 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
4688 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
4689 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
4690 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
4691 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
4693 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4695 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
4696 error. Only quit immediately if parsing is complete.
4697 (BEGIN): Initialize fatal_err and parse_done.
4698 (begin fpu, end fpu): Check number of arguments.
4699 (begin arch, end arch): Likewise.
4700 (begin cpu, end cpu): Likewise.
4701 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
4702 (optalias): Likewise.
4704 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
4706 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
4707 * config/arm/arm-isa.h: Delete. Move definitions to ...
4708 * arm-cpus.in: ... here. Use new feature and fgroup values.
4709 * config/arm/arm.c (arm_option_override): Use lower case for feature
4711 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
4712 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
4713 * config/arm/parsecpu.awk (END): Add new command 'isa'.
4715 (print_isa_bits_for): New function.
4716 (gen_isa): New function.
4717 (gen_comm_data): Use print_isa_bits_for.
4718 (define feature): New keyword.
4719 (define fgroup): New keyword.
4720 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
4721 (arm-isa.h): Add rule to generate file.
4722 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
4723 case for feature bit names.
4725 2017-09-06 Richard Biener <rguenther@suse.de>
4727 * tree-ssa-pre.c (NECESSARY): Remove.
4728 (create_expression_by_pieces): Do not touch pass-local flags.
4729 (insert_into_preds_of_block): Likewise.
4730 (do_pre_regular_insertion): Likewise.
4731 (eliminate_insert): Likewise.
4732 (eliminate_dom_walker::before_dom_children): Likewise.
4733 (fini_eliminate): Do not look at inserted_exprs.
4734 (mark_operand_necessary): Remove.
4735 (remove_dead_inserted_code): Replace with simple work-list
4736 algorithm based on inserted_exprs and SSA uses.
4737 (pass_pre::execute): Re-order fini_eliminate and
4738 remove_dead_inserted_code.
4740 2017-09-06 Olivier Hainque <hainque@adacore.com>
4742 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
4743 for VxWorks 7. Adjust surrounding comments.
4745 2017-09-06 Richard Biener <rguenther@suse.de>
4747 * gimple-ssa-strength-reduction.c
4748 (find_candidates_dom_walker::before_dom_children): Also allow
4751 2017-09-06 Richard Biener <rguenther@suse.de>
4753 PR tree-optimization/82108
4754 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
4755 for gap in the non-permutation SLP case.
4757 2017-09-06 Martin Jambor <mjambor@suse.cz>
4759 PR tree-optimization/82078
4760 * tree-sra.c (sort_and_splice_var_accesses): Move call to
4761 add_access_to_work_queue...
4762 (build_accesses_from_assign): ...here.
4763 (propagate_all_subaccesses): Make sure racc is the group
4764 representative, if there is one.
4766 2017-09-06 Jakub Jelinek <jakub@redhat.com>
4769 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
4772 2017-09-06 Richard Biener <rguenther@suse.de>
4774 * gimple-ssa-strength-reduction.c
4775 (find_candidates_dom_walker::before_doom_children): Use a
4776 type and not a mode check.
4778 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
4781 * config/arm/predicates.md (arm_general_adddi_operand): Create new
4783 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
4785 2017-09-05 Jeff Law <law@redhat.com>
4787 PR tree-optimization/64910
4788 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
4789 cases where we have 3 or more operands.
4791 2017-09-05 Jakub Jelinek <jakub@redhat.com>
4794 * omp-low.c (lower_omp_for): Recompute tree invariant if
4795 gimple_omp_for_initial/final is ADDR_EXPR.
4798 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
4799 into gimple val before gimplification fo the COND_EXPR.
4801 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
4803 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
4804 REGION_COPY argument.
4805 (thread_through_all_blocks): Remove unused argument to
4806 duplicate_thread_path.
4808 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4809 Alan Hayward <alan.hayward@arm.com>
4810 David Sherwood <david.sherwood@arm.com>
4812 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
4813 Take a scalar_mode rather than a machine_mode.
4814 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4815 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
4816 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
4817 (aarch64_gen_adjusted_ldpstp): Likewise.
4818 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
4820 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4821 Alan Hayward <alan.hayward@arm.com>
4822 David Sherwood <david.sherwood@arm.com>
4824 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
4825 Take a scalar_int_mode instead of a machine_mode.
4826 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4827 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4828 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4829 (aarch64_simd_attr_length_rglist): Delete.
4830 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
4831 a scalar_int_mode instead of a machine_mode.
4832 (aarch64_add_offset): Likewise.
4833 (aarch64_internal_mov_immediate): Likewise
4834 (aarch64_add_constant_internal): Likewise.
4835 (aarch64_add_constant): Likewise.
4836 (aarch64_movw_imm): Likewise.
4837 (aarch64_rtx_arith_op_extract_p): Likewise.
4838 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
4839 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
4840 Remove assert that the mode isn't a vector.
4841 (aarch64_output_scalar_simd_mov_immediate): Likewise.
4842 (aarch64_expand_mov_immediate): Update calls after above changes.
4843 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
4844 (aarch64_and_bitmask_imm): Check for scalar integer modes.
4845 (aarch64_move_imm): Likewise.
4846 (aarch64_can_const_movi_rtx_p): Likewise.
4847 (aarch64_strip_extend): Likewise.
4848 (aarch64_extr_rtx_p): Likewise.
4849 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
4850 a CONST_INT when the mode parameter is VOIDmode.
4851 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
4853 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4855 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
4856 * stor-layout.c (bitwise_mode_for_mode): Likewise.
4857 (bitwise_type_for_mode): Update accordingly.
4859 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4861 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
4862 * stor-layout.c (mode_for_size_tree): Likewise.
4863 (mode_for_array): Update accordingly.
4864 (layout_decl): Likewise.
4865 (compute_record_mode): Likewise. Only set the mode once.
4867 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4869 * target.def (get_mask_mode): Change return type to opt_mode.
4871 * doc/tm.texi: Regenerate.
4872 * targhooks.h (default_get_mask_mode): Return an opt_mode.
4873 * targhooks.c (default_get_mask_mode): Likewise.
4874 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
4875 * optabs-query.c (can_vec_mask_load_store_p): Update use of
4876 targetm.get_mask_mode.
4877 * tree.c (build_truth_vector_type): Likewise.
4879 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4881 * machmode.h (mode_for_vector): Return an opt_mode.
4882 * stor-layout.c (mode_for_vector): Likewise.
4883 (mode_for_int_vector): Update accordingly.
4884 (layout_type): Likewise.
4885 * config/i386/i386.c (emit_memmov): Likewise.
4886 (ix86_expand_set_or_movmem): Likewise.
4887 (ix86_expand_vector_init): Likewise.
4888 (ix86_get_mask_mode): Likewise.
4889 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
4891 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
4892 * expmed.c (extract_bit_field_1): Likewise.
4893 * expr.c (expand_expr_real_2): Likewise.
4894 * optabs-query.c (can_vec_perm_p): Likewise.
4895 (can_vec_mask_load_store_p): Likewise.
4896 * optabs.c (expand_vec_perm): Likewise.
4897 * targhooks.c (default_get_mask_mode): Likewise.
4898 * tree-vect-stmts.c (vectorizable_store): Likewise.
4899 (vectorizable_load): Likewise.
4900 (get_vectype_for_scalar_type_and_size): Likewise.
4902 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4904 * machmode.h (mode_for_int_vector): New function.
4905 * stor-layout.c (mode_for_int_vector): Likewise.
4906 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
4907 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
4908 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
4909 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
4910 (s390_expand_vcond): Likewise.
4912 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4914 * machmode.h (opt_machine_mode): New type.
4915 (opt_mode<T>): Allow construction from anything that can be
4917 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
4918 (mode_for_size): Return an opt_machine_mode.
4919 * stor-layout.c (mode_for_size): Likewise.
4920 (mode_for_size_tree): Update call accordingly.
4921 (bitwise_mode_for_mode): Likewise.
4922 (make_fract_type): Likewise.
4923 (make_accum_type): Likewise.
4924 * caller-save.c (replace_reg_with_saved_mem): Update call
4926 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4927 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4928 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4929 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
4930 * expmed.c (extract_bit_field_1): Likewise.
4931 * reload.c (get_secondary_mem): Likewise.
4932 * varasm.c (assemble_integer): Likewise.
4933 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
4936 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4938 * machmode.h (decimal_float_mode_for_size): New function.
4939 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
4940 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
4941 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
4942 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
4943 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
4944 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
4946 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4948 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
4949 (get_builtin_sync_mode): Likewise.
4950 (expand_ifn_atomic_compare_exchange): Likewise.
4951 (expand_builtin_atomic_clear): Likewise.
4952 (expand_builtin_atomic_test_and_set): Likewise.
4953 (fold_builtin_atomic_always_lock_free): Likewise.
4954 * calls.c (compute_argument_addresses): Likewise.
4955 (emit_library_call_value_1): Likewise.
4956 (store_one_arg): Likewise.
4957 * combine.c (combine_instructions): Likewise.
4958 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
4959 * config/arm/arm.c (arm_function_value): Likewise.
4960 (aapcs_allocate_return_reg): Likewise.
4961 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
4962 * config/i386/i386.c (construct_container): Likewise.
4963 (ix86_gimplify_va_arg): Likewise.
4964 (ix86_expand_sse_cmp): Likewise.
4965 (emit_memmov): Likewise.
4966 (emit_memset): Likewise.
4967 (expand_small_movmem_or_setmem): Likewise.
4968 (ix86_expand_pextr): Likewise.
4969 (ix86_expand_pinsr): Likewise.
4970 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
4971 * config/microblaze/microblaze.c (microblaze_block_move_straight):
4973 * config/mips/mips.c (mips_function_value_1) Likewise.
4974 (mips_block_move_straight): Likewise.
4975 (mips_expand_ins_as_unaligned_store): Likewise.
4976 * config/powerpcspe/powerpcspe.c
4977 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4978 (rs6000_darwin64_record_arg_flush): Likewise.
4979 * config/rs6000/rs6000.c
4980 (rs6000_darwin64_record_arg_advance_flush): Likewise.
4981 (rs6000_darwin64_record_arg_flush): Likewise.
4982 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
4983 (sparc_function_value_1): Likewise.
4984 * config/spu/spu.c (adjust_operand): Likewise.
4985 (spu_emit_branch_or_set): Likewise.
4986 (arith_immediate_p): Likewise.
4987 * emit-rtl.c (gen_lowpart_common): Likewise.
4988 * expr.c (expand_expr_real_1): Likewise.
4989 * function.c (assign_parm_setup_block): Likewise.
4990 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
4991 * reload1.c (alter_reg): Likewise.
4992 * stor-layout.c (mode_for_vector): Likewise.
4993 (layout_type): Likewise.
4995 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
4997 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
4998 (spu_convert_move): Likewise.
4999 * lower-subreg.c (resolve_simple_move): Likewise.
5001 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5004 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
5005 define_insn to a define_expand.
5006 (altivec_vsum2sws_direct): New define_insn.
5007 (altivec_vsumsws): Convert from a define_insn to a define_expand.
5009 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5011 * config/arm/arm.c (arm_option_params_internal): Improve setting of
5014 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
5019 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
5020 realign stack if stack alignment needed is less than incoming
5023 2017-09-05 Marek Polacek <polacek@redhat.com>
5026 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
5029 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
5031 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
5033 2017-09-05 Richard Biener <rguenther@suse.de>
5035 PR tree-optimization/82084
5036 * fold-const.c (can_native_encode_string_p): Handle wide characters.
5038 2017-09-05 Richard Biener <rguenther@suse.de>
5040 PR tree-optimization/82102
5041 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
5043 2017-09-05 Martin Liska <mliska@suse.cz>
5045 PR tree-optimization/82032
5046 * tree-cfg.c (generate_range_test): New function.
5047 * tree-cfg.h (generate_range_test): Declared here.
5048 * tree-cfgcleanup.c (convert_single_case_switch): New function.
5049 (cleanup_control_expr_graph): Use it.
5050 * tree-switch-conversion.c (try_switch_expansion): Remove
5052 (emit_case_nodes): Use generate_range_test.
5054 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5057 * config/i386/i386.md (*<btsc><mode>_mask): Add
5058 TARGET_USE_BT to insn constraint.
5059 (*btr<mode>_mask): Ditto.
5061 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
5063 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
5064 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
5066 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
5069 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
5070 TARGET_NEON and TARGET_IWMMXT.
5071 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
5072 TARGET_NEON and TARGET_IWMMXT.
5073 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
5075 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
5077 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
5078 (ix86_rewrite_tls_address): Ditto.
5079 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
5080 (ix86_rewrite_tls_address_1): Ditto.
5081 (ix86_rewrite_tls_address): Ditto.
5082 * config/i386/predicates.md (tls_address_pattern): New predicate.
5083 * config/i386/i386.md (TLS address splitter): New splitter.
5085 2017-09-04 Richard Biener <rguenther@suse.de>
5087 PR tree-optimization/82084
5088 * fold-const.h (can_native_encode_string_p): Declare.
5089 * fold-const.c (can_native_encode_string_p): Factor out from ...
5090 (native_encode_string): ... here.
5091 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
5092 vectorizing stores from constants we later cannot handle.
5094 2017-09-04 Marek Polacek <polacek@redhat.com>
5097 * doc/invoke.texi: Update -Wtautological-compare documentation.
5099 2017-09-04 Jeff Law <law@redhat.com>
5101 PR tree-optimization/64910
5102 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
5103 swap the first and last operand if the last is a constant.
5105 2017-09-04 Marek Polacek <polacek@redhat.com>
5108 * convert.c (do_narrow): When sanitizing signed integer overflows,
5109 bail out for signed types.
5110 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
5112 2017-09-04 Richard Biener <rguenther@suse.de>
5114 PR tree-optimization/82060
5115 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
5116 Move devirtualization after stmt folding and before EH/AB/noreturn
5117 cleanup to get the stmt refs canonicalized. Use a bool instead
5118 of gimple_modified_p since that doesn't work for NOPs. Schedule
5119 NOPs generated by folding for removal.
5121 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5122 Alan Hayward <alan.hayward@arm.com>
5123 David Sherwood <david.sherwood@arm.com>
5125 * coretypes.h (pad_direction): New enum.
5126 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
5127 (FUNCTION_ARG_PADDING): Likewise.
5128 * target.def (function_arg_padding): New hook.
5129 * targhooks.h (default_function_arg_padding): Declare.
5130 * targhooks.c (default_function_arg_padding): New function.
5131 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
5132 (TARGET_FUNCTION_ARG_PADDING): ...this.
5133 * doc/tm.texi: Regenerate.
5134 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
5135 instead of direction.
5136 (compute_argument_addresses): Likewise.
5137 (load_register_parameters): Likewise.
5138 (emit_library_call_value_1): Likewise.
5139 (store_one_arg): Use targetm.calls.function_arg_padding instead
5140 of FUNCTION_ARG_PADDING.
5141 (must_pass_in_stack_var_size_or_pad): Likewise.
5142 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
5143 (emit_group_store): Likewise.
5144 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
5145 instead of FUNCTION_ARG_PADDING.
5146 (emit_push_insn): Likewise, and propagate enum change throughout
5148 * function.h (direction): Delete.
5149 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
5151 * function.c (assign_parm_find_stack_rtl): Likewise.
5152 (assign_parm_setup_block_p): Likewise.
5153 (assign_parm_setup_block): Likewise.
5154 (gimplify_parameters): Likewise.
5155 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
5156 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
5158 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
5159 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5160 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
5161 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
5162 (aarch64_function_arg_padding): ...this new function.
5163 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
5164 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5165 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
5166 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5167 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
5168 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5169 (arm_pad_arg_upward): Replace with...
5170 (arm_function_arg_padding): ...this new function.
5171 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
5173 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
5174 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
5175 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5176 (ia64_hpux_function_arg_padding): Replace with...
5177 (ia64_function_arg_padding): ...this new function. Use pad_direction
5178 instead of direction. Check for TARGET_HPUX.
5179 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
5180 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5181 (iq2000_function_arg_padding): New function.
5182 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
5183 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
5184 (mips_function_arg_padding): ...this new function.
5185 (mips_pad_reg_upward): Update accordingly.
5186 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5187 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
5188 targetm.calls.function_arg_padding.
5189 (FUNCTION_ARG_PADDING): Delete.
5190 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
5191 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
5192 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5193 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
5194 (nios2_block_reg_padding): Return pad_direction instead of direction.
5195 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
5196 instead of direction.
5197 (nios2_function_arg_padding): Likewise. Make static.
5198 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5199 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
5200 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
5201 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
5202 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5203 (pa_function_arg_padding): Make static. Return pad_direction instead
5205 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
5206 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5207 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
5208 instead of direction. Use targetm.calls.function_arg_padding.
5209 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
5210 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5211 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
5212 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
5213 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
5215 (function_arg_padding): Rename to...
5216 (rs6000_function_arg_padding): ...this. Make static. Return
5217 pad_direction instead of direction.
5218 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5219 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
5220 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
5221 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
5222 instead of direction. Use targetm.calls.function_arg_padding.
5223 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
5224 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
5225 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
5226 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
5227 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5228 (function_arg_padding): Rename to...
5229 (rs6000_function_arg_padding): ...this. Make static. Return
5230 pad_direction instead of direction.
5231 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
5232 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
5233 * config/s390/s390.c (s390_function_arg_padding): New function.
5234 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5235 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
5236 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
5237 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
5238 (function_arg_padding): Rename to...
5239 (sparc_function_arg_padding): ...this. Make static. Return
5240 pad_direction instead of direction.
5241 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
5242 * config/spu/spu.c (spu_function_arg_padding): New function.
5243 (TARGET_FUNCTION_ARG_PADDING): Redefine.
5244 * system.h (FUNCTION_ARG_PADDING): Poison.
5246 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5247 Alan Hayward <alan.hayward@arm.com>
5248 David Sherwood <david.sherwood@arm.com>
5250 * target.def (modes_tieable_p): New hook.
5251 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
5252 (TARGET_MODES_TIEABLE_P): ...this.
5253 * doc/tm.texi.in: Regenerate.
5254 * hooks.h (hook_bool_mode_mode_true): Declare.
5255 * hooks.c (hook_bool_mode_mode_true): New function.
5256 * combine.c (subst): Use targetm.modes_tieable_p instead of
5258 * dse.c (find_shift_sequence): Likewise.
5259 * expmed.c (extract_low_bits): Likewise.
5260 * lower-subreg.c: Include target.h.
5261 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
5263 * rtlanal.c (rtx_cost): Likewise.
5264 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
5265 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
5266 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
5267 (TARGET_MODES_TIEABLE_P): Redefine.
5268 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
5269 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
5270 (TARGET_MODES_TIEABLE_P): Redefine.
5271 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
5272 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
5273 (arc_modes_tieable_p): New function.
5274 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
5275 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
5276 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
5277 (arm_modes_tieable_p): Make static.
5278 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
5279 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
5280 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
5281 (TARGET_MODES_TIEABLE_P): Redefine.
5282 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
5283 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
5284 (TARGET_MODES_TIEABLE_P): Redefine.
5285 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
5286 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
5287 (cr16_modes_tieable_p): New function.
5288 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
5289 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
5290 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
5291 (TRULY_NOOP_TRUNCATION): Update comment.
5292 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
5293 (TRULY_NOOP_TRUNCATION): Update comment.
5294 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
5295 (frv_modes_tieable_p): New function.
5296 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
5297 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
5298 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
5299 (TARGET_MODES_TIEABLE_P): Redefine.
5300 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
5301 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
5302 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
5303 (TARGET_MODES_TIEABLE_P): Redefine.
5304 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
5305 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
5306 (ia64_modes_tieable_p): New function.
5307 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
5308 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
5309 (iq2000_modes_tieable_p): New function.
5310 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
5311 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
5312 (lm32_modes_tieable_p): New function.
5313 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
5314 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
5315 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
5316 (TARGET_MODES_TIEABLE_P): Redefine.
5317 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
5318 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
5319 (m32r_modes_tieable_p): New function.
5320 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
5321 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
5322 (m68k_modes_tieable_p): New function.
5323 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
5324 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
5325 (mcore_modes_tieable_p): New function.
5326 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
5327 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
5329 (TARGET_MODES_TIEABLE_P): Redefine.
5330 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
5331 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
5332 * config/mips/mips.c (mips_modes_tieable_p): Make static.
5333 (TARGET_MODES_TIEABLE_P): Redefine.
5334 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
5335 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
5336 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
5337 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
5338 (mn10300_modes_tieable_p): ...this and make static.
5339 (TARGET_MODES_TIEABLE_P): Redefine.
5340 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
5341 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
5342 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
5343 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
5344 (msp430_modes_tieable_p): Make static.
5345 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
5346 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
5347 (TARGET_MODES_TIEABLE_P): Redefine.
5348 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
5349 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
5350 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
5351 (TARGET_MODES_TIEABLE_P): Redefine.
5352 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
5353 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
5354 * config/pa/pa.c (pa_modes_tieable_p): Make static.
5355 (TARGET_MODES_TIEABLE_P): Redefine.
5356 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
5357 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
5358 (pdp11_modes_tieable_p): New function.
5359 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
5360 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
5361 (rs6000_modes_tieable_p): New function.
5362 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5363 * config/powerpcspe/powerpcspe.md: Update comment.
5364 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
5365 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
5366 (TARGET_MODES_TIEABLE_P): Redefine.
5367 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
5368 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
5369 (rl78_modes_tieable_p): New function.
5370 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
5371 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
5372 (rs6000_modes_tieable_p): New function.
5373 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
5374 * config/rs6000/rs6000.md: Update comment.
5375 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
5376 * config/rx/rx.c (rx_modes_tieable_p): New function.
5377 (TARGET_MODES_TIEABLE_P): Redefine.
5378 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
5379 * config/s390/s390.c (s390_modes_tieable_p): New function.
5380 (TARGET_MODES_TIEABLE_P): Redefine.
5381 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
5382 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
5383 (sh_modes_tieable_p): New function.
5384 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
5385 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
5386 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
5387 (sparc_modes_tieable_p): Make static.
5388 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
5389 * config/spu/spu.c (spu_modes_tieable_p): New function.
5390 (TARGET_MODES_TIEABLE_P): Redefine.
5391 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
5392 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
5393 (TARGET_MODES_TIEABLE_P): Redefine.
5394 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
5395 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
5396 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
5397 * config/v850/v850.c (v850_modes_tieable_p): New function.
5398 (TARGET_MODES_TIEABLE_P): Redefine.
5399 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
5400 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
5401 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
5402 (visium_modes_tieable_p): New function.
5403 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
5404 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
5405 (xtensa_modes_tieable_p): New function.
5406 * system.h (MODES_TIEABLE_P): Poison.
5408 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5409 Alan Hayward <alan.hayward@arm.com>
5410 David Sherwood <david.sherwood@arm.com>
5412 * target.def (hard_regno_mode_ok): New hook.
5413 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
5414 (TARGET_HARD_REGNO_MODE_OK): ...this.
5415 * doc/tm.texi.in: Regenerate.
5416 * hooks.h (hook_bool_uint_mode_true): Declare.
5417 * hooks.c (hook_bool_uint_mode_true): New function.
5418 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
5420 * genpreds.c (write_insn_preds_c): Add an include of target.h.
5421 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
5422 instead of HARD_REGNO_MODE_OK.
5423 * caller-save.c: Include target.h.
5424 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
5426 * combine.c (can_combine_p): Likewise.
5427 (combinable_i3pat): Likewise.
5428 (can_change_dest_mode): Likewise.
5429 * expr.c (init_expr_target): Likewise.
5430 (convert_move): Likewise.
5431 (convert_modes): Likewise.
5432 * ira.c (setup_prohibited_class_mode_regs): Likewise.
5433 (setup_prohibited_mode_move_regs): Likewise.
5434 * ira.h (target_ira): Likewise.
5435 * lra-assigns.c (find_hard_regno_for_1): Likewise.
5436 * lra-constraints.c (process_alt_operands): Likewise.
5437 (split_reg): Likewise.
5438 * recog.c (peep2_find_free_register): Likewise.
5439 * ree.c (combine_reaching_defs): Likewise.
5440 * regcprop.c (maybe_mode_change): Likewise.
5441 * reginfo.c (init_reg_sets_1): Likewise.
5442 (choose_hard_reg_mode): Likewise.
5443 (simplifiable_subregs): Likewise.
5444 * regrename.c (check_new_reg_p): Likewise.
5445 * reload.c (find_valid_class): Likewise.
5446 (find_valid_class_1): Likewise.
5447 (reload_inner_reg_of_subreg): Likewise.
5448 (push_reload): Likewise.
5449 (combine_reloads): Likewise.
5450 (find_dummy_reload): Likewise.
5451 (find_reloads): Likewise.
5452 * reload1.c (find_reg): Likewise.
5453 (set_reload_reg): Likewise.
5454 (allocate_reload_reg): Likewise.
5455 (choose_reload_regs): Likewise.
5456 (reload_adjust_reg_for_temp): Likewise.
5457 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
5458 (simplify_subreg_regno): Likewise.
5459 * sel-sched.c (init_regs_for_mode): Likewise.
5460 * varasm.c (make_decl_rtl): Likewise.
5461 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
5462 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
5464 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
5465 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
5466 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5467 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
5468 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
5469 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5470 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
5471 (arc_mode_class): Delete.
5472 (HARD_REGNO_MODE_OK): Delete.
5473 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5474 (arc_hard_regno_mode_ok): Rename old array to...
5475 (arc_hard_regno_mode_ok_modes): ...this.
5476 (arc_conditional_register_usage): Update accordingly.
5477 (arc_mode_class): Make static.
5478 (arc_hard_regno_mode_ok): New function.
5479 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
5480 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
5481 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5482 (arm_hard_regno_mode_ok): Make static.
5483 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
5485 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
5486 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
5487 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
5489 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5490 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
5491 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
5492 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
5493 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
5494 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5495 * config/bfin/predicates.md (valid_reg_operand): Use
5496 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
5497 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
5498 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
5499 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5500 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
5501 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
5502 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5503 (cr16_hard_regno_mode_ok): Make static and return a bool.
5504 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
5505 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5506 (cris_hard_regno_mode_ok): New function.
5507 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
5508 (epiphany_mode_class): Delete.
5509 (HARD_REGNO_MODE_OK): Delete.
5510 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
5511 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5512 (hard_regno_mode_ok): Rename to...
5513 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
5514 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
5515 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
5517 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
5518 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
5519 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5520 (frv_hard_regno_mode_ok): Make static and return a bool.
5521 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
5523 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
5524 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
5525 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
5526 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
5528 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5529 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
5530 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
5531 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
5533 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5534 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
5535 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5536 (ia64_hard_regno_mode_ok): New function.
5537 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
5538 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5539 (iq2000_hard_regno_mode_ok): New function.
5540 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
5541 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5542 (lm32_hard_regno_mode_ok): New function.
5543 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
5544 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
5545 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
5546 instead of HARD_REGNO_MODE_OK.
5547 (m32c_hard_regno_ok): Rename to...
5548 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
5549 (m32c_cannot_change_mode_class): Update accordingly.
5550 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5551 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
5552 (m32r_mode_class): Delete.
5553 (HARD_REGNO_MODE_OK): Delete.
5554 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5555 (m32r_hard_regno_mode_ok): Rename to...
5556 (m32r_hard_regno_modes): ...this.
5557 (m32r_mode_class): Make static.
5558 (m32r_hard_regno_mode_ok): New function.
5559 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
5560 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
5561 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5562 (m68k_hard_regno_mode_ok): Make static.
5563 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
5564 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5565 (mcore_hard_regno_mode_ok): New function.
5566 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
5567 (HARD_REGNO_MODE_OK): Delete.
5568 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
5570 (microblaze_hard_regno_mode_ok_p): ...this and make static.
5571 (microblaze_hard_regno_mode_ok): New function.
5572 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5573 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
5574 (mips_hard_regno_mode_ok): Delete.
5575 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
5576 (mips_hard_regno_mode_ok_p): ...this and make static.
5577 (mips_hard_regno_mode_ok_p): Rename to...
5578 (mips_hard_regno_mode_ok_uncached): ...this.
5579 (mips_hard_regno_mode_ok): New function.
5580 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
5581 of HARD_REGNO_MODE_OK.
5582 (mips_option_override): Update after above name changes.
5583 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5584 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
5585 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
5586 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
5587 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
5588 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5589 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
5590 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
5591 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
5592 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5593 (msp430_hard_regno_mode_ok): Make static and return a bool.
5594 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
5595 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
5596 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
5598 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5599 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
5600 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
5601 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
5602 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
5603 (PA_HARD_REGNO_MODE_OK): ...this
5604 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
5605 (PA_HARD_REGNO_MODE_OK): ...this.
5606 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5607 (pa_hard_regno_mode_ok): New function.
5608 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
5609 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5610 (pdp11_hard_regno_mode_ok): New function.
5611 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
5612 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
5614 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
5616 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5617 (rs6000_hard_regno_mode_ok): Rename to...
5618 (rs6000_hard_regno_mode_ok_uncached): ...this.
5619 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5620 (rs6000_hard_regno_mode_ok): New function.
5621 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
5622 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
5623 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
5624 (riscv_hard_regno_mode_ok): ...this and make static.
5625 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5626 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
5627 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
5628 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5629 (rl78_hard_regno_mode_ok): Make static and return bool.
5630 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
5631 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
5633 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
5634 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5635 (rs6000_hard_regno_mode_ok): Rename to...
5636 (rs6000_hard_regno_mode_ok_uncached): ...this.
5637 (rs6000_init_hard_regno_mode_ok): Update accordingly.
5638 (rs6000_hard_regno_mode_ok): New function.
5639 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
5640 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
5641 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5642 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
5643 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
5644 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
5645 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5646 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
5647 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
5648 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5649 (sh_hard_regno_mode_ok): Make static.
5650 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
5651 instead of HARD_REGNO_MODE_OK.
5652 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
5653 (sparc_mode_class): Delete.
5654 (HARD_REGNO_MODE_OK): Delete.
5655 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5656 (hard_regno_mode_classes): Make static.
5657 (sparc_mode_class): Likewise.
5658 (sparc_hard_regno_mode_ok): New function.
5659 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
5660 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
5661 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
5663 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5664 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
5665 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
5666 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
5667 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
5668 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5669 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
5670 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
5671 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
5672 (visium_hard_regno_mode_ok): New function.
5673 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
5674 instead of HARD_REGNO_MODE_OK.
5675 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
5676 (HARD_REGNO_MODE_OK): Delete.
5677 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
5678 (xtensa_hard_regno_mode_ok_p): ...this and make static.
5679 (xtensa_option_override): Update accordingly.
5680 (TARGET_HARD_REGNO_MODE_OK): Redefine.
5681 (xtensa_hard_regno_mode_ok): New function.
5682 * system.h (HARD_REGNO_MODE_OK): Poison.
5684 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5685 Alan Hayward <alan.hayward@arm.com>
5686 David Sherwood <david.sherwood@arm.com>
5688 * target.def (hard_regno_call_part_clobbered): New hook.
5689 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
5690 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
5691 * doc/tm.texi: Regenerate.
5692 * hooks.h (hook_bool_uint_mode_false): Declare.
5693 * hooks.c (hook_bool_uint_mode_false): New function.
5694 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5695 * cselib.c (cselib_process_insn): Use
5696 targetm.hard_regno_call_part_clobbered instead of
5697 HARD_REGNO_CALL_PART_CLOBBERED.
5698 * ira-conflicts.c (ira_build_conflicts): Likewise.
5699 * ira-costs.c (ira_tune_allocno_costs): Likewise.
5700 * lra-constraints.c (need_for_call_save_p): Likewise.
5701 * lra-lives.c: Include target.h.
5702 (check_pseudos_live_through_calls): Use
5703 targetm.hard_regno_call_part_clobbered instead of
5704 HARD_REGNO_CALL_PART_CLOBBERED.
5705 * regcprop.c: Include target.h.
5706 (copyprop_hardreg_forward_1): Use
5707 targetm.hard_regno_call_part_clobbered instead of
5708 HARD_REGNO_CALL_PART_CLOBBERED.
5709 * reginfo.c (choose_hard_reg_mode): Likewise.
5710 * regrename.c (check_new_reg_p): Likewise.
5711 * reload.c (find_equiv_reg): Likewise.
5712 * reload1.c (emit_reload_insns): Likewise.
5713 * sched-deps.c (deps_analyze_insn): Likewise.
5714 * sel-sched.c (init_regs_for_mode): Likewise.
5715 (mark_unavailable_hard_regs): Likewise.
5716 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
5717 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5718 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
5720 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5721 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5722 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
5724 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
5726 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5727 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5728 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
5730 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5731 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5732 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
5734 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5735 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
5737 * config/powerpcspe/powerpcspe.c
5738 (rs6000_hard_regno_call_part_clobbered): New function.
5739 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5740 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5741 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
5743 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5744 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5745 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
5747 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
5748 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
5749 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
5751 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5752 Alan Hayward <alan.hayward@arm.com>
5753 David Sherwood <david.sherwood@arm.com>
5755 * rtl.h (subreg_memory_offset): Declare.
5756 * emit-rtl.c (subreg_memory_offset): New function.
5757 * expmed.c (store_bit_field_1): Use it.
5758 * expr.c (undefined_operand_subword_p): Likewise.
5759 * simplify-rtx.c (simplify_subreg): Likewise.
5761 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
5763 PR rtl-optimization/57448
5766 * optabs.c (expand_atomic_load): Place compiler memory barriers if
5767 using atomic_load pattern.
5768 (expand_atomic_store): Likewise.
5770 2017-09-04 Jakub Jelinek <jakub@redhat.com>
5773 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
5774 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
5775 handling. Use replace_call_with_value with NULL instead of
5776 gsi_replace, unlink_stmt_vdef and release_defs.
5778 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
5781 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
5783 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
5786 * rtl.h (emit_library_call_value_1): Declare.
5787 (emit_library_call): Replace declaration with a series of overloads.
5788 Remove the parameter count argument.
5789 (emit_library_call_value): Likewise.
5790 * calls.c (emit_library_call_value_1): Make global. Replace varargs
5791 with an "rtx_mode_t *".
5792 (emit_library_call_value): Delete.
5793 (emit_library_call): Likewise.
5794 * asan.c (asan_emit_stack_protection): Update calls accordingly.
5795 (asan_emit_allocas_unpoison): Likewise.
5796 * builtins.c (expand_builtin_powi): Likewise.
5797 (expand_asan_emit_allocas_unpoison): Likewise.
5798 * cfgexpand.c (expand_main_function): Likewise.
5799 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
5800 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
5801 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
5802 * config/arm/arm.c (arm_trampoline_init): Likewise.
5803 (arm_call_tls_get_addr): Likewise.
5804 (arm_expand_divmod_libfunc): Likewise.
5805 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
5806 (smulsi3_highpart): Likewise.
5807 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
5808 (c6x_expand_compare): Likewise.
5809 (c6x_expand_movmem): Likewise.
5810 * config/frv/frv.c (frv_trampoline_init): Likewise.
5811 * config/i386/i386.c (ix86_trampoline_init): Likewise.
5812 (ix86_expand_divmod_libfunc): Likewise.
5813 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
5814 (ia64_expand_compare): Likewise.
5815 (ia64_profile_hook): Likewise.
5816 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
5817 (nonlocal_goto): Likewise.
5818 (restore_stack_nonlocal): Likewise.
5819 * config/m32r/m32r.c (block_move_call): Likewise.
5820 (m32r_trampoline_init): Likewise.
5821 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
5822 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
5823 (m68k_call_m68k_read_tp): Likewise.
5824 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
5825 (microblaze_expand_divide): Likewise.
5826 * config/mips/mips.h (mips_args): Likewise.
5827 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
5828 (MIPS_ICACHE_SYNC): Likewise.
5829 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
5830 (nios2_trampoline_init): Likewise.
5831 * config/pa/pa.c (hppa_tls_call): Likewise.
5832 (pa_trampoline_init): Likewise.
5833 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
5834 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
5835 (expand_strn_compare): Likewise.
5836 (rs6000_generate_compare): Likewise.
5837 (rs6000_expand_float128_convert): Likewise.
5838 (output_profile_hook): Likewise.
5839 (rs6000_trampoline_init): Likewise.
5840 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
5841 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
5842 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
5843 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
5844 (rs6000_generate_compare): Likewise.
5845 (rs6000_expand_float128_convert): Likewise.
5846 (output_profile_hook): Likewise.
5847 (rs6000_trampoline_init): Likewise.
5848 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
5849 * config/sh/sh.c (sh_trampoline_init): Likewise.
5850 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
5851 (sparc_emit_float_lib_cmp): Likewise.
5852 (sparc32_initialize_trampoline): Likewise.
5853 (sparc64_initialize_trampoline): Likewise.
5854 (sparc_profile_hook): Likewise.
5855 * config/spu/spu.c (ea_load_store): Likewise.
5856 * config/spu/spu.md (floatunssidf2): Likewise.
5857 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
5858 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
5859 * config/visium/visium.c (expand_block_move_4): Likewise.
5860 (expand_block_move_2): Likewise.
5861 (expand_block_move_1): Likewise.
5862 (expand_block_set_4): Likewise.
5863 (expand_block_set_2): Likewise.
5864 (expand_block_set_1): Likewise.
5865 (visium_trampoline_init): Likewise.
5866 (visium_profile_hook): Likewise.
5867 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
5868 (xtensa_setup_frame_addresses): Likewise.
5869 (xtensa_trampoline_init): Likewise.
5870 * except.c (sjlj_emit_function_enter): Likewise.
5871 (sjlj_emit_function_exit): Likewise.
5872 * explow.c (allocate_dynamic_stack_space): Likewise.
5873 (probe_stack_range): Likewise.
5874 * expr.c (convert_mode_scalar): Likewise.
5875 * optabs.c (expand_binop): Likewise.
5876 (expand_twoval_binop_libfunc): Likewise.
5877 (expand_unop): Likewise.
5878 (prepare_cmp_insn): Likewise.
5879 (prepare_float_lib_cmp): Likewise.
5880 (expand_float): Likewise.
5881 (expand_fix): Likewise.
5882 (expand_fixed_convert): Likewise.
5883 (maybe_emit_sync_lock_test_and_set): Likewise.
5884 (expand_atomic_compare_and_swap): Likewise.
5885 (expand_mem_thread_fence): Likewise.
5886 (expand_atomic_fetch_op): Likewise.
5888 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
5890 * doc/generic.texi (OpenACC): Adjust URL.
5891 * doc/invoke.texi (C Dialect Options): Ditto.
5893 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
5895 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
5896 predicate for operand 1. Add (m,<S>) constraint.
5897 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
5898 Prevent memory operand 1 with register operand 2.
5900 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
5902 PR rtl-optimization/82024
5903 * combine.c (try_combine): If the combination result is a PARALLEL,
5904 and we only need to retain the SET in there that would be placed
5905 at I2, check that we can place that at I3 instead, before doing so.
5907 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5910 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
5912 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
5913 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
5916 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
5917 Jeff Law <law@redhat.com>
5919 * varasm.c (bss_initializer_p): Do not put constants into .bss
5920 (categorize_decl_for_section): Handle bss_initializer_p returning
5921 false when DECL_INITIAL is NULL.
5923 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5926 * config/s390/s390.c (s390_can_inline_p): New function.
5928 2017-09-01 Jeff Law <law@redhat.com>
5930 PR tree-optimization/82052
5931 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
5932 Always initialize the returned slot after a hash table miss
5933 when INSERT is true.
5935 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
5937 * config/s390/s390.md (mem_signal_fence): Remove.
5938 * doc/md.texi (mem_signal_fence): Remove.
5939 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
5941 * target-insns.def (mem_signal_fence): Remove.
5943 2017-09-01 Jakub Jelinek <jakub@redhat.com>
5946 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
5949 * asan.c (create_odr_indicator): Strip name encoding from assembler
5950 name before appending it after __odr_asan_.
5952 2017-09-01 Martin Liska <mliska@suse.cz>
5954 PR tree-optimization/82059
5955 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
5956 frequency only when an edge is redirected.
5958 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5960 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
5961 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
5962 (arc_conditional_register_usage): Remove ARC600 lp_count
5964 (arc_file_start): Emit Tag_ARC_CPU_variation.
5965 (arc_can_use_doloop_p): New conditions to use ZOLs.
5966 (hwloop_fail): New function.
5967 (hwloop_optimize): Likewise.
5968 (hwloop_pattern_reg): Likewise.
5969 (arc_doloop_hooks): New struct, to be used with reorg_loops.
5970 (arc_reorg_loops): New function, calls reorg_loops.
5971 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
5972 (arc600_corereg_hazard): Remove ZOL checking, case handled by
5974 (arc_loop_hazard): Remove function, functionality moved into
5976 (arc_hazard): Remove arc_loop_hazard call.
5977 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
5978 into hwloop_optimize.
5979 (arc_label_align): Remove ZOL handling.
5980 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
5981 * config/arc/arc.md (doloop_begin): Remove pattern.
5982 (doloop_begin_i): Likewise.
5983 (doloop_end_i): Likewise.
5984 (doloop_fallback): Likewise.
5985 (doloop_fallback_m): Likewise.
5986 (doloop_end): Reimplement expand.
5987 (arc_lp): New pattern for LP instruction.
5988 (loop_end): New pattern.
5989 (loop_fail): Likewise.
5990 (decrement_and_branch_until_zero): Likewise.
5991 * config/arc/arc.opt (mlpc-width): New option.
5992 * doc/invoke.texi (mlpc-width): Document option.
5994 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
5996 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
5997 (arc_ccfsm_advance): Fix checking for delay slots.
5998 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
6000 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6002 * config/arc/arc.md (movqi_insn): Add stores to save constant long
6004 (movhi_insn): Update store instruction constraint which are saving
6005 6-bit short immediates.
6006 (movsi_insn): Consider also short scaled load operations.
6007 (zero_extendhisi2_i): Use Usd constraint instead of T.
6008 (extendhisi2_i): Add q constraint.
6009 (arc_clzsi2): Add type and length attributes.
6010 (arc_ctzsi2): Likewise.
6011 * config/arc/constraints.md (Usc): Update constraint, the
6012 assembler can parse two relocations for a single instruction.
6014 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
6016 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
6017 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
6019 2017-08-31 Olivier Hainque <hainque@adacore.com>
6021 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
6022 match as powerpc-wrs-vxworks*.
6024 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
6026 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
6027 register constraint for by-element operand.
6028 (aarch64_mls_elt_merge<mode>): Likewise.
6030 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6032 * config/arc/arc.c (arc_can_follow_jump): Check for short
6035 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6037 * config.gcc: Use g.opt for arc.
6038 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
6039 functionality moved to ...
6040 (legitimate_scaled_address_p): New function, ...here.
6041 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
6042 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
6043 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
6045 (arc_override_options): Handle G option.
6046 (arc_output_pic_addr_const): Correct function definition.
6047 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
6048 (arc_decl_anon_ns_mem_p): Delete.
6049 (arc_in_small_data_p): Overhaul this function to take into
6050 consideration the value given via G option.
6051 (arc_rewrite_small_data_1): Renamed and corrected old
6052 arc_rewrite_small_data function.
6053 (arc_rewrite_small_data): New function.
6054 (small_data_pattern): Don't use pic_offset_table_rtx.
6055 * config/arc/arc.h (CC1_SPEC): Recognize G option.
6056 * config/arc/simdext.md (movmisalignv2hi): Use
6057 prepare_move_operands function.
6059 (movmisalign*): Likewise.
6060 * doc/invoke.texi (ARC options): Document -G option.
6062 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
6064 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
6066 * config/arc/arc.c (arc_print_operand): Output scalled address for
6067 sdata whenever is possible.
6068 (arc_in_small_data_p): Allow sdata for 64bit datum when double
6069 load/stores are available.
6070 (compact_sda_memory_operand): Check for the alignment required by
6071 code density instructions.
6072 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
6074 * config/arc/constraints.md (Usd): Update constraint.
6075 (Us0): New constraint.
6076 (Usc): Update constraint.
6078 2017-08-31 Richard Biener <rguenther@suse.de>
6081 * dwarf2out.c (dwarf2out_early_global_decl): Process each
6084 2017-08-31 Tamar Christina <tamar.christina@arm.com>
6086 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
6087 Resize type_signature.
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.c (aarch64_base_register_rtx_p): Only allow
6094 subregs whose inner modes can be stored in GPRs.
6095 (aarch64_classify_index): Likewise.
6097 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6098 Alan Hayward <alan.hayward@arm.com>
6099 David Sherwood <david.sherwood@arm.com>
6101 * config/aarch64/iterators.md (V_cmp_result): Rename to...
6102 (V_INT_EQUIV): ...this.
6103 (v_cmp_result): Rename to...
6104 (v_int_equiv): ...this.
6105 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
6106 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
6107 (copysign<mode>3): Likewise.
6108 (aarch64_simd_bsl<mode>_internal): Likewise.
6109 (aarch64_simd_bsl<mode>): Likewise.
6110 (vec_cmp<mode><mode>): Likewise.
6111 (vcond<mode><mode>): Likewise.
6112 (vcond<v_cmp_mixed><mode>): Likewise.
6113 (vcondu<mode><v_cmp_mixed>): Likewise.
6114 (aarch64_cm<optab><mode>): Likewise.
6115 (aarch64_cmtst<mode>): Likewise.
6116 (aarch64_fac<optab><mode>): Likewise.
6117 (vec_perm_const<mode>): Likewise.
6118 (vcond_mask_<mode><v_cmp_result>): Rename to...
6119 (vcond_mask_<mode><v_int_equiv>): ...this.
6120 (vec_cmp<mode><v_cmp_result>): Rename to...
6121 (vec_cmp<mode><v_int_equiv>): ...this.
6123 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
6124 Alan Hayward <alan.hayward@arm.com>
6125 David Sherwood <david.sherwood@arm.com>
6127 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
6129 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
6130 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
6131 (UNSPEC_LD4_DREG): New unspecs.
6132 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
6133 (aarch64_ld2<mode>_dreg_be): Replace with...
6134 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
6136 (aarch64_ld3<mode>_dreg_le)
6137 (aarch64_ld3<mode>_dreg_be): Replace with...
6138 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
6140 (aarch64_ld4<mode>_dreg_le)
6141 (aarch64_ld4<mode>_dreg_be): Replace with...
6142 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
6145 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6147 PR tree-optimization/81987
6148 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
6149 insert an initializer in a location not dominated by the stride
6152 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6154 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
6155 on the entire header of the finally block in the fallthru case.
6157 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
6159 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
6161 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
6163 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
6164 rs6000_emit_move_from_cr and call renamed function.
6165 (rs6000_emit_prologue): Call renamed functions.
6166 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
6167 movesi_from_cr, remove volatile CRs.
6169 2017-08-30 Jon Beniston <jon@beniston.com>
6170 Richard Biener <rguenther@suse.de>
6172 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
6173 of VECTOR_MODE_P check.
6174 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
6175 element vector types.
6177 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6179 * df.h (df_read_modify_subreg_p): Remove in favor of...
6180 * rtl.h (read_modify_subreg_p): ...this new function. Take a
6181 const_rtx instead of an rtx.
6182 * cprop.c (local_cprop_find_used_regs): Update accordingly.
6183 * df-problems.c (df_word_lr_mark_ref): Likewise.
6184 * ira-lives.c (mark_pseudo_reg_live): Likewise.
6185 (mark_pseudo_reg_dead): Likewise.
6186 (mark_ref_dead): Likewise.
6187 * reginfo.c (init_subregs_of_mode): Likewise.
6188 * sched-deps.c (sched_analyze_1): Likewise.
6189 * df-scan.c (df_def_record_1): Likewise.
6190 (df_uses_record): Likewise.
6191 (df_read_modify_subreg_p): Remove in favor of...
6192 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
6193 const_rtx instead of an rtx.
6195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6196 Alan Hayward <alan.hayward@arm.com>
6197 David Sherwood <david.sherwood@arm.com>
6199 * rtl.h (partial_subreg_p): New function.
6200 * caller-save.c (save_call_clobbered_regs): Use it.
6201 * calls.c (expand_call): Likewise.
6202 * combine.c (combinable_i3pat): Likewise.
6203 (simplify_set): Likewise.
6204 (make_extraction): Likewise.
6205 (make_compound_operation_int): Likewise.
6206 (gen_lowpart_or_truncate): Likewise.
6207 (force_to_mode): Likewise.
6208 (make_field_assignment): Likewise.
6209 (reg_truncated_to_mode): Likewise.
6210 (record_truncated_value): Likewise.
6211 (move_deaths): Likewise.
6212 * cse.c (record_jump_cond): Likewise.
6213 (cse_insn): Likewise.
6214 * cselib.c (cselib_lookup_1): Likewise.
6215 * expmed.c (extract_bit_field_using_extv): Likewise.
6216 * function.c (assign_parm_setup_reg): Likewise.
6217 * ifcvt.c (noce_convert_multiple_sets): Likewise.
6218 * ira-build.c (create_insn_allocnos): Likewise.
6219 * lra-coalesce.c (merge_pseudos): Likewise.
6220 * lra-constraints.c (match_reload): Likewise.
6221 (simplify_operand_subreg): Likewise.
6222 (curr_insn_transform): Likewise.
6223 * lra-lives.c (process_bb_lives): Likewise.
6224 * lra.c (new_insn_reg): Likewise.
6225 (lra_substitute_pseudo): Likewise.
6226 * regcprop.c (mode_change_ok): Likewise.
6227 (maybe_mode_change): Likewise.
6228 (copyprop_hardreg_forward_1): Likewise.
6229 * reload.c (push_reload): Likewise.
6230 (find_reloads): Likewise.
6231 (find_reloads_subreg_address): Likewise.
6232 * reload1.c (alter_reg): Likewise.
6233 (eliminate_regs_1): Likewise.
6234 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
6236 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
6238 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
6239 back to if statements, including unpack.
6241 2017-08-30 Martin Liska <mliska@suse.cz>
6244 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
6246 (func_checker::compare_asm_inputs_outputs): ... this function.
6247 (func_checker::compare_gimple_asm): Use the function to compare
6248 also ASM constrains.
6249 * ipa-icf-gimple.h: Rename the function.
6251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6252 Alan Hayward <alan.hayward@arm.com>
6253 David Sherwood <david.sherwood@arm.com>
6255 * coretypes.h (complex_mode): New type.
6256 * gdbhooks.py (build_pretty_printer): Handle it.
6257 * machmode.h (complex_mode): New class.
6258 (complex_mode::includes_p): New function.
6259 (is_complex_int_mode): Likewise.
6260 (is_complex_float_mode): Likewise.
6261 * genmodes.c (get_mode_class): Handle complex mode classes.
6262 * function.c (expand_function_end): Use is_complex_int_mode.
6264 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6265 Alan Hayward <alan.hayward@arm.com>
6266 David Sherwood <david.sherwood@arm.com>
6268 * coretypes.h (scalar_mode_pod): New typedef.
6269 * gdbhooks.py (build_pretty_printer): Handle it.
6270 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
6271 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
6272 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
6273 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
6276 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6277 Alan Hayward <alan.hayward@arm.com>
6278 David Sherwood <david.sherwood@arm.com>
6280 * machmode.h (mode_for_vector): Take a scalar_mode instead
6282 * stor-layout.c (mode_for_vector): Likewise.
6283 * explow.c (promote_mode): Use as_a <scalar_mode>.
6284 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
6286 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6287 Alan Hayward <alan.hayward@arm.com>
6288 David Sherwood <david.sherwood@arm.com>
6290 * target.def (preferred_simd_mode): Take a scalar_mode
6291 instead of a machine_mode.
6292 * targhooks.h (default_preferred_simd_mode): Likewise.
6293 * targhooks.c (default_preferred_simd_mode): Likewise.
6294 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
6295 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
6296 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
6297 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
6298 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
6299 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
6300 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
6301 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
6303 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
6304 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
6305 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
6306 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
6307 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
6308 * doc/tm.texi: Regenerate.
6309 * optabs-query.c (can_vec_mask_load_store_p): Return false for
6312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6313 Alan Hayward <alan.hayward@arm.com>
6314 David Sherwood <david.sherwood@arm.com>
6316 * target.def (scalar_mode_supported_p): Take a scalar_mode
6317 instead of a machine_mode.
6318 * targhooks.h (default_scalar_mode_supported_p): Likewise.
6319 * targhooks.c (default_scalar_mode_supported_p): Likewise.
6320 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
6321 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
6322 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
6323 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
6324 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
6325 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
6326 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
6327 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
6328 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
6329 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
6330 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
6331 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
6333 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
6334 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
6335 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
6336 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
6337 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
6339 * doc/tm.texi: Regenerate.
6341 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6342 Alan Hayward <alan.hayward@arm.com>
6343 David Sherwood <david.sherwood@arm.com>
6345 * coretypes.h (opt_scalar_mode): New typedef.
6346 * gdbhooks.py (build_pretty_printers): Handle it.
6347 * machmode.h (mode_iterator::get_2xwider): Add overload for
6349 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
6351 * expr.c (convert_mode_scalar): Likewise.
6352 * omp-low.c (omp_clause_aligned_alignment): Likewise.
6353 * optabs.c (expand_float): Likewise.
6354 (expand_fix): Likewise.
6355 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6358 Alan Hayward <alan.hayward@arm.com>
6359 David Sherwood <david.sherwood@arm.com>
6361 * optabs.c (expand_float): Explicitly check for scalars before
6362 using a branching expansion.
6363 (expand_fix): Likewise.
6365 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6366 Alan Hayward <alan.hayward@arm.com>
6367 David Sherwood <david.sherwood@arm.com>
6369 * expr.c (convert_mode): Split scalar handling out into...
6370 (convert_mode_scalar): ...this new function. Treat the modes
6373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6374 Alan Hayward <alan.hayward@arm.com>
6375 David Sherwood <david.sherwood@arm.com>
6377 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
6379 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
6381 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6382 Alan Hayward <alan.hayward@arm.com>
6383 David Sherwood <david.sherwood@arm.com>
6385 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
6386 rather than a machine_mode.
6387 (fixed_from_string): Likewise.
6388 (fixed_convert): Likewise.
6389 (fixed_convert_from_int): Likewise.
6390 (fixed_convert_from_real): Likewise.
6391 (real_convert_from_fixed): Likewise.
6392 * fixed-value.c (fixed_from_double_int): Likewise.
6393 (fixed_from_string): Likewise.
6394 (fixed_convert): Likewise.
6395 (fixed_convert_from_int): Likewise.
6396 (fixed_convert_from_real): Likewise.
6397 (real_convert_from_fixed): Likewise.
6398 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
6400 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6401 Alan Hayward <alan.hayward@arm.com>
6402 David Sherwood <david.sherwood@arm.com>
6404 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
6405 of separate mode class checks. Do not allow vector modes here.
6406 (immed_wide_int_const): Use as_a <scalar_mode>.
6407 * explow.c (trunc_int_for_mode): Likewise.
6408 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
6409 (wi::shwi): Likewise.
6410 (wi::min_value): Likewise.
6411 (wi::max_value): Likewise.
6412 * dwarf2out.c (loc_descriptor): Likewise.
6413 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
6416 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6417 Alan Hayward <alan.hayward@arm.com>
6418 David Sherwood <david.sherwood@arm.com>
6420 * tree.h (SCALAR_TYPE_MODE): New macro.
6421 * expr.c (expand_expr_addr_expr_1): Use it.
6422 (expand_expr_real_2): Likewise.
6423 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
6424 (fold_convert_const_fixed_from_int): Likewise.
6425 (fold_convert_const_fixed_from_real): Likewise.
6426 (native_encode_fixed): Likewise
6427 (native_encode_complex): Likewise
6428 (native_encode_vector): Likewise.
6429 (native_interpret_fixed): Likewise.
6430 (native_interpret_real): Likewise.
6431 (native_interpret_complex): Likewise.
6432 (native_interpret_vector): Likewise.
6433 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
6434 (simd_clone_adjust_argument_types): Likewise.
6435 (simd_clone_init_simd_arrays): Likewise.
6436 (simd_clone_adjust): Likewise.
6437 * stor-layout.c (layout_type): Likewise.
6438 * tree.c (build_minus_one_cst): Likewise.
6439 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
6440 * tree-inline.c (estimate_move_cost): Likewise.
6441 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
6442 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
6443 (vectorizable_reduction): Likewise.
6444 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
6445 (vect_recog_mixed_size_cond_pattern): Likewise.
6446 (check_bool_pattern): Likewise.
6447 (adjust_bool_pattern): Likewise.
6448 (search_type_for_mask_1): Likewise.
6449 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
6450 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
6451 (vectorizable_load): Likewise.
6452 (vectorizable_store): Likewise.
6453 * ubsan.c (ubsan_encode_value): Likewise.
6454 * varasm.c (output_constant): Likewise.
6456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6457 Alan Hayward <alan.hayward@arm.com>
6458 David Sherwood <david.sherwood@arm.com>
6460 * coretypes.h (scalar_mode): New class.
6461 * machmode.h (scalar_mode): Likewise.
6462 (scalar_mode::includes_p): New function.
6463 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
6464 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
6465 * genmodes.c (get_mode_class): Handle remaining scalar modes.
6466 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
6467 * expmed.c (store_bit_field_1): Likewise.
6468 (extract_bit_field_1): Likewise.
6469 * expr.c (write_complex_part): Likewise.
6470 (read_complex_part): Likewise.
6471 (emit_move_complex_push): Likewise.
6472 (expand_expr_real_2): Likewise.
6473 * function.c (assign_parm_setup_reg): Likewise.
6474 (assign_parms_unsplit_complex): Likewise.
6475 * optabs.c (expand_binop): Likewise.
6476 * rtlanal.c (subreg_get_info): Likewise.
6477 * simplify-rtx.c (simplify_immed_subreg): Likewise.
6478 * varasm.c (output_constant_pool_2): Likewise.
6480 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6481 Alan Hayward <alan.hayward@arm.com>
6482 David Sherwood <david.sherwood@arm.com>
6484 * expmed.c (extract_high_half): Use scalar_int_mode and remove
6486 (expmed_mult_highpart_optab): Likewise.
6487 (expmed_mult_highpart): Likewise.
6489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6490 Alan Hayward <alan.hayward@arm.com>
6491 David Sherwood <david.sherwood@arm.com>
6493 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
6494 instead of a machine_mode.
6495 (builtin_memset_read_str): Likewise.
6496 * builtins.c (c_readstr): Likewise.
6497 (builtin_memcpy_read_str): Likewise.
6498 (builtin_strncpy_read_str): Likewise.
6499 (builtin_memset_read_str): Likewise.
6500 (builtin_memset_gen_str): Likewise.
6501 (expand_builtin_signbit): Use scalar_int_mode for local variables.
6502 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
6503 instead of a machine_mode.
6504 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
6506 (make_extraction): Likewise.
6507 (try_widen_shift_mode): Take and return scalar_int_modes instead
6509 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
6510 a scalar_int_mode instead of a machine_mode.
6511 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
6512 (avr_addr_space_pointer_mode): Likewise.
6513 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
6514 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
6515 (msp430_unwind_word_mode): Likewise.
6516 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
6517 (spu_addr_space_pointer_mode): Likewise.
6518 (spu_addr_space_address_mode): Likewise.
6519 (spu_libgcc_cmp_return_mode): Likewise.
6520 (spu_libgcc_shift_count_mode): Likewise.
6521 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
6522 (rl78_addr_space_pointer_mode): Likewise.
6523 (fl78_unwind_word_mode): Likewise.
6524 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
6526 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
6527 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
6528 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
6529 (mips_valid_pointer_mode): Likewise.
6530 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
6531 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
6532 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
6534 (ft32_addr_space_address_mode): Likewise.
6535 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
6536 scalar_int_mode instead of a machine_mode.
6537 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
6539 (m32c_addr_space_address_mode): Likewise.
6540 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
6541 (rs6000_eh_return_filter_mode): Likewise.
6542 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
6543 (rs6000_eh_return_filter_mode): Likewise.
6544 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
6545 (s390_libgcc_shift_count_mode): Likewise.
6546 (s390_unwind_word_mode): Likewise.
6547 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
6549 * target.def (mode_rep_extended): Likewise.
6550 (valid_pointer_mode): Likewise.
6551 (addr_space.valid_pointer_mode): Likewise.
6552 (eh_return_filter_mode): Return a scalar_int_mode rather than
6554 (libgcc_cmp_return_mode): Likewise.
6555 (libgcc_shift_count_mode): Likewise.
6556 (unwind_word_mode): Likewise.
6557 (addr_space.pointer_mode): Likewise.
6558 (addr_space.address_mode): Likewise.
6559 * doc/tm.texi: Regenerate.
6560 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
6562 (do_jump): Use scalar_int_mode for local variables.
6563 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
6564 rather than a machine_mode.
6565 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
6566 (scompare_loc_descriptor_wide): Likewise.
6567 (scompare_loc_descriptor_narrow): Likewise.
6568 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
6570 * except.c (sjlj_emit_dispatch_table): Likewise.
6571 (expand_builtin_eh_copy_values): Likewise.
6572 * explow.c (convert_memory_address_addr_space_1): Likewise.
6573 Take a scalar_int_mode rather than a machine_mode.
6574 (convert_memory_address_addr_space): Take a scalar_int_mode rather
6575 than a machine_mode.
6576 (memory_address_addr_space): Use scalar_int_mode for local variables.
6577 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
6578 rather than a machine_mode.
6579 * expmed.c (mask_rtx): Likewise.
6580 (init_expmed_one_conv): Likewise.
6581 (expand_mult_highpart_adjust): Likewise.
6582 (extract_high_half): Likewise.
6583 (expmed_mult_highpart_optab): Likewise.
6584 (expmed_mult_highpart): Likewise.
6585 (expand_smod_pow2): Likewise.
6586 (expand_sdiv_pow2): Likewise.
6587 (emit_store_flag_int): Likewise.
6588 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
6590 (extract_low_bits): Likewise.
6591 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
6593 * expr.c (pieces_addr::adjust): Likewise.
6594 (can_store_by_pieces): Likewise.
6595 (store_by_pieces): Likewise.
6596 (clear_by_pieces_1): Likewise.
6597 (expand_expr_addr_expr_1): Likewise.
6598 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
6599 (expand_expr_real_1): Likewise.
6600 (try_casesi): Likewise.
6601 * final.c (shorten_branches): Likewise.
6602 * fold-const.c (fold_convert_const_int_from_fixed): Change the
6603 type of "mode" to machine_mode.
6604 * internal-fn.c (expand_arith_overflow_result_store): Take a
6605 scalar_int_mode rather than a machine_mode.
6606 (expand_mul_overflow): Use scalar_int_mode for local variables.
6607 * loop-doloop.c (doloop_modify): Likewise.
6608 (doloop_optimize): Likewise.
6609 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
6610 than a machine_mode.
6611 (expand_doubleword_shift_condmove): Likewise.
6612 (expand_doubleword_shift): Likewise.
6613 (expand_doubleword_clz): Likewise.
6614 (expand_doubleword_popcount): Likewise.
6615 (expand_doubleword_parity): Likewise.
6616 (expand_absneg_bit): Use scalar_int_mode for local variables.
6617 (prepare_float_lib_cmp): Likewise.
6618 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
6619 rather than a machine_mode.
6620 (convert_memory_address_addr_space): Likewise.
6621 (get_mode_bounds): Likewise.
6622 (get_address_mode): Return a scalar_int_mode rather than a
6624 * rtlanal.c (get_address_mode): Likewise.
6625 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
6626 than a machine_mode.
6627 * targhooks.c (default_mode_rep_extended): Likewise.
6628 (default_valid_pointer_mode): Likewise.
6629 (default_addr_space_valid_pointer_mode): Likewise.
6630 (default_eh_return_filter_mode): Return a scalar_int_mode rather
6631 than a machine_mode.
6632 (default_libgcc_cmp_return_mode): Likewise.
6633 (default_libgcc_shift_count_mode): Likewise.
6634 (default_unwind_word_mode): Likewise.
6635 (default_addr_space_pointer_mode): Likewise.
6636 (default_addr_space_address_mode): Likewise.
6637 * targhooks.h (default_eh_return_filter_mode): Likewise.
6638 (default_libgcc_cmp_return_mode): Likewise.
6639 (default_libgcc_shift_count_mode): Likewise.
6640 (default_unwind_word_mode): Likewise.
6641 (default_addr_space_pointer_mode): Likewise.
6642 (default_addr_space_address_mode): Likewise.
6643 (default_mode_rep_extended): Take a scalar_int_mode rather than
6645 (default_valid_pointer_mode): Likewise.
6646 (default_addr_space_valid_pointer_mode): Likewise.
6647 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
6649 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
6650 rather than a machine_mode.
6651 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
6652 for local variables.
6653 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
6654 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
6655 than a machine_mode.
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 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
6662 the mode argument to scalar_int_mode.
6663 (do_jump_by_parts_zero_rtx): Likewise.
6664 (do_jump_by_parts_equality_rtx): Likewise.
6665 (do_jump_by_parts_greater): Take a mode argument.
6666 (do_jump_by_parts_equality): Likewise.
6667 (do_jump_1): Update calls accordingly.
6669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6670 Alan Hayward <alan.hayward@arm.com>
6671 David Sherwood <david.sherwood@arm.com>
6673 * is-a.h (safe_dyn_cast): New function.
6674 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
6675 (jump_table_for_label): Likewise.
6676 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
6677 instead of an rtx_insn *.
6678 (shorten_branches): Use dyn_cast instead of LABEL_P and
6679 JUMP_TABLE_DATA_P. Use jump_table_for_label and
6680 rtx_jump_table_data::get_data_mode.
6681 (final_scan_insn): Likewise.
6683 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6684 Alan Hayward <alan.hayward@arm.com>
6685 David Sherwood <david.sherwood@arm.com>
6687 * combine.c (try_combine): Use is_a <scalar_int_mode> when
6688 trying to combine a full-register integer set with a subreg
6691 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6692 Alan Hayward <alan.hayward@arm.com>
6693 David Sherwood <david.sherwood@arm.com>
6695 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
6696 that is always either address_mode or pointer_mode.
6698 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6699 Alan Hayward <alan.hayward@arm.com>
6700 David Sherwood <david.sherwood@arm.com>
6702 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
6703 when the two are known to be equal.
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 * simplify-rtx.c (simplify_const_unary_operation): Use
6710 is_a <scalar_int_mode> instead of checking for a nonzero
6711 precision. Forcibly convert op_mode to a scalar_int_mode
6712 in that case. More clearly differentiate the operand and
6713 result modes and use the former when deciding what the value
6714 of a count-bits operation should be. Use is_int_mode instead
6715 of checking for a MODE_INT. Remove redundant check for whether
6716 this mode has a zero precision.
6718 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6719 Alan Hayward <alan.hayward@arm.com>
6720 David Sherwood <david.sherwood@arm.com>
6722 * optabs.c (widen_leading): Change the type of the mode argument
6723 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
6724 (widen_bswap): Likewise.
6725 (expand_parity): Likewise.
6726 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
6727 (expand_ffs): Likewise.
6728 (epand_unop): Check for scalar integer modes before calling the
6731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6732 Alan Hayward <alan.hayward@arm.com>
6733 David Sherwood <david.sherwood@arm.com>
6735 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
6737 (expand_expr_real_1): Update call accordingly.
6739 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6740 Alan Hayward <alan.hayward@arm.com>
6741 David Sherwood <david.sherwood@arm.com>
6743 * expmed.c (store_bit_field_using_insv): Add op0_mode and
6744 value_mode arguments. Use scalar_int_mode internally.
6745 (store_bit_field_1): Rename the new integer mode from imode
6746 to op0_mode and use it instead of GET_MODE (op0). Update calls
6747 to store_split_bit_field, store_bit_field_using_insv and
6748 store_fixed_bit_field.
6749 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
6750 Use scalar_int_mode internally. Use a bit count rather than a mode
6751 when calculating the largest bit size for get_best_mode.
6752 Update calls to store_split_bit_field and store_fixed_bit_field_1.
6753 (store_fixed_bit_field_1): Add mode and value_mode arguments.
6754 Remove assertion that OP0 has a scalar integer mode.
6755 (store_split_bit_field): Add op0_mode and value_mode arguments.
6756 Update calls to extract_fixed_bit_field.
6757 (extract_bit_field_using_extv): Add an op0_mode argument.
6758 Use scalar_int_mode internally.
6759 (extract_bit_field_1): Rename the new integer mode from imode to
6760 op0_mode and use it instead of GET_MODE (op0). Update calls to
6761 extract_split_bit_field, extract_bit_field_using_extv and
6762 extract_fixed_bit_field.
6763 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
6764 to extract_split_bit_field and extract_fixed_bit_field_1.
6765 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
6766 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
6768 (extract_split_bit_field): Add an op0_mode argument. Update call
6769 to extract_fixed_bit_field.
6771 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6772 Alan Hayward <alan.hayward@arm.com>
6773 David Sherwood <david.sherwood@arm.com>
6775 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
6776 * explow.c (hard_function_value): Likewise.
6777 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
6778 convert_to_mode call outside the loop.
6779 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
6780 for the mode iterator. Require the mode specified by max_pieces
6782 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
6784 (copy_blkmode_to_reg): Likewise.
6785 (set_storage_via_setmem): Likewise.
6786 * optabs.c (prepare_cmp_insn): Likewise.
6787 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
6788 * stor-layout.c (finish_bitfield_representative): Likewise.
6790 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6791 Alan Hayward <alan.hayward@arm.com>
6792 David Sherwood <david.sherwood@arm.com>
6794 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
6795 * expr.c (convert_move): Use them.
6796 (convert_modes): Likewise.
6797 (store_expr_with_bounds): Likewise.
6799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6800 Alan Hayward <alan.hayward@arm.com>
6801 David Sherwood <david.sherwood@arm.com>
6803 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
6804 parameter for the mode of "x". Remove the "known_x", "known_mode"
6805 and "known_ret" arguments. Change the type of the mode argument
6807 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
6808 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
6809 (reg_num_sign_bit_copies_for_combine): Likewise.
6810 * rtlanal.c (nonzero_bits1): Likewise.
6811 (num_sign_bit_copies1): Likewise.
6812 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
6813 (reg_num_sign_bit_copies_general): Likewise.
6814 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
6815 (reg_nonzero_bits_general): Likewise.
6817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6818 Alan Hayward <alan.hayward@arm.com>
6819 David Sherwood <david.sherwood@arm.com>
6821 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
6822 than in subroutines. Return 1 for non-integer modes.
6823 (cached_num_sign_bit_copies): Change the type of the mode parameter
6825 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
6826 classes. Handle CONST_INT_P first and then check whether X also
6827 has a scalar integer mode. Check the same thing for inner registers
6828 of a SUBREG and for values that are being extended or truncated.
6830 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6831 Alan Hayward <alan.hayward@arm.com>
6832 David Sherwood <david.sherwood@arm.com>
6834 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
6835 in subroutines. Return the mode mask for non-integer modes.
6836 (cached_nonzero_bits): Change the type of the mode parameter
6838 (nonzero_bits1): Likewise. Remove early exit for other mode
6839 classes. Handle CONST_INT_P first and then check whether X
6840 also has a scalar integer mode.
6842 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6843 Alan Hayward <alan.hayward@arm.com>
6844 David Sherwood <david.sherwood@arm.com>
6846 * expr.c (widest_int_mode_for_size): Make the comment match the code.
6847 Return a scalar_int_mode and assert that the size is greater than
6849 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
6850 (op_by_pieces_d::op_by_pieces_d): Likewise.
6851 (op_by_pieces_d::run): Likewise.
6852 (can_store_by_pieces): Likewise.
6854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6855 Alan Hayward <alan.hayward@arm.com>
6856 David Sherwood <david.sherwood@arm.com>
6858 * combine.c (extract_left_shift): Add a mode argument and update
6860 (make_compound_operation_int): Change the type of the mode parameter
6861 to scalar_int_mode and update the call to extract_left_shift.
6863 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6864 Alan Hayward <alan.hayward@arm.com>
6865 David Sherwood <david.sherwood@arm.com>
6867 * combine.c (simplify_and_const_int): Change the type of the mode
6868 parameter to scalar_int_mode.
6869 (simplify_and_const_int_1): Likewise. Update recursive call.
6871 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6872 Alan Hayward <alan.hayward@arm.com>
6873 David Sherwood <david.sherwood@arm.com>
6875 * combine.c (simplify_compare_const): Check that the mode is a
6876 scalar_int_mode (rather than VOIDmode) before testing its
6878 (simplify_comparison): Move COMPARISON_P handling out of the
6879 loop and restrict the latter part of the loop to scalar_int_modes.
6880 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
6881 and when considering SUBREG_REGs. Use is_int_mode instead of
6882 checking GET_MODE_CLASS against MODE_INT.
6884 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6885 Alan Hayward <alan.hayward@arm.com>
6886 David Sherwood <david.sherwood@arm.com>
6888 * combine.c (try_widen_shift_mode): Move check for equal modes to...
6889 (simplify_shift_const_1): ...here. Use scalar_int_mode for
6890 shift_unit_mode and for modes involved in scalar shifts.
6892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6893 Alan Hayward <alan.hayward@arm.com>
6894 David Sherwood <david.sherwood@arm.com>
6896 * combine.c (force_int_to_mode): New function, split out from...
6897 (force_to_mode): ...here. Keep xmode up-to-date and use it
6898 instead of GET_MODE (x).
6900 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6901 Alan Hayward <alan.hayward@arm.com>
6902 David Sherwood <david.sherwood@arm.com>
6904 * optabs-query.h (extraction_insn::struct_mode): Change type to
6905 opt_scalar_int_mode and update comment.
6906 (extraction_insn::field_mode): Change type to scalar_int_mode.
6907 (extraction_insn::pos_mode): Likewise.
6908 * combine.c (make_extraction): Update accordingly.
6909 * optabs-query.c (get_traditional_extraction_insn): Likewise.
6910 (get_optab_extraction_insn): Likewise.
6911 * recog.c (simplify_while_replacing): Likewise.
6912 * expmed.c (narrow_bit_field_mem): Change the type of the mode
6913 parameter to opt_scalar_int_mode.
6915 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6916 Alan Hayward <alan.hayward@arm.com>
6917 David Sherwood <david.sherwood@arm.com>
6919 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
6920 to a scalar_int_mode instead of a machine_mode.
6921 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
6922 (get_best_mode): Return a boolean and use a pointer argument to store
6923 the selected mode. Replace the limit mode parameter with a bit limit.
6924 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
6925 for the values returned by bit_field_mode_iterator::next_mode.
6926 (store_bit_field): Update call to get_best_mode.
6927 (store_fixed_bit_field): Likewise.
6928 (extract_fixed_bit_field): Likewise.
6929 * expr.c (optimize_bitfield_assignment_op): Likewise.
6930 * fold-const.c (optimize_bit_field_compare): Likewise.
6931 (fold_truth_andor_1): Likewise.
6932 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
6933 Update for new type of m_mode.
6934 (get_best_mode): As above.
6936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6937 Alan Hayward <alan.hayward@arm.com>
6938 David Sherwood <david.sherwood@arm.com>
6940 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
6941 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
6942 (store_bit_field): Check is_a <scalar_int_mode> before calling
6943 strict_volatile_bitfield_p.
6944 (extract_bit_field): Likewise.
6946 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6947 Alan Hayward <alan.hayward@arm.com>
6948 David Sherwood <david.sherwood@arm.com>
6950 * target.def (cstore_mode): Return a scalar_int_mode.
6951 * doc/tm.texi: Regenerate.
6952 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
6953 * targhooks.h (default_cstore_mode): Likewise.
6954 * targhooks.c (default_cstore_mode): Likewise, using a forced
6956 * expmed.c (emit_cstore): Expect the target of the cstore to be
6959 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
6960 Alan Hayward <alan.hayward@arm.com>
6961 David Sherwood <david.sherwood@arm.com>
6963 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
6965 (niter_desc): Likewise mode.
6966 (iv_analyze): Add a mode parameter.
6968 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
6969 and change its type to scalar_int_mode.
6970 * loop-iv.c: Update commentary at head of file.
6971 (iv_constant): Pass the mode paraeter before the rtx it describes
6972 and change its type to scalar_int_mode. Remove VOIDmode handling.
6973 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
6974 (iv_extend): Likewise.
6975 (shorten_into_mode): Likewise.
6976 (iv_add): Use scalar_int_mode.
6977 (iv_mult): Likewise.
6978 (iv_shift): Likewise.
6979 (canonicalize_iv_subregs): Likewise.
6980 (get_biv_step_1): Pass the outer_mode parameter before the rtx
6981 it describes and change its mode to scalar_int_mode. Also change
6982 the type of the returned inner_mode to scalar_int_mode.
6983 (get_biv_step): Likewise, turning outer_mode from a pointer
6984 into a direct parameter. Update call to get_biv_step_1.
6985 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
6986 iv_constant and get_biv_step.
6987 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
6988 and change its type to scalar_int_mode. Don't initialise iv->mode
6989 to VOIDmode and remove later checks for its still being VOIDmode.
6990 Update calls to iv_analyze_op and iv_analyze_expr. Check
6991 is_a <scalar_int_mode> when changing the mode under consideration.
6992 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
6993 Update call to iv_analyze_expr.
6994 (iv_analyze_op): Add a mode parameter. Reject subregs whose
6995 inner register is not also a scalar_int_mode. Update call to
6997 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
6998 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
6999 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
7000 separate mode class checks. Update calls to iv_analyze. Remove
7001 fix-up of VOIDmodes after iv_analyze_biv.
7002 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
7003 don't have a scalar_int_mode. Update call to biv_p.
7005 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7006 Alan Hayward <alan.hayward@arm.com>
7007 David Sherwood <david.sherwood@arm.com>
7009 * cfgexpand.c (convert_debug_memory_address): Use
7010 as_a <scalar_int_mode>.
7011 * combine.c (expand_compound_operation): Likewise.
7012 (make_extraction): Likewise.
7013 (change_zero_ext): Likewise.
7014 (simplify_comparison): Likewise.
7015 * cse.c (cse_insn): Likewise.
7016 * dwarf2out.c (minmax_loc_descriptor): Likewise.
7017 (mem_loc_descriptor): Likewise.
7018 (loc_descriptor): Likewise.
7019 * expmed.c (init_expmed_one_mode): Likewise.
7020 (synth_mult): Likewise.
7021 (emit_store_flag_1): Likewise.
7022 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
7023 of a comparison with size.
7024 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
7025 (reduce_to_bit_field_precision): Likewise.
7026 * function.c (expand_function_end): Likewise.
7027 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
7028 * loop-doloop.c (doloop_modify): Likewise.
7029 * optabs.c (expand_binop): Likewise.
7030 (expand_unop): Likewise.
7031 (expand_copysign_absneg): Likewise.
7032 (prepare_cmp_insn): Likewise.
7033 (maybe_legitimize_operand): Likewise.
7034 * recog.c (const_scalar_int_operand): Likewise.
7035 * rtlanal.c (get_address_mode): Likewise.
7036 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7037 (simplify_cond_clz_ctz): Likewise.
7038 * tree-nested.c (get_nl_goto_field): Likewise.
7039 * tree.c (build_vector_type_for_mode): Likewise.
7040 * var-tracking.c (use_narrower_mode): Likewise.
7042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7043 Alan Hayward <alan.hayward@arm.com>
7044 David Sherwood <david.sherwood@arm.com>
7046 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
7047 * builtins.c (expand_builtin_signbit): Use it.
7048 * cfgexpand.c (expand_debug_expr): Likewise.
7049 * dojump.c (do_jump): Likewise.
7050 (do_compare_and_jump): Likewise.
7051 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
7052 * expmed.c (make_tree): Likewise.
7053 * expr.c (expand_expr_real_2): Likewise.
7054 (expand_expr_real_1): Likewise.
7055 (try_casesi): Likewise.
7056 * fold-const-call.c (fold_const_call_ss): Likewise.
7057 * fold-const.c (unextend): Likewise.
7058 (extract_muldiv_1): Likewise.
7059 (fold_single_bit_test): Likewise.
7060 (native_encode_int): Likewise.
7061 (native_encode_string): Likewise.
7062 (native_interpret_int): Likewise.
7063 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
7064 * internal-fn.c (expand_addsub_overflow): Likewise.
7065 (expand_neg_overflow): Likewise.
7066 (expand_mul_overflow): Likewise.
7067 (expand_arith_overflow): Likewise.
7068 * match.pd: Likewise.
7069 * stor-layout.c (layout_type): Likewise.
7070 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
7071 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
7072 * tree-ssanames.c (get_range_info): Likewise.
7073 * tree-switch-conversion.c (array_value_type) Likewise.
7074 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
7075 (vect_recog_divmod_pattern): Likewise.
7076 (vect_recog_mixed_size_cond_pattern): Likewise.
7077 * tree-vrp.c (extract_range_basic): Likewise.
7078 (simplify_float_conversion_using_ranges): Likewise.
7079 * tree.c (int_fits_type_p): Likewise.
7080 * ubsan.c (instrument_bool_enum_load): Likewise.
7081 * varasm.c (mergeable_string_section): Likewise.
7082 (narrowing_initializer_constant_valid_p): Likewise.
7083 (output_constant): Likewise.
7085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7086 Alan Hayward <alan.hayward@arm.com>
7087 David Sherwood <david.sherwood@arm.com>
7089 * machmode.h (NARROWEST_INT_MODE): New macro.
7090 * expr.c (alignment_for_piecewise_move): Use it instead of
7091 GET_CLASS_NARROWEST_MODE (MODE_INT).
7092 (push_block): Likewise.
7093 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
7095 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7098 Alan Hayward <alan.hayward@arm.com>
7099 David Sherwood <david.sherwood@arm.com>
7101 * postreload.c (move2add_valid_value_p): Change the type of the
7102 mode parameter to scalar_int_mode.
7103 (move2add_use_add2_insn): Add a mode parameter and use it instead
7105 (move2add_use_add3_insn): Likewise.
7106 (reload_cse_move2add): Update accordingly.
7108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7109 Alan Hayward <alan.hayward@arm.com>
7110 David Sherwood <david.sherwood@arm.com>
7112 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
7114 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
7115 * optabs.c (expand_unop): Likewise.
7117 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7118 Alan Hayward <alan.hayward@arm.com>
7119 David Sherwood <david.sherwood@arm.com>
7121 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
7122 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
7123 (popcount_loc_descriptor): Likewise.
7124 (bswap_loc_descriptor): Likewise.
7125 (rotate_loc_descriptor): Likewise.
7126 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
7127 calling the functions above.
7129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7130 Alan Hayward <alan.hayward@arm.com>
7131 David Sherwood <david.sherwood@arm.com>
7133 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
7135 (try_combine): Likewise.
7136 (simplify_if_then_else): Likewise.
7137 * cse.c (cse_insn): Likewise.
7138 * dwarf2out.c (mem_loc_descriptor): Likewise.
7139 * emit-rtl.c (gen_lowpart_common): Likewise.
7140 * simplify-rtx.c (simplify_truncation): Likewise.
7141 (simplify_binary_operation_1): Likewise.
7142 (simplify_const_relational_operation): Likewise.
7143 (simplify_ternary_operation): Likewise.
7144 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
7146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7147 Alan Hayward <alan.hayward@arm.com>
7148 David Sherwood <david.sherwood@arm.com>
7150 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
7151 * reload.c (push_reload): Likewise.
7152 (find_reloads): Likewise.
7154 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7155 Alan Hayward <alan.hayward@arm.com>
7156 David Sherwood <david.sherwood@arm.com>
7158 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
7159 (make_compound_operation_int): Likewise.
7160 (change_zero_ext): Likewise.
7161 * expr.c (convert_move): Likewise.
7162 (convert_modes): Likewise.
7163 * fwprop.c (forward_propagate_subreg): Likewise.
7164 * loop-iv.c (get_biv_step_1): Likewise.
7165 * optabs.c (widen_operand): Likewise.
7166 * postreload.c (move2add_valid_value_p): Likewise.
7167 * recog.c (simplify_while_replacing): Likewise.
7168 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7169 (simplify_binary_operation_1): Likewise. Remove redundant
7170 mode equality check.
7172 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7173 Alan Hayward <alan.hayward@arm.com>
7174 David Sherwood <david.sherwood@arm.com>
7176 * combine.c (combine_simplify_rtx): Add checks for
7177 is_a <scalar_int_mode>.
7178 (simplify_if_then_else): Likewise.
7179 (make_field_assignment): Likewise.
7180 (simplify_comparison): Likewise.
7181 * ifcvt.c (noce_try_bitop): Likewise.
7182 * loop-invariant.c (canonicalize_address_mult): Likewise.
7183 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7186 Alan Hayward <alan.hayward@arm.com>
7187 David Sherwood <david.sherwood@arm.com>
7189 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
7190 is_a <scalar_int_mode> instead of != BLKmode.
7192 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7193 Alan Hayward <alan.hayward@arm.com>
7194 David Sherwood <david.sherwood@arm.com>
7196 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
7197 instead of != VOIDmode.
7198 * combine.c (if_then_else_cond): Likewise.
7199 (change_zero_ext): Likewise.
7200 * dwarf2out.c (mem_loc_descriptor): Likewise.
7201 (loc_descriptor): Likewise.
7202 * rtlanal.c (canonicalize_condition): Likewise.
7203 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
7205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7206 Alan Hayward <alan.hayward@arm.com>
7207 David Sherwood <david.sherwood@arm.com>
7209 * simplify-rtx.c (simplify_binary_operation_1): Use
7210 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
7212 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7213 Alan Hayward <alan.hayward@arm.com>
7214 David Sherwood <david.sherwood@arm.com>
7216 * wide-int.h (int_traits<unsigned char>) New class.
7217 (int_traits<unsigned short>) Likewise.
7218 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
7219 Use GET_MODE_UNIT_PRECISION and remove redundant test for
7221 * combine.c (set_nonzero_bits_and_sign_copies): Use
7222 is_a <scalar_int_mode>.
7223 (find_split_point): Likewise.
7224 (combine_simplify_rtx): Likewise.
7225 (simplify_logical): Likewise.
7226 (expand_compound_operation): Likewise.
7227 (expand_field_assignment): Likewise.
7228 (make_compound_operation): Likewise.
7229 (extended_count): Likewise.
7230 (change_zero_ext): Likewise.
7231 (simplify_comparison): Likewise.
7232 * dwarf2out.c (scompare_loc_descriptor): Likewise.
7233 (ucompare_loc_descriptor): Likewise.
7234 (minmax_loc_descriptor): Likewise.
7235 (mem_loc_descriptor): Likewise.
7236 (loc_descriptor): Likewise.
7237 * expmed.c (init_expmed_one_mode): Likewise.
7238 * lra-constraints.c (lra_constraint_offset): Likewise.
7239 * optabs.c (prepare_libcall_arg): Likewise.
7240 * postreload.c (move2add_note_store): Likewise.
7241 * reload.c (operands_match_p): Likewise.
7242 * rtl.h (load_extend_op): Likewise.
7243 * rtlhooks.c (gen_lowpart_general): Likewise.
7244 * simplify-rtx.c (simplify_truncation): Likewise.
7245 (simplify_unary_operation_1): Likewise.
7246 (simplify_binary_operation_1): Likewise.
7247 (simplify_const_binary_operation): Likewise.
7248 (simplify_const_relational_operation): Likewise.
7249 (simplify_subreg): Likewise.
7250 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7251 * var-tracking.c (adjust_mems): Likewise.
7252 (prepare_call_arguments): Likewise.
7254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7255 Alan Hayward <alan.hayward@arm.com>
7256 David Sherwood <david.sherwood@arm.com>
7258 * machmode.h (is_int_mode): New fuction.
7259 * combine.c (find_split_point): Use it.
7260 (combine_simplify_rtx): Likewise.
7261 (simplify_if_then_else): Likewise.
7262 (simplify_set): Likewise.
7263 (simplify_shift_const_1): Likewise.
7264 (simplify_comparison): Likewise.
7265 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
7266 * cse.c (notreg_cost): Likewise.
7267 (cse_insn): Likewise.
7268 * cselib.c (cselib_lookup_1): Likewise.
7269 * dojump.c (do_jump_1): Likewise.
7270 (do_compare_rtx_and_jump): Likewise.
7271 * dse.c (get_call_args): Likewise.
7272 * dwarf2out.c (rtl_for_decl_init): Likewise.
7273 (native_encode_initializer): Likewise.
7274 * expmed.c (emit_store_flag_1): Likewise.
7275 (emit_store_flag): Likewise.
7276 * expr.c (convert_modes): Likewise.
7277 (store_field): Likewise.
7278 (expand_expr_real_1): Likewise.
7279 * fold-const.c (fold_read_from_constant_string): Likewise.
7280 * gimple-ssa-sprintf.c (get_format_string): Likewise.
7281 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
7282 * optabs.c (expand_binop): Likewise.
7283 (expand_unop): Likewise.
7284 (expand_abs_nojump): Likewise.
7285 (expand_one_cmpl_abs_nojump): Likewise.
7286 * simplify-rtx.c (mode_signbit_p): Likewise.
7287 (val_signbit_p): Likewise.
7288 (val_signbit_known_set_p): Likewise.
7289 (val_signbit_known_clear_p): Likewise.
7290 (simplify_relational_operation_1): Likewise.
7291 * tree.c (vector_type_mode): Likewise.
7293 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7294 Alan Hayward <alan.hayward@arm.com>
7295 David Sherwood <david.sherwood@arm.com>
7297 * machmode.h (smallest_mode_for_size): Fix formatting.
7298 (smallest_int_mode_for_size): New function.
7299 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
7300 instead of smallest_mode_for_size.
7301 * combine.c (make_extraction): Likewise.
7302 * config/arc/arc.c (arc_expand_movmem): Likewise.
7303 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
7304 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7305 * config/s390/s390.c (s390_expand_insv): Likewise.
7306 * config/sparc/sparc.c (assign_int_registers): Likewise.
7307 * config/spu/spu.c (spu_function_value): Likewise.
7308 (spu_function_arg): Likewise.
7309 * coverage.c (get_gcov_type): Likewise.
7310 (get_gcov_unsigned_t): Likewise.
7311 * dse.c (find_shift_sequence): Likewise.
7312 * expmed.c (store_bit_field_1): Likewise.
7313 * expr.c (convert_move): Likewise.
7314 (store_field): Likewise.
7315 * internal-fn.c (expand_arith_overflow): Likewise.
7316 * optabs-query.c (get_best_extraction_insn): Likewise.
7317 * optabs.c (expand_twoval_binop_libfunc): Likewise.
7318 * stor-layout.c (layout_type): Likewise.
7319 (initialize_sizetypes): Likewise.
7320 * targhooks.c (default_get_mask_mode): Likewise.
7321 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
7323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7324 Alan Hayward <alan.hayward@arm.com>
7325 David Sherwood <david.sherwood@arm.com>
7327 * machmode.h (opt_mode::else_blk): New function.
7328 (int_mode_for_mode): Declare.
7329 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
7330 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
7332 * cfgexpand.c (expand_debug_expr): Likewise.
7333 * combine.c (gen_lowpart_or_truncate): Likewise.
7334 (gen_lowpart_for_combine): Likewise.
7335 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
7336 * config/avr/avr.c (avr_to_int_mode): Likewise.
7337 (avr_out_plus_1): Likewise.
7338 (avr_out_plus): Likewise.
7339 (avr_out_round): Likewise.
7340 * config/i386/i386.c (ix86_split_to_parts): Likewise.
7341 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7342 (s390_expand_vcond): Likewise.
7343 * config/spu/spu.c (spu_split_immediate): Likewise.
7344 (spu_expand_mov): Likewise.
7345 * dse.c (get_stored_val): Likewise.
7346 * expmed.c (store_bit_field_1): Likewise.
7347 (convert_extracted_bit_field): Use int_mode_for_mode instead of
7349 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
7350 (extract_low_bits): Likewise.
7351 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
7352 handling rather than repeating the check.
7353 (emit_group_store): Likewise.
7354 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
7355 * optabs.c (expand_absneg_bit): Likewise.
7356 (expand_copysign_absneg): Likewise.
7357 (expand_copysign_bit): Likewise.
7358 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
7359 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
7360 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
7361 * var-tracking.c (prepare_call_arguments): Likewise.
7362 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
7363 int_mode_for_mode instead of mode_for_size.
7364 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7366 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7367 Alan Hayward <alan.hayward@arm.com>
7368 David Sherwood <david.sherwood@arm.com>
7370 * machmode.h (int_mode_for_size): New function.
7371 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
7372 instead of mode_for_size.
7373 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
7375 * combine.c (expand_field_assignment): Use int_mode_for_size
7376 instead of mode_for_size.
7377 (make_extraction): Likewise.
7378 (simplify_shift_const_1): Likewise.
7379 (simplify_comparison): Likewise.
7380 * dojump.c (do_jump): Likewise.
7381 * dwarf2out.c (mem_loc_descriptor): Likewise.
7382 * emit-rtl.c (init_derived_machine_modes): Likewise.
7383 * expmed.c (flip_storage_order): Likewise.
7384 (convert_extracted_bit_field): Likewise.
7385 * expr.c (copy_blkmode_from_reg): Likewise.
7386 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
7387 * internal-fn.c (expand_mul_overflow): Likewise.
7388 * lower-subreg.c (simple_move): Likewise.
7389 * optabs-libfuncs.c (init_optabs): Likewise.
7390 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
7391 * tree.c (vector_type_mode): Likewise.
7392 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
7393 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
7394 * tree-vect-generic.c (expand_vector_parallel): Likewise.
7395 * tree-vect-stmts.c (vectorizable_load): Likewise.
7396 (vectorizable_store): Likewise.
7398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7399 Alan Hayward <alan.hayward@arm.com>
7400 David Sherwood <david.sherwood@arm.com>
7402 * coretypes.h (pod_mode): New type.
7403 (scalar_int_mode_pod): New typedef.
7404 * machmode.h (pod_mode): New class.
7405 (int_n_data_t::m): Change type to scalar_int_mode_pod.
7406 * genmodes.c (emit_mode_int_n): Update accordingly.
7407 * lower-subreg.h (target_lower_subreg): Change type to
7408 scalar_int_mode_pod.
7409 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
7410 scalar_int_mode_pod.
7412 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7413 Alan Hayward <alan.hayward@arm.com>
7414 David Sherwood <david.sherwood@arm.com>
7416 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
7417 machine_mode to scalar_int_mode.
7418 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
7419 (rs6000_option_override_internal): Remove cast to int.
7420 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
7421 machine_mode to scalar_int_mode.
7422 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
7423 (rs6000_option_override_internal): Remove cast to int.
7424 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
7425 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
7427 * config/s390/s390.c (s390_expand_builtin): Likewise.
7428 * coretypes.h (scalar_int_mode): New type.
7429 (opt_scalar_int_mode): New typedef.
7430 * machmode.h (scalar_int_mode): New class.
7431 (scalar_int_mode::includes_p): New function.
7432 (byte_mode): Change type to scalar_int_mode.
7433 (word_mode): Likewise.
7434 (ptr_mode): Likewise.
7435 * emit-rtl.c (byte_mode): Likewise.
7436 (word_mode): Likewise.
7437 (ptr_mode): Likewise.
7438 (init_derived_machine_modes): Update accordingly.
7439 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
7440 and MODE_PARTIAL_INT.
7441 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
7442 opt_scalar_int_mode.
7444 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7445 Alan Hayward <alan.hayward@arm.com>
7446 David Sherwood <david.sherwood@arm.com>
7448 * target.def (libgcc_floating_mode_supported_p): Take a
7450 * doc/tm.texi: Regenerate.
7451 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
7453 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
7454 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
7457 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7458 Alan Hayward <alan.hayward@arm.com>
7459 David Sherwood <david.sherwood@arm.com>
7461 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
7462 * doc/tm.texi: Regenerate.
7463 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
7464 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
7465 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
7466 * targhooks.h (default_floatn_mode): Likewise.
7467 * targhooks.c (default_floatn_mode): Likewise.
7468 * tree.c (build_common_tree_nodes): Update accordingly.
7470 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7471 Alan Hayward <alan.hayward@arm.com>
7472 David Sherwood <david.sherwood@arm.com>
7474 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
7475 (mode_iterator::iterate_p): Likewise.
7476 (mode_iterator::get_wider): Likewise.
7477 * expr.c (init_expr_target): Use opt_scalar_float_mode.
7479 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7480 Alan Hayward <alan.hayward@arm.com>
7481 David Sherwood <david.sherwood@arm.com>
7483 * coretypes.h (opt_scalar_float_mode): New typedef.
7484 * machmode.h (float_mode_for_size): New function.
7485 * emit-rtl.c (double_mode): Delete.
7486 (init_emit_once): Use float_mode_for_size.
7487 * stor-layout.c (layout_type): Likewise.
7488 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
7490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7491 Alan Hayward <alan.hayward@arm.com>
7492 David Sherwood <david.sherwood@arm.com>
7494 * output.h (assemble_real): Take a scalar_float_mode.
7495 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
7496 * config/arm/arm.md (consttable_4): Likewise.
7497 (consttable_8): Likewise.
7498 (consttable_16): Likewise.
7499 * config/mips/mips.md (consttable_float): Likewise.
7500 * config/s390/s390.c (s390_output_pool_entry): Likewise.
7501 * varasm.c (assemble_real): Take a scalar_float_mode.
7502 (output_constant_pool_2): Update accordingly.
7503 (output_constant): Likewise.
7505 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7506 Alan Hayward <alan.hayward@arm.com>
7507 David Sherwood <david.sherwood@arm.com>
7509 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
7510 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
7511 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
7512 (native_encode_real): Likewise.
7513 (native_interpret_real): Likewise.
7514 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
7515 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7518 Alan Hayward <alan.hayward@arm.com>
7519 David Sherwood <david.sherwood@arm.com>
7521 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
7522 <scalar_float_mode>. Simplify.
7523 (gen_extend_conv_libfunc): Likewise.
7525 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7526 Alan Hayward <alan.hayward@arm.com>
7527 David Sherwood <david.sherwood@arm.com>
7529 * coretypes.h (scalar_float_mode): New type.
7530 * machmode.h (mode_traits::from_int): Use machine_mode if
7531 USE_ENUM_MODES is defined.
7532 (is_a): New function.
7534 (dyn_cast): Likewise.
7535 (scalar_float_mode): New class.
7536 (scalar_float_mode::includes_p): New function.
7537 (is_float_mode): Likewise.
7538 * gdbhooks.py (MachineModePrinter): New class.
7539 (build_pretty_printer): Use it for scalar_float_mode.
7540 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
7541 (format_helper::format_helper): Turn into a template.
7542 * genmodes.c (get_mode_class): New function.
7543 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
7544 or machine_mode if none.
7545 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
7546 as_a <scalar_float_mode>.
7547 * dwarf2out.c (mem_loc_descriptor): Likewise.
7548 (insert_float): Likewise.
7549 (add_const_value_attribute): Likewise.
7550 * simplify-rtx.c (simplify_immed_subreg): Likewise.
7551 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
7552 (expand_unop): Update accordingly.
7553 (expand_abs_nojump): Likewise.
7554 (expand_copysign_absneg): Take a scalar_float_mode.
7555 (expand_copysign_bit): Likewise.
7556 (expand_copysign): Update accordingly.
7558 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7559 Alan Hayward <alan.hayward@arm.com>
7560 David Sherwood <david.sherwood@arm.com>
7562 * coretypes.h (opt_mode): New class.
7563 * machmode.h (opt_mode): Likewise.
7564 (opt_mode::else_void): New function.
7565 (opt_mode::require): Likewise.
7566 (opt_mode::exists): Likewise.
7567 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
7568 (GET_MODE_2XWIDER_MODE): Likewise.
7569 (mode_iterator::get_wider): Update accordingly.
7570 (mode_iterator::get_2xwider): Likewise.
7571 (mode_iterator::get_known_wider): Likewise, turning into a template.
7572 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
7573 forcing a wider mode to exist.
7574 * config/cr16/cr16.h (LONG_REG_P): Likewise.
7575 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7576 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
7577 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
7578 * lower-subreg.c (init_lower_subreg): Likewise.
7579 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
7580 on the final iteration.
7581 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
7582 a wider mode exists before asking for a move pattern.
7583 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
7584 forcing a wider mode to exist.
7585 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
7586 returning false if no such mode exists.
7587 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
7588 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
7589 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
7590 Avoid checking for a MODE_INT if we already know the mode is not a
7592 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
7593 forcing a wider mode to exist.
7594 (expmed_mult_highpart_optab): Likewise.
7595 (expmed_mult_highpart): Likewise.
7596 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
7598 * lto-streamer-in.c (lto_input_mode_table): Likewise.
7599 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
7600 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7601 * internal-fn.c (expand_mul_overflow): Update use of
7602 GET_MODE_2XWIDER_MODE.
7603 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7604 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
7605 GET_MODE_WIDER_MODE.
7606 (convert_plusminus_to_widen): Likewise.
7607 * tree-switch-conversion.c (array_value_type): Likewise.
7608 * var-tracking.c (emit_note_insn_var_location): Likewise.
7609 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
7610 Return false inside rather than outside the loop if no wider mode
7612 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
7613 and GET_MODE_2XWIDER_MODE
7614 (can_compare_p): Use else_void.
7615 * gdbhooks.py (OptMachineModePrinter): New class.
7616 (build_pretty_printer): Use it for opt_mode.
7618 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7619 Alan Hayward <alan.hayward@arm.com>
7620 David Sherwood <david.sherwood@arm.com>
7622 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
7623 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
7625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7626 Alan Hayward <alan.hayward@arm.com>
7627 David Sherwood <david.sherwood@arm.com>
7629 * machmode.h (mode_traits): New structure.
7630 (get_narrowest_mode): New function.
7631 (mode_iterator::start): Likewise.
7632 (mode_iterator::iterate_p): Likewise.
7633 (mode_iterator::get_wider): Likewise.
7634 (mode_iterator::get_known_wider): Likewise.
7635 (mode_iterator::get_2xwider): Likewise.
7636 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
7637 (FOR_EACH_MODE): Likewise.
7638 (FOR_EACH_MODE_FROM): Likewise.
7639 (FOR_EACH_MODE_UNTIL): Likewise.
7640 (FOR_EACH_WIDER_MODE): Likewise.
7641 (FOR_EACH_2XWIDER_MODE): Likewise.
7642 * builtins.c (expand_builtin_strlen): Use new mode iterators.
7643 * combine.c (simplify_comparison): Likewise
7644 * config/i386/i386.c (type_natural_mode): Likewise.
7645 * cse.c (cse_insn): Likewise.
7646 * dse.c (find_shift_sequence): Likewise.
7647 * emit-rtl.c (init_derived_machine_modes): Likewise.
7648 (init_emit_once): Likewise.
7649 * explow.c (hard_function_value): Likewise.
7650 * expmed.c (extract_fixed_bit_field_1): Likewise.
7651 (extract_bit_field_1): Likewise.
7652 (expand_divmod): Likewise.
7653 (emit_store_flag_1): Likewise.
7654 * expr.c (init_expr_target): Likewise.
7655 (convert_move): Likewise.
7656 (alignment_for_piecewise_move): Likewise.
7657 (widest_int_mode_for_size): Likewise.
7658 (emit_block_move_via_movmem): Likewise.
7659 (copy_blkmode_to_reg): Likewise.
7660 (set_storage_via_setmem): Likewise.
7661 (compress_float_constant): Likewise.
7662 * omp-low.c (omp_clause_aligned_alignment): Likewise.
7663 * optabs-query.c (get_best_extraction_insn): Likewise.
7664 * optabs.c (expand_binop): Likewise.
7665 (expand_twoval_unop): Likewise.
7666 (expand_twoval_binop): Likewise.
7667 (widen_leading): Likewise.
7668 (widen_bswap): Likewise.
7669 (expand_parity): Likewise.
7670 (expand_unop): Likewise.
7671 (prepare_cmp_insn): Likewise.
7672 (prepare_float_lib_cmp): Likewise.
7673 (expand_float): Likewise.
7674 (expand_fix): Likewise.
7675 (expand_sfix_optab): Likewise.
7676 * postreload.c (move2add_use_add2_insn): Likewise.
7677 * reg-stack.c (reg_to_stack): Likewise.
7678 * reginfo.c (choose_hard_reg_mode): Likewise.
7679 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
7680 * stor-layout.c (mode_for_size): Likewise.
7681 (smallest_mode_for_size): Likewise.
7682 (mode_for_vector): Likewise.
7683 (finish_bitfield_representative): Likewise.
7684 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
7685 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
7686 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
7687 * var-tracking.c (prepare_call_arguments): Likewise.
7689 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7690 Alan Hayward <alan.hayward@arm.com>
7691 David Sherwood <david.sherwood@arm.com>
7693 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
7694 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
7695 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
7696 * machmode.h (mode_size): Move earlier in file.
7697 (mode_precision): Likewise.
7698 (mode_inner): Likewise.
7699 (mode_nunits): Likewise.
7700 (mode_unit_size): Likewise.
7701 (unit_unit_precision): Likewise.
7702 (mode_wider): Likewise.
7703 (mode_2xwider): Likewise.
7704 (machine_mode): New class.
7705 (mode_to_bytes): New function.
7706 (mode_to_bits): Likewise.
7707 (mode_to_precision): Likewise.
7708 (mode_to_inner): Likewise.
7709 (mode_to_unit_size): Likewise.
7710 (mode_to_unit_precision): Likewise.
7711 (mode_to_nunits): Likewise.
7712 (GET_MODE_SIZE): Use mode_to_bytes.
7713 (GET_MODE_BITSIZE): Use mode_to_bits.
7714 (GET_MODE_PRECISION): Use mode_to_precision.
7715 (GET_MODE_INNER): Use mode_to_inner.
7716 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
7717 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
7718 (GET_MODE_NUNITS): Use mode_to_nunits.
7719 * system.h (ALWAYS_INLINE): New macro.
7720 * config/powerpcspe/powerpcspe-c.c
7721 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
7722 int for arg1_mode and arg2_mode.
7724 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
7725 Alan Hayward <alan.hayward@arm.com>
7726 David Sherwood <david.sherwood@arm.com>
7728 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
7729 Prefix mode names with E_ in case statements.
7730 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7731 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
7732 (aarch64_split_simd_move): Likewise.
7733 (aarch64_gen_storewb_pair): Likewise.
7734 (aarch64_gen_loadwb_pair): Likewise.
7735 (aarch64_gen_store_pair): Likewise.
7736 (aarch64_gen_load_pair): Likewise.
7737 (aarch64_get_condition_code_1): Likewise.
7738 (aarch64_constant_pool_reload_icode): Likewise.
7739 (get_rsqrte_type): Likewise.
7740 (get_rsqrts_type): Likewise.
7741 (get_recpe_type): Likewise.
7742 (get_recps_type): Likewise.
7743 (aarch64_gimplify_va_arg_expr): Likewise.
7744 (aarch64_simd_container_mode): Likewise.
7745 (aarch64_emit_load_exclusive): Likewise.
7746 (aarch64_emit_store_exclusive): Likewise.
7747 (aarch64_expand_compare_and_swap): Likewise.
7748 (aarch64_gen_atomic_cas): Likewise.
7749 (aarch64_emit_bic): Likewise.
7750 (aarch64_emit_atomic_swap): Likewise.
7751 (aarch64_emit_atomic_load_op): Likewise.
7752 (aarch64_evpc_trn): Likewise.
7753 (aarch64_evpc_uzp): Likewise.
7754 (aarch64_evpc_zip): Likewise.
7755 (aarch64_evpc_ext): Likewise.
7756 (aarch64_evpc_rev): Likewise.
7757 (aarch64_evpc_dup): Likewise.
7758 (aarch64_gen_ccmp_first): Likewise.
7759 (aarch64_gen_ccmp_next): Likewise.
7760 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
7761 (alpha_emit_xfloating_libcall): Likewise.
7762 (emit_insxl): Likewise.
7763 (alpha_arg_type): Likewise.
7764 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
7765 (arc_preferred_simd_mode): Likewise.
7766 (arc_secondary_reload): Likewise.
7767 (get_arc_condition_code): Likewise.
7768 (arc_print_operand): Likewise.
7769 (arc_legitimate_constant_p): Likewise.
7770 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7771 * config/arc/arc.md (casesi_load): Likewise.
7772 (casesi_compact_jump): Likewise.
7773 * config/arc/predicates.md (proper_comparison_operator): Likewise.
7774 (cc_use_register): Likewise.
7775 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7776 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
7777 (arm_init_iwmmxt_builtins): Likewise.
7778 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
7779 (neon_expand_vector_init): Likewise.
7780 (arm_attr_length_move_neon): Likewise.
7781 (maybe_get_arm_condition_code): Likewise.
7782 (arm_emit_vector_const): Likewise.
7783 (arm_preferred_simd_mode): Likewise.
7784 (arm_output_iwmmxt_tinsr): Likewise.
7785 (thumb1_output_casesi): Likewise.
7786 (thumb2_output_casesi): Likewise.
7787 (arm_emit_load_exclusive): Likewise.
7788 (arm_emit_store_exclusive): Likewise.
7789 (arm_expand_compare_and_swap): Likewise.
7790 (arm_evpc_neon_vuzp): Likewise.
7791 (arm_evpc_neon_vzip): Likewise.
7792 (arm_evpc_neon_vrev): Likewise.
7793 (arm_evpc_neon_vtrn): Likewise.
7794 (arm_evpc_neon_vext): Likewise.
7795 (arm_validize_comparison): Likewise.
7796 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
7797 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
7798 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
7799 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
7800 (c6x_preferred_simd_mode): Likewise.
7801 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
7802 (epiphany_rtx_costs): Likewise.
7803 * config/epiphany/predicates.md (proper_comparison_operator):
7805 * config/frv/frv.c (condexec_memory_operand): Likewise.
7806 (frv_emit_move): Likewise.
7807 (output_move_single): Likewise.
7808 (output_condmove_single): Likewise.
7809 (frv_hard_regno_mode_ok): Likewise.
7810 (frv_matching_accg_mode): Likewise.
7811 * config/h8300/h8300.c (split_adds_subs): Likewise.
7812 (h8300_rtx_costs): Likewise.
7813 (h8300_print_operand): Likewise.
7814 (compute_mov_length): Likewise.
7815 (output_logical_op): Likewise.
7816 (compute_logical_op_length): Likewise.
7817 (compute_logical_op_cc): Likewise.
7818 (h8300_shift_needs_scratch_p): Likewise.
7819 (output_a_shift): Likewise.
7820 (compute_a_shift_length): Likewise.
7821 (compute_a_shift_cc): Likewise.
7822 (expand_a_rotate): Likewise.
7823 (output_a_rotate): Likewise.
7824 * config/i386/i386.c (classify_argument): Likewise.
7825 (function_arg_advance_32): Likewise.
7826 (function_arg_32): Likewise.
7827 (function_arg_64): Likewise.
7828 (function_value_64): Likewise.
7829 (ix86_gimplify_va_arg): Likewise.
7830 (ix86_legitimate_constant_p): Likewise.
7831 (put_condition_code): Likewise.
7832 (split_double_mode): Likewise.
7833 (ix86_avx256_split_vector_move_misalign): Likewise.
7834 (ix86_expand_vector_logical_operator): Likewise.
7835 (ix86_split_idivmod): Likewise.
7836 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
7837 (ix86_build_const_vector): Likewise.
7838 (ix86_build_signbit_mask): Likewise.
7839 (ix86_match_ccmode): Likewise.
7840 (ix86_cc_modes_compatible): Likewise.
7841 (ix86_expand_branch): Likewise.
7842 (ix86_expand_sse_cmp): Likewise.
7843 (ix86_expand_sse_movcc): Likewise.
7844 (ix86_expand_int_sse_cmp): Likewise.
7845 (ix86_expand_vec_perm_vpermi2): Likewise.
7846 (ix86_expand_vec_perm): Likewise.
7847 (ix86_expand_sse_unpack): Likewise.
7848 (ix86_expand_int_addcc): Likewise.
7849 (ix86_split_to_parts): Likewise.
7850 (ix86_vectorize_builtin_gather): Likewise.
7851 (ix86_vectorize_builtin_scatter): Likewise.
7852 (avx_vpermilp_parallel): Likewise.
7853 (inline_memory_move_cost): Likewise.
7854 (ix86_tieable_integer_mode_p): Likewise.
7855 (x86_maybe_negate_const_int): Likewise.
7856 (ix86_expand_vector_init_duplicate): Likewise.
7857 (ix86_expand_vector_init_one_nonzero): Likewise.
7858 (ix86_expand_vector_init_one_var): Likewise.
7859 (ix86_expand_vector_init_concat): Likewise.
7860 (ix86_expand_vector_init_interleave): Likewise.
7861 (ix86_expand_vector_init_general): Likewise.
7862 (ix86_expand_vector_set): Likewise.
7863 (ix86_expand_vector_extract): Likewise.
7864 (emit_reduc_half): Likewise.
7865 (ix86_emit_i387_round): Likewise.
7866 (ix86_mangle_type): Likewise.
7867 (ix86_expand_round_sse4): Likewise.
7868 (expand_vec_perm_blend): Likewise.
7869 (canonicalize_vector_int_perm): Likewise.
7870 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
7871 (expand_vec_perm_1): Likewise.
7872 (expand_vec_perm_interleave3): Likewise.
7873 (expand_vec_perm_even_odd_pack): Likewise.
7874 (expand_vec_perm_even_odd_1): Likewise.
7875 (expand_vec_perm_broadcast_1): Likewise.
7876 (ix86_vectorize_vec_perm_const_ok): Likewise.
7877 (ix86_expand_vecop_qihi): Likewise.
7878 (ix86_expand_mul_widen_hilo): Likewise.
7879 (ix86_expand_sse2_abs): Likewise.
7880 (ix86_expand_pextr): Likewise.
7881 (ix86_expand_pinsr): Likewise.
7882 (ix86_preferred_simd_mode): Likewise.
7883 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
7884 * config/i386/sse.md (*andnot<mode>3): Likewise.
7885 (<mask_codefor><code><mode>3<mask_name>): Likewise.
7886 (*<code><mode>3): Likewise.
7887 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
7888 (ia64_expand_atomic_op): Likewise.
7889 (ia64_arg_type): Likewise.
7890 (ia64_mode_to_int): Likewise.
7891 (ia64_scalar_mode_supported_p): Likewise.
7892 (ia64_vector_mode_supported_p): Likewise.
7893 (expand_vec_perm_broadcast): Likewise.
7894 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
7895 (iq2000_function_arg_advance): Likewise.
7896 (iq2000_function_arg): Likewise.
7897 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
7898 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
7899 (m68k_libcall_value): Likewise.
7900 (m68k_function_value): Likewise.
7901 (sched_attr_op_type): Likewise.
7902 * config/mcore/mcore.c (mcore_output_move): Likewise.
7903 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
7905 (microblaze_function_arg): Likewise.
7906 * config/mips/mips.c (mips16_build_call_stub): Likewise.
7907 (mips_print_operand): Likewise.
7908 (mips_mode_ok_for_mov_fmt_p): Likewise.
7909 (mips_vector_mode_supported_p): Likewise.
7910 (mips_preferred_simd_mode): Likewise.
7911 (mips_expand_vpc_loongson_even_odd): Likewise.
7912 (mips_expand_vec_unpack): Likewise.
7913 (mips_expand_vi_broadcast): Likewise.
7914 (mips_expand_vector_init): Likewise.
7915 (mips_expand_vec_reduc): Likewise.
7916 (mips_expand_msa_cmp): Likewise.
7917 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
7918 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
7919 (cc_flags_for_mode): Likewise.
7920 * config/msp430/msp430.c (msp430_print_operand): Likewise.
7921 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
7922 (nds32_output_casesi_pc_relative): Likewise.
7923 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
7924 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
7925 (nvptx_gen_unpack): Likewise.
7926 (nvptx_gen_pack): Likewise.
7927 (nvptx_gen_shuffle): Likewise.
7928 (nvptx_gen_wcast): Likewise.
7929 (nvptx_preferred_simd_mode): Likewise.
7930 * config/pa/pa.c (pa_secondary_reload): Likewise.
7931 * config/pa/predicates.md (base14_operand): Likewise.
7932 * config/powerpcspe/powerpcspe-c.c
7933 (altivec_resolve_overloaded_builtin): Likewise.
7934 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
7936 (rs6000_preferred_simd_mode): Likewise.
7937 (output_vec_const_move): Likewise.
7938 (rs6000_expand_vector_extract): Likewise.
7939 (rs6000_split_vec_extract_var): Likewise.
7940 (reg_offset_addressing_ok_p): Likewise.
7941 (rs6000_legitimate_offset_address_p): Likewise.
7942 (rs6000_legitimize_address): Likewise.
7943 (rs6000_emit_set_const): Likewise.
7944 (rs6000_const_vec): Likewise.
7945 (rs6000_emit_move): Likewise.
7946 (spe_build_register_parallel): Likewise.
7947 (rs6000_darwin64_record_arg_recurse): Likewise.
7948 (swap_selector_for_mode): Likewise.
7949 (spe_init_builtins): Likewise.
7950 (paired_init_builtins): Likewise.
7951 (altivec_init_builtins): Likewise.
7952 (do_load_for_compare): Likewise.
7953 (rs6000_generate_compare): Likewise.
7954 (rs6000_expand_float128_convert): Likewise.
7955 (emit_load_locked): Likewise.
7956 (emit_store_conditional): Likewise.
7957 (rs6000_output_function_epilogue): Likewise.
7958 (rs6000_handle_altivec_attribute): Likewise.
7959 (rs6000_function_value): Likewise.
7960 (emit_fusion_gpr_load): Likewise.
7961 (emit_fusion_p9_load): Likewise.
7962 (emit_fusion_p9_store): Likewise.
7963 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
7964 (fusion_gpr_mem_load): Likewise.
7965 (fusion_addis_mem_combo_load): Likewise.
7966 (fusion_addis_mem_combo_store): Likewise.
7967 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
7968 (fusion_gpr_mem_load): Likewise.
7969 (fusion_addis_mem_combo_load): Likewise.
7970 (fusion_addis_mem_combo_store): Likewise.
7971 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
7973 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
7974 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
7975 (rs6000_preferred_simd_mode): Likewise.
7976 (output_vec_const_move): Likewise.
7977 (rs6000_expand_vector_extract): Likewise.
7978 (rs6000_split_vec_extract_var): Likewise.
7979 (reg_offset_addressing_ok_p): Likewise.
7980 (rs6000_legitimate_offset_address_p): Likewise.
7981 (rs6000_legitimize_address): Likewise.
7982 (rs6000_emit_set_const): Likewise.
7983 (rs6000_const_vec): Likewise.
7984 (rs6000_emit_move): Likewise.
7985 (rs6000_darwin64_record_arg_recurse): Likewise.
7986 (swap_selector_for_mode): Likewise.
7987 (paired_init_builtins): Likewise.
7988 (altivec_init_builtins): Likewise.
7989 (rs6000_expand_float128_convert): Likewise.
7990 (emit_load_locked): Likewise.
7991 (emit_store_conditional): Likewise.
7992 (rs6000_output_function_epilogue): Likewise.
7993 (rs6000_handle_altivec_attribute): Likewise.
7994 (rs6000_function_value): Likewise.
7995 (emit_fusion_gpr_load): Likewise.
7996 (emit_fusion_p9_load): Likewise.
7997 (emit_fusion_p9_store): Likewise.
7998 * config/rx/rx.c (rx_gen_move_template): Likewise.
7999 (flags_from_mode): Likewise.
8000 * config/s390/predicates.md (s390_alc_comparison): Likewise.
8001 (s390_slb_comparison): Likewise.
8002 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
8003 (s390_vector_mode_supported_p): Likewise.
8004 (s390_cc_modes_compatible): Likewise.
8005 (s390_match_ccmode_set): Likewise.
8006 (s390_canonicalize_comparison): Likewise.
8007 (s390_emit_compare_and_swap): Likewise.
8008 (s390_branch_condition_mask): Likewise.
8009 (s390_rtx_costs): Likewise.
8010 (s390_secondary_reload): Likewise.
8011 (__SECONDARY_RELOAD_CASE): Likewise.
8012 (s390_expand_cs): Likewise.
8013 (s390_preferred_simd_mode): Likewise.
8014 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
8015 * config/sh/sh.c (sh_print_operand): Likewise.
8016 (dump_table): Likewise.
8017 (sh_secondary_reload): Likewise.
8018 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8019 * config/sh/sh.md (casesi_worker_1): Likewise.
8020 (casesi_worker_2): Likewise.
8021 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
8022 (fcc_comparison_operator): Likewise.
8023 * config/sparc/sparc.c (sparc_expand_move): Likewise.
8024 (emit_soft_tfmode_cvt): Likewise.
8025 (sparc_preferred_simd_mode): Likewise.
8026 (output_cbranch): Likewise.
8027 (sparc_print_operand): Likewise.
8028 (sparc_expand_vec_perm_bmask): Likewise.
8029 (vector_init_bshuffle): Likewise.
8030 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
8031 (spu_vector_mode_supported_p): Likewise.
8032 (spu_expand_insv): Likewise.
8033 (spu_emit_branch_or_set): Likewise.
8034 (spu_handle_vector_attribute): Likewise.
8035 (spu_builtin_splats): Likewise.
8036 (spu_builtin_extract): Likewise.
8037 (spu_builtin_promote): Likewise.
8038 (spu_expand_sign_extend): Likewise.
8039 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
8040 (tilegx_simd_int): Likewise.
8041 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
8042 (tilepro_simd_int): Likewise.
8043 * config/v850/v850.c (const_double_split): Likewise.
8044 (v850_print_operand): Likewise.
8045 (ep_memory_offset): Likewise.
8046 * config/vax/vax.c (vax_rtx_costs): Likewise.
8047 (vax_output_int_move): Likewise.
8048 (vax_output_int_add): Likewise.
8049 (vax_output_int_subtract): Likewise.
8050 * config/visium/predicates.md (visium_branch_operator): Likewise.
8051 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
8052 (visium_print_operand_address): Likewise.
8053 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
8054 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
8055 (xtensa_expand_conditional_branch): Likewise.
8056 (xtensa_copy_incoming_a7): Likewise.
8057 (xtensa_output_literal): Likewise.
8058 * dfp.c (decimal_real_maxval): Likewise.
8059 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
8061 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8062 Alan Hayward <alan.hayward@arm.com>
8063 David Sherwood <david.sherwood@arm.com>
8065 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
8066 (mode_nunits_inline): Likewise.
8067 (mode_inner_inline): Likewise.
8068 (mode_unit_size_inline): Likewise.
8069 (mode_unit_precision_inline): Likewise.
8070 (emit_insn_modes_h): Likewise. Also emit a #define of the
8072 (emit_mode_wider): Add an E_ prefix to mode names.
8073 (emit_mode_complex): Likewise.
8074 (emit_mode_inner): Likewise.
8075 (emit_mode_adjustments): Likewise.
8076 (emit_mode_int_n): Likewise.
8077 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
8078 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
8079 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
8080 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
8081 (CRC32_BUILTIN, ENTRY): Likewise.
8082 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
8083 (aarch64_pop_regs): Likewise.
8084 (aarch64_process_components): Likewise.
8085 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
8086 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
8087 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
8088 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
8089 * config/arm/arm.c (arm_init_libfuncs): Likewise.
8090 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
8092 * config/i386/i386-builtin.def (pcmpestr): Likewise.
8093 (pcmpistr): Likewise.
8094 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
8095 * config/mmix/mmix.c (mmix_output_condition): Likewise.
8096 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
8098 * config/rl78/rl78.c (mduc_regs): Likewise.
8099 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8100 (htm_expand_builtin): Likewise.
8101 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
8102 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
8103 * config/xtensa/xtensa.c (print_operand): Likewise.
8104 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
8105 (NUM_MODE_VECTOR_INT): Likewise.
8106 * genoutput.c (null_operand): Likewise.
8107 (output_operand_data): Likewise.
8108 * genrecog.c (print_parameter_value): Likewise.
8109 * lra.c (debug_operand_data): Likewise.
8111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8113 * dwarf2out.c (scompare_loc_descriptor_wide)
8114 (scompare_loc_descriptor_narrow): New functions, split out from...
8115 (scompare_loc_descriptor): ...here.
8116 * expmed.c (emit_store_flag_int): New function, split out from...
8117 (emit_store_flag): ...here.
8119 2017-08-30 Richard Biener <rguenther@suse.de>
8121 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
8122 (dwarf2out_early_finish): Move setting of AT_pubnames from
8123 early debug output to early finish.
8125 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8127 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
8128 and -mdata-region to the assembler.
8130 2017-08-30 Richard Biener <rguenther@suse.de>
8132 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
8134 (gen_subprogram_die): Add DW_AT_object_pointer only early.
8135 (dwarf2out_early_global_decl): Only generate a DIE for the
8136 abstract origin if it doesn't already exist or is a declaration DIE.
8137 (resolve_addr): Do not add the linkage name twice when
8138 generating a stub DIE for the DW_TAG_GNU_call_site target.
8140 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8142 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
8143 Use machine_mode rather than int for arg1_mode.
8145 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
8148 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
8149 that the second argument of the built-in functions to unpack
8150 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
8151 switch statement instead a lot of if statements.
8152 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
8153 Allow 64-bit values to be in Altivec registers as well as
8154 traditional floating point registers.
8155 (pack<mode>, FMOVE128_VSX iterator): Likewise.
8157 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
8159 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
8160 MAX_REGS_PER_ADDRESS == 1.
8162 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
8164 * config/i386/i386.opt (flag_fentry): Do not init to -1.
8165 * config/i386/i386.c (ix86_option_override_internal): Simplify
8166 setting of opts->x_flag_entry.
8168 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8169 Jakub Jelinek <jakub@redhat.com>
8170 Richard Biener <rguenther@suse.de>
8172 PR tree-optimization/81503
8173 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
8174 folded constant fits in the target type; reorder tests for clarity.
8176 2017-08-29 Martin Liska <mliska@suse.cz>
8178 * passes.def: Include pass_lower_switch.
8179 * stmt.c (dump_case_nodes): Remove and move to
8180 tree-switch-conversion.
8181 (case_values_threshold): Likewise.
8182 (expand_switch_as_decision_tree_p): Likewise.
8183 (emit_case_decision_tree): Likewise.
8184 (expand_case): Likewise.
8185 (balance_case_nodes): Likewise.
8186 (node_has_low_bound): Likewise.
8187 (node_has_high_bound): Likewise.
8188 (node_is_bounded): Likewise.
8189 (emit_case_nodes): Likewise.
8190 (struct simple_case_node): New struct.
8191 (add_case_node): Remove.
8192 (emit_case_dispatch_table): Use vector instead of case_list.
8193 (reset_out_edges_aux): Remove.
8194 (compute_cases_per_edge): Likewise.
8195 (expand_case): Build list of simple_case_node.
8196 (expand_sjlj_dispatch_table): Use it.
8197 * tree-switch-conversion.c (struct case_node): Moved from
8198 stmt.c and adjusted.
8199 (emit_case_nodes): Likewise.
8200 (node_has_low_bound): Likewise.
8201 (node_has_high_bound): Likewise.
8202 (node_is_bounded): Likewise.
8203 (case_values_threshold): Likewise.
8204 (reset_out_edges_aux): Likewise.
8205 (compute_cases_per_edge): Likewise.
8206 (add_case_node): Likewise.
8207 (dump_case_nodes): Likewise.
8208 (balance_case_nodes): Likewise.
8209 (expand_switch_as_decision_tree_p): Likewise.
8210 (emit_jump): Likewise.
8211 (emit_case_decision_tree): Likewise.
8212 (try_switch_expansion): Likewise.
8213 (do_jump_if_equal): Likewise.
8214 (emit_cmp_and_jump_insns): Likewise.
8215 (fix_phi_operands_for_edge): New function.
8216 (record_phi_operand_mapping): Likewise.
8217 (class pass_lower_switch): New pass.
8218 (pass_lower_switch::execute): New function.
8219 (make_pass_lower_switch): Likewise.
8220 (conditional_probability):
8221 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
8222 * tree-pass.h: Add make_pass_lower_switch.
8224 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
8227 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
8230 2017-08-29 Richard Biener <rguenther@suse.de>
8232 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
8233 we do not add a DW_AT_inline attribute twice.
8234 (gen_subprogram_die): Remove code setting DW_AT_inline on
8235 DECL_ABSTRACT_P nodes.
8237 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
8239 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
8240 calls to internal functions.
8241 (gimplify_modify_expr): Likewise.
8242 * tree-call-cdce.c (use_internal_fn): Likewise.
8243 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
8244 (convert_to_divmod): Set the nothrow flag.
8245 * tree-if-conv.c (predicate_mem_writes): Likewise.
8246 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
8247 (vectorizable_call): Likewise.
8248 (vectorizable_store): Likewise.
8249 (vectorizable_load): Likewise.
8250 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
8251 (vect_recog_mask_conversion_pattern): Likewise.
8253 2017-08-29 Martin Liska <mliska@suse.cz>
8256 * gcc.c (driver_handle_option): Add new argument.
8257 * opts-common.c (handle_option): Pass
8258 target_option_override_hook.
8259 * opts-global.c (lang_handle_option): Add new option.
8260 (set_default_handlers): Add new argument.
8261 (decode_options): Likewise.
8262 * opts.c (target_handle_option): Likewise.
8263 (common_handle_option): Call target_option_override_hook.
8264 * opts.h (struct cl_option_handler_func): Add hook for
8265 target option override.
8266 (struct cl_option_handlers): Likewise.
8267 (set_default_handlers): Add new argument.
8268 (decode_options): Likewise.
8269 (common_handle_option): Likewise.
8270 (target_handle_option): Likewise.
8271 * toplev.c (toplev::main): Pass targetm.target_option.override
8274 2017-08-29 Richard Biener <rguenther@suse.de>
8275 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
8277 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
8278 life to the active subtree.
8280 2017-08-28 Jeff Law <law@redhat.com>
8282 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
8283 derive_equivalences.
8284 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
8286 (edge_info::derive_equivalences): New private member function
8288 * tree-ssa-dom.c (class edge_info): Changed from a struct
8289 to a class. Add ctor/dtor, methods and data members.
8290 (edge_info::edge_info): Renamed from allocate_edge_info.
8291 Initialize additional members.
8292 (edge_info::~edge_info): New.
8293 (free_dom_edge_info): Delete the edge info.
8294 (record_edge_info): Use new class & associated member functions.
8295 Tighten forms for testing for edge equivalences.
8296 (record_temporary_equivalences): Iterate over the simple
8297 equivalences rather than assuming there's only one per edge.
8298 (cprop_into_successor_phis): Iterate over the simple
8299 equivalences rather than assuming there's only one per edge.
8300 (optimize_stmt): Use operand_equal_p rather than pointer
8301 equality for mini-DSE code.
8303 2017-08-28 Nathan Sidwell <nathan@acm.org>
8305 * gcc.c (execute): Fold SIGPIPE handling into switch
8306 statement. Adjust internal error message.
8308 2017-08-28 Richard Biener <rguenther@suse.de>
8311 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
8312 Do nothing for removed DIEs.
8314 2017-08-28 Richard Biener <rguenther@suse.de>
8316 PR tree-optimization/81977
8317 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
8320 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
8323 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
8324 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
8325 using targetm.gen_mem_thread_fence.
8327 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
8330 * config/i386/i386.md (*<btsc><mode>): Change operand 2
8331 predicate to register_operand. Reorder operands.
8332 (*btr<mode>): Ditto.
8333 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
8334 (*btr<mode>_mask): Ditto.
8336 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
8338 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
8339 * config/rs6000/xmmintrin.h: New file.
8340 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
8342 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8345 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
8346 parameter and_insn and return it.
8347 (recombine_lvx_pattern): Insert a copy to ensure availability of
8348 the base register of the copied masking operation at the point of
8349 the instruction replacement.
8350 (recombine_stvx_pattern): Likewise.
8352 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
8354 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
8355 undocumented switches.
8356 (-mpower9-dform-vector): Likewise.
8357 (-mpower9-dform): Likewise.
8358 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
8359 comments to delete references to -mpower9-dform* switches.
8360 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
8361 Delete reference to -mpower9-dform* switches, test for
8362 -mpower9-vector instead.
8363 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
8364 (OTHER_P9_VECTOR_MASKS): Likewise.
8365 (POWERPC_MASKS): Likewise.
8366 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
8367 tests against -mpower9-dform* to -mpower9-vector. Delete code
8368 that checked for -mpower9-dform* consistancy with other options.
8369 Add test for -mpower9-misc to enable other power9 switches.
8370 (rs6000_init_hard_regno_mode_ok): Likewise.
8371 (rs6000_option_override_internal): Likewise.
8372 (rs6000_emit_prologue): Likewise.
8373 (rs6000_emit_epilogue): Likewise.
8374 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
8375 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
8376 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
8378 (emit_fusion_p9_store): Likewise.
8379 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
8380 resetting these macros if the assembler does not support ISA 3.0
8382 (TARGET_P9_DFORM_VECTOR): Likewise.
8383 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
8384 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
8386 2017-08-25 Alan Modra <amodra@gmail.com>
8388 PR rtl-optimization/81747
8389 * cse.c (cse_extended_basic_block): Don't attempt to record
8390 equivalences for degenerate conditional branches that branch
8391 to their fall-through.
8393 2017-08-24 Martin Sebor <msebor@redhat.com>
8396 * gimple-fold.c (size_must_be_zero_p): New function.
8397 (gimple_fold_builtin_memory_op): Call it.
8399 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
8401 * config/rs6000/mm_malloc.h: New file.
8403 2017-08-24 Bin Cheng <bin.cheng@arm.com>
8405 PR tree-optimization/81913
8406 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
8407 analysis when either IVs in condition can wrap.
8409 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
8411 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
8412 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
8414 2017-08-24 Richard Biener <rguenther@suse.de>
8417 * targhooks.c (default_target_can_inline_p): Properly
8418 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
8419 is present and always compare.
8420 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
8421 infer -mfpmath=sse from TARGET_SSE_P.
8422 (ix86_can_inline_p): Properly use target_option_default_node when
8423 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
8425 2017-08-24 Richard Biener <rguenther@suse.de>
8428 * dwarf2out.c (output_die): Handle flag_generate_offload like
8430 (output_comp_unit): Likewise.
8431 (gen_array_type_die): Likewise.
8432 (dwarf2out_early_finish): Likewise.
8433 (note_variable_value_in_expr): Likewise.
8434 (dwarf2out_finish): Likewise. Adjust assert.
8435 * cgraphunit.c (symbol_table::compile): Move setting of
8436 flag_generate_offload earlier ...
8437 (symbol_table::finalize_compilation_unit): ... here, before
8438 early debug finalization.
8440 2017-08-24 Richard Biener <rguenther@suse.de>
8442 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
8443 and ipa-fnsummary.h.
8444 (ix86_can_inline_p): When ix86_fpmath flags do not match
8445 check whether the callee uses FP math at all.
8447 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
8450 * tree-ssanames.c (get_nonzero_bits): Use element_precision
8451 instead of TYPE_PRECISION.
8453 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
8454 Alan Hayward <alan.hayward@arm.com>
8455 David Sherwood <david.sherwood@arm.com>
8457 * combine.c (make_extraction): Use subreg_offset_from_lsb.
8459 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
8461 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
8463 (ix86_frame::stack_realign_allocate): New field.
8464 (struct machine_frame_state): Modify comments.
8465 (machine_frame_state::sp_realigned_fp_end): New field.
8466 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
8468 (sp_valid_at): Add assertion to assure no attempt to access invalid
8469 offset of a realigned stack.
8470 (fp_valid_at): Likewise.
8471 (choose_baseaddr): Modify comments.
8472 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
8473 ix86_expand_prologue.
8474 (ix86_expand_prologue): Modify stack realignment and allocation.
8475 (ix86_expand_epilogue): Modify comments.
8476 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
8477 avx2_runtime, avx512f, and avx512f_runtime.
8479 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
8481 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
8482 (mstackrealign): Do not init to -1.
8483 * config/i386/i386.c (ix86_option_override_internal):
8484 Check opts_set, not opts when setting default value of
8485 opts->x_ix86_force_align_arg_pointer.
8487 2017-08-23 Richard Biener <rguenther@suse.de>
8489 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
8490 lang_hooks.decl_printable_name.
8491 * print-rtl-function.c (print_rtx_function): Likewise.
8492 * tree-pretty-print.c (dump_function_header): Likewise.
8494 2017-08-23 Richard Biener <rguenther@suse.de>
8497 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
8498 -g0 at compile-time.
8500 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8503 * doc/sourcebuild.texi (Other hardware attributes):
8506 2017-08-23 Tamar Christina <tamar.christina@arm.com>
8509 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
8510 Fix single-use check.
8512 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
8514 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
8516 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8518 * doc/install.texi: Modify to add more details on running selected
8521 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
8523 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
8524 is combined with -mabi=ms.
8525 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
8528 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8530 PR tree-optimization/81488
8531 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
8532 and cached_basis fields.
8533 (MAX_SPREAD): New constant.
8534 (alloc_cand_and_find_basis): Initialize new fields.
8535 (clear_visited): New function.
8536 (create_phi_basis_1): Rename from create_phi_basis, set visited
8537 and cached_basis fields.
8538 (create_phi_basis): New wrapper function.
8539 (phi_add_costs_1): Rename from phi_add_costs, add spread
8540 parameter, set visited field, short-circuit when limits reached.
8541 (phi_add_costs): New wrapper function.
8542 (record_phi_increments_1): Rename from record_phi_increments, set
8544 (record_phi_increments): New wrapper function.
8545 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
8546 (phi_incr_cost): New wrapper function.
8547 (all_phi_incrs_profitable_1): Rename from
8548 all_phi_incrs_profitable, set visited field.
8549 (all_phi_incrs_profitable): New wrapper function.
8551 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8552 Alan Hayward <alan.hayward@arm.com>
8553 David Sherwood <david.sherwood@arm.com>
8555 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
8556 that takes the outer and inner modes.
8557 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
8558 comparison as the canonical test for a paradoxical subreg.
8559 * combine.c (simplify_set): Use paradoxical_subreg_p.
8560 (make_extraction): Likewise.
8561 (force_to_mode): Likewise.
8562 (rtx_equal_for_field_assignment_p): Likewise.
8563 (gen_lowpart_for_combine): Likewise.
8564 (simplify_comparison): Likewise.
8565 * cse.c (equiv_constant): Likewise.
8566 * expmed.c (store_bit_field_1): Likewise.
8567 * final.c (alter_subreg): Likewise.
8568 * fwprop.c (propagate_rtx): Likewise.
8569 (forward_propagate_subreg): Likewise.
8570 * ira-conflicts.c (ira_build_conflicts): Likewise.
8571 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
8572 * lra-constraints.c (curr_insn_transform): Likewise.
8573 (split_reg): Likewise.
8574 * lra-eliminations.c (move_plus_up): Likewise.
8575 (lra_eliminate_regs_1): Likewise.
8576 * recog.c (general_operand): Likewise.
8577 * ree.c (combine_reaching_defs): Likewise.
8578 * reload.c (push_reload): Likewise.
8579 (find_reloads): Likewise.
8580 * reload1.c (elimination_effects): Likewise.
8581 (compute_reload_subreg_offset): Likewise.
8582 (choose_reload_regs): Likewise.
8583 * rtlanal.c (subreg_lsb_1): Likewise.
8584 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8585 (simplify_subreg): Likewise.
8586 * var-tracking.c (track_loc_p): Likewise.
8587 * emit-rtl.c (byte_lowpart_offset): Likewise.
8588 (paradoxical_subreg_p): Delete out-of-line definition.
8590 2017-08-22 Jeff Law <law@redhat.com>
8592 PR tree-optimization/81741
8593 PR tree-optimization/71947
8594 * tree-ssa-dom.c: Include tree-inline.h.
8595 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
8596 equivalences if one is more expensive to compute than the other.
8597 * tree-ssa-scopedtables.h (class const_or_copies): Make
8598 record_const_or_copy_raw method private.
8599 (class avail_exprs_stack): New method simplify_binary_operation.
8600 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
8601 avail_exprs_stack::simplify_binary_operation as needed.
8602 (avail_exprs_stack::simplify_binary_operation): New function.
8604 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
8606 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
8607 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
8608 (DOT_SYMBOLS): Likewise.
8609 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
8610 (RELOCATABLE_NEEDS_FIXUP): Likewise.
8611 (RS6000_ABI_NAME): Likewise.
8612 (TARGET_CMODEL): Likewise.
8613 (TOC_SECTION_ASM_OP): Likewise.
8614 (SET_CMODEL): New macro.
8615 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
8617 2017-08-22 Richard Biener <rguenther@suse.de>
8619 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
8620 to allow for free-lang-data replacements similar to verify_type_variant.
8622 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
8623 Alan Hayward <alan.hayward@arm.com>
8624 David Sherwood <david.sherwood@arm.com>
8626 * config/aarch64/aarch64.md (casesi): Use DImode rather than
8627 VOIDmode for the LABEL_REF.
8629 2017-08-22 Richard Biener <rguenther@suse.de>
8631 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
8633 2017-08-22 Richard Biener <rguenther@suse.de>
8635 * common.opt (feliminate-dwarf2-dups): Ignore.
8636 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
8637 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
8638 same_die_p_wrap, compute_section_prefix,
8639 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
8640 (comdat_symbol_id, comdat_symbol_number): Likewise.
8641 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
8643 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
8644 (output_die): Mark unreachable path unreachable.
8645 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
8646 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
8647 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
8648 (dwarf2out_early_finish): Likewise.
8650 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
8652 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
8654 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
8657 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
8658 not VAR_P. Filter attribute warnings with OPT_Wattributes.
8659 (avr_attribute_table) <io, io_low, address>: Initialize
8660 .decl_required with true.
8662 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
8664 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
8665 undocumented debugging options.
8666 (-mvsx-scalar-double): Likewise.
8667 (-mallow-df-permute): Likewise.
8668 (-mvectorize-builtins): Likewise.
8669 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
8670 (rs6000_builtin_vectorized_function): Likewise.
8671 (rs6000_builtin_md_vectorized_function): Likewise.
8672 (rs6000_opt_vars): Likewise.
8674 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
8677 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
8678 (*btrq_imm): Rename from *btrq.
8679 (*btcq_imm): Rename from *btcq.
8680 (btsc): New code attribute.
8681 (*<btsc><mode>): New insn pattern.
8682 (*btr<mode>): Ditto.
8683 (*<btsc><mode>_mask): New insn_and_split pattern.
8684 (*btr<mode>_mask): Ditto.
8686 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8687 Alan Hayward <alan.hayward@arm.com>
8688 David Sherwood <david.sherwood@arm.com>
8690 * function.c (pad_below): Simplify padding calculation.
8692 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8693 Alan Hayward <alan.hayward@arm.com>
8694 David Sherwood <david.sherwood@arm.com>
8696 * target.def (function_prologue): Remove frame size argument.
8697 (function_epilogue): Likewise.
8698 * doc/tm.texi: Regenerate.
8699 * final.c (final_start_function): Update call to function_prologue.
8700 (final_end_function): Update call to function_epilogue.
8701 (default_function_pro_epilogue): Remove frame size argument.
8702 * output.h (default_function_pro_epilogue): Likewise.
8703 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
8704 (arm_output_function_prologue): Likewise.
8705 * config/frv/frv.c (frv_function_prologue): Likewise.
8706 (frv_function_epilogue): Likewise.
8707 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
8708 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
8709 (ia64_output_function_epilogue): Likewise.
8710 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
8711 (m32r_output_function_epilogue): Likewise.
8712 * config/microblaze/microblaze.c (microblaze_function_prologue)
8713 (microblaze_function_epilogue): Likewise.
8714 * config/mips/mips.c (mips_output_function_prologue): Likewise.
8715 (mips_output_function_epilogue): Likewise.
8716 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
8717 (mmix_target_asm_function_epilogue): Likewise.
8718 * config/msp430/msp430.c (msp430_start_function): Likewise.
8719 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
8720 (nds32_asm_function_epilogue): Likewise.
8721 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
8722 * config/pa/pa.c (pa_output_function_prologue): Likewise.
8723 (pa_output_function_epilogue): Likewise.
8724 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
8725 (rs6000_output_function_epilogue): Likewise.
8726 * config/rl78/rl78.c (rl78_start_function): Likewise.
8727 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
8728 (rs6000_output_function_epilogue): Likewise.
8729 * config/rx/rx.c (rx_output_function_prologue): Likewise.
8730 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
8731 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
8732 (sparc_asm_function_epilogue): Likewise.
8734 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8736 * tree.h (type_has_mode_precision_p): New function.
8737 * convert.c (convert_to_integer_1): Use it.
8738 * expr.c (expand_expr_real_2): Likewise.
8739 (expand_expr_real_1): Likewise.
8740 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
8741 * match.pd: Likewise.
8742 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
8743 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
8744 * tree-tailcall.c (process_assignment): Likewise.
8745 * tree-vect-loop.c (vectorizable_reduction): Likewise.
8746 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
8747 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
8748 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
8749 (vectorizable_assignment): Likewise.
8750 (vectorizable_shift): Likewise.
8751 (vectorizable_operation): Likewise.
8752 * tree-vrp.c (register_edge_assert_for_2): Likewise.
8754 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
8756 * match.pd: Add pow (C, x) simplification.
8758 2017-08-21 Richard Biener <rguenther@suse.de>
8760 PR tree-optimization/81900
8761 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
8762 for blocks with abnormal predecessors.
8763 (compute_antic): Do not set visited flag prematurely.
8765 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
8768 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
8770 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8772 * stor-layout.h (vector_type_mode): Move to...
8773 * tree.h (vector_type_mode): ...here.
8774 * stor-layout.c (vector_type_mode): Move to...
8775 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
8777 2017-08-21 Richard Biener <rguenther@suse.de>
8779 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
8780 register_external_die hooks.
8781 (debug_false_tree_charstarstar_uhwistar): Declare.
8782 (debug_nothing_tree_charstar_uhwi): Likewise.
8783 * debug.c (do_nothing_debug_hooks): Adjust.
8784 (debug_false_tree_charstarstar_uhwistar): New do nothing.
8785 (debug_nothing_tree_charstar_uhwi): Likewise.
8786 * dbxout.c (dbx_debug_hooks): Adjust.
8787 (xcoff_debug_hooks): Likewise.
8788 * sdbout.c (sdb_debug_hooks): Likewise.
8789 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
8790 * dwarf2out.c (macinfo_label_base): New global.
8791 (dwarf2out_register_external_die): New function for the
8792 register_external_die hook.
8793 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
8794 (dwarf2_debug_hooks): Use them.
8795 (dwarf2_lineno_debug_hooks): Adjust.
8796 (struct die_struct): Add with_offset flag.
8797 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
8798 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
8799 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
8800 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
8801 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
8802 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
8803 defining section names for the early LTO debug variants.
8804 (reset_indirect_string): New helper.
8805 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
8806 (print_dw_val): Add support for offsetted symbol references.
8807 (get_ultimate_context): Split out from is_cxx.
8808 (is_cxx): Use get_ultimate_context.
8809 (is_fortran): Add decl overload.
8810 (compute_comp_unit_symbol): Split out worker from
8811 compute_section_prefix.
8812 (compute_section_prefix): Call compute_comp_unit_symbol and
8813 set comdat_type_p here.
8814 (output_die): Skip DIE symbol output for the LTO added one.
8815 Handle DIE symbol references with offset.
8816 (output_comp_unit): Guard section name mangling properly.
8817 For LTO debug sections emit a symbol at the section beginning
8818 which we use to refer to its DIEs.
8819 (add_abstract_origin_attribute): For DIEs registered via
8820 dwarf2out_register_external_die directly refer to the early
8821 DIE rather than indirectly through the shadow one we created.
8822 Remove obsolete call to dwarf2out_abstract_function for
8823 non-function/block origins.
8824 (gen_array_type_die): When generating early LTO debug do
8825 not emit DW_AT_string_length.
8826 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
8827 late when in LTO. As suggested place a gcc_unreachable for
8828 the DECL_ABSTRACT_P case.
8829 (gen_subprogram_die): Avoid another specification DIE
8830 for early built declarations/definitions for the late LTO case.
8831 (gen_variable_die): Add type references for late duplicated VLA dies
8833 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
8834 we have the abstract instance already.
8835 (process_scope_var): Adjust decl DIE contexts in LTO which
8836 first puts them in limbo.
8837 (gen_decl_die): Do not generate type DIEs late apart from
8838 types for VLAs or for decls we do not yet have a DIE. Do not
8839 call dwarf2out_abstract_function late.
8840 (dwarf2out_early_global_decl): Make sure to create DIEs
8841 for abstract instances of a decl first.
8842 (dwarf2out_late_global_decl): Adjust comment.
8843 (output_macinfo_op): With multiple macro sections use
8844 macinfo_label_base to distinguish labels.
8845 (output_macinfo): Likewise. Update macinfo_label_base.
8846 Pass in the line info label.
8847 (note_variable_value_in_expr): When generating LTO resolve
8848 all variable values here by generating DIEs as needed.
8849 (init_sections_and_labels): Add early LTO debug flag parameter
8850 and generate different sections and names if set. Add generation
8851 counter for the labels so we can have multiple of them.
8852 (reset_dies): Helper to allow DIEs to be output multiple times.
8853 (dwarf2out_finish): When outputting DIEs to the fat part of an
8854 LTO object first reset DIEs.
8855 (dwarf2out_early_finish): Output early DIEs when generating LTO.
8856 (modified_type_die): Check for decl_ultimate_origin being self
8858 (gen_type_die_with_usage): Likewise.
8859 (gen_typedef_die): Allow decl_ultimate_origin being self.
8860 (set_decl_abstract_flags): Remove.
8861 (set_block_abstract_flags): Likewise.
8862 (dwarf2out_abstract_function): Treat the early generated DIEs
8863 as the abstract copy and only add DW_AT_inline and
8864 DW_AT_artificial here and call set_decl_origin_self.
8865 If the DIE has an abstract origin don't do anything.
8866 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
8867 if we have none yet (Go fails to build one, PR78628).
8868 (variably_modified_type_p): Prevent endless recursion for Ada
8869 cyclic pointer types.
8870 * lto-streamer-in.c: Include debug.h.
8871 (dref_queue): New global.
8872 (lto_read_tree_1): Stream in DIE references.
8873 (lto_input_tree): Register DIE references.
8874 (input_function): Stream DECL_DEBUG_ARGS.
8875 * lto-streamer-out.c: Include debug.h.
8876 (lto_write_tree_1): Output DIE references.
8877 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
8878 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
8879 (output_function): Stream DECL_DEBUG_ARGS.
8880 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
8881 Stream DECL_ABSTRACT_ORIGIN.
8882 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
8883 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
8884 DECL_CONTEXT for file-scope decls.
8885 * lto-streamer.h (struct dref_entry): Declare.
8886 (dref_queue): Likewise.
8887 * cfgexpand.c (pass_expand::execute): Do not call the
8888 outlining_inline_function hook here.
8889 * lto-wrapper.c (debug_obj): New global.
8890 (tool_cleanup): Unlink it if required.
8891 (debug_objcopy): New function.
8892 (run_gcc): Handle early debug sections in the IL files by
8893 extracting them to separate files, partially linkin them and
8894 feeding the result back as result to the linker.
8895 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
8896 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
8897 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
8898 sections into a separate segment.
8899 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
8901 (darwin_asm_dwarf_section): Likewise.
8902 (darwin_asm_output_dwarf_offset): Likewise.
8903 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
8905 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8906 Alan Hayward <alan.hayward@arm.com>
8907 David Sherwood <david.sherwood@arm.com>
8909 * read-md.h (md_reader::record_potential_iterator_use): Replace
8910 pointer argument with an rtx and an index.
8911 * read-rtl.c (iterator_group::apply_iterator): Likewise.
8912 (apply_mode_iterator): Likewise.
8913 (apply_code_iterator): Likewise.
8914 (apply_int_iterator): Likewise.
8915 (apply_subst_iterator): Likewise.
8916 (record_iterator_use): Likewise.
8917 (record_attribute_use): Likewise.
8918 (md_reader::record_potential_iterator_use): Likewise. Update calls
8919 to record_iterator_use and apply_iterator.
8920 (iterator_use): Replace ptr with x and index.
8921 (attribute_use): Likewise.
8922 (apply_attribute_uses): Update calls to apply_iterator.
8923 (apply_iterators): Likewise. Update initialization of iterator_use.
8924 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
8925 and record_potential_iterator_use.
8926 (rtx_reader::read_rtx_operand): Likewise.
8928 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
8929 Alan Hayward <alan.hayward@arm.com>
8930 David Sherwood <david.sherwood@arm.com>
8932 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
8935 2017-08-21 Richard Biener <rguenther@suse.de>
8938 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
8939 at struct end conservatively when comparing common bases.
8941 2017-08-21 Richard Biener <rguenther@suse.de>
8943 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
8944 (mem_ref_in_stmt): Remove.
8945 (determine_max_movement): Use ref index to get at the reference.
8946 (invariantness_dom_walker::before_dom_children): Deal with
8947 lim data already initialized.
8948 (gather_mem_refs_stmt): Initialize lim data and record ref index.
8950 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8952 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
8953 (TARGET_ISA_ROUND): Ditto.
8954 (TARGET_ROUND): Ditto.
8955 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
8956 * config/i386/i386.md: Ditto.
8957 * config/i386/sse.md: Ditto.
8958 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
8959 with OPTION_MASK_ISA_SSE4_1.
8961 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
8964 * doc/extend.texi (x86 Built-in Functions): Correct the name of
8965 __builtin_ia32_lzcnt_u16.
8967 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
8970 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
8971 (rs6000_set_current_function): Rewrite function to use it.
8973 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
8976 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
8977 and TYPE_WARN_IF_NOT_ALIGN.
8978 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
8979 (handle_warn_if_not_align): New.
8980 (place_union_field): Call handle_warn_if_not_align.
8981 (place_field): Call handle_warn_if_not_align.
8982 Copy TYPE_WARN_IF_NOT_ALIGN.
8983 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
8984 (layout_type): Likewise.
8985 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
8987 (tree_decl_common): Add warn_if_not_align.
8988 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
8989 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
8990 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
8991 (DECL_WARN_IF_NOT_ALIGN): Likewise.
8992 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
8993 * doc/extend.texi: Document warn_if_not_aligned attribute.
8994 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
8996 2017-08-17 Martin Liska <mliska@suse.cz>
8999 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
9000 (get_data_dependence): Use it as pointer type.
9001 (distribute_loop): Likewise.
9003 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9005 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
9006 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
9007 (p8_vmrgow_<mode>_direct): New define_insn.
9008 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
9009 handle endianness for vmrgew and vmrgow permute patterns.
9011 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9013 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
9014 * config/rs6000/rs6000-cpus.def: Remove comment.
9015 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
9016 (POWERPC_MASKS): Likewise.
9017 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
9018 use of TARGET_VSX_TIMODE.
9019 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
9020 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
9021 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
9022 (rs6000_option_override_internal): Remove dead code.
9023 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
9024 (rs6000_legitimize_reload_address): Likewise.
9025 (rs6000_legitimate_address_p): Likewise.
9026 (rs6000_opt_masks): Delete "vsx-timode".
9027 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
9028 from function comment.
9029 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
9030 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
9031 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
9033 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
9034 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
9035 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
9038 (define_peephole2): Likewise.
9040 2017-08-17 Martin Sebor <msebor@redhat.com>
9043 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
9044 past the end of an array.
9045 (test_pp_format): Add test cases.
9047 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
9049 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
9050 missing ECF_NOTHROW flags.
9052 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
9055 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
9056 operands residing in integer registers.
9057 (*vsx_le_perm_load_<mode>): Likewise.
9058 (*vsx_le_perm_store_<mode>): Likewise.
9059 (define_peephole2): Add peepholes to optimize the above.
9061 2017-08-17 Marek Polacek <polacek@redhat.com>
9064 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
9065 to mimic what shorten_compare did. Change the return type to bool.
9066 (fold_cond_expr_with_comparison): Update call to
9067 operand_equal_for_comparison_p.
9068 (fold_ternary_loc): Likewise.
9070 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
9072 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
9074 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
9076 2017-08-17 Richard Biener <rguenther@suse.de>
9078 * tree-ssa-structalias.c (solve_graph): When propagating
9079 to successors update the graphs succ edges and avoid duplicate work.
9081 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
9084 * config/i386/i386.c (ix86_option_override_internal): Save target
9085 specific options after ix86_stack_protector_guard_reg was changed.
9087 2017-08-17 Richard Biener <rguenther@suse.de>
9089 PR tree-optimization/81827
9090 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
9091 (new_var_info): Initialize it conservatively.
9092 (get_call_vi): Mark register vars.
9093 (new_scalar_tmp_constraint_exp): Likewise.
9094 (handle_rhs_call): Likewise.
9095 (handle_const_call): Likewise.
9096 (create_function_info_for): Likewise.
9097 (solve_constraints): Sort varinfos to separate register from
9098 non-register vars to pack points-to solution bitmaps during
9101 2017-08-17 Marek Polacek <polacek@redhat.com>
9103 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
9105 2017-08-17 Richard Biener <rguenther@suse.de>
9107 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
9108 to true when overflow is undefined and we saturated the result.
9110 2017-08-17 Alan Modra <amodra@gmail.com>
9113 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
9114 Don't use store multiple if only one reg needs saving.
9115 (interesting_frame_related_regno): New function.
9116 (rs6000_frame_related): Don't emit frame info for regs that
9118 (rs6000_emit_epilogue): Likewise.
9120 2017-08-16 Nathan Sidwell <nathan@acm.org>
9122 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
9123 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
9124 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
9125 * tree.c (free_lang_data_in_type): Use else-if chain. Always
9126 clear TYPE_LANG_1. Remove obsolete member-function stripping.
9127 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
9128 (verify_type): Adjust for TYPE_BINFO move.
9129 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
9130 process TYPE_BINFO directly.
9131 (hash_tree): Likewise.
9132 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
9134 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
9137 2017-08-16 David Malcolm <dmalcolm@redhat.com>
9139 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
9141 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
9144 * config/i386/i386.md (*anddi_1_btr): Change predicates of
9145 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
9146 Add ix86_binary_operator_ok to insn constraint.
9147 (*iordi_1_bts): Ditto.
9148 (*xordi_1_btc): Ditto.
9149 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
9150 Update corresponding peephole2 pattern.
9154 2017-08-16 Bin Cheng <bin.cheng@arm.com>
9156 PR tree-optimization/81832
9157 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
9158 copy loop header which has IFN_LOOP_DIST_ALIAS call.
9160 2017-08-16 Marek Polacek <polacek@redhat.com>
9163 * fold-const.c (fold_indirect_ref_1): Restore original behavior
9164 regarding size_zero_node.
9166 2017-08-16 Martin Liska <mliska@suse.cz>
9169 * config.gcc: Respect previously set extra_objs in case
9172 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
9174 PR tree-optimization/81835
9175 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
9176 the phi SSA_NAME. Check that the condition in a COND_EXPR does
9177 not depend on the phi.
9179 2017-08-16 Alan Modra <amodra@gmail.com>
9181 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
9184 2017-08-16 Alan Modra <amodra@gmail.com>
9186 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
9187 (save_reg_p): ..into this. Update all callers.
9188 (first_reg_to_save): Simplify.
9190 2017-08-16 Alan Modra <amodra@gmail.com>
9192 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
9195 2017-08-15 Joseph Myers <joseph@codesourcery.com>
9199 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
9200 constant count if that count is less than 32.
9202 2017-08-15 Nathan Sidwell <nathan@acm.org>
9204 * gcc.c (execute): Emit friendlier message if inferior is killed
9205 by an external cause.
9207 2017-08-15 Richard Biener <rguenther@suse.de>
9209 PR tree-optimization/81790
9210 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
9211 CONSTRUCTORs from simplifying and VN.
9213 2017-08-14 Martin Sebor <msebor@redhat.com>
9215 * builtin-attrs.def: Add comments.
9217 2017-08-14 Martin Sebor <msebor@redhat.com>
9220 * doc/extend.texi (attribute nonstring): Document new attribute.
9222 2017-08-14 Martin Sebor <msebor@redhat.com>
9225 * tree-diagnostic.c (default_tree_printer): Handle %G.
9226 * gimple-pretty-print.h (percent_G_format): Declare new function.
9227 * gimple-pretty-print.c (percent_G_format): Define.
9228 * tree-pretty-print.c (percent_K_format): Add argument.
9230 2017-08-14 Martin Sebor <msebor@redhat.com>
9232 PR translation/79998
9233 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
9234 Remove a stray space.
9236 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
9239 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
9240 (*iordi_1_bts): Ditto.
9241 (*xordi_1_btc): Ditto.
9243 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9246 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
9247 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9249 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
9250 quoted strings, and make more translator-friendly.
9251 (darwin_rs6000_override_options): Likewise.
9252 (rs6000_option_override_internal): Likewise.
9253 (rs6000_return_in_memory): Fix overlong line.
9254 (init_cmulative_args): Use quoted strings, and make more
9255 translator-friendly.
9256 (rs6000_pass_by_reference): Fix overlong line.
9257 (def_builtin): Use quoted strings.
9258 (altivec_expand_predicate_builtin): Use quoted strings, and make
9259 more translator-friendly.
9260 (htm_expand_builtin): Use quoted strings.
9261 (cpu_expand_builtin): Use quoted strings, and make more
9262 translator-friendly.
9263 (altivec_expand_builtin): Likewise.
9264 (paired_expand_predicate_builtin): Likewise.
9265 (rs6000_invalid_builtin): Likewise.
9266 (builtin_function_type): Use quoted strings.
9267 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
9268 more translator-friendly.
9269 (rs6000_trampoline_init): Likewise.
9270 (rs6000_handle_altivec_attribute): Likewise.
9271 (rs6000_inner_target_options): Use quoted strings.
9272 (rs6000_disable_incompatible_switches): Likewise.
9273 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
9274 strings, and make more translator-friendly.
9275 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
9277 2017-08-14 Bin Cheng <bin.cheng@arm.com>
9279 PR tree-optimization/81799
9280 * tree-loop-distribution.c (version_loop_by_alias_check): Force
9281 cond_expr to simple gimple operand.
9283 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
9286 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
9288 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
9292 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
9293 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
9294 TARGET_GASISR_PROLOGUES.
9295 * config/avr/avr.c (avr_option_override): Same.
9296 (avr_pass_pre_proep::execute): Same.
9298 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
9301 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
9302 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
9304 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
9306 * config/i386/i386.md (*load_tp_<mode>): Redefine as
9307 define_insn_and_split. Split to a memory load from 0 in
9308 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
9309 using PTR mode iterator.
9310 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
9311 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
9312 (*add_tp_<mode>): Redefine as define_insn_and_split.
9313 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
9314 address space. Merge with *add_tp_x32 using PTR mode iterator.
9315 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
9316 Split to an add with a memory load from 0 in
9317 DEFAULT_TLS_SEG_REG address space.
9319 2017-08-12 Andrew Pinski <apinski@cavium.com>
9321 * config/aarch64/aarch64-option-extensions.def (rdma):
9322 Fix feature string to what Linux prints out in /proc/cpuinfo.
9324 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
9327 * dwarf2out.c (modified_type_die): For C typedef types that have
9328 an ultimate origin, process the ultimate origin instead of the
9330 (gen_typedef_die): Assert that input DECLs have no ultimate
9332 (gen_type_die_with_usage): For typedef variants that have an
9333 ultimate origin, just call gen_decl_die on the original DECL.
9334 (process_scope_var): Avoid creating DIEs for local typedefs and
9335 concrete static variables.
9337 2017-08-12 Alan Modra <amodra@gmail.com>
9341 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
9342 match gnu-user.h startfile.
9343 (ENDFILE_LINUX_SPEC): Similarly.
9345 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
9348 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
9350 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
9352 2017-08-11 Tamar Christina <tamar.christina@arm.com>
9353 * config/aarch64/aarch64.md (mov<mode>): Change.
9354 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
9355 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
9356 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
9358 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
9360 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
9361 for storage order barriers.
9363 2017-08-11 Martin Liska <mliska@suse.cz>
9366 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
9367 variables of void type.
9369 2017-08-11 Martin Liska <mliska@suse.cz>
9371 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
9372 TARGET_SUPPORTS_ALIASES.
9373 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
9374 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
9375 (optimize_weakref): Likewise.
9376 * symtab.c (symtab_node::noninterposable_alias): Likewise.
9377 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
9378 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
9380 2017-08-11 Martin Liska <mliska@suse.cz>
9383 * config/i386/i386.c (make_resolver_func): Do complete
9384 refactoring of the function.
9386 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9389 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
9390 * config/i386/i386.c (ix86_stack_protect_guard): Use
9391 ix86_stack_protect_guard_symbol_str to generate varible declaration.
9392 * doc/invoke.texi (x86 Options): Document
9393 -mstack-protector-guard-symbol= option.
9395 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
9397 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
9398 * config/i386/i386.c (ix86_split_stack_guard): New function.
9399 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
9400 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
9401 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
9402 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
9403 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
9404 (split_stack_space_check): Call ix86_split_stack_guard.
9406 2017-08-10 Martin Sebor <msebor@redhat.com>
9408 * print-tree.c (print_node): Print location using the established
9410 Replace spaces with colons.
9411 (debug_raw, debug): Ditto.
9413 2017-08-10 Martin Sebor <msebor@redhat.com>
9416 * pretty-print.c (pp_format): Correct the handling of %s precision.
9418 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
9421 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9423 (ix86_finalize_stack_frame_flags): This. Also clear
9424 frame_pointer_needed if -fno-omit-frame-pointer is used without
9426 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9427 with ix86_finalize_stack_frame_flags.
9428 (ix86_expand_epilogue): Likewise.
9429 (ix86_expand_split_stack_prologue): Likewise.
9430 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
9432 2017-08-10 Martin Liska <mliska@suse.cz>
9435 * c-attribs.c (handle_target_attribute):
9436 Report warning for an empty string argument of target attribute.
9438 2017-08-09 Jakub Jelinek <jakub@redhat.com>
9441 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
9443 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
9444 or DECL_NONLOCAL labels.
9445 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
9446 or DECL_NONLOCAL labels here.
9448 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
9450 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
9451 to indicate when early gimple folding has been disabled.
9452 (rs6000_gimple_fold_builtin): Add debug content.
9453 (rs6000_invalid_builtin): Fix whitespace.
9454 (rs6000_expand_builtin): Fix whitespace.
9455 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
9457 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
9460 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
9461 SAVE_MULTIPLE if not all the registers that saves, should be saved.
9463 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
9465 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
9466 (qdf24xx): Likewise.
9467 * config/aarch64/aarch64.md: Include falkor.md.
9468 * config/aarch64/falkor.md: New.
9470 2017-08-09 Marek Polacek <polacek@redhat.com>
9473 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
9474 * diagnostic.c (emit_diagnostic): Add a comment.
9475 (emit_diagnostic_valist): New function.
9477 2017-08-09 Marek Polacek <polacek@redhat.com>
9480 * input.c (make_location): New overload.
9481 * input.h (make_location): Declare.
9483 2017-08-08 Alan Modra <amodra@gmail.com>
9484 H.J. Lu <hongjiu.lu@intel.com>
9487 * gcc.c (NO_PIE_SPEC): Delete.
9488 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
9490 (LINK_PIE_SPEC): ..to here.
9491 (LINK_COMMAND_SPEC): Support -no-pie.
9492 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
9493 chain of crtbegin*.o selection, update for PIE_SPEC changes and
9495 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
9496 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
9497 (ENDFILE_CRTEND_SPEC): Similarly.
9499 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
9502 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
9503 (mstack-protector-guard-offset=): Ditto.
9504 * config/i386/i386.c (ix86_option_override): Handle
9505 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
9507 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
9508 ix86_stack_protect_guard_offset variables.
9509 (TARGET_STACK_PROTECT_GUARD): Always define.
9510 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
9511 and -mstack-protector-guard-offset= options.
9513 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9515 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
9516 boundary case for the last candidate.
9518 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9520 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
9521 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
9523 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9526 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
9527 * config/aarch64/aarch64-builtins.c
9528 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
9529 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
9530 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
9532 2017-08-08 Tamar Christina <tamar.christina@arm.com>
9533 Andrew Pinski <pinskia@gmail.com>
9536 * internal-fn.def (XORSIGN): New.
9537 * optabs.def (xorsign_optab): New.
9538 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
9539 (convert_expand_mult_copysign): New.
9540 (pass_optimize_widening_mul::execute): Call
9541 convert_expand_mult_copysign.
9543 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9545 PR tree-optimization/81354
9546 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
9547 Insert on edges rather than explicitly creating landing pads.
9548 (analyze_candidates_and_replace): Commit edge inserts.
9550 2017-08-08 Richard Biener <rguenther@suse.de>
9553 * tree-ssa-loop-niter.c: Include tree-dfa.h.
9554 (expand_simple_operations): Also look through ADDR_EXPRs with
9555 MEM_REF bases treating them as POINTER_PLUS_EXPR.
9557 2017-08-08 Richard Biener <rguenther@suse.de>
9559 PR tree-optimization/81723
9560 * tree-vect-slp.c (struct bst_traits): New hash traits.
9561 (bst_fail): New global.
9562 (vect_build_slp_tree_2): New worker, split out from ...
9563 (vect_build_slp_tree): ... this now wrapping it with using
9564 bst_fail set to cache SLP tree build fails. Properly handle
9566 (vect_analyze_slp_instance): Allocate and free bst_fail.
9568 2017-08-08 Martin Liska <mliska@suse.cz>
9571 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
9572 LABEL_DECLs that can be from a different function.
9574 2017-08-08 Bin Cheng <bin.cheng@arm.com>
9576 PR tree-optimization/81744
9577 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
9578 loop's number of iterations.
9580 2017-08-08 Martin Liska <mliska@suse.cz>
9582 * asan.c: Include header files.
9583 * attribs.c (build_decl_attribute_variant): New function moved
9585 (build_type_attribute_qual_variant): Likewise.
9586 (cmp_attrib_identifiers): Likewise.
9587 (simple_cst_list_equal): Likewise.
9588 (omp_declare_simd_clauses_equal): Likewise.
9589 (attribute_value_equal): Likewise.
9590 (comp_type_attributes): Likewise.
9591 (build_type_attribute_variant): Likewise.
9592 (lookup_ident_attribute): Likewise.
9593 (remove_attribute): Likewise.
9594 (merge_attributes): Likewise.
9595 (merge_type_attributes): Likewise.
9596 (merge_decl_attributes): Likewise.
9597 (merge_dllimport_decl_attributes): Likewise.
9598 (handle_dll_attribute): Likewise.
9599 (attribute_list_equal): Likewise.
9600 (attribute_list_contained): Likewise.
9601 * attribs.h (lookup_attribute): New function moved from tree.[ch].
9602 (lookup_attribute_by_prefix): Likewise.
9603 * bb-reorder.c: Include header files.
9604 * builtins.c: Likewise.
9605 * calls.c: Likewise.
9606 * cfgexpand.c: Likewise.
9607 * cgraph.c: Likewise.
9608 * cgraphunit.c: Likewise.
9609 * convert.c: Likewise.
9610 * dwarf2out.c: Likewise.
9611 * final.c: Likewise.
9612 * fold-const.c: Likewise.
9613 * function.c: Likewise.
9614 * gimple-expr.c: Likewise.
9615 * gimple-fold.c: Likewise.
9616 * gimple-pretty-print.c: Likewise.
9617 * gimple.c: Likewise.
9618 * gimplify.c: Likewise.
9619 * hsa-common.c: Likewise.
9620 * hsa-gen.c: Likewise.
9621 * internal-fn.c: Likewise.
9622 * ipa-chkp.c: Likewise.
9623 * ipa-cp.c: Likewise.
9624 * ipa-devirt.c: Likewise.
9625 * ipa-fnsummary.c: Likewise.
9626 * ipa-inline.c: Likewise.
9627 * ipa-visibility.c: Likewise.
9629 * lto-cgraph.c: Likewise.
9630 * omp-expand.c: Likewise.
9631 * omp-general.c: Likewise.
9632 * omp-low.c: Likewise.
9633 * omp-offload.c: Likewise.
9634 * omp-simd-clone.c: Likewise.
9635 * opts-global.c: Likewise.
9636 * passes.c: Likewise.
9637 * predict.c: Likewise.
9638 * sancov.c: Likewise.
9639 * sanopt.c: Likewise.
9640 * symtab.c: Likewise.
9641 * toplev.c: Likewise.
9642 * trans-mem.c: Likewise.
9643 * tree-chkp.c: Likewise.
9644 * tree-eh.c: Likewise.
9645 * tree-into-ssa.c: Likewise.
9646 * tree-object-size.c: Likewise.
9647 * tree-parloops.c: Likewise.
9648 * tree-profile.c: Likewise.
9649 * tree-ssa-ccp.c: Likewise.
9650 * tree-ssa-live.c: Likewise.
9651 * tree-ssa-loop.c: Likewise.
9652 * tree-ssa-sccvn.c: Likewise.
9653 * tree-ssa-structalias.c: Likewise.
9654 * tree-ssa.c: Likewise.
9655 * tree-streamer-in.c: Likewise.
9656 * tree-vectorizer.c: Likewise.
9657 * tree-vrp.c: Likewise.
9659 * ubsan.c: Likewise.
9660 * varasm.c: Likewise.
9661 * varpool.c: Likewise.
9662 * tree.c: Remove functions moved to attribs.[ch].
9664 * config/aarch64/aarch64.c: Add attrs.h header file.
9665 * config/alpha/alpha.c: Likewise.
9666 * config/arc/arc.c: Likewise.
9667 * config/arm/arm.c: Likewise.
9668 * config/avr/avr.c: Likewise.
9669 * config/bfin/bfin.c: Likewise.
9670 * config/c6x/c6x.c: Likewise.
9671 * config/cr16/cr16.c: Likewise.
9672 * config/cris/cris.c: Likewise.
9673 * config/darwin.c: Likewise.
9674 * config/epiphany/epiphany.c: Likewise.
9675 * config/fr30/fr30.c: Likewise.
9676 * config/frv/frv.c: Likewise.
9677 * config/ft32/ft32.c: Likewise.
9678 * config/h8300/h8300.c: Likewise.
9679 * config/i386/winnt.c: Likewise.
9680 * config/ia64/ia64.c: Likewise.
9681 * config/iq2000/iq2000.c: Likewise.
9682 * config/lm32/lm32.c: Likewise.
9683 * config/m32c/m32c.c: Likewise.
9684 * config/m32r/m32r.c: Likewise.
9685 * config/m68k/m68k.c: Likewise.
9686 * config/mcore/mcore.c: Likewise.
9687 * config/microblaze/microblaze.c: Likewise.
9688 * config/mips/mips.c: Likewise.
9689 * config/mmix/mmix.c: Likewise.
9690 * config/mn10300/mn10300.c: Likewise.
9691 * config/moxie/moxie.c: Likewise.
9692 * config/msp430/msp430.c: Likewise.
9693 * config/nds32/nds32-isr.c: Likewise.
9694 * config/nds32/nds32.c: Likewise.
9695 * config/nios2/nios2.c: Likewise.
9696 * config/nvptx/nvptx.c: Likewise.
9697 * config/pa/pa.c: Likewise.
9698 * config/pdp11/pdp11.c: Likewise.
9699 * config/powerpcspe/powerpcspe.c: Likewise.
9700 * config/riscv/riscv.c: Likewise.
9701 * config/rl78/rl78.c: Likewise.
9702 * config/rx/rx.c: Likewise.
9703 * config/s390/s390.c: Likewise.
9704 * config/sh/sh.c: Likewise.
9705 * config/sol2.c: Likewise.
9706 * config/sparc/sparc.c: Likewise.
9707 * config/spu/spu.c: Likewise.
9708 * config/stormy16/stormy16.c: Likewise.
9709 * config/tilegx/tilegx.c: Likewise.
9710 * config/tilepro/tilepro.c: Likewise.
9711 * config/v850/v850.c: Likewise.
9712 * config/vax/vax.c: Likewise.
9713 * config/visium/visium.c: Likewise.
9714 * config/xtensa/xtensa.c: Likewise.
9716 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
9719 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
9720 constraints since the -mupper-regs-* switches have been
9722 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
9723 into a vector from a double word element that was extracted from
9724 another vector, and eliminate extra XXPERMDI instructions.
9725 (vsx_concat_<mode>_2): Likewise.
9726 (vsx_concat_<mode>_3): Likewise.
9727 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
9728 concat to allow optimizing inserts from previous extracts.
9730 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
9732 * config/i386/i386.c (ix86_stack_protect_guard): Generate
9733 memory reference to a SSP offset in TLS address space.
9734 (ix86_print_operand) <case '@'>: Remove.
9735 (ix86_print_operand_punct_valid_p): Remove '@' code.
9736 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
9738 (stack_tls_protect_set_<mode>): Remove.
9739 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
9740 (stack_tls_protect_test_<mode>): Remove.
9741 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
9743 2017-08-07 Olivier Hainque <hainque@adacore.com>
9746 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
9748 2017-08-07 Douglas Rupp <rupp@adacore.com>
9750 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
9751 variable was referenced as multidir in command.
9753 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9756 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
9759 2017-08-07 Martin Liska <mliska@suse.cz>
9761 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
9762 * config/rl78/rl78.c: Add include of attribs.h.
9763 * config/sh/sh.c: Likewise.
9764 * config/v850/v850.c: Likewise.
9766 2017-08-07 Tom de Vries <tom@codesourcery.com>
9769 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
9771 2017-08-07 Martin Liska <mliska@suse.cz>
9773 * config/mips/mips.c: Include attribs.h.
9775 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
9778 * doc/invoke.texi: Document change in behvaior for -Ofast for
9781 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
9783 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
9785 (aarch64_pop_regs): Likewise.
9786 (aarch64_gen_load_pair): Likewise.
9787 (aarch64_save_callee_saves): Likewise.
9788 (aarch64_restore_callee_saves): Likewise.
9790 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9792 * config/i386/i386.c: Revert the last change.
9794 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9797 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
9799 (ix86_finalize_stack_frame_flags): This. Also clear
9800 frame_pointer_needed if -fno-omit-frame-pointer is used without
9802 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
9803 with ix86_finalize_stack_frame_flags.
9804 (ix86_expand_epilogue): Likewise.
9805 (ix86_expand_split_stack_prologue): Likewise.
9807 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
9810 * config/i386/i386.c (get_builtin_code_for_version): Set priority
9811 to P_AES for Westmere.
9813 2017-08-07 Jonathan Yong <10walls@gmail.com>
9815 * config/i386/mingw.opt (fset-stack-executable): Removed.
9816 * config/i386/cygming.opt (fset-stack-executable): Moved
9818 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
9820 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
9822 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
9824 2017-08-07 Marek Polacek <polacek@redhat.com>
9827 * fold-const.c (fold_indirect_ref_1): Check type_domain.
9829 2017-08-07 Martin Liska <mliska@suse.cz>
9831 * attribs.h (canonicalize_attr_name): New function.
9832 (cmp_attribs): Move from c-format.c and adjusted.
9833 (is_attribute_p): Moved from tree.h.
9834 * tree-inline.c: Add new includes.
9835 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
9836 (private_is_attribute_p): Remove.
9837 (private_lookup_attribute): Likewise.
9838 (private_lookup_attribute_by_prefix): Simplify.
9839 (remove_attribute): Use is_attribute_p.
9840 * tree.h: Remove removed declarations.
9842 2017-08-07 Jakub Jelinek <jakub@redhat.com>
9845 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
9846 instead of computing it in the function. Formatting fix.
9847 (expand_case): Don't rely on default_edge being the first edge,
9848 clear it if removing it, pass default_edge to
9849 emit_case_dispatch_table.
9850 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
9853 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
9855 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
9856 insn in the function, emit NOP after the insn.
9858 2017-08-06 Tom de Vries <tom@codesourcery.com>
9860 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
9863 2017-08-06 Tom de Vries <tom@codesourcery.com>
9865 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
9868 2017-08-04 Yury Gribov <tetra2005@gmail.com>
9870 PR tree-optimization/57371
9871 * match.pd: New pattern.
9873 2017-08-04 Marek Polacek <polacek@redhat.com>
9876 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
9877 perform the computation in offset_int.
9879 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9881 PR tree-optimization/81136
9882 * tree-vectorizer.h: Include tree-hash-traits.h.
9883 (vec_base_alignments): New typedef.
9884 (vec_info): Add a base_alignments field.
9885 (vect_record_base_alignments): Declare.
9886 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
9888 (DR_IS_CONDITIONAL_IN_STMT): New macro.
9889 (create_data_ref): Add an is_conditional_in_stmt argument.
9890 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
9891 the is_conditional_in_stmt field.
9892 (data_ref_loc): Add an is_conditional_in_stmt field.
9893 (get_references_in_stmt): Set the is_conditional_in_stmt field.
9894 (find_data_references_in_stmt): Update call to create_data_ref.
9895 (graphite_find_data_references_in_stmt): Likewise.
9896 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
9897 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
9898 (vect_record_base_alignment): New function.
9899 (vect_record_base_alignments): Likewise.
9900 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
9901 for nested statements even if we fail to compute a misalignment.
9902 Use pooled base alignments for unconditional references.
9903 (vect_find_same_alignment_drs): Compare base addresses instead
9905 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
9906 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
9908 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9910 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
9911 Add an explicit name for the enum. Use auto_vec for slp_instances
9913 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
9915 (_bb_vec_info): Add a constructor and destructor.
9916 (vinfo_for_stmt): Return NULL for uids of -1 as well.
9917 (destroy_loop_vec_info): Delete.
9918 (vect_destroy_datarefs): Likewise.
9919 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
9920 (vec_info::vec_info): New function.
9921 (vec_info::~vec_info): Likewise.
9922 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
9923 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
9924 destroy_loop_vec_info.
9925 * tree-vect-loop.c (new_loop_vec_info): Replace with...
9926 (_loop_vec_info::_loop_vec_info): ...this.
9927 (destroy_loop_vec_info): Replace with...
9928 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
9929 the stmt_vec_infos. Leave handling of vec_info information to its
9930 destructor. Remove explicit vector releases.
9931 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
9932 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
9933 * tree-vect-slp.c (new_bb_vec_info): Replace with...
9934 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
9935 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
9936 (destroy_bb_vec_info): Replace with...
9937 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
9938 information to its destructor.
9939 (vect_slp_analyze_bb_1): Use new and delete instead of
9940 new_bb_vec_info and destroy_bb_vec_info.
9941 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
9944 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
9946 * tree-data-ref.h (subscript): Add access_fn field.
9947 (data_dependence_relation): Add could_be_independent_p.
9948 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
9949 (same_access_functions): Move to tree-data-ref.c.
9950 * tree-data-ref.c (ref_contains_union_access_p): New function.
9951 (access_fn_component_p): Likewise.
9952 (access_fn_components_comparable_p): Likewise.
9953 (dr_analyze_indices): Add a reference to access_fn_component_p.
9954 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
9956 (constant_access_functions): Likewise.
9957 (add_other_self_distances): Likewise.
9958 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
9959 (initialize_data_dependence_relation): Use XCNEW and remove
9960 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
9961 of access functions that have the same type. Allow the
9962 subsequence to end with different bases in some circumstances.
9963 Record the chosen access functions in SUB_ACCESS_FN.
9964 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
9965 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
9966 (subscript_dependence_tester_1): Likewise dra and drb.
9967 (build_classic_dist_vector): Update calls accordingly.
9968 (subscript_dependence_tester): Likewise.
9969 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
9970 DDR_COULD_BE_INDEPENDENT_P.
9971 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
9972 comp_alias_ddrs instead of may_alias_ddrs.
9973 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
9975 (vect_analyze_data_ref_dependence): Use it if
9976 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
9977 distance vectors if that fails.
9978 (dependence_distance_ge_vf): New function.
9979 (vect_prune_runtime_alias_test_list): Use it. Don't clear
9980 LOOP_VINFO_MAY_ALIAS_DDRS.
9982 2017-08-04 Richard Biener <rguenther@suse.de>
9985 * fold-const.c (fold_binary_loc): Properly restrict
9986 minus_var0 && minus_var1 case when associating undefined overflow
9989 2017-08-04 Tom de Vries <tom@codesourcery.com>
9991 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
9993 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9995 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
9996 Don't start diagnostic messages with a capital letter.
9997 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9999 (rs6000_invalid_builtin): Likewise.
10000 (rs6000_trampoline_init): Likewise.
10002 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10005 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
10006 after setting changeable df flags.
10008 2017-08-03 Richard Biener <rguenther@suse.de>
10010 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
10011 up if the use is in USE - X.
10013 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
10015 * toplev.c (dumpfile.h): New include.
10016 (internal_error_reentered): New static function. Use it...
10017 (internal_error_function): ...here to handle reentered internal_error.
10019 2017-08-03 Richard Biener <rguenther@suse.de>
10021 PR middle-end/81148
10022 * fold-const.c (split_tree): Add minus_var and minus_con
10023 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
10024 here but always use minus_*.
10025 (associate_trees): Assert we never associate with MINUS_EXPR
10026 and NULL first operand. Do not recurse for PLUS_EXPR operands
10027 when associating as MINUS_EXPR either.
10028 (fold_binary_loc): Track minus_var and minus_con.
10030 2017-08-03 Tom de Vries <tom@codesourcery.com>
10033 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
10034 ACCEL_COMPILER, apply finish_options on
10035 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
10037 2017-08-03 Tom de Vries <tom@codesourcery.com>
10040 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
10041 function_entry_patch_area_size > 0.
10043 2017-08-03 Jakub Jelinek <jakub@redhat.com>
10046 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
10047 instead of 10??LU, perform unit multiplication in wide_int,
10048 don't change alloc_object_size_limit if the limit is larger
10051 PR tree-optimization/81655
10052 PR tree-optimization/81588
10053 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
10054 the case when ranges[i].low and high are 1 for unsigned type with
10057 PR middle-end/81052
10058 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
10059 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
10061 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10063 * tree-vrp.h: Add include guard.
10065 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
10068 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
10069 (ud2): New insn pattern.
10070 * config/i386/i386.c (ix86_expand_epilogue):
10071 For naked functions, generate ud2 instead of trap insn.
10073 2017-08-02 Marek Polacek <polacek@redhat.com>
10076 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
10078 2017-08-02 Tom de Vries <tom@codesourcery.com>
10079 Cesar Philippidis <cesar@codesourcery.com>
10081 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
10082 Add missing edge probabilities.
10084 2017-08-02 Tamar Christina <tamar.christina@arm.com>
10086 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
10087 Correct endianness.
10089 2017-08-02 Jakub Jelinek <jakub@redhat.com>
10091 PR middle-end/79499
10092 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
10093 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
10094 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
10095 prologue_seq sequences - if any.
10097 2017-08-02 Richard Biener <rguenther@suse.de>
10099 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
10100 via vectors if supported, integer extracts via punning if supported
10101 or otherwise vector extracts.
10103 2017-08-02 Richard Biener <rguenther@suse.de>
10105 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
10107 (bitmap_insert_into_set): ... this.
10109 2017-08-02 Richard Biener <rguenther@suse.de>
10111 PR tree-optimization/81633
10113 2015-08-17 Alan Hayward <alan.hayward@arm.com>
10115 PR tree-optimization/71752
10116 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
10118 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
10120 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
10121 (machine_function::call_ms2sysv_pad_out): Remove field.
10122 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
10123 (ix86_compute_frame_layout): Likewise.
10125 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
10128 * config/i386/i386.c (ix86_set_func_type): Disallow naked
10129 attribute with interrupt attribute.
10131 2017-08-01 Andrew Pinski <apinski@cavium.com>
10133 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
10134 BIT_INSERT_EXPR's operand 1
10135 to see if the types precision matches.
10137 2017-08-01 Martin Liska <mliska@suse.cz>
10139 PR middle-end/70140
10140 * builtins.c (expand_builtin_memcpy_args): Remove.
10141 (expand_builtin_memcpy): Call newly added function
10142 expand_builtin_memory_copy_args.
10143 (expand_builtin_memcpy_with_bounds): Likewise.
10144 (expand_builtin_mempcpy): Remove last argument.
10145 (expand_builtin_mempcpy_with_bounds): Likewise.
10146 (expand_builtin_memory_copy_args): New function created from
10147 expand_builtin_mempcpy_args with small modifications.
10148 (expand_builtin_mempcpy_args): Remove.
10149 (expand_builtin_stpcpy): Remove unused argument.
10150 (expand_builtin): Likewise.
10151 (expand_builtin_with_bounds): Likewise.
10153 2017-08-01 Martin Liska <mliska@suse.cz>
10156 Make mempcpy more optimal (PR middle-end/70140).
10158 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10161 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
10162 __builtin_vec_cmpne verify both arguments are compatible vectors
10163 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
10164 verify arg1_type is a pointer or array type. For __builtin_vec_st,
10165 move computation of aligned to after checking the argument types.
10169 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10170 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
10173 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
10174 Doug Rupp <rupp@adacore.com>
10175 Olivier Hainque <hainque@adacore.com>
10177 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
10178 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
10180 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
10181 for TARGET_OS_CPP_BUILTIN.
10182 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
10183 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
10185 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
10186 passing required abi options to the assembler for EABI configurations.
10187 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
10188 of .text.hot and .text.unlikely sections for kernel modules when
10189 using ARM style exceptions.
10190 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
10191 options. Add EXTRA_CC1_SPEC.
10192 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
10194 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
10196 (ARM_TARGET2_DWARF_FORMAT): Define.
10197 * config/arm/t-vxworks: Adjust multilib control to removal of the
10198 Diab command line options.
10200 2017-08-01 Martin Liska <mliska@suse.cz>
10202 PR gcov-profile/81561
10203 * gcov.c (unblock): Make unblocking safe as we need to preserve
10204 index correspondence of blocks and block_lists.
10206 2017-08-01 Richard Biener <rguenther@suse.de>
10208 PR tree-optimization/81181
10209 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
10210 (compute_antic): ... end of iteration here.
10212 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
10214 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
10215 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
10216 (ftree-slp-vectorize): Likewise.
10217 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
10218 can no longer be set independent of flag_tree_loop_vectorize.
10219 * omp-general.c (emp_max_vf): Likewise.
10220 * opts.c (enable_fdo_optimizations): Remove references to
10221 flag_tree_vectorize, these are now implicit.
10222 (common_handle_option): Remove handling for OPT_ftree_vectorize,
10223 and leave it for the options machinery.
10225 2017-08-01 Martin Liska <mliska@suse.cz>
10227 PR middle-end/70140
10228 * builtins.c (expand_builtin_memcpy_args): Remove.
10229 (expand_builtin_memcpy): Call newly added function
10230 expand_builtin_memory_copy_args.
10231 (expand_builtin_memcpy_with_bounds): Likewise.
10232 (expand_builtin_mempcpy): Remove last argument.
10233 (expand_builtin_mempcpy_with_bounds): Likewise.
10234 (expand_builtin_memory_copy_args): New function created from
10235 expand_builtin_mempcpy_args with small modifications.
10236 (expand_builtin_mempcpy_args): Remove.
10237 (expand_builtin_stpcpy): Remove unused argument.
10238 (expand_builtin): Likewise.
10239 (expand_builtin_with_bounds): Likewise.
10241 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10244 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
10245 print "ds:" only for immediates in generic address space.
10247 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
10250 * config/i386/i386.c (ix86_funciton_naked): New prototype.
10251 (ix86_function_ok_for_sibcall): Return false for naked functions.
10253 2017-08-01 Richard Biener <rguenther@suse.de>
10255 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
10256 (compute_antic): Seed worklist with exit block predecessors.
10257 * cfganal.c (dfs_find_deadend): For a cycle return the source
10258 of the edge closing it.
10260 2017-08-01 Tamar Christina <tamar.christina@arm.com>
10262 * config/aarch64/aarch64.c
10263 (aarch64_can_const_movi_rtx_p): Move 0 check.
10265 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10267 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
10268 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
10270 * match.pd: Ditto in address comparison pattern.
10272 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10274 PR tree-optimization/81627
10275 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
10276 closed ssa form for store-store chain.
10278 2017-08-01 Bin Cheng <bin.cheng@arm.com>
10280 PR tree-optimization/81620
10281 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
10282 for store-store chain.
10284 2017-08-01 Jakub Jelinek <jakub@redhat.com>
10286 PR tree-optimization/81588
10287 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
10288 ranges[i].in_p, invert comparison code ccode. For >/>=,
10289 swap rhs1 and rhs2 and comparison code unconditionally,
10290 for </<= don't do that. Don't swap rhs1/rhs2 again if
10291 ranges[i].in_p, instead invert comparison code ccode if
10292 opcode or oe->rank is BIT_IOR_EXPR.
10295 * optabs.def (vec_extract_optab, vec_init_optab): Change from
10296 a direct optab to conversion optab.
10297 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
10298 with GET_MODE_INNER as last argument instead of optab_handler.
10299 * expmed.c (extract_bit_field_1): Likewise. Use vector from
10300 vector extraction if possible and optab is available.
10301 * expr.c (store_constructor): Use convert_optab_handler instead
10302 of optab_handler. Use vector initialization from smaller
10303 vectors if possible and optab is available.
10304 * tree-vect-stmts.c (vectorizable_load): Likewise.
10305 * doc/md.texi (vec_extract, vec_init): Document that the optabs
10306 now have two modes.
10307 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
10308 of vec_init from half-sized vectors with the same element mode.
10309 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
10310 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
10311 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
10312 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
10313 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
10314 after mode in gen_vec_extract* calls.
10315 (vec_extract<mode>): Renamed to ...
10316 (vec_extract<mode><ssescalarmodelower>): ... this.
10317 (vec_extract<mode><ssehalfvecmodelower>): New expander.
10318 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
10319 element mode after mode in gen_vec_init* calls.
10320 (VEC_INIT_HALF_MODE): New mode iterator.
10321 (vec_init<mode>): Renamed to ...
10322 (vec_init<mode><ssescalarmodelower>): ... this.
10323 (vec_init<mode><ssehalfvecmodelower>): New expander.
10324 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
10325 (vec_extractv2sfsf): ... this.
10326 (vec_initv2sf): Renamed to ...
10327 (vec_initv2sfsf): ... this.
10328 (vec_extractv2si): Renamed to ...
10329 (vec_extractv2sisi): ... this.
10330 (vec_initv2si): Renamed to ...
10331 (vec_initv2sisi): ... this.
10332 (vec_extractv4hi): Renamed to ...
10333 (vec_extractv4hihi): ... this.
10334 (vec_initv4hi): Renamed to ...
10335 (vec_initv4hihi): ... this.
10336 (vec_extractv8qi): Renamed to ...
10337 (vec_extractv8qiqi): ... this.
10338 (vec_initv8qi): Renamed to ...
10339 (vec_initv8qiqi): ... this.
10340 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
10341 (vec_init<mode>): Renamed to ...
10342 (vec_init<mode><VEC_base_l>): ... this.
10343 (vec_extract<mode>): Renamed to ...
10344 (vec_extract<mode><VEC_base_l>): ... this.
10345 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
10346 (vec_initv2sfsf): ... this.
10347 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
10348 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10349 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
10350 element mode after mode in gen_vec_init* calls.
10351 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
10352 (vec_init<mode><Vel>): ... this.
10353 (vec_extract<mode>): Renamed to ...
10354 (vec_extract<mode><Vel>): ... this.
10355 * config/aarch64/iterators.md (Vel): New mode attribute.
10356 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
10357 Add element mode after mode in gen_vec_extract* calls.
10358 * config/s390/vector.md (non_vec_l): New mode attribute.
10359 (vec_extract<mode>): Renamed to ...
10360 (vec_extract<mode><non_vec_l>): ... this.
10361 (vec_init<mode>): Renamed to ...
10362 (vec_init<mode><non_vec_l>): ... this.
10363 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
10364 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
10366 * config/arm/iterators.md (V_elem_l): New mode attribute.
10367 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
10368 (vec_extract<mode><V_elem_l>): ... this.
10369 (vec_extractv2di): Renamed to ...
10370 (vec_extractv2didi): ... this.
10371 (vec_init<mode>): Renamed to ...
10372 (vec_init<mode><V_elem_l>): ... this.
10373 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
10374 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
10375 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
10376 Add element mode after gen_vec_extract* calls.
10377 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
10378 (vec_init<mode><unitmode>): ... this.
10379 (vec_extract<mode>): Renamed to ...
10380 (vec_extract<mode><unitmode>): ... this.
10381 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
10382 (vec_init<mode><unitmode>): ... this.
10383 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
10384 (vec_initv2sfsf): ... this.
10385 (vec_extractv2sf): Renamed to ...
10386 (vec_extractv2sfsf): ... this.
10387 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
10388 Add element mode after gen_vec_extract* calls.
10389 * config/mips/mips.md (unitmode): New mode iterator.
10390 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
10391 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
10392 * config/spu/spu.md (inner_l): New mode attribute.
10393 (vec_init<mode>): Renamed to ...
10394 (vec_init<mode><inner_l>): ... this.
10395 (vec_extract<mode>): Renamed to ...
10396 (vec_extract<mode><inner_l>): ... this.
10397 * config/sparc/sparc.md (veltmode): New mode iterator.
10398 (vec_init<VMALL:mode>): Renamed to ...
10399 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
10400 * config/ia64/vect.md (vec_initv2si): Renamed to ...
10401 (vec_initv2sisi): ... this.
10402 (vec_initv2sf): Renamed to ...
10403 (vec_initv2sfsf): ... this.
10404 (vec_extractv2sf): Renamed to ...
10405 (vec_extractv2sfsf): ... this.
10406 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
10407 (vec_init<mode>): Renamed to ...
10408 (vec_init<mode><VEC_base_l>): ... this.
10409 (vec_extract<mode>): Renamed to ...
10410 (vec_extract<mode><VEC_base_l>): ... this.
10411 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
10412 (vec_initv2sfsf): ... this.
10413 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
10414 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
10415 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
10416 gen_vec_init* calls.
10418 2017-08-01 Richard Biener <rguenther@suse.de>
10420 PR tree-optimization/81297
10421 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
10422 TREE_OVERFLOW from INTEGER_CSTs.
10424 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
10426 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
10428 2017-07-31 Carl Love <cel@us.ibm.com>
10430 * config/rs6000/rs6000-c: Add support for built-in functions
10431 vector signed char vec_xl_be (signed long long, signed char *);
10432 vector unsigned char vec_xl_be (signed long long, unsigned char *);
10433 vector signed int vec_xl_be (signed long long, signed int *);
10434 vector unsigned int vec_xl_be (signed long long, unsigned int *);
10435 vector signed long long vec_xl_be (signed long long, signed long long *);
10436 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
10437 vector signed short vec_xl_be (signed long long, signed short *);
10438 vector unsigned short vec_xl_be (signed long long, unsigned short *);
10439 vector double vec_xl_be (signed long long, double *);
10440 vector float vec_xl_be (signed long long, float *);
10441 * config/rs6000/altivec.h (vec_xl_be): Add #define.
10442 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
10443 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
10445 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
10446 (altivec_expand_builtin): Add switch statement to call
10447 altivec_expand_xl_be for each builtin.
10448 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
10449 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
10450 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
10451 __builtin_vsx_le_be_v16qi.
10452 * doc/extend.texi: Update the built-in documentation file for the
10453 new built-in functions.
10455 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10458 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
10460 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
10462 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10464 * config.gcc: Add z14.
10465 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
10466 CPU model numbers for z13s and z14.
10467 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
10469 * config/s390/s390-opts.h (enum processor_type): Rename
10470 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10471 * config/s390/s390.c (processor_table): Add field for CPU name to
10472 be passed to Binutils.
10473 (s390_asm_output_machine_for_arch): Use the new field in
10474 processor_table for Binutils.
10475 (s390_expand_builtin): Replace arch12 with z14.
10476 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
10477 (s390_get_sched_attrmask): Likewise.
10478 (s390_get_unit_mask): Likewise.
10479 * config/s390/s390.opt: Add z14 to processor_type enum.
10481 2017-07-31 Martin Jambor <mjambor@suse.cz>
10484 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
10485 regardless of optimization level.
10487 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
10488 Martin Liska <mliska@suse.cz>
10490 * predict.def: Remove old comment and adjust probability.
10491 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
10492 PREDICT statements.
10494 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
10497 * config/i386/i386.c (ix86_function_naked): New function.
10498 (ix86_can_use_return_insn_p): Return false for naked functions.
10499 (ix86_expand_prologue): Skip prologue for naked functions.
10500 (ix86_expand_epilogue): Skip epilogue for naked functions
10501 and emit trap instruction.
10502 (ix86_warn_func_return): New function.
10503 (ix86_attribute_table): Add "naked" attribute specification.
10504 (TARGET_WARN_FUNC_RETURN): Define.
10505 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
10507 2017-07-31 Martin Liska <mliska@suse.cz>
10509 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
10510 (dump_gimple_bb_header): Always dump BB info.
10511 (pp_cfg_jump): Do not append info about BB when dumping a jump.
10513 2017-07-31 Martin Liska <mliska@suse.cz>
10516 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
10517 also with current_function_decl non-null equality.
10519 2017-07-31 Jakub Jelinek <jakub@redhat.com>
10522 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
10523 change type to the element type, instead add eltype variable and
10524 use it where we are interested in the element type.
10526 PR tree-optimization/81603
10527 * ipa-polymorphic-call.c
10528 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
10529 offset arithmetic in offset_int, bail out if the resulting bit offset
10530 doesn't fit into shwi.
10532 2017-07-31 Martin Liska <mliska@suse.cz>
10534 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
10535 (gimplify_save_expr): Fix comment.
10537 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
10540 * config/i386/i386.c (ix86_function_arg): Update arguments for
10542 (ix86_compute_frame_layout): Set the initial stack offset to
10543 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
10544 INCOMING_FRAME_SP_OFFSET.
10545 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
10546 stack before exception handler returns.
10547 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
10548 the 'ERROR_CODE' for exception handler.
10550 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
10552 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
10553 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
10554 (ASM_OUTPUT_REG_POP): Ditto.
10555 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
10556 instead of asm_fprintf to output pure string.
10558 2017-07-29 Jakub Jelinek <jakub@redhat.com>
10560 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
10561 to imported_module_or_decl hook.
10562 (debug_nothing_tree_tree_tree_bool): Remove.
10563 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
10564 * debug.c (do_nothing_debug_hooks): Use
10565 debug_nothing_tree_tree_tree_bool_bool instead of
10566 debug_nothing_tree_tree_tree_bool.
10567 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
10568 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
10569 * sdbout.c (sdb_debug_hooks): Likewise.
10570 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
10571 (gen_namespace_die): Add DW_AT_export_symbols attribute if
10573 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
10574 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
10575 attribute, don't add anything.
10577 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10579 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
10580 (fold_build2_stat_loc): Likewise.
10581 (fold_build3_stat_loc): Likewise.
10582 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
10583 (fold_build1_loc): Remove macro.
10584 (fold_build2_loc): Likewise.
10585 (fold_build3_loc): Likewise.
10587 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10589 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
10590 (gimple_build_debug_bind_source_stat): Likewise.
10591 * gimple.h (gimple_build_debug_bind): Remove macro.
10592 (gimple_build_debug_bind_source): Likewise.
10594 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10596 * bitmap.c (bitmap_alloc): Adjust.
10597 (bitmap_gc_alloc): Likewise.
10598 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
10600 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10602 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
10603 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
10604 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
10605 (bitmap_gc_alloc_stat): Likewise.
10606 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
10608 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10610 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
10611 * rtl.h (shallow_copy_rtx): Remove macro.
10613 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10615 * emit-rtl.c (gen_raw_REG): Adjust.
10616 * gengenrtl.c (gendef): Likewise.
10617 * rtl.c (rtx_alloc_stat): Remove _stat from name.
10618 * rtl.h (rtx_alloc): Remove macro.
10620 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10622 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
10623 (build_tree_list_stat): Likewise.
10624 * tree.h (build_tree_list): Remove macro.
10625 (build_tree_list_vec): Likewise.
10627 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10629 * tree.c (make_vector_stat): Remove _stat from name.
10630 (build_vector_stat): Likewise.
10631 * tree.h (make_vector_stat): Remove macro.
10632 (build_vector_stat): Likewise.
10634 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10636 * tree.h (build_var_debug_value): Remove prototype.
10638 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10640 * tree.c (tree_cons_stat): Remove _stat from name.
10641 * tree.h (tree_cons): Remove macro.
10643 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10645 * tree.c (build_vl_exp_stat): Remove _stat from name.
10646 * tree.h (build_vl_exp): Remove macro.
10648 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10650 * tree.c (build_decl_stat): Remove _stat from name.
10651 * tree.h (build_decl): Remove macro.
10653 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10655 * gimple.c (gimple_build_with_ops_stat): Adjust.
10656 (gimple_alloc_stat): Remove _stat from name.
10657 * gimple.h (gimple_alloc): Remove macro.
10659 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10661 * tree.c (make_tree_vec_stat): Remove _stat from name.
10662 (grow_tree_vec_stat): Likewise.
10663 * tree.h (make_tree_vec_stat): Adjust prototype.
10664 (grow_tree_vec_stat): Likewise.
10665 (make_tree_vec): Remove macro.
10666 (grow_tree_vec): Likewise.
10668 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10670 * fold-const.c (fold_build1_stat_loc): Adjust.
10671 (fold_build2_stat_loc): Likewise.
10672 (fold_build3_stat_loc): Likewise.
10673 * tree.c (build0_stat): Remove _stat from name.
10674 (build1_stat): Likewise.
10675 (build2_stat): Likewise.
10676 (build3_stat): Likewise.
10677 (build4_stat): Likewise.
10678 (build5_stat): Likewise.
10679 * tree.h (build1_loc): Remove macro, and rename _stat function
10681 (build2_loc): Likewise.
10682 (build3_loc): Likewise.
10683 (build4_loc): Likewise.
10684 (build5_loc): Likewise.
10686 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10688 * tree.c (make_int_cst_stat): Remove _stat from name.
10689 * tree.h (make_int_cst_stat): Adjust prototype.
10690 (make_int_cst): Remove macro.
10692 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10694 * tree.c (make_tre_binfo_stat): Remove _stat from name.
10695 * tree.h (make_tree_binfo_stat): Adjust prototype.
10696 (make_tree_binfo): Remove.
10698 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10700 * tree.c (copy_node_stat): Rename to copy_node.
10701 (build_distinct_type_copy): Adjust.
10702 * tree.h (copy_node_stat): Adjust prototype.
10703 (copy_node): Remove macro.
10705 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
10707 * tree.c (make_node_stat): rename to make_node.
10708 (build_tree_list_stat): Adjust.
10709 (build0_stat): Likewise.
10710 (build2_stat): Likewise.
10711 (build3_stat): Likewise.
10712 (build4_stat): Likewise.
10713 (build5_stat): Likewise.
10714 (build_decl_stat): Likewise.
10715 * tree.h (make_node_stat): Adjust prototype.
10716 (make_node): remove macro.
10718 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
10720 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
10721 (PPC_FEATURE2_SCV): Likewise.
10722 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
10724 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10726 * config/aarch64/aarch64.c
10727 (aarch64_internal_mov_immediate): Add new special pattern.
10728 * config/aarch64/aarch64.md (*movdi_aarch64):
10729 Add reg/32bit const mov case.
10731 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10732 Richard Sandiford <richard.sandiford@linaro.org>
10734 * config/aarch64/aarch64.md (mov<mode>): Generalize.
10735 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
10736 Add integer and movi cases.
10737 (movi-split-hf-df-sf split, fp16): New.
10738 (enabled): Added TARGET_FP_F16INST.
10739 * config/aarch64/iterators.md (GPF_HF): New.
10740 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
10742 2017-07-28 Tamar Christina <tamar.christina@arm.com>
10744 * config/aarch64/aarch64.c
10745 (aarch64_simd_container_mode): Add prototype.
10746 (aarch64_expand_mov_immediate): Add HI support.
10747 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
10748 (aarch64_can_const_movi_rtx_p): New.
10749 (aarch64_preferred_reload_class):
10750 Remove restrictions of using FP registers for certain SIMD operations.
10751 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
10752 (aarch64_valid_floating_const): Add integer move validation.
10753 (aarch64_simd_imm_scalar_p): Remove.
10754 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
10755 (aarch64_legitimate_constant_p): Expand list of supported cases.
10756 * config/aarch64/aarch64-protos.h
10757 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
10758 (aarch64_reinterpret_float_as_int): New.
10759 (aarch64_simd_imm_scalar_p): Remove.
10760 * config/aarch64/constraints.md (Uvi): New.
10761 (Dd): Split into Ds and new Dd.
10762 * config/aarch64/aarch64.md (*movsi_aarch64):
10764 (*movdi_aarch64): Add SIMD mov case.
10766 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10768 * tree-predcom.c: (struct chain): Handle store-store chain in which
10769 stores for elimination only store loop invariant values.
10770 (execute_pred_commoning_chain): Ditto.
10771 (prepare_initializers_chain_store_elim): Ditto.
10772 (prepare_finalizers): Ditto.
10773 (is_inv_store_elimination_chain): New function.
10774 (initialize_root_vars_store_elim_1): New function.
10776 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10778 * tree-predcom.c: Revise general description of the pass.
10779 (enum chain_type): New enum type for store elimination.
10780 (struct chain): New field supporting store elimination.
10781 (struct component): Ditto.
10782 (dump_chain): Dump store-stores chain.
10783 (release_chain): Release resources.
10784 (split_data_refs_to_components): Compute and create component
10785 contains only stores for elimination.
10786 (get_chain_last_ref_at): New function.
10787 (make_invariant_chain): Initialization.
10788 (make_rooted_chain): Specify chain type in parameter and record it.
10789 (add_looparound_copies): Skip for store-stores chain.
10790 (determine_roots_comp): Compute type of chain and pass it to
10792 (initialize_root_vars_store_elim_2): New function.
10793 (finalize_eliminated_stores): New function.
10794 (remove_stmt): Handle store for elimination.
10795 (execute_pred_commoning_chain): Execute predictive commoning on
10796 store-store chains.
10797 (determine_unroll_factor): Skip unroll for store-stores chain.
10798 (prepare_initializers_chain_store_elim): New function.
10799 (prepare_initializers_chain): Hanlde store-store chain.
10800 (prepare_finalizers_chain, prepare_finalizers): New function.
10801 (tree_predictive_commoning_loop): Return integer value indicating
10802 if loop is unrolled or lcssa form is corrupted.
10803 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
10805 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10807 * tree-predcom.c (initialize_root): Delete.
10808 (execute_pred_commoning_chain): Initialize root vars and replace
10809 reference of non-combined chain directly, rather than call above
10812 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10814 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
10815 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
10817 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10819 * tree-predcom.c (struct chain): New field init_seq.
10820 (release_chain): Release init_seq.
10821 (prepare_initializers_chain): Record intialization stmts in above
10823 (insert_init_seqs): New function.
10824 (tree_predictive_commoning_loop): Call insert_init_seqs.
10826 2017-07-28 Bin Cheng <bin.cheng@arm.com>
10828 * tree-predcom.c (determine_roots_comp): Skip trivial components.
10830 2017-07-28 Richard Biener <rguenther@suse.de>
10832 * match.pd: Remove superfluous :c.
10833 * genmatch.c (simplify::id): Add member.
10834 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
10836 (current_id): New global.
10837 (dt_node::parent): Move from ...
10838 (dt_operand::parent): ... here. Add for_id member.
10839 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
10840 (decision_tree::find_node): Relax order requirement when
10841 merging DT_TRUE nodes to ones inbetween the current simplify
10842 and the one we try to merge with. Add diagnostic whenever
10843 we need to enforce pattern order by not merging.
10844 (decision_tree::insert): Set current_id.
10845 (decision_tree::print_node): Dump parent node and for_id.
10846 (parser::last_id): Add member.
10847 (parser::push_simplify): Assign unique id.
10848 (parser::parser): Initialize last_id.
10850 2017-07-28 Martin Liska <mliska@suse.cz>
10853 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
10854 gimple_build_debug_bind.
10856 2017-07-28 Richard Biener <rguenther@suse.de>
10858 PR tree-optimization/81502
10859 * match.pd: Add pattern combining BIT_INSERT_EXPR with
10861 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
10863 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
10864 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
10865 for BIT_FIELD_REF args.
10866 * fold-const.c (make_bit_field_ref): Likewise.
10867 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
10869 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10872 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
10873 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
10874 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
10875 Or it into SANITIZER_UNDEFINED.
10876 * ubsan.c: Include gimple-fold.h and varasm.h.
10877 (ubsan_expand_ptr_ifn): New function.
10878 (instrument_pointer_overflow): New function.
10879 (maybe_instrument_pointer_overflow): New function.
10880 (instrument_object_size): Formatting fix.
10881 (pass_ubsan::execute): Call instrument_pointer_overflow
10882 and maybe_instrument_pointer_overflow.
10883 * internal-fn.c (expand_UBSAN_PTR): New function.
10884 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
10885 * sanitizer.def (__ubsan_handle_pointer_overflow,
10886 __ubsan_handle_pointer_overflow_abort): New builtins.
10887 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
10888 * internal-fn.def (UBSAN_PTR): New internal function.
10889 * opts.c (sanitizer_opts): Add pointer-overflow.
10890 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
10891 * fold-const.c (build_range_check): Compute pointer range check in
10892 integral type if pointer arithmetics would be needed. Formatting
10895 2017-07-28 Martin Liska <mliska@suse.cz>
10898 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
10899 parameters that are of a variable-length.
10901 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
10903 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
10905 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
10906 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
10907 (CRT_CALL_STATIC_FUNCTION): Likewise.
10908 (ASM_DEFAULT_SPEC): New define.
10909 (ASM_SPEC32): Likewise.
10910 (ASM_SPEC64): Likewise.
10911 (ASM_SPEC_COMMON): Likewise.
10912 (ASM_SPEC): Likewise.
10913 (INVALID_64BIT): Likewise.
10914 (LINK_OS_DEFAULT_SPEC): Likewise.
10915 (LINK_OS_SPEC32): Likewise.
10916 (LINK_OS_SPEC64): Likewise.
10917 (POWERPC_LINUX): Likewise.
10918 (PTRDIFF_TYPE): Likewise.
10919 (RESTORE_FP_PREFIX): Likewise.
10920 (RESTORE_FP_SUFFIX): Likewise.
10921 (SAVE_FP_PREFIX): Likewise.
10922 (SAVE_FP_SUFFIX): Likewise.
10923 (SIZE_TYPE): Likewise.
10924 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
10925 (TARGET_64BIT): Likewise.
10926 (TARGET_64BIT): Likewise.
10927 (TARGET_AIX): Likewise.
10928 (WCHAR_TYPE_SIZE): Likewise.
10929 (WCHAR_TYPE): Undefine.
10930 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
10931 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
10932 (CPP_OS_RTEMS_SPEC): Delete.
10933 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
10934 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
10936 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
10938 2017-07-28 Jakub Jelinek <jakub@redhat.com>
10940 PR tree-optimization/81578
10941 * tree-parloops.c (build_new_reduction): Bail out if
10942 reduction_code isn't one of the standard OpenMP reductions.
10943 Move the details printing after that decision.
10945 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10947 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
10948 related to reload_in_progress.
10949 (splat_input_operand): Likewise.
10950 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
10952 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
10954 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
10955 (TARGET_INSTANTIATE_DECLS): Likewise.
10956 (legitimate_indexed_address_p): Delete reload_in_progress code.
10957 (rs6000_debug_legitimate_address_p): Likewise.
10958 (rs6000_eliminate_indexed_memrefs): Likewise.
10959 (rs6000_emit_le_vsx_store): Likewise.
10960 (rs6000_emit_move_si_sf_subreg): Likewise.
10961 (rs6000_emit_move): Likewise.
10962 (register_to_reg_type): Likewise.
10963 (rs6000_pre_atomic_barrier): Likewise.
10964 (rs6000_machopic_legitimize_pic_address): Likewise.
10965 (rs6000_allocate_stack_temp): Likewise.
10966 (rs6000_address_for_fpconvert): Likewise.
10967 (rs6000_address_for_altivec): Likewise.
10968 (rs6000_secondary_memory_needed_rtx): Delete function.
10969 (rs6000_check_sdmode): Likewise.
10970 (rs6000_alloc_sdmode_stack_slot): Likewise.
10971 (rs6000_instantiate_decls): Likewise.
10972 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
10973 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
10974 Delete reload_in_progress.
10975 (*vec_reload_and_plus_<mptrsize>): Likewise.
10976 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
10977 (vsx_div_v2di): Likewise.
10978 (vsx_udiv_v2di): Likewise.
10980 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
10982 * config/rs6000/rs6000.opt (mlra): Replace with stub.
10983 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
10984 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
10985 (rs6000_debug_reg_global): Delete print of LRA status.
10986 (rs6000_option_override_internal): Delete dead LRA related code.
10987 (rs6000_lra_p): Delete function.
10988 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
10990 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
10992 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
10993 * config/riscv/rtems.h: New file.
10995 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10996 Sudakshina Das <sudi.das@arm.com>
10998 * config/aarch64/aarch64.md
10999 (define_split for and<mode>3nr_compare): Move
11000 non aarch64_logical_operand to a register.
11001 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
11002 register immediate operand to a register.
11003 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
11005 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
11007 PR middle-end/81564
11008 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
11010 2017-07-27 Richard Biener <rguenther@suse.de>
11012 PR tree-optimization/81573
11013 PR tree-optimization/81494
11014 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
11015 multi defuse cycle case.
11017 2017-07-27 Richard Biener <rguenther@suse.de>
11019 PR tree-optimization/81571
11020 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
11023 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
11025 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
11026 earlier and only if MASK_FPU is set. Adjust formatting.
11028 2017-07-27 Martin Liska <mliska@suse.cz>
11030 * opt-functions.awk: Add validation of value of Init.
11031 * optc-gen.awk: Pass new argument.
11033 2017-07-27 Martin Liska <mliska@suse.cz>
11035 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
11036 Fix wrong condition.
11038 2017-07-27 Martin Liska <mliska@suse.cz>
11040 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
11041 BBs and edges seen by autoFDO.
11043 2017-07-27 Richard Biener <rguenther@suse.de>
11045 PR tree-optimization/81502
11046 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
11047 with incompatible but same sized type.
11048 (execute_update_addresses_taken): Likewise.
11050 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
11052 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
11053 flag_tree_loop_vectorize rather than flag_tree_vectorize.
11055 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11058 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
11059 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
11060 Change s_operand to memory_operand.
11062 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
11064 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
11065 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
11066 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
11067 Emit instructions rather than returning an expression. Handle TFmode
11068 and KFmode by casting to TImode.
11069 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
11070 (rs6000_emit_le_vsx_store): Likewise.
11071 * config/rs6000/vsx.md (VSX_TI): New iterator.
11072 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
11073 (*vsx_le_undo_permute_<mode>): Likewise.
11074 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
11075 emit the split sequence.
11076 (*vsx_le_perm_store_<mode>): Likewise.
11078 2017-07-27 Jakub Jelinek <jakub@redhat.com>
11080 PR tree-optimization/81555
11081 PR tree-optimization/81556
11082 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
11083 if true, force CHANGED for the recursive invocation.
11084 (reassociate_bb): Remember original length of ops array, pass
11085 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
11087 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
11088 attributes for noipa attribute. For naked attribute use
11089 lookup_attribute first before lookup_attribute_spec.
11090 * final.c (rest_of_handle_final): Disable IPA RA for functions with
11092 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
11093 for functions with noipa attribute.
11094 (cgraph_externally_visible_p): Return true for functions with noipa
11096 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
11097 for functions with noipa attribute.
11098 * doc/extend.texi: Document noipa function attribute.
11099 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
11100 also for functions with noipa attribute.
11101 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
11103 2017-07-26 Andrew Pinski <apinski@cavium.com>
11105 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
11106 vec_unalign_load_cost and vec_unalign_store_cost.
11108 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
11110 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
11111 -mvsx-small-integer option.
11112 (ISA_3_0_MASKS_IEEE): Likewise.
11113 (OTHER_VSX_VECTOR_MASKS): Likewise.
11114 (POWERPC_MASKS): Likewise.
11115 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
11116 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
11117 code, only testing for DImode being allowed in non-VSX floating
11119 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
11120 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
11122 (rs6000_option_override_internal): Delete -mvsx-small-integer.
11123 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
11124 TARGET_P8_VECTOR test.
11125 (rs6000_secondary_reload_simple_move): Likewise.
11126 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
11127 since TARGET_P9_VECTOR was already tested.
11128 (rs6000_opt_masks): Remove -mvsx-small-integer.
11129 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
11130 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11132 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
11133 test for TARGET_VEXTRACTUB was used, and that uses
11135 (p9 extract splitter): Likewise.
11136 (vsx_extract_<mode>_di_p9): Likewise.
11137 (vsx_extract_<mode>_store_p9): Likewise.
11138 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
11139 for TARGET_P9_VECTOR was used. Delete code that is now dead with
11140 the elimination of TARGET_VSX_SMALL_INTEGER.
11141 (vsx_extract_<mode>_p8): Likewise.
11142 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
11143 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
11144 (vsx_set_<mode>_p9): Likewise.
11145 (vsx_set_v4sf_p9): Likewise.
11146 (vsx_set_v4sf_p9_zero): Likewise.
11147 (vsx_insert_extract_v4sf_p9): Likewise.
11148 (vsx_insert_extract_v4sf_p9_2): Likewise.
11149 * config/rs6000/rs6000.md (sign extend splitter): Change
11150 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
11151 (floatsi<mode>2_lfiwax_mem): Likewise.
11152 (floatunssi<mode>2_lfiwzx_mem): Likewise.
11153 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
11154 since a test for TARGET_P9_VECTOR was used.
11155 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11156 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
11157 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11158 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11159 TARGET_P8_VECTOR test.
11160 (fix_trunc<mode>si2_stfiwx): Likewise.
11161 (fix_trunc<mode>si2_internal): Likewise.
11162 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
11163 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11165 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11166 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
11167 TARGET_P8_VECTOR test.
11168 (fixuns_trunc<mode>si2_stfiwx): Likewise.
11169 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
11170 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
11172 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
11173 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
11174 since a test for TARGET_P9_VECTOR was used.
11175 (splitter for loading small constants): Likewise.
11177 2017-07-26 Andrew Pinski <apinski@cavium.com>
11179 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
11182 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
11185 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
11186 (fp_valid_at): Likewise.
11188 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11190 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
11191 (qdf24xx_addrcost_table): Likewise.
11192 (cortexa57_tunings): Update to use generic_branch_cost.
11193 (cortexa72_tunings): Likewise.
11194 (cortexa73_tunings): Likewise.
11195 (qdf24xx_tunings): Likewise.
11197 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
11199 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
11200 (thunderx2t99_branch_cost): Likewise.
11201 (cortexa35_tunings): Update to use generic_branch_cost.
11202 (cortexa53_tunings): Likewise.
11203 (cortexa57_tunings): Likewise.
11204 (cortexa72_tunings): Likewise.
11205 (cortexa73_tunings): Likewise.
11206 (thunderx2t99_tunings): Likewise.
11208 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11210 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
11211 (sparc_option_override): Honour MASK_FSMULD.
11212 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
11213 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
11214 * config/sparc/sparc.opt (mfsmuld): New option.
11215 * doc/invoke.texi (mfsmuld): Document option.
11217 2017-07-26 Marek Polacek <polacek@redhat.com>
11219 PR middle-end/70992
11220 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
11222 2017-07-26 Richard Biener <rguenther@suse.de>
11224 * gimple-match-head.c (do_valueize): Return OP if valueize
11226 (get_def): New helper to get at the def stmt of a SSA name
11227 if valueize allows.
11228 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
11229 do_valueize to get at the def stmt.
11230 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
11232 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
11234 PR middle-end/46932
11235 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
11237 2017-07-26 Martin Liska <mliska@suse.cz>
11240 * function.c (expand_function_start): Make expansion of
11241 nonlocal_goto_save_area after parm_birth_insn.
11243 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
11245 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
11246 from all CPU target flags enable members.
11248 2017-07-26 Richard Biener <rguenther@suse.de>
11250 * genmatch.c (dt_simplify::gen): Make iterator vars const.
11251 (decision_tree::gen): Make 'type' const.
11252 (write_predicate): Likewise.
11254 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11256 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
11257 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
11258 (rs6000_option_override_internal): Likewise.
11259 (rs6000_expand_vector_set): Likewise.
11260 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
11261 (TARGET_UPPER_REGS_SF): Likewise.
11262 (TARGET_UPPER_REGS_DI): Likewise.
11263 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
11264 (TARGET_DIRECT_MOVE_64BIT): Likewise.
11265 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
11266 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
11267 (Splitters for DI constants in Altivec registers): Likewise.
11268 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
11269 (vsx_set_v4sf_p9): Likewise.
11270 (vsx_set_v4sf_p9_zero): Likewise.
11271 (vsx_insert_extract_v4sf_p9): Likewise.
11272 (vsx_insert_extract_v4sf_p9_2): Likewise.
11274 2017-07-25 Carl Love <cel@us.ibm.com>
11276 * doc/extend.texi: Update the built-in documentation file for the
11277 existing built-in functions
11278 vector signed char vec_cnttz (vector signed char);
11279 vector unsigned char vec_cnttz (vector unsigned char);
11280 vector signed short vec_cnttz (vector signed short);
11281 vector unsigned short vec_cnttz (vector unsigned short);
11282 vector signed int vec_cnttz (vector signed int);
11283 vector unsigned int vec_cnttz (vector unsigned int);
11284 vector signed long long vec_cnttz (vector signed long long);
11285 vector unsigned long long vec_cnttz (vector unsigned long long);
11287 2017-07-25 Andrew Pinski <apinski@cavium.com>
11289 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
11290 accesses where the use is for the first operand of a BIT_INSERT.
11292 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11295 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11296 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11298 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
11300 * config/i386/gstabs.h: Delete.
11301 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
11303 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
11305 * config/i386/i386.c (ix86_decompose_address): Do not check for
11306 register RTX when looking at index_reg or base_reg.
11307 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
11309 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
11311 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
11312 to update EH info here.
11314 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11316 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
11318 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11320 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
11322 2017-07-25 Torsten Duwe <duwe@suse.de>
11324 * common.opt: Introduce -fpatchable-function-entry
11325 command line option, and its variables function_entry_patch_area_size
11326 and function_entry_patch_area_start.
11327 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
11328 including a two-value parser.
11329 * target.def (print_patchable_function_entry): New target hook.
11330 * targhooks.h (default_print_patchable_function_entry): New function.
11331 * targhooks.c (default_print_patchable_function_entry): Likewise.
11332 * toplev.c (process_options): Switch off IPA-RA if
11333 patchable function entries are being generated.
11334 * varasm.c (assemble_start_function): Look at the
11335 patchable-function-entry command line switch and current
11336 function attributes and maybe generate NOP instructions by
11337 calling the print_patchable_function_entry hook.
11338 * doc/extend.texi: Document patchable_function_entry attribute.
11339 * doc/invoke.texi: Document -fpatchable_function_entry
11340 command line option.
11341 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
11343 * doc/tm.texi: Re-generate.
11345 2017-07-25 Jakub Jelinek <jakub@redhat.com>
11348 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
11349 TARGET_AVX512DQ rather than TARGET_AVX512BW.
11351 2017-07-25 Tamar Christina <tamar.christina@arm.com>
11353 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
11355 2017-07-25 Richard Biener <rguenther@suse.de>
11357 PR tree-optimization/81455
11358 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
11359 not walk in cycles when looking for guards.
11361 2017-07-25 Richard Biener <rguenther@suse.de>
11363 PR tree-optimization/81529
11364 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
11365 when optimizing backedge uses.
11367 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
11369 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
11371 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
11372 to dl_section_ref. On AIX, append an expression to subtract
11373 the size of the section length to dl_section_ref.
11375 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
11377 * configure.ac: If any of the config.* scripts fail, exit 1.
11378 * configure: Regenerate.
11380 2017-07-25 Richard Biener <rguenther@suse.de>
11382 PR middle-end/81546
11383 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
11384 of immediate uses, be more verbose on errors.
11386 2017-07-25 Richard Biener <rguenther@suse.de>
11388 PR tree-optimization/81510
11389 * tree-vect-loop.c (vect_is_simple_reduction): When the
11390 reduction stmt is not inside the loop bail out.
11392 2017-07-25 Richard Biener <rguenther@suse.de>
11394 PR tree-optimization/81303
11395 * tree-vect-loop-manip.c (vect_loop_versioning): Build
11396 profitability check against LOOP_VINFO_NITERSM1.
11398 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
11400 * domwalk.c (cmp_bb_postorder): Simplify.
11401 (sort_bbs_postorder): New function. Use it...
11402 (dom_walker::walk): ...here to optimize common cases.
11404 2017-07-25 Martin Liska <mliska@suse.cz>
11407 * ipa-visibility.c (function_and_variable_visibility): Make the
11408 redirection just on target that supports aliasing.
11409 Fix GNU coding style.
11411 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
11414 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
11416 * config/arm/rtems.h: Likewise.
11417 * config/bfin/rtems.h: Likewise.
11418 * config/i386/rtemself.h: Likewise.
11419 * config/lm32/rtems.h: Likewise.
11420 * config/m32c/rtems.h: Likewise.
11421 * config/m68k/rtemself.h: Likewise.
11422 * config/microblaze/rtems.h: Likewise.
11423 * config/mips/rtems.h: Likewise.
11424 * config/moxie/rtems.h: Likewise.
11425 * config/nios2/rtems.h: Likewise.
11426 * config/powerpcspe/rtems.h: Likewise.
11427 * config/rs6000/rtems.h: Likewise.
11428 * config/rtems.h: Likewise.
11429 * config/sh/rtems.h: Likewise.
11430 * config/sh/rtemself.h: Likewise.
11431 * config/sparc/rtemself.h: Likewise.
11433 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11436 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
11437 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
11438 * tree-ssa-structalias.c (alias_get_name): Same.
11440 2017-07-25 Bin Cheng <bin.cheng@arm.com>
11443 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
11444 instructions if no du chain is found.
11446 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
11448 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
11450 2017-07-25 Richard Biener <rguenther@suse.de>
11452 PR middle-end/81505
11453 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
11456 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
11458 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
11459 upper-regs options.
11460 (ISA_2_7_MASKS_SERVER): Likewise.
11461 (ISA_3_0_MASKS_IEEE): Likewise.
11462 (OTHER_P8_VECTOR_MASKS): Likewise.
11463 (OTHER_VSX_VECTOR_MASKS): Likewise.
11464 (POWERPC_MASKS): Likewise.
11465 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
11466 duplicate list of options.
11467 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
11468 explicit -mupper-regs options.
11469 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
11470 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
11471 alias for -mupper-regs-df.
11472 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11473 (rs6000_init_hard_regno_mode_ok): Likewise.
11474 (rs6000_option_override_internal): Likewise.
11475 (rs6000_opt_masks): Likewise.
11476 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
11477 options in terms of whether -mvsx or -mpower8-vector was used.
11478 (TARGET_UPPER_REGS_DI): Likewise.
11479 (TARGET_UPPER_REGS_SF): Likewise.
11480 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
11481 -mupper-regs-* options.
11483 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11485 * passes.c (emergency_dump_function): Print some empty lines and a
11486 header before the RTL dump.
11488 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
11490 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
11492 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
11495 * config/aarch64/aarch64.c (aarch64_classify_symbol):
11496 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
11498 2017-07-24 Carl Love <cel@us.ibm.com>
11500 * config/rs6000/rs6000-c.c: Add support for built-in functions
11501 vector float vec_extract_fp32_from_shorth (vector unsigned short);
11502 vector float vec_extract_fp32_from_shortl (vector unsigned short);
11503 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
11504 vec_extract_fp_from_shortl): Add defines for the two builtins.
11505 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
11506 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
11508 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
11509 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
11510 * doc/extend.texi: Update the built-in documentation file for the
11511 new built-in function.
11513 2017-07-24 Jakub Jelinek <jakub@redhat.com>
11516 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
11518 * doc/generic.texi: Likewise.
11519 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
11520 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
11522 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
11524 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
11525 (aarch64_mls_elt_merge<mode>): Likewise.
11527 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
11529 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
11530 having __cxa_atexit.
11532 2017-07-23 Michael Collison <michael.collison@arm.com>
11534 * config/arm/arm.c (arm_option_override): Deprecate
11535 use of -mstructure-size-boundary.
11536 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
11537 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
11539 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11542 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
11543 Reduce cost estimate for direct moves.
11545 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
11548 * config/i386/i386.c (ix86_option_override_internal): Disable
11549 BMI, BMI2 and TBM instructions for -m16.
11551 2017-07-21 Carl Love <cel@us.ibm.com>
11553 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11554 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11555 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11556 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11557 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11558 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11559 VMULOSW): New enum "unspec" values.
11560 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11561 altivec_vmulosw): New patterns.
11562 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11563 VMULOSW): Add definitions.
11565 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
11567 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
11568 (qdf24xx): Likewise.
11569 * config/aarch64/aarch64-options-extensions.def (rdma); New.
11570 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
11571 (AARCH64_FL_V8_1): Renumber.
11572 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
11573 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
11574 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
11575 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
11576 rdma to feature modifiers list.
11578 2017-07-21 Yury Gribov <tetra2005@gmail.com>
11580 PR middle-end/56727
11581 * ipa-visibility (function_and_variable_visibility): Convert
11582 recursive PLT call to direct call if appropriate.
11584 2017-07-21 Andrew Pinski <apinski@cavium.com>
11586 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
11587 operand 1 to see if the types precision matches.
11588 * fold-const.c (operand_equal_p): Likewise.
11590 2017-07-21 Richard Biener <rguenther@suse.de>
11592 PR tree-optimization/81303
11593 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
11594 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
11595 (vect_peeling_hash_get_lowest_cost): Adjust.
11596 (vect_enhance_data_refs_alignment): Likewise. Use
11597 vect_get_peeling_costs_all_drs to compute the penalty for no
11598 peeling to match up costs.
11600 2017-07-21 Richard Biener <rguenther@suse.de>
11602 PR tree-optimization/81500
11603 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
11604 we didn't identify a reduction path.
11606 2017-07-21 Tom de Vries <tom@codesourcery.com>
11607 Cesar Philippidis <cesar@codesourcery.com>
11609 PR gcov-profile/81442
11610 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
11613 2017-07-21 Tom de Vries <tom@codesourcery.com>
11616 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
11618 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
11619 nvptx_override_options_after_change.
11621 2017-07-21 Ulrich Drepper <drepper@redhat.com>
11623 * dwarf2out.c (output_file_names): Avoid double testing for
11624 dwarf_version >= 5.
11626 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
11628 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
11630 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11632 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
11634 (try_crossjump_to_edge): Do not punt on partitioned functions.
11636 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
11638 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
11639 Put all BBs reachable only via paths crossing cold region to cold
11641 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
11643 2016-07-21 Richard Biener <rguenther@suse.de>
11645 PR tree-optimization/81303
11646 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
11647 into account prologue and epilogue iterations when raising
11648 min_profitable_iters to sth at least covering one vector iteration.
11650 2017-07-21 Tamar Christina <tamar.christina@arm.com>
11652 * config/arm/arm.c (arm_test_cpu_arch_dat):
11655 2017-07-20 Nathan Sidwell <nathan@acm.org>
11657 Remove TYPE_METHODS.
11658 * tree.h (TYPE_METHODS): Delete.
11659 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
11660 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
11661 (dbxout_type_methods): Scan TYPE_FIELDS.
11662 (dbxout_type): Don't check TYPE_METHODS here.
11663 * function.c (use_register_for_decl): Always ignore register for
11664 class types when not optimizing.
11665 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
11666 * tree.c (free_lang_data_in_type): Stitch out member functions and
11667 templates from TYPE_FIELDS.
11668 (build_distinct_type_copy, verify_type_variant,
11669 verify_type): Member fns are on TYPE_FIELDS.
11670 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
11671 * tree-pretty-print.c (dump_generic_node): Likewise.
11673 2017-07-20 Jakub Jelinek <jakub@redhat.com>
11676 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
11677 V2TImode and V4TImode.
11678 (ix86_expand_vector_extract): Likewise.
11679 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
11680 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
11681 (ssescalarmode): Handle V4TImode and V2TImode.
11682 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
11683 (*vec_extractv2ti, *vec_extractv4ti): New insns.
11684 (VEXTRACTI128_MODE): New mode iterator.
11685 (splitter for *vec_extractv?ti first element): New.
11686 (VEC_INIT_MODE): New mode iterator.
11687 (vec_init<mode>): Consolidate 3 expanders into one using
11688 VEC_INIT_MODE mode iterator.
11690 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11692 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
11693 non_spilled_static_chain_regno_p.
11695 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
11697 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
11699 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
11701 * bb-reorder.c (connect_traces): Allow copying of blocks within
11704 2017-07-20 Richard Biener <rguenther@suse.de>
11706 * gimple.h (gimple_phi_result): Add gphi * overload.
11707 (gimple_phi_result_ptr): Likewise.
11708 (gimple_phi_arg): Likewise. Adjust index assert to only
11709 allow actual argument accesses rather than all slots available
11711 (gimple_phi_arg_def): Add gphi * overload.
11712 * tree-phinodes.c (make_phi_node): Initialize only actual
11714 (resize_phi_node): Clear memory not covered by old node,
11715 do not initialize excess argument slots.
11716 (reserve_phi_args_for_new_edge): Initialize new argument slot
11719 2017-07-20 Bin Cheng <bin.cheng@arm.com>
11721 PR tree-optimization/81388
11723 2016-07-21 Bin Cheng <bin.cheng@arm.com>
11725 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
11726 by removing computation of may_be_zero.
11728 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11729 Tom de Vries <tom@codesourcery.com>
11731 PR middle-end/81030
11732 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
11733 when gimple level profile disagrees with what RTL expander did.
11735 2017-07-20 Richard Biener <rguenther@suse.de>
11737 PR tree-optimization/61171
11738 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
11739 (vect_analyze_stmt): Add slp instance parameter.
11740 (vectorizable_reduction): Likewise.
11741 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
11742 (vect_is_simple_reduction): Deal with chains not detected
11743 as SLP reduction chain, specifically not properly associated
11744 chains containing a mix of plus/minus.
11745 (get_reduction_op): Remove.
11746 (get_initial_defs_for_reduction): Simplify, pass in whether
11747 this is a reduction chain, pass in the SLP node for the PHIs.
11748 (vect_create_epilog_for_reduction): Get the SLP instance as
11750 (vectorizable_reduction): Get the SLP instance as arg.
11751 During analysis remember the SLP node with the PHIs in the
11752 instance. Simplify getting at the vectorized reduction PHIs.
11753 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
11754 through SLP instance.
11755 (vect_slp_analyze_operations): Likewise.
11756 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
11757 (vect_transform_stmt): Likewise.
11759 2017-07-20 Tom de Vries <tom@codesourcery.com>
11761 PR tree-optimization/81489
11762 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
11763 read of phi arg location to before loop that modifies phi.
11765 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
11767 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
11770 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11772 PR middle-end/81331
11773 * except.c (execute): Fix ordering issue.
11775 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11777 PR rtl-optimization/81423
11778 * combine.c (make_compound_operation_int): Don't try to optimize
11779 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
11781 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
11783 PR rtl-optimization/81423
11784 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
11785 with a constant that is -1 in the truncated to mode.
11787 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11789 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
11790 (determine_unlikely_bbs): ... here.
11791 * predict.h (propagate_unlikely_bbs_forward): Declare.
11792 * cfgexpand.c (pass_expand::execute): Use it.
11793 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
11795 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
11796 propagate_unlikely_bbs_forward.
11798 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
11800 PR middle-end/81331
11801 * except.c (maybe_add_nop_after_section_switch): New function.
11804 2017-07-19 Tom de Vries <tom@codesourcery.com>
11806 * gimple.h (gimple_phi_set_arg): Make assert more strict.
11808 2017-07-19 Tom de Vries <tom@codesourcery.com>
11810 * gimple.h (gimple_phi_arg): Make assert more strict.
11812 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
11814 * config.gcc (powerpc*-*-*): Add mmintrin.h.
11815 * config/rs6000/mmintrin.h: New file.
11816 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
11818 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11820 PR tree-optimization/81346
11821 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
11823 2017-07-19 Tom de Vries <tom@codesourcery.com>
11825 * config/nvptx/nvptx.md (VECIM): Add V2DI.
11827 2017-07-19 Tom de Vries <tom@codesourcery.com>
11829 * config/nvptx/nvptx-modes.def: Add V2DImode.
11830 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
11831 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
11832 (nvptx_output_mov_insn): Handle lack of mov.b128.
11833 (nvptx_print_operand): Handle 'H' and 'L' codes.
11834 (nvptx_vector_mode_supported): Allow V2DImode.
11835 (nvptx_preferred_simd_mode): New function.
11836 (nvptx_data_alignment): New function.
11837 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
11838 nvptx_preferred_simd_mode.
11839 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
11841 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
11843 2017-07-19 Tom de Vries <tom@codesourcery.com>
11845 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
11846 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
11847 (nvptx_vector_mode_supported): New function. Allow V2SImode.
11848 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
11849 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
11850 (mov<VECIM>_insn): New define_insn.
11851 (define_expand "mov<VECIM>): New define_expand.
11853 2017-07-19 Tom de Vries <tom@codesourcery.com>
11855 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
11857 2017-07-19 Jakub Jelinek <jakub@redhat.com>
11859 PR tree-optimization/81346
11860 * fold-const.h (fold_div_compare, range_check_type): Declare.
11861 * fold-const.c (range_check_type): New function.
11862 (build_range_check): Use range_check_type.
11863 (fold_div_compare): No longer static, rewritten into
11864 a match.pd helper function.
11865 (fold_comparison): Don't call fold_div_compare here.
11866 * match.pd (X / C1 op C2): New optimization using fold_div_compare
11867 as helper function.
11869 2017-07-19 Nathan Sidwell <nathan@acm.org>
11871 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
11872 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
11873 * tree.c (find_decls_types_r, verify_type): Use
11874 TYPE_{MIN,MAX}_VALUE_RAW.
11875 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
11876 (hash_tree): Likewise.
11877 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11879 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11882 2017-07-18 Tom de Vries <tom@codesourcery.com>
11884 PR middle-end/81464
11885 * omp-expand.c (expand_omp_for_static_chunk): Handle
11886 equal-argument loop exit phi.
11888 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
11891 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
11892 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
11893 operand 2 predicate.
11894 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
11895 operand 2 predicate.
11896 (ror,rol -> rorx splitters): Use const_int_operand as
11897 operand 2 predicate.
11899 2017-06-18 Richard Biener <rguenther@suse.de>
11901 PR tree-optimization/81410
11902 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
11903 the gap in the ! slp_perm SLP case after each group.
11905 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11907 PR middle-end/81463
11908 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
11911 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
11913 PR middle-end/81462
11914 * predict.c (set_even_probabilities): Cleanup; do not affect
11915 probabilities that are already known.
11916 (combine_predictions_for_bb): Call even when count is set.
11918 2017-07-18 Nathan Sidwell <nathan@acm.org>
11920 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
11923 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11926 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
11927 optimization for loop niter analysis.
11929 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
11932 * config/avr/avr.c (avr_optimize_casesi): Don't use
11933 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
11935 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
11937 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
11938 body_cost_vec from _vect_peel_extended_info.
11939 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
11940 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
11943 2017-07-18 Bin Cheng <bin.cheng@arm.com>
11945 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
11947 2017-07-18 Richard Biener <rguenther@suse.de>
11949 PR tree-optimization/80620
11950 PR tree-optimization/81403
11951 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
11952 info when re-using a VN table entry.
11954 2017-07-18 Richard Biener <rguenther@suse.de>
11956 PR tree-optimization/81418
11957 * tree-vect-loop.c (vectorizable_reduction): Properly compute
11958 vectype_in. Verify that with lane-reducing reduction operations
11959 we have a single def-use cycle.
11961 2017-07-17 Carl Love <cel@us.ibm.com>
11963 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
11965 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
11966 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
11967 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
11968 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
11969 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
11970 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
11971 VMULOSW): New enum "unspec" values.
11972 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
11973 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
11974 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
11975 altivec_vmulosw): New patterns.
11976 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
11977 VMULOSW): Add definitions.
11978 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
11980 * config/alpha/alpha.c: Include predict.h.
11982 2017-07-17 Yury Gribov <tetra2005@gmail.com>
11984 * tree-vrp.c (compare_assert_loc): Fix comparison function
11985 to return predictable results.
11987 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11989 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
11991 (subdi3): Likewise.
11992 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
11993 * doc/invoke.texi (mexpand-adddi): Update text.
11995 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
11997 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
11998 that also clobbers the CC register. The old expand code is moved
12000 (*arc_clzsi2): ... here.
12001 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
12002 the CC register. The old expand code is moved to ...
12003 (arc_ctzsi2): ... here.
12005 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12007 * config/arc/arc.opt (mindexed-loads): Use initial value
12008 TARGET_INDEXED_LOADS_DEFAULT.
12009 (mauto-modify-reg): Use initial value
12010 TARGET_AUTO_MODIFY_REG_DEFAULT.
12011 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12012 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12013 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
12014 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
12016 2017-07-17 Martin Liska <mliska@suse.cz>
12019 * opts.c (finish_options): Do not allow -fgnu-tm
12020 w/ -fsanitize={kernel-,}address. Say sorry.
12022 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12025 * tree-loop-distribution.c (classify_partition): Only assert on
12026 numer of iterations.
12027 (merge_dep_scc_partitions): Delete prameter. Update function call.
12028 (distribute_loop): Remove code handling loop with unknown niters.
12029 (pass_loop_distribution::execute): Skip loop with unknown niters.
12031 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12034 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
12035 function sort_partitions_by_post_order.
12037 2017-07-17 Bin Cheng <bin.cheng@arm.com>
12039 PR tree-optimization/81374
12040 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
12041 the max index of basic blocks, rather than number of basic blocks.
12043 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12045 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
12047 (arc_legitimate_pic_operand_p): Likewise.
12048 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
12050 (arc_needs_pcl_p): Likewise.
12051 (arc_legitimate_pc_offset_p): Likewise.
12052 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
12053 function is also used in constrains.md.
12054 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
12055 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
12056 PLUS. Only return true/false in known cases, otherwise assert.
12057 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
12058 is already called in arc_legitimate_constant_p.
12059 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
12061 (LEGITIMATE_PIC_OPERAND_P): Use
12062 arc_raw_symbolic_reference_mentioned_p function.
12063 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
12068 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
12069 Andrew Burgess <andrew.burgess@embecosm.com>
12071 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
12072 (arc_return_address_register): New function.
12073 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
12074 (arc_handle_fndecl_attribute): Add naked attribute.
12075 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
12076 (TARGET_WARN_FUNC_RETURN): Likewise.
12077 (arc_allocate_stack_slots_for_args): New function.
12078 (arc_warn_func_return): Likewise.
12079 (machine_function): Change type fn_type.
12080 (arc_compute_function_type): Consider new naked function type,
12081 change function return type.
12082 (arc_must_save_register): Adapt to handle new
12083 arc_compute_function_type's return type.
12084 (arc_expand_prologue): Likewise.
12085 (arc_expand_epilogue): Likewise.
12086 (arc_return_address_regs): Delete.
12087 (arc_return_address_register): New function.
12088 (arc_epilogue_uses): Use above function.
12089 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
12090 (arc_function_type): Change encoding, add naked type.
12091 (ARC_INTERRUPT_P): Change to handle the new encoding.
12092 (ARC_FAST_INTERRUPT_P): Likewise.
12093 (ARC_NORMAL_P): Define.
12094 (ARC_NAKED_P): Likewise.
12095 (arc_compute_function_type): Delete prototype.
12096 * config/arc/arc.md (in_ret_delay_slot): Use
12097 arc_return_address_register function.
12098 (simple_return): Likewise.
12099 (p_return_i): Likewise.
12101 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12103 PR tree-optimization/81428
12104 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
12105 can't be built for those types.
12107 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12109 Remove stuff dead since r239246.
12111 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
12112 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
12113 (avr_inform_devices): Remove dead stuff.
12115 2017-07-17 Tamar Christina <tamar.christina@arm.com>
12117 * config/arm/arm_neon.h: Fix softp typo.
12119 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12121 PR tree-optimization/81365
12122 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
12123 aggregate moves onto bb predecessor edges, make sure there are no
12124 loads that could alias the lhs in between the start of bb and the
12127 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
12130 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
12131 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
12132 [LSHIFTRT, outer_code = TRUNCATE]: Same.
12134 2017-07-17 Jakub Jelinek <jakub@redhat.com>
12136 PR tree-optimization/81396
12137 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
12138 (init_symbolic_number): Initialize it to 1.
12139 (perform_symbolic_merge): Add n_ops from both operands into the new
12141 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
12142 without base_addr as useless if they need more than one operation.
12143 (bswap_replace): Handle !bswap case for NULL base_addr.
12145 2017-07-17 Tom de Vries <tom@codesourcery.com>
12148 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
12151 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
12153 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
12154 conditional builtin define __FIX_LEON3FT_B2BST.
12156 2017-07-17 Daniel Cederman <cederman@gaisler.com>
12158 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
12159 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
12162 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12164 PR rtl-optimization/81424
12165 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
12166 to remove potential trapping from operands if -fnon-call-exceptions.
12168 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12170 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
12171 profile_proability for scalling.
12172 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
12174 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12176 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
12178 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12180 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
12181 fixpoint arithmetics.
12183 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12185 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
12186 fixpoint arithmetics.
12188 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12190 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
12191 fixpoint arithmetics.
12193 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
12195 * profile-count.h (profile_probability::from_reg_br_prob_note,
12196 profile_probability::to_reg_br_prob_note): New functions.
12197 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
12198 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
12199 * predict.c (probability_reliable_p): Update.
12200 (edge_probability_reliable_p): Update.
12201 (br_prob_note_reliable_p): Update.
12202 (invert_br_probabilities): Update.
12203 (add_reg_br_prob_note): New function.
12204 (combine_predictions_for_insn): Update.
12205 * asan.c (asan_clear_shadow): Update.
12206 * cfgbuild.c (compute_outgoing_frequencies): Update.
12207 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
12208 (update_br_prob_note): Update.
12209 (rtl_verify_edges): Update.
12210 (purge_dead_edges): Update.
12211 (fixup_reorder_chain): Update.
12212 * emit-rtl.c (try_split): Update.
12213 * ifcvt.c (cond_exec_process_insns): Update.
12214 (cond_exec_process_if_block): Update.
12215 (dead_or_predicable): Update.
12216 * internal-fn.c (expand_addsub_overflow): Update.
12217 (expand_neg_overflow): Update.
12218 (expand_mul_overflow): Update.
12219 * loop-doloop.c (doloop_modify): Update.
12220 * loop-unroll.c (compare_and_jump_seq): Update.
12221 * optabs.c (emit_cmp_and_jump_insn_1): Update.
12222 * predict.h: Update.
12223 * reorg.c (mostly_true_jump): Update.
12225 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
12226 * config/alpha/alpha.c (emit_unlikely_jump): Update.
12227 * config/arc/arc.c: (emit_unlikely_jump): Update.
12228 * config/arm/arm.c: (emit_unlikely_jump): Update.
12229 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
12230 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
12231 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
12232 (ix86_print_operand): Update.
12233 (ix86_split_fp_branch): Update.
12234 (predict_jump): Update.
12235 * config/ia64/ia64.c (ia64_print_operand): Update.
12236 * config/mmix/mmix.c (mmix_print_operand): Update.
12237 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
12238 (rs6000_expand_split_stack_prologue): Update.
12239 * config/rs6000/rs6000.c: Update.
12240 * config/s390/s390.c (s390_expand_vec_strlen): Update.
12241 (s390_expand_vec_movstr): Update.
12242 (s390_expand_cs_tdsi): Update.
12243 (s390_expand_split_stack_prologue): Update.
12244 * config/sh/sh.c (sh_print_operand): Update.
12245 (expand_cbranchsi4): Update.
12246 (expand_cbranchdi4): Update.
12247 * config/sparc/sparc.c (output_v9branch): Update.
12248 * config/spu/spu.c (get_branch_target): Update.
12249 (ea_load_store_inline): Update.
12250 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
12251 * config/tilepro/tilepro.c: Update.
12253 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
12255 * gimplify.c (mostly_copy_tree_r): Revert latest change.
12256 (gimplify_save_expr): Likewise.
12258 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12260 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
12262 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12264 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
12266 * timevar.def (TV_IPA_FNSUMMARY): Define.
12268 2017-07-16 Daniel Cederman <cederman@gaisler.com>
12270 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
12271 to back store errata sensitive sequence from being generated.
12272 (sqrtdf2_fix): Likewise.
12274 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12276 * tree-ssa-threadupdate.c (compute_path_counts,
12277 update_joiner_offpath_counts): Use profile_probability.
12279 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
12282 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12284 * config/arm/arm-c.c (arm_cpu_builtins): Define
12285 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12287 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
12289 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
12290 array entries to represent __ieee128 versions of the
12291 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
12292 scalar_extract_sig, and scalar_insert_exp built-in functions.
12293 (altivec_resolve_overloaded_builtin): Add special case handling
12294 for the __builtin_scalar_insert_exp function, as represented by
12295 the P9V_BUILTIN_VEC_VSIEDP constant.
12296 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
12297 exponent support for __ieee128 argument.
12298 (VSESQP): Add scalar extract signature support for __ieee128
12300 (VSTDCNQP): Add scalar test negative support for __ieee128
12302 (VSIEQP): Add scalar insert exponent support for __int128 argument
12303 with __ieee128 result.
12304 (VSIEQPF): Add scalar insert exponent support for __ieee128
12305 argument with __ieee128 result.
12306 (VSTDCQP): Add scalar test data class support for __ieee128
12308 (VSTDCNQP): Add overload support for scalar test negative with
12309 __ieee128 argument.
12310 (VSTDCQP): Add overload support for scalar test data class
12311 __ieee128 argument.
12312 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
12313 UNSPEC_VSX_SXSIGDP.
12314 (UNSPEC_VSX_SIEXPQP): New constant.
12315 (xsxexpqp): New insn for VSX scalar extract exponent quad
12317 (xsxsigqp): New insn for VSX scalar extract significand quad
12319 (xsiexpqpf): New insn for VSX scalar insert exponent quad
12320 precision with floating point argument.
12321 (xststdcqp): New expand for VSX scalar test data class quad
12323 (xststdcnegqp): New expand for VSX scalar test negative quad
12325 (xststdcqp): New insn to match expansions for VSX scalar test data
12326 class quad precision and VSX scalar test negative quad precision.
12327 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
12328 special case operand checking to enforce that second operand of
12329 VSX scalar test data class with quad precision argument is a 7-bit
12331 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
12332 prototypes and descriptions of __ieee128 versions of
12333 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
12334 scalar_test_data_class, and scalar_test_neg built-in functions.
12336 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12338 PR tree-optimization/81162
12339 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
12340 replace a negate with an add.
12342 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
12344 * doc/invoke.texi (arm/-mcpu): Document +crypto.
12346 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12348 * config/arm/arm-c.c (arm_cpu_builtins): Define
12349 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
12351 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12353 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
12354 (armv8-r): Set ARM Cortex-R52 as default CPU.
12355 * config/arm/arm-tables.opt: Regenerate.
12356 * config/arm/arm-tune.md: Regenerate.
12357 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
12359 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
12360 extension for -mcpu=cortex-r52.
12362 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
12364 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
12365 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
12366 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
12367 (fp-armv8): Define it as FP_ARMv8 only.
12368 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
12369 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
12371 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
12372 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
12373 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
12374 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
12375 than TARGET_FPU_ARMV8.
12376 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
12377 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
12378 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
12380 * config/arm/neon.md (neon_vrint): Likewise.
12381 (neon_vcvt): Likewise.
12382 (neon_<fmaxmin_op><mode>): Likewise.
12383 (<fmaxmin><mode>3): Likewise.
12384 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
12385 * config/arm/predicates.md (arm_cond_move_operator): Check against
12386 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
12388 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
12390 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
12391 to top of function.
12393 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12395 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
12396 loop in comment with memset.
12398 2017-07-14 Martin Liska <mliska@suse.cz>
12400 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
12401 * dwarf2out.c (is_java): Remove the function.
12402 (output_pubname): Remove usage of the function.
12403 (lower_bound_default): Remove usage of DW_LANG_Java.
12404 (gen_compile_unit_die): Likewise.
12405 * gcc.c: Remove compiler defaults for .java and .zip files.
12406 * gimple-expr.c (remove_suffix): Change as there's no longer
12407 extension than 4-letter one.
12408 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
12409 (gimplify_save_expr): Likewise.
12410 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
12411 as it's possible even for other languages than Java.
12412 * langhooks.h (struct lang_hooks): Remove Java from a comment.
12413 * lto-opts.c (lto_write_options): Remove reference to Java.
12414 * opts.c (strip_off_ending): Update file extension handling.
12415 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
12416 * tree-eh.c (lower_resx): Likewise.
12417 * tree.c (free_lang_data_in_type): Remove dead code.
12418 (find_decls_types_r): Likewise.
12419 (build_common_builtin_nodes): Remove Java from a comment.
12420 (verify_type): Remove dead code.
12421 * varasm.c (assemble_external): Remove Java from a comment.
12423 2017-07-14 Martin Liska <mliska@suse.cz>
12425 * opts.c (finish_options): Add quotes.
12426 (common_handle_option): Likewise.
12428 2017-07-14 Martin Liska <mliska@suse.cz>
12430 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
12431 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
12432 Remove N_SO_PASCAL.
12433 * dwarf2out.c (lower_bound_default): Do not handle
12435 (gen_compile_unit_die): Likewise.
12436 * gcc.c: Remove default extension binding for GNU Pascal.
12437 * stmt.c: Remove Pascal language from a comment.
12438 * xcoffout.c: Likewise.
12440 2017-07-13 David Malcolm <dmalcolm@redhat.com>
12443 * diagnostic-show-locus.c (fixit_cmp): New function.
12444 (layout::layout): Sort m_fixit_hints.
12445 (column_range::column_range): Assert that the values are valid.
12446 (struct char_span): New struct.
12447 (correction::overwrite): New method.
12448 (struct source_line): New struct.
12449 (line_corrections::add_hint): Add assertions. Reimplement memcpy
12450 calls in terms of classes source_line and char_span, and
12451 correction::overwrite.
12452 (selftest::test_overlapped_fixit_printing_2): New function.
12453 (selftest::diagnostic_show_locus_c_tests): Call it.
12455 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
12457 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
12458 early if there is no lhs.
12460 2017-07-13 Martin Liska <mliska@suse.cz>
12462 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
12463 (gen_reference_type_die): Likewise.
12464 * stor-layout.c: Remove Pascal-related comment.
12466 2017-07-13 Martin Liska <mliska@suse.cz>
12468 * opts.c (finish_options): Add quotes to error messages.
12469 (parse_sanitizer_options): Likewise.
12471 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
12473 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
12475 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
12477 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
12479 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
12481 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
12483 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
12485 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
12488 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
12489 provides the hardware capability bits, define the macro
12490 __BUILTIN_CPU_SUPPORTS__.
12491 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
12492 if GLIBC does not provide the hardware capability bits. Add a
12493 gcc_unreachable call if the built-in cpu function is neither
12494 __builtin_cpu_is nor __builtin_cpu_supports.
12495 (rs6000_get_function_versions_dispatcher): Change the warning
12496 that an old GLIBC is used which does not export the capability
12497 bits to be an error.
12498 * doc/extend.texi (target_clones attribute): Document the
12499 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
12500 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
12501 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
12502 the macros defined by GCC if the newer GLIBC is available.
12504 2017-07-12 Jeff Law <law@redhat.com>
12506 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
12507 remaining includes slightly.
12508 * config/riscv/riscv-builtins.c: Include profile-count.h.
12510 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12513 * config/avr/avr.c (avr_set_current_function): In diagnostic
12514 messages: Quote keywords and (parts of) identifiers.
12515 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
12518 2017-07-12 Carl Love <cel@us.ibm.com>
12520 * config/rs6000/rs6000-c.c: Add support for built-in functions
12521 vector bool char vec_revb (vector bool char);
12522 vector bool short vec_revb (vector short char);
12523 vector bool int vec_revb (vector bool int);
12524 vector bool long long vec_revb (vector bool long long);
12525 * doc/extend.texi: Update the built-in documentation file for the
12526 new built-in functions.
12528 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12530 * config/s390/s390.md: Remove movcc splitter.
12532 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12534 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
12535 load/store on condition.
12537 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
12540 * config/avr/avr.c (avr_encode_section_info)
12541 [progmem && !TREE_READONLY]: Error if progmem object needs
12544 2017-07-11 Michael Collison <michael.collison@arm.com>
12546 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
12549 2017-07-11 Carl Love <cel@us.ibm.com>
12551 * config/rs6000/rs6000-c.c: Add support for builtins
12552 vector unsigned int vec_parity_lsbb (vector signed int);
12553 vector unsigned int vec_parity_lsbb (vector unsigned int);
12554 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
12555 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
12556 vector unsigned long long vec_parity_lsbb (vector signed long long);
12557 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
12558 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
12559 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
12560 * doc/extend.texi: Update the built-in documentation file for the
12561 new built-in functions.
12563 2017-07-11 David Malcolm <dmalcolm@redhat.com>
12565 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
12566 (layout::m_primary_loc): New field.
12567 (layout::layout): Initialize new field. Move location filtering
12568 logic from here to...
12569 (layout::maybe_add_location_range): ...this new method. Add
12570 support for filtering to just the lines already specified by other
12572 (layout::will_show_line_p): New method.
12573 (gcc_rich_location::add_location_if_nearby): New method.
12574 (selftest::test_add_location_if_nearby): New test function.
12575 (selftest::diagnostic_show_locus_c_tests): Call it.
12576 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
12579 2017-07-11 Tom de Vries <tom@codesourcery.com>
12581 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
12582 (bb_first_real_insn): New function.
12583 (nvptx_single): Add extra initialization of broadcasted condition
12586 2017-07-11 Nathan Sidwell <nathan@acm.org>
12588 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
12590 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
12592 * doc/extend.texi (AVR Function Attributes): Remove weblink to
12593 Binutils doc as TEXI will mess them up.
12594 * doc/invoke.texi (AVR Options): Same here.
12596 2017-07-11 Daniel Cederman <cederman@gaisler.com>
12598 * config/sparc/sparc.opt (mfix-ut700): New option.
12599 (mfix-gr712rc): Likewise.
12600 (sparc_fix_b2bst): New variable.
12601 * doc/invoke.texi (SPARC options): Document them.
12602 (ARM options): Fix warnings.
12603 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
12604 instructions to prevent sequences that can trigger the store-store
12605 errata for certain LEON3FT processors.
12606 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
12607 (sparc_option_override): Set sparc_fix_b2bst appropriately.
12608 * config/sparc/sparc.md (fix_b2bst): New attribute.
12609 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
12611 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
12614 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
12616 (*rsqrtsf2_sse): Ditto.
12618 (div<mode>3): Macroize insn from divdf3 and divsf3
12619 using MODEF mode iterator.
12621 2017-07-10 Martin Sebor <msebor@redhat.com>
12623 PR tree-optimization/80397
12624 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
12625 instead of testing for equality to INTEGER_TYPE.
12627 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
12629 * config.gcc: Remove uclibc from arc target spec.
12631 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
12633 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
12635 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
12638 * lto-wrapper.c (remove_option): New function.
12639 (merge_and_complain): Merge PIC/PIE options more realistically.
12641 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12643 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
12647 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
12648 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
12649 * config.in: Regenerate.
12650 * configure: Regenerate.
12651 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
12652 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
12653 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
12654 (TARGET_GASISR_PROLOGUES): ...target mask.
12655 * common/config/avr/avr-common.c
12656 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
12657 Set -mgas-isr-prologues.
12658 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
12659 INSERT_PASS_BEFORE for it.
12660 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
12661 * config/avr/avr.c (avr_option_override)
12662 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
12663 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
12664 (avr_attribute_table) <no_gccisr>: Add new function attribute.
12665 (avr_set_current_function) <is_no_gccisr>: Init machine field.
12666 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
12668 (make_avr_pass_pre_proep): New function.
12669 (emit_push_sfr) <treg>: Add argument to function and use it
12670 instead of TMP_REG.
12671 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
12672 and set machine->gasisr.yes.
12673 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
12674 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
12675 __gcc_isr.n_pushed to .L__stack_usage.
12676 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
12677 (avr_asm_final_postscan_insn): ...this new static function.
12678 * config/avr/avr.h (machine_function)
12679 <is_no_gccisr, use_L__stack_usage>: New fields.
12680 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
12681 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
12682 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
12683 (gasisr, *gasisr): New expander and insn.
12684 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
12685 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
12686 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
12688 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
12690 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
12693 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
12695 Move jump-tables out of .text again.
12698 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
12699 (ASM_OUTPUT_ADDR_VEC): New function.
12700 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
12701 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
12702 INSN_ADDRESSes as asm comment.
12703 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
12704 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
12705 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
12706 * config/avr/avr.md (*tablejump): Adjust comment.
12707 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
12708 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
12710 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
12711 (avr_output_addr_vec): New proto.
12712 (avr_log_t) <insn_addresses>: New field.
12714 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
12717 * config/i386/i386.c (ix86_function_arg_advance): Set
12718 outgoing_args_on_stack to true if there are outgoing arguments
12720 (ix86_function_arg): Likewise.
12721 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
12722 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
12723 * config/i386/i386.h (machine_function): Add
12724 outgoing_args_on_stack.
12726 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
12728 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
12729 supporting pthreds.
12730 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
12732 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
12734 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
12735 (REAL_H): Remove $(MACHMODE_H).
12736 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
12738 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
12739 $(MACHMODE_H) and double-int.h.
12740 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
12742 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
12745 2017-07-07 Andrew Pinski <apinski@cavium.com>
12747 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
12748 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
12750 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12752 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
12753 Add warning if GCC was not configured to link against a GLIBC that
12754 exports the hardware capability bits.
12755 (make_resolver_func): Make resolver function private and not a
12756 COMDAT function. Create the name with clone_function_name instead
12757 of make_unique_name.
12760 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
12761 correct operand in doing the split.
12763 2017-07-07 Carl Love <cel@us.ibm.com>
12765 * config/rs6000/rs6000-c: Add support for built-in function
12766 vector unsigned short vec_pack_to_short_fp32 (vector float,
12768 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
12769 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
12770 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
12771 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
12772 (convert_4f32_8i16): Add define_expand.
12773 * doc/extend.texi: Update the built-in documentation file for the
12774 new built-in function.
12776 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12778 * config/sparc/m8.md: New file.
12779 * config/sparc/sparc.md: Include m8.md.
12781 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12783 * config/sparc/sparc.opt: New option -mvis4b.
12784 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
12785 (sparc_option_override): Handle VIS4B.
12786 (enum sparc_builtins): Define
12787 SPARC_BUILTIN_DICTUNPACK{8,16,32},
12788 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
12789 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
12790 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
12791 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
12792 (check_constant_argument): New function.
12793 (sparc_vis_init_builtins): Define builtins
12794 __builtin_vis_dictunpack{8,16,32},
12795 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
12796 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
12797 __builtin_vis_fpcmpde{8,16,32}shl and
12798 __builtin_vis_fpcmpur{8,16,32}shl.
12799 (sparc_expand_builtin): Check that the constant operands to
12800 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
12801 constant and in range.
12802 * config/sparc/sparc-c.c (sparc_target_macros): Handle
12804 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
12805 (SPARC_IMM5_P): Likewise.
12806 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
12807 (enabled): Handle vis4b.
12808 (UNSPEC_DICTUNPACK): New unspec.
12809 (UNSPEC_FPCMPSHL): Likewise.
12810 (UNSPEC_FPUCMPSHL): Likewise.
12811 (UNSPEC_FPCMPDESHL): Likewise.
12812 (UNSPEC_FPCMPURSHL): Likewise.
12813 (cpu_feature): New CPU feature `vis4b'.
12814 (dictunpack{8,16,32}): New insns.
12815 (FPCSMODE): New mode iterator.
12816 (fpcscond): New code iterator.
12817 (fpcsucond): Likewise.
12818 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
12819 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
12820 (fpcmpde{8,16,32}{si,di}shl): Likewise.
12821 (fpcmpur{8,16,32}{si,di}shl): Likewise.
12822 * config/sparc/constraints.md: Define constraints `q' for unsigned
12823 2-bit integer constants and `t' for unsigned 5-bit integer
12825 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
12827 (imm5_operand_dictunpack16): Likewise.
12828 (imm5_operand_dictunpack32): Likewise.
12829 (imm2_operand): Likewise.
12830 * doc/invoke.texi (SPARC Options): Document -mvis4b.
12831 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
12832 ditunpack* and fpcmp*shl builtins.
12834 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12836 * config.gcc: Handle m8 in --with-{cpu,tune} options.
12837 * config.in: Add HAVE_AS_SPARC6 define.
12838 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
12840 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
12842 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
12843 (CPP_CPU_SPEC): Handle m8.
12844 (ASM_CPU_SPEC): Likewise.
12845 * config/sparc/sparc-opts.h (enum processor_type): Add
12847 * config/sparc/sparc.c (m8_costs): New struct.
12848 (sparc_option_override): Handle TARGET_CPU_m8.
12849 (sparc32_initialize_trampoline): Likewise.
12850 (sparc64_initialize_trampoline): Likewise.
12851 (sparc_issue_rate): Likewise.
12852 (sparc_register_move_cost): Likewise.
12853 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
12854 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
12855 (ASM_CPU64_DEFAULT_SPEC): Likewise.
12856 (CPP_CPU_SPEC): Handle M8.
12857 (ASM_CPU_SPEC): Likewise.
12858 (AS_M8_FLAG): Define.
12859 * config/sparc/sparc.md: Add m8 to the cpu attribute.
12860 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
12861 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
12863 * configure: Regenerate.
12864 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
12867 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12869 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
12871 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
12872 ("*movdi_insn_sp32"): Do not set v3pipe.
12873 ("*movsi_insn"): Likewise.
12874 ("*movdi_insn_sp64"): Likewise.
12875 ("*movsf_insn"): Likewise.
12876 ("*movdf_insn_sp32"): Likewise.
12877 ("*movdf_insn_sp64"): Likewise.
12878 ("*zero_extendsidi2_insn_sp64"): Likewise.
12879 ("*sign_extendsidi2_insn"): Likewise.
12880 ("*mov<VM32:mode>_insn"): Likewise.
12881 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12882 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12883 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12884 ("<vlop:code><VL:mode>3"): Likewise.
12885 ("*not_<vlop:code><VL:mode>3"): Likewise.
12886 ("*nand<VL:mode>_vis"): Likewise.
12887 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
12888 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
12889 ("one_cmpl<VL:mode>2"): Likewise.
12890 ("faligndata<VM64:mode>_vis"): Likewise.
12891 ("alignaddrsi_vis"): Likewise.
12892 ("alignaddrdi_vis"): Likweise.
12893 ("alignaddrlsi_vis"): Likewise.
12894 ("alignaddrldi_vis"): Likewise.
12895 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12896 ("bmaskdi_vis"): Likewise.
12897 ("bmasksi_vis"): Likewise.
12898 ("bshuffle<VM64:mode>_vis"): Likewise.
12899 ("cmask8<P:mode>_vis"): Likewise.
12900 ("cmask16<P:mode>_vis"): Likewise.
12901 ("cmask32<P:mode>_vis"): Likewise.
12902 ("pdistn<P:mode>_vis"): Likewise.
12903 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12905 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12907 * config/sparc/sparc.md ("subtype"): New insn attribute.
12908 ("*wrgsr_sp64"): Set insn subtype.
12909 ("*rdgsr_sp64"): Likewise.
12910 ("alignaddrsi_vis"): Likewise.
12911 ("alignaddrdi_vis"): Likewise.
12912 ("alignaddrlsi_vis"): Likewise.
12913 ("alignaddrldi_vis"): Likewise.
12914 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
12915 ("fexpand_vis"): Likewise.
12916 ("fpmerge_vis"): Likewise.
12917 ("faligndata<VM64:mode>_vis"): Likewise.
12918 ("bshuffle<VM64:mode>_vis"): Likewise.
12919 ("cmask8<P:mode>_vis"): Likewise.
12920 ("cmask16<P:mode>_vis"): Likewise.
12921 ("cmask32<P:mode>_vis"): Likewise.
12922 ("fchksm16_vis"): Likewise.
12923 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
12924 ("fmean16_vis"): Likewise.
12925 ("fp<plusminus_insn>64_vis"): Likewise.
12926 ("<plusminus_insn>v8qi3"): Likewise.
12927 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
12928 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
12929 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
12930 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
12931 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
12932 ("*movqi_insn"): Likewise.
12933 ("*movhi_insn"): Likewise.
12934 ("*movsi_insn"): Likewise.
12935 ("movsi_pic_gotdata_op"): Likewise.
12936 ("*movdi_insn_sp32"): Likewise.
12937 ("*movdi_insn_sp64"): Likewise.
12938 ("movdi_pic_gotdata_op"): Likewise.
12939 ("*movsf_insn"): Likewise.
12940 ("*movdf_insn_sp32"): Likewise.
12941 ("*movdf_insn_sp64"): Likewise.
12942 ("*zero_extendhisi2_insn"): Likewise.
12943 ("*zero_extendqihi2_insn"): Likewise.
12944 ("*zero_extendqisi2_insn"): Likewise.
12945 ("*zero_extendqidi2_insn"): Likewise.
12946 ("*zero_extendhidi2_insn"): Likewise.
12947 ("*zero_extendsidi2_insn_sp64"): Likewise.
12948 ("ldfsr"): Likewise.
12949 ("prefetch_64"): Likewise.
12950 ("prefetch_32"): Likewise.
12951 ("tie_ld32"): Likewise.
12952 ("tie_ld64"): Likewise.
12953 ("*tldo_ldub_sp32"): Likewise.
12954 ("*tldo_ldub1_sp32"): Likewise.
12955 ("*tldo_ldub2_sp32"): Likewise.
12956 ("*tldo_ldub_sp64"): Likewise.
12957 ("*tldo_ldub1_sp64"): Likewise.
12958 ("*tldo_ldub2_sp64"): Likewise.
12959 ("*tldo_ldub3_sp64"): Likewise.
12960 ("*tldo_lduh_sp32"): Likewise.
12961 ("*tldo_lduh1_sp32"): Likewise.
12962 ("*tldo_lduh_sp64"): Likewise.
12963 ("*tldo_lduh1_sp64"): Likewise.
12964 ("*tldo_lduh2_sp64"): Likewise.
12965 ("*tldo_lduw_sp32"): Likewise.
12966 ("*tldo_lduw_sp64"): Likewise.
12967 ("*tldo_lduw1_sp64"): Likewise.
12968 ("*tldo_ldx_sp64"): Likewise.
12969 ("*mov<VM32:mode>_insn"): Likewise.
12970 ("*mov<VM64:mode>_insn_sp64"): Likewise.
12971 ("*mov<VM64:mode>_insn_sp32"): Likewise.
12973 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12975 * config/sparc/sparc.md ("type"): New insn type viscmp.
12976 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
12978 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
12979 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
12980 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
12981 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
12983 ("n7_vis_logical_11cycle"): Likewise.
12984 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
12985 * config/sparc/niagara2.md ("niag3_vis": Likewise.
12986 * config/sparc/niagara.md ("niag_vis"): Likewise.
12987 * config/sparc/ultra3.md ("us3_fga"): Likewise.
12988 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
12990 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
12992 * config/sparc/sparc.md: New instruction type `bmask'.
12993 (bmaskdi_vis): Use the `bmask' type.
12994 (bmasksi_vis): Likewise.
12995 * config/sparc/ultra3.md (us3_array): Likewise.
12996 * config/sparc/niagara7.md (n7_array): Likewise.
12997 * config/sparc/niagara4.md (n4_array): Likewise.
12998 * config/sparc/niagara2.md (niag2_vis): Likewise.
12999 (niag3_vis): Likewise.
13000 * config/sparc/niagara.md (niag_vis): Likewise.
13002 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13004 * ipa-comdats.c: Remove optimize check from gate.
13005 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
13006 for functions not optimized.
13007 (ipa_fn_summary_read): Skip optimize check.
13008 (ipa_fn_summary_write): Likewise.
13009 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
13011 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
13013 (can_inline_edge_p): Check flag_pcc_struct_return for match.
13014 (check_callers): Give up on caller which is not optimized.
13015 (inline_small_functions): Likewise.
13016 (ipa_inline): Do not give up when not optimizing.
13017 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
13018 away unoptimizes cdtors.
13019 (whole_program_function_and_variable_visibility): Do
13020 ipa_discover_readonly_nonaddressable_vars in LTO mode.
13021 * ipa.c (process_references): Do not check optimize.
13022 (symbol_table::remove_unreachable_nodes): Update optimize check.
13023 (set_writeonly_bit): Update optimize check.
13024 (pass_ipa_cdtor_merge::gate): Do not check optimize.
13025 (pass_ipa_single_use::gate): Remove.
13027 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13029 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
13030 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
13031 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
13032 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
13033 permute_load, permute_store, adjust_extract, adjust_splat,
13034 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
13035 replace_swap_with_copy, dump_swap_insn_table,
13036 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
13037 recombine_lvx_pattern, recombine_stvx_pattern,
13038 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
13039 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
13040 to file rs6000-p8swap.c.
13041 * config/rs6000/rs6000-p8swap.c: New file.
13042 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
13043 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
13044 and rs6000*-*-* targets.
13046 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13048 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
13050 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13052 * lto-wrapper.c (merge_and_complain): Do not merge
13053 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
13054 fsigned_zeros, ftrapping_math, fwrapv.
13055 (append_compiler_options): Do not track these options.
13056 (append_linker_options): Likewie
13058 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13060 * cgraphunit.c (cgraph_node::finalize_function): When
13061 !flag_toplevel_reorde set no_reorder flag.
13062 (varpool_node::finalize_decl): Likewise.
13063 (symbol_table::compile): Drop no toplevel reorder path.
13065 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13067 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
13068 edges; zero probability is not better than uninitialized.
13070 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13072 * asan.h (asan_sanitize_allocas_p): Declare.
13073 * asan.c (asan_sanitize_allocas_p): New function.
13074 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
13075 (handle_builtin_alloca): Likewise.
13076 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
13077 if !asan_sanitize_allocas_p.
13078 * params.def (asan-instrument-allocas): Add new option.
13079 * params.h (ASAN_PROTECT_ALLOCAS): Define.
13080 * opts.c (common_handle_option): Disable allocas sanitization for
13083 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
13085 * asan.c: Include gimple-fold.h.
13086 (get_last_alloca_addr): New function.
13087 (handle_builtin_stackrestore): Likewise.
13088 (handle_builtin_alloca): Likewise.
13089 (asan_emit_allocas_unpoison): Likewise.
13090 (get_mem_refs_of_builtin_call): Add new parameter, remove const
13091 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
13092 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
13093 (instrument_builtin_call): Pass gimple iterator to
13094 get_mem_refs_of_builtin_call.
13095 (last_alloca_addr): New global.
13096 * asan.h (asan_emit_allocas_unpoison): Declare.
13097 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
13098 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
13099 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
13100 if function calls alloca.
13101 * gimple-fold.c (replace_call_with_value): Remove static keyword.
13102 * gimple-fold.h (replace_call_with_value): Declare.
13103 * internal-fn.c: Include asan.h.
13104 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
13105 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
13107 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13109 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
13110 (C_SELFTEST_FLAGS): New.
13111 (CPP_SELFTEST_FLAGS): New.
13112 (SELFTEST_DEPS): New, from deps of s-selftest.
13113 (C_SELFTEST_DEPS): New, from deps of s-selftest.
13114 (CPP_SELFTEST_DEPS): New.
13115 (selftest): Add dependency on s-selftest-c++.
13116 (s-selftest): Rename to...
13117 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
13118 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
13119 than SELFTEST_FLAGS.
13120 (selftest-gdb): Rename to...
13121 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
13123 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
13124 (selftest-valgrind): Rename to...
13125 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
13127 (selftest-valgrind): Reintroduce as an alias for
13128 selftest-c-valgrind.
13129 (s-selftest-c++): New.
13130 (selftest-c++-gdb): New.
13131 (selftest-c++-valgrind): New.
13133 2017-07-06 Olivier Hainque <hainque@adacore.com>
13135 * gcc.c (process_command): When deciding if undefined variables
13136 should be ignored when processing specs, accept "gcc -v" as well.
13138 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
13140 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
13141 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
13143 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
13145 * config/arm/arm-cpus.in (armv8-r): Add new entry.
13146 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
13147 * config/arm/arm-tables.opt: Regenerate.
13148 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
13150 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
13152 2017-07-06 Carl Love <cel@us.ibm.com>
13154 * ChangeLog: Clean up from mid air collision
13156 2017-07-06 Carl Love <cel@us.ibm.com>
13158 * config/rs6000/rs6000-c.c: Add support for built-in functions
13159 vector signed int vec_subc (vector signed int, vector signed int);
13160 vector signed __int128 vec_subc (vector signed __int128,
13161 vector signed __int128);
13162 vector unsigned __int128 vec_subc (vector unsigned __int128,
13163 vector unsigned __int128);
13164 vector signed int vec_sube (vector signed int, vector signed int,
13165 vector signed int);
13166 vector unsigned int vec_sube (vector unsigned int,
13167 vector unsigned int,
13168 vector unsigned int);
13169 vector signed __int128 vec_sube (vector signed __int128,
13170 vector signed __int128,
13171 vector signed__int128);
13172 vector unsigned __int128 vec_sube (vector unsigned __int128,
13173 vector unsigned __int128,
13174 vector unsigned __int128);
13175 vector signed int vec_subec (vector signed int, vector signed int,
13176 vector signed int);
13177 vector unsigned int vec_subec (vector unsigned int,
13178 vector unsigned int,
13179 vector unsigned int);
13180 vector signed __int128 vec_subec (vector signed __int128,
13181 vector signed __int128,
13182 vector signed__int128);
13183 vector unsigned __int128 vec_subec (vector unsigned __int128,
13184 vector unsigned __int128,
13185 vector unsigned __int128);
13186 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
13187 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
13188 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
13189 BU_ALTIVEC_OVERLOAD_X definitions.
13190 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
13191 * doc/extend.texi: Update the built-in documentation file for the new
13192 built-in functions.
13194 2017-07-06 David Malcolm <dmalcolm@redhat.com>
13197 * diagnostic-show-locus.c (layout::layout): Use start and finish
13198 spelling location for the start and finish of each range.
13199 * genmatch.c (linemap_client_expand_location_to_spelling_point):
13200 Add unused aspect param.
13201 * input.c (expand_location_1): Add "aspect" param, and use it
13202 to access the correct part of the location.
13203 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
13205 (expand_location_to_spelling_point): Likewise.
13206 (linemap_client_expand_location_to_spelling_point): Add "aspect"
13207 param, and pass it to expand_location_1.
13209 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
13211 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
13212 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
13213 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
13214 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
13215 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
13216 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
13217 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
13218 _mm_maskz_getmant_ss): New intrinsics.
13219 (__builtin_ia32_getexpss128_mask): Changed to ...
13220 __builtin_ia32_getexpss128_round ... this.
13221 (__builtin_ia32_getexpsd128_mask): Changed to ...
13222 __builtin_ia32_getexpsd128_round ... this.
13223 * config/i386/i386-builtin-types.def
13224 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
13225 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
13226 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
13227 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
13228 __builtin_ia32_getmantss_mask_round): New builtins.
13229 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
13230 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
13231 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
13232 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
13233 * config/i386/sse.md
13234 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
13235 avx512f_sgetexp<mode><mask_scalar_name>
13236 <round_saeonly_scalar_name> ... this.
13237 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
13238 %0, %1, %2<round_saeonly_op3>}): Changed to ...
13239 vgetexp<ssescalarmodesuffix>
13240 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13241 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
13242 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
13243 avx512f_vgetmant<mode><mask_scalar_name>
13244 <round_saeonly_scalar_name> ... this.
13245 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
13246 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
13247 vgetmant<ssescalarmodesuffix>
13248 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
13249 %0<mask_scalar_operand4>, %1, %2
13250 <round_saeonly_scalar_mask_op4>, %3} ... this.
13251 * config/i386/subst.md (mask_scalar_operand4,
13252 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
13253 round_saeonly_scalar_nimm_predicate): New subst attributes.
13255 2017-07-06 Julia Koval <julia.koval@intel.com>
13257 * config/i386/i386.c (ix86_erase_embedded_rounding):
13258 Remove code for old rounding pattern.
13260 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
13262 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
13264 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
13266 * doc/sourcebuild.texi (Test Directives, Variants of
13267 dg-require-support): Add documentation for dg-require-stack-check.
13269 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
13271 * config/i386/subst.md (mask_scalar, round_scalar,
13272 round_saeonly_scalar): New meta-templates.
13273 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
13274 round_scalar_mask_operand3, round_scalar_mask_op3,
13275 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
13276 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
13277 round_saeonly_scalar_constraint,
13278 round_saeonly_scalar_prefix): New subst attribute.
13279 * config/i386/sse.md
13280 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
13281 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
13282 <round_scalar_name> ... this.
13283 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
13284 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
13285 <round_scalar_name> ... this.
13286 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
13287 <sse>_vm<code><mode>3<mask_scalar_name>
13288 <round_saeonly_scalar_name> ... this.
13289 (v<plusminus_mnemonic><ssescalarmodesuffix>
13290 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13291 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13292 v<plusminus_mnemonic><ssescalarmodesuffix>
13293 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13294 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13295 (v<multdiv_mnemonic><ssescalarmodesuffix>
13296 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
13297 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
13298 v<multdiv_mnemonic><ssescalarmodesuffix>
13299 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13300 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
13301 (v<maxmin_float><ssescalarmodesuffix>
13302 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
13303 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
13304 v<maxmin_float><ssescalarmodesuffix>
13305 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
13306 %0<mask_scalar_operand3>, %1, %<iptr>2
13307 <round_saeonly_scalar_mask_op3>} ... this.
13309 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
13311 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
13312 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
13314 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
13315 Alan Hayward <alan.hayward@arm.com>
13316 David Sherwood <david.sherwood@arm.com>
13318 * combine.c (simplify_if_then_else): Remove "enum" before
13320 * compare-elim.c (can_eliminate_compare): Likewise.
13321 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
13323 (aarch64_lookup_simd_builtin_type): Likewise.
13324 (aarch64_simd_builtin_type): Likewise.
13325 (aarch64_init_simd_builtin_types): Likewise.
13326 (aarch64_simd_expand_args): Likewise.
13327 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
13329 (aarch64_reverse_mask): Likewise.
13330 (aarch64_simd_emit_reg_reg_move): Likewise.
13331 (aarch64_gen_adjusted_ldpstp): Likewise.
13332 (aarch64_ccmp_mode_to_code): Likewise.
13333 (aarch64_operands_ok_for_ldpstp): Likewise.
13334 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13335 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
13337 (aarch64_min_divisions_for_recip_mul): Likewise.
13338 (aarch64_reassociation_width): Likewise.
13339 (aarch64_get_condition_code_1): Likewise.
13340 (aarch64_simd_emit_reg_reg_move): Likewise.
13341 (aarch64_simd_attr_length_rglist): Likewise.
13342 (aarch64_reverse_mask): Likewise.
13343 (aarch64_operands_ok_for_ldpstp): Likewise.
13344 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
13345 (aarch64_gen_adjusted_ldpstp): Likewise.
13346 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
13348 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
13349 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
13350 (arm_lookup_simd_builtin_type): Likewise.
13351 (arm_simd_builtin_type): Likewise.
13352 (arm_init_simd_builtin_types): Likewise.
13353 (arm_expand_builtin_args): Likewise.
13354 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
13355 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
13356 (ft32_setup_incoming_varargs): Likewise.
13357 (ft32_function_arg): Likewise.
13358 (ft32_function_arg_advance): Likewise.
13359 (ft32_pass_by_reference): Likewise.
13360 (ft32_arg_partial_bytes): Likewise.
13361 (ft32_valid_pointer_mode): Likewise.
13362 (ft32_addr_space_pointer_mode): Likewise.
13363 (ft32_addr_space_legitimate_address_p): Likewise.
13364 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
13366 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
13367 (ix86_emit_outlined_ms2sysv_restore): Likewise.
13368 (iamcu_alignment): Likewise.
13369 (canonicalize_vector_int_perm): Likewise.
13370 (ix86_noce_conversion_profitable_p): Likewise.
13371 (ix86_mpx_bound_mode): Likewise.
13372 (ix86_operands_ok_for_move_multiple): Likewise.
13373 * config/microblaze/microblaze-protos.h
13374 (microblaze_expand_conditional_branch_reg): Likewise.
13375 * config/microblaze/microblaze.c
13376 (microblaze_expand_conditional_branch_reg): Likewise.
13377 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
13379 (rs6000_reassociation_width): Likewise.
13380 (rs6000_invalid_binary_op): Likewise.
13381 (fusion_p9_p): Likewise.
13382 (emit_fusion_p9_load): Likewise.
13383 (emit_fusion_p9_store): Likewise.
13384 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
13386 (riscv_hard_regno_mode_ok_p): Likewise.
13387 (riscv_address_insns): Likewise.
13388 (riscv_split_symbol): Likewise.
13389 (riscv_legitimize_move): Likewise.
13390 (riscv_function_value): Likewise.
13391 (riscv_hard_regno_nregs): Likewise.
13392 (riscv_expand_builtin): Likewise.
13393 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
13394 (riscv_build_integer): Likewise.
13395 (riscv_split_integer): Likewise.
13396 (riscv_legitimate_constant_p): Likewise.
13397 (riscv_cannot_force_const_mem): Likewise.
13398 (riscv_regno_mode_ok_for_base_p): Likewise.
13399 (riscv_valid_base_register_p): Likewise.
13400 (riscv_valid_offset_p): Likewise.
13401 (riscv_valid_lo_sum_p): Likewise.
13402 (riscv_classify_address): Likewise.
13403 (riscv_legitimate_address_p): Likewise.
13404 (riscv_address_insns): Likewise.
13405 (riscv_load_store_insns): Likewise.
13406 (riscv_force_binary): Likewise.
13407 (riscv_split_symbol): Likewise.
13408 (riscv_force_address): Likewise.
13409 (riscv_legitimize_address): Likewise.
13410 (riscv_move_integer): Likewise.
13411 (riscv_legitimize_const_move): Likewise.
13412 (riscv_legitimize_move): Likewise.
13413 (riscv_address_cost): Likewise.
13414 (riscv_subword): Likewise.
13415 (riscv_output_move): Likewise.
13416 (riscv_canonicalize_int_order_test): Likewise.
13417 (riscv_emit_int_order_test): Likewise.
13418 (riscv_function_arg_boundary): Likewise.
13419 (riscv_pass_mode_in_fpr_p): Likewise.
13420 (riscv_pass_fpr_single): Likewise.
13421 (riscv_pass_fpr_pair): Likewise.
13422 (riscv_get_arg_info): Likewise.
13423 (riscv_function_arg): Likewise.
13424 (riscv_function_arg_advance): Likewise.
13425 (riscv_arg_partial_bytes): Likewise.
13426 (riscv_function_value): Likewise.
13427 (riscv_pass_by_reference): Likewise.
13428 (riscv_setup_incoming_varargs): Likewise.
13429 (riscv_print_operand): Likewise.
13430 (riscv_elf_select_rtx_section): Likewise.
13431 (riscv_save_restore_reg): Likewise.
13432 (riscv_for_each_saved_reg): Likewise.
13433 (riscv_register_move_cost): Likewise.
13434 (riscv_hard_regno_mode_ok_p): Likewise.
13435 (riscv_hard_regno_nregs): Likewise.
13436 (riscv_class_max_nregs): Likewise.
13437 (riscv_memory_move_cost): Likewise.
13438 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
13439 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
13440 (rl78_addr_space_address_mode): Likewise.
13441 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13443 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
13444 (rs6000_reassociation_width): Likewise.
13445 (rs6000_invalid_binary_op): Likewise.
13446 (fusion_p9_p): Likewise.
13447 (emit_fusion_p9_load): Likewise.
13448 (emit_fusion_p9_store): Likewise.
13449 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
13450 (ok_for_simple_move_operands): Likewise.
13451 (ok_for_simple_move_strict_operands): Likewise.
13452 (ok_for_simple_arith_logic_operands): Likewise.
13453 (visium_legitimize_reload_address): Likewise.
13454 (visium_select_cc_mode): Likewise.
13455 (output_cbranch): Likewise.
13456 (visium_split_double_move): Likewise.
13457 (visium_expand_copysign): Likewise.
13458 (visium_expand_int_cstore): Likewise.
13459 (visium_expand_fp_cstore): Likewise.
13460 * config/visium/visium.c (visium_pass_by_reference): Likewise.
13461 (visium_function_arg): Likewise.
13462 (visium_function_arg_advance): Likewise.
13463 (visium_libcall_value): Likewise.
13464 (visium_setup_incoming_varargs): Likewise.
13465 (visium_legitimate_constant_p): Likewise.
13466 (visium_legitimate_address_p): Likewise.
13467 (visium_legitimize_address): Likewise.
13468 (visium_secondary_reload): Likewise.
13469 (visium_register_move_cost): Likewise.
13470 (visium_memory_move_cost): Likewise.
13471 (prepare_move_operands): Likewise.
13472 (ok_for_simple_move_operands): Likewise.
13473 (ok_for_simple_move_strict_operands): Likewise.
13474 (ok_for_simple_arith_logic_operands): Likewise.
13475 (visium_function_value_1): Likewise.
13476 (rtx_ok_for_offset_p): Likewise.
13477 (visium_legitimize_reload_address): Likewise.
13478 (visium_split_double_move): Likewise.
13479 (visium_expand_copysign): Likewise.
13480 (visium_expand_int_cstore): Likewise.
13481 (visium_expand_fp_cstore): Likewise.
13482 (visium_split_cstore): Likewise.
13483 (visium_select_cc_mode): Likewise.
13484 (visium_split_cbranch): Likewise.
13485 (output_cbranch): Likewise.
13486 (visium_print_operand_address): Likewise.
13487 * expmed.c (flip_storage_order): Likewise.
13488 * expmed.h (emit_cstore): Likewise.
13489 (flip_storage_order): Likewise.
13490 * genrecog.c (validate_pattern): Likewise.
13491 * hsa-gen.c (gen_hsa_addr): Likewise.
13492 * internal-fn.c (expand_arith_overflow): Likewise.
13493 * ira-color.c (allocno_copy_cost_saving): Likewise.
13494 * lra-assigns.c (find_hard_regno_for_1): Likewise.
13495 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
13496 (process_invariant_for_inheritance): Likewise.
13497 * lra-eliminations.c (move_plus_up): Likewise.
13498 * omp-low.c (lower_oacc_reductions): Likewise.
13499 * simplify-rtx.c (simplify_subreg): Likewise.
13500 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
13501 (TARGET_CHKP_BOUND_MODE): Likewise..
13502 * targhooks.c (default_chkp_bound_mode): Likewise.
13503 (default_setup_incoming_vararg_bounds): Likewise.
13504 * targhooks.h (default_chkp_bound_mode): Likewise.
13505 (default_setup_incoming_vararg_bounds): Likewise.
13506 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
13507 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
13508 (have_whole_vector_shift): Likewise.
13509 * tree-vect-stmts.c (vectorizable_load): Likewise.
13510 * doc/tm.texi: Regenerate.
13512 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13514 Graceful degrade if Binutils PR21472 is not available.
13517 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
13518 .rodata in flash test fails.
13519 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
13520 * confgure: Regenerate.
13521 * config.in: Regenerate.
13522 * config/avr/avr.c (avr_asm_named_section)
13523 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
13524 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
13525 (avr_asm_init_sections): Same.
13527 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
13529 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
13530 (fma<VH:mode>4_intrinsic): Likewise.
13531 (*fmsub<VCVTF:mode>4): Likewise.
13532 (*fmsub<VH:mode>4_intrinsic): Likewise.
13534 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
13537 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
13538 Don't depend on "optimize > 0".
13539 (out_movhi_r_mr, out_movqi_mr_r): Same.
13540 (out_movhi_mr_r, out_movqi_r_mr): Same.
13541 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
13542 io_address_operand on "optimize > 0".
13544 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13546 * tree-loop-distribution.c: Add general explanantion on the pass.
13547 (generate_loops_for_partition): Mark distributed loop.
13548 (pg_add_dependence_edges): New parameter. Handle alias data
13549 dependence specially and record it in the parameter if asked.
13550 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
13551 (init_partition_graph_vertices, add_partition_graph_edge): New.
13552 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
13553 (free_partition_graph_vdata, build_partition_graph): New.
13554 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
13555 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
13556 (data_ref_segment_size, latch_dominated_by_data_ref): New.
13557 (compute_alias_check_pairs, version_loop_by_alias_check): New.
13558 (version_for_distribution_p, finalize_partitions): New.
13559 (distribute_loop): Handle alias data dependence specially. Factor
13560 out loop fusion code as functions and call these functions.
13562 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13564 * tree-loop-distribution.c (classify_partition): New parameter and
13565 better handle reduction statement.
13566 (rdg_build_partitions): Revise comment.
13567 (distribute_loop): Compute statements in all partitions and pass it
13568 to classify_partition.
13570 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13572 * tree-loop-distribution.c (enum partition_type): New.
13573 (struct partition): New field type.
13574 (partition_merge_into): Add parameter. Update partition type.
13575 (data_dep_in_cycle_p, update_type_for_merge): New functions.
13576 (build_rdg_partition_for_vertex): Compute partition type.
13577 (rdg_build_partitions): Dump partition type.
13578 (distribute_loop): Update calls to partition_merge_into.
13580 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13582 * tree-loop-distribution.c (struct ddr_hasher): New.
13583 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
13585 (classify_partition): Call get_data_dependence.
13586 (pg_add_dependence_edges): Ditto.
13587 (distribute_loop): Release data dependence hash table.
13589 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13591 * tree-loop-distribution.c (ref_base_address): Delete.
13592 (similar_memory_accesses): Rename ...
13593 (share_memory_accesses): ... to this. Check if partitions access
13594 the same memory reference.
13595 (distribute_loop): Call share_memory_accesses.
13597 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13599 * tree-loop-distribution.c (struct partition): New field recording
13600 its data reference.
13601 (partition_alloc, partition_free): Init and release data refs.
13602 (partition_merge_into): Merge data refs.
13603 (build_rdg_partition_for_vertex): Collect data refs for partition.
13604 (pg_add_dependence_edges): Change parameters from vector to bitmap.
13606 (distribute_loop): Remve data refs from vertice data of partition
13609 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13611 * tree-loop-distribution.c (params.h): Include header file.
13612 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
13613 (datarefs_vec): New global var.
13614 (create_rdg_vertices): Use datarefs_vec directly.
13615 (free_rdg): Don't free data references.
13616 (build_rdg): Update use. Don't free data references.
13617 (distribute_loop): Compute global variable for data references.
13618 Bail out if there are too many data references.
13620 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13622 * tree-loop-distribution.c (loop_nest): New global var.
13623 (build_rdg): Use loop directly, rather than loop nest.
13624 (pg_add_dependence_edges): Remove loop nest parameter. Use global
13626 (distribute_loop): Compute global variable loop nest. Update use.
13628 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13630 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
13631 (partition_merge_into): New parameter. Dump reason for fusion.
13632 (distribute_loop): Update use of partition_merge_into.
13634 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13636 * tree-loop-distribution.c (bb_top_order_index): New.
13637 (bb_top_order_index_size, bb_top_order_cmp): New.
13638 (stmts_from_loop): Use topological order.
13639 (pass_loop_distribution::execute): Compute and release topological
13640 order for basic blocks.
13642 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13644 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
13647 2017-07-05 Bin Cheng <bin.cheng@arm.com>
13649 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
13650 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
13651 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
13652 * internal-fn.def (LOOP_DIST_ALIAS): New.
13653 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
13654 (fold_loop_internal_call): ... this.
13655 (vect_loop_dist_alias_call): New function.
13656 (set_uid_loop_bbs): Call fold_loop_internal_call.
13657 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
13660 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13663 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
13664 Require dead FLAGS_REG at the beginning of a peephole.
13666 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
13669 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
13670 arguments in the call to __builtin_ia32_sbb_u32.
13671 (_subborrow_u64): Swap _X and _Y arguments in the call to
13672 __builtin_ia32_sbb_u64.
13674 2017-07-04 Jakub Jelinek <jakub@redhat.com>
13677 * tree-vrp.c (compare_assert_loc): Turn into a function template
13678 with stable template parameter. Only test if a->e is NULL,
13679 !a->e == !b->e has been verified already. Use e == NULL or
13680 e != NULL instead of e or ! e tests. If stable is true, don't use
13681 iterative_hash_expr, on the other side allow a or b or both NULL
13682 and sort the NULLs last.
13683 (process_assert_insertions): Sort using compare_assert_loc<false>
13684 instead of compare_assert_loc, later sort using
13685 compare_assert_loc<true> before calling process_assert_insertions_for
13686 in a loop. Use break instead of continue once seen NULL pointer.
13688 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
13690 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
13691 Cortex-R7 and Cortex-R8 processors.
13693 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13695 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
13696 uninitialized while src is not.
13698 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
13700 * common/config/arm/arm-common.c: Adjust include path for
13702 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
13703 (arm-cpu.h): Create in build directory. Adjust dependency rules.
13704 (arm-cpu-data.h): Likewise.
13705 (arm-cpu-cdata.h): Likewise.
13706 * config/arm/arm-cpu.h: Delete.
13707 * config/arm/arm-cpu-cdata.h: Delete.
13708 * config/arm/arm-cpu-data.h: Delete.
13710 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
13712 * config/arm/arm-cpus.in (cortex-a55): New.
13713 (cortex-a75): Likewise.
13714 (cortex-a75.cortex-a55): Likewise.
13715 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
13717 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
13718 * config/arm/arm-cpu-cdata.h: Regenerate.
13719 * config/arm/arm-cpu-data.h: Regenerate.
13720 * config/arm/arm-cpu.h: Regenerate.
13721 * config/arm/arm-tables.opt: Regenerate.
13722 * config/arm/arm-tune.md: Regenerate.
13724 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13726 * haifa-sched.c (sched_create_recovery_edges): Update profile.
13728 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
13730 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
13733 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
13735 PR tree-optimization/81292
13736 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
13737 full_string_p, also call adjust_related_strinfos if the adjustment
13738 is simple, otherwise invalidate related strinfos.
13740 2017-07-04 Martin Liska <mliska@suse.cz>
13743 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
13744 newly created variable as DECL_IGNORED_P.
13746 2017-07-04 Martin Liska <mliska@suse.cz>
13749 * ipa-inline.c (inline_small_functions):
13750 Use xstrdup_for_dump.
13752 2017-07-04 Tom de Vries <tom@codesourcery.com>
13754 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
13756 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13759 * config/darwin.c (darwin_function_switched_text_sections):
13762 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
13764 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
13766 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13768 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
13770 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13772 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
13773 min_profitable_iters, and th as inclusive lower bounds.
13774 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
13775 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
13776 for min_profitable_iters and min_profitable_estimate.
13777 (vect_transform_loop): Treat th as an inclusive lower bound.
13778 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
13780 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
13783 * config/darwin.c (darwin_function_switched_text_sections):
13784 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
13785 in two pieces, and suppress the use of buf.
13787 2017-07-03 Nathan Sidwell <nathan@acm.org>
13789 * hash-table.h (hash_table_mod1): Fix indentation.
13791 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13793 PR middle-end/81290
13794 * predict.c (force_edge_cold): Be more careful about propagation
13796 * profile-count.h (profile_probability::guessed,
13797 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
13799 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
13801 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
13803 * doc/invoke.texi (rcpc architecture extension): Document it.
13805 2017-07-03 Richard Biener <rguenther@suse.de>
13807 PR tree-optimization/60510
13808 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
13809 the scalar reduction PHI and use it.
13810 (vectorizable_reduction): Properly guard the single_defuse_cycle
13811 path for non-SLP reduction chains where we cannot use it.
13812 Rework reduc_def/index and vector type deduction. Rework
13813 vector operand gathering during reduction op code-gen.
13814 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
13815 chains dissolve the chain and leave it to non-SLP reduction
13818 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13820 * tree-data-ref.h (dr_alignment): Declare.
13821 * tree-data-ref.c (dr_alignment): New function.
13822 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
13823 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
13825 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
13827 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13829 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
13830 and base_misalignment fields.
13831 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
13832 * tree-data-ref.c: Include builtins.h.
13833 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
13834 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
13835 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
13836 * tree-vect-data-refs.c: Include tree-cfg.h.
13837 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
13838 fields instead of calculating an alignment here.
13839 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
13840 innermost_loop_behavior fields.
13842 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13844 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
13846 (DR_STEP_ALIGNMENT): New macro.
13847 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
13848 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
13849 (create_data_ref): Print it.
13850 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
13851 to tell whether the step preserves vector (mis)alignment.
13852 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13853 Move the check for an integer step and generalise to all INTEGER_CST.
13854 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
13855 Print the outer step alignment.
13857 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13859 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
13860 with offset_alignment.
13861 (DR_ALIGNED_TO): Delete.
13862 (DR_OFFSET_ALIGNMENT): New macro.
13863 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
13864 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
13865 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
13866 (create_data_ref): Likewise.
13867 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
13868 (vect_analyze_data_refs): Likewise.
13869 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
13870 creating dummy innermost behavior.
13872 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13874 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
13875 with a "innermost_loop_behavior *" and refeence tree.
13876 * tree-data-ref.c (dr_analyze_innermost): Likewise.
13877 (create_data_ref): Update call accordingly.
13878 * tree-predcom.c (find_looparound_phi): Likewise.
13880 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13882 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
13883 fields with dr_wrt_vec_loop.
13884 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
13885 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
13886 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
13887 (vect_dr_behavior): New function.
13888 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13889 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
13890 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
13891 track whether the step preserves the misalignment.
13892 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
13893 Use vect_dr_behavior.
13894 (vect_setup_realignment): Update call accordingly.
13895 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
13896 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
13897 call to vect_create_addr_base_for_vector_ref.
13898 (vect_create_cond_for_align_checks): Likewise.
13899 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
13900 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
13901 (vect_recog_mask_conversion_pattern): Likewise.
13902 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
13903 (new_stmt_vec_info): Remove redundant zeroing.
13905 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
13907 * common/config/arm/arm-common.c (arm_be8_option): New function.
13908 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
13909 (ISA_ARMv6): Add isa_bit_be8.
13910 * config/arm/arm.h (arm_be8_option): Add prototype.
13911 (BE8_SPEC_FUNCTION): New define.
13912 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
13913 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
13914 (mlittle-endian): Similarly.
13915 (mbe8, mbe32): New options.
13916 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
13917 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
13919 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13921 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
13923 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13925 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
13926 (cleanup_tree_cfg_bb): Use it.
13927 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
13929 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
13931 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
13934 * loop-doloop.c (add_test): Update profile.
13936 2017-07-03 Martin Liska <mliska@suse.cz>
13939 * sanopt.c (rewrite_usage_of_param): New function.
13940 (sanitize_rewrite_addressable_params): Likewise.
13941 (pass_sanopt::execute): Call rewrite_usage_of_param.
13943 2017-07-03 Richard Biener <rguenther@suse.de>
13945 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
13946 back to using VIEW_CONVERT_EXPR.
13948 2017-07-03 Martin Liska <mliska@suse.cz>
13951 * doc/extend.texi: Document when a resolver function is
13952 generated for target_clones.
13954 2017-07-03 Martin Liska <mliska@suse.cz>
13956 * asan.c (asan_emit_stack_protection): Unpoison just red zones
13957 and shadow memory of auto variables which are subject of
13958 use-after-scope sanitization.
13959 (asan_expand_mark_ifn): Add do set only when is_poison.
13961 2016-07-03 Richard Biener <rguenther@suse.de>
13963 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
13965 (vect_force_simple_reduction): Record reduction def -> phi mapping.
13966 (vectorizable_reduction): Perform reduction PHI creation when
13967 visiting a reduction PHI and adjust and simplify code generation
13968 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
13969 (vect_transform_loop): Visit reduction PHIs.
13970 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
13971 defs into the SLP tree.
13972 (vect_build_slp_tree): Reduction defs terminate the recursion.
13973 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
13975 (vect_get_vec_defs_for_stmt_copy): Export.
13976 (vect_get_vec_defs): Likewise.
13977 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
13979 (vect_get_vec_defs_for_stmt_copy): Declare.
13980 (vect_get_vec_defs): Likewise.
13982 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
13984 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
13985 parameter with a "loop" parameter and use it instead of the
13986 loop containing DR_STMT. Don't check simple_iv when doing
13987 BB analysis. Describe the two analysis modes in the comment.
13989 2017-07-03 Tom de Vries <tom@codesourcery.com>
13991 PR tree-optimization/69468
13992 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
13993 (find_same_succ_bb): Handle ignore_edge_flags.
13995 2017-07-03 Tom de Vries <tom@codesourcery.com>
13997 PR tree-optimization/81192
13998 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
14000 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
14002 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
14004 2017-07-03 Tom de Vries <tom@codesourcery.com>
14006 PR tree-optimization/81192
14007 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
14008 BB_SAME_SUCC (bb) == NULL.
14010 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14012 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
14015 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14017 * dumpfile.c: Include profile-count.h
14018 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
14020 (insert_cond_bb): Update profile.
14021 * tree-cfg.h (insert_cond_bb): Update prototype.
14022 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
14023 * tree-dump.c: Do not include tree-cfg.
14025 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14027 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
14029 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14031 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
14034 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
14036 * tree-complex.c (expand_complex_div_wide): update profile.
14038 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14039 Alan Hayward <alan.hayward@arm.com>
14040 David Sherwood <david.sherwood@arm.com>
14042 * Makefile.in (MACHMODE_H): Remove insn-modes.h
14043 (CORETYPES_H): New define.
14044 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
14045 (insn-modes-inline.h, s-modes-inline-h): New rules.
14046 (generated_files): Add insn-modes-inline.h.
14047 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
14048 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
14049 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
14050 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
14051 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
14052 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
14053 (build/gencodes.o, build/genconditions.o): Likewise.
14054 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
14055 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
14056 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
14057 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
14058 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
14059 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
14060 * coretypes.h: Include everything up to real.h for generators.
14061 Include insn-modes.h first. Include wide-int-print.h after
14062 wide-int.h. Include insn-modes-inline.h and then machmode.h.
14063 * machmode.h: Don't include insn-modes.h here.
14064 * function-tests.c: Remove includes of signop.h, machmode.h,
14065 double-int.h and wide-int.h.
14067 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
14069 * optc-save-gen.awk: Likewise.
14070 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
14071 * godump.c: Remove include of wide-int-print.h.
14072 * pretty-print.h: Likewise.
14073 * wide-int-print.cc: Likewise.
14074 * wide-int.cc: Likewise.
14075 * hash-map-tests.c: Remove include of signop.h.
14076 * hash-set-tests.c: Likewise.
14077 * rtl-tests.c: Likewise.
14078 * mkconfig.sh: Remove include of machmode.h.
14079 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
14081 (emit_insn_modes_inline_h): ...this new function. Emit the code
14082 into an insn-modes-inline.h header file, adding appropriate
14083 include guards and end comments.
14084 (emit_insn_modes_c_header): Remove include of machmode.h.
14085 (emit_min_insn_modes_c_header): Include coretypes.h rather than
14087 (main): Handle -i flag and call emit_insn_modes_inline_h when
14090 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14092 * tree-ssa-strlen.c (strinfo): Rename the length field to
14093 nonzero_chars. Add a full_string_p field.
14094 (compare_nonzero_chars, zero_length_string_p): New functions.
14095 (get_addr_stridx): Add an offset_out parameter.
14096 Use compare_nonzero_chars.
14097 (get_stridx): Update accordingly. Use compare_nonzero_chars.
14098 (new_strinfo): Update after above changes to strinfo.
14099 (set_endptr_and_length): Set full_string_p.
14100 (get_string_length): Update after above changes to strinfo.
14101 (unshare_strinfo): Update call to new_strinfo.
14102 (maybe_invalidate): Likewise.
14103 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
14104 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
14105 as a uhwi instead of an shwi. Update after above changes to
14106 strinfo and new_strinfo.
14107 (zero_length_string): Assert that chainsi contains full strings.
14108 Use zero_length_string_p. Update call to new_strinfo.
14109 (adjust_related_strinfos): Update after above changes to strinfo.
14110 Copy full_string_p from origsi.
14111 (adjust_last_stmt): Use zero_length_string_p.
14112 (handle_builtin_strlen): Update after above changes to strinfo and
14113 new_strinfo. Install the lhs as the string length if the previous
14114 entry didn't describe a full string.
14115 (handle_builtin_strchr): Update after above changes to strinfo
14117 (handle_builtin_strcpy): Likewise.
14118 (handle_builtin_strcat): Likewise.
14119 (handle_builtin_malloc): Likewise.
14120 (handle_pointer_plus): Likewise.
14121 (handle_builtin_memcpy): Likewise. Track nonzero characters
14122 that aren't necessarily followed by a nul terminator.
14123 (handle_char_store): Likewise.
14125 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14127 PR tree-optimization/80769
14128 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
14129 for malloc and calloc. Document the new invariant that all related
14130 strinfos have delayed lengths or none do.
14131 (verify_related_strinfos): Move earlier in file.
14132 (set_endptr_and_length): New function, split out from...
14133 (get_string_length): ...here. Also set the lengths of related
14135 (zero_length_string): Assert that chainsi has known (rather than
14137 (adjust_related_strinfos): Likewise.
14139 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
14141 PR tree-optimization/81136
14142 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
14143 assert that two references with the same misalignment have the same
14144 compile-time misalignment if those compile-time misalignments
14147 2017-07-01 Andi Kleen <ak@linux.intel.com>
14149 * print-tree.c (print_node): Print all attributes.
14151 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14153 * cfg.c (scale_bbs_frequencies): New function.
14154 * cfg.h (scale_bbs_frequencies): Declare it.
14155 * cfgloopanal.c (single_likely_exit): Cleanup.
14156 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
14158 (scale_loop_profile): Likewise.
14159 (loop_version): Likewise.
14160 (create_empty_loop_on_edge): Update.
14161 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
14162 scale_loop_frequencies, scale_loop_profile, loopify,
14163 loop_version): Update prototypes.
14164 * modulo-sched.c (sms_schedule): Update.
14165 * predict.c (unlikely_executed_edge_p): Also check probability.
14166 (probably_never_executed_edge_p): Fix typo.
14167 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14168 * tree-parloops.c (gen_parallel_loop): Update.
14169 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
14170 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14171 * tree-ssa-loop-split.c (split_loop): Update.
14172 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14173 * tree-vect-loop-manip.c (vect_do_peeling): Update.
14174 (vect_loop_versioning): Update.
14175 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14177 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14179 * trans-mem.c (split_bb_make_tm_edge): Update profile.
14181 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14183 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
14184 to keep profile consistent.
14186 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14188 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
14189 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
14190 * profile-count.h (max_safe_multiplier): Make unsigned.
14191 (profile_count::guessed_zero): New.
14193 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14195 * bb-reorder.c (fix_up_crossing_landing_pad,
14196 fix_crossing_conditional_branches): Use make_single_succ_edge
14197 to keep profile consistent.
14199 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
14201 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
14204 2017-07-01 Jakub Jelinek <jakub@redhat.com>
14207 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
14208 the right scopes, make sure cond_jump isn't preserved between multiple
14209 iterations. Search for fallthru edge whenever there are 3+ edges and
14210 use find_fallthru_edge for it.
14212 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14214 Patch by Alexander Monakov <amonakov@ispras.ru>
14215 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
14216 probabilities consistently.
14218 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14220 * pa.c (pa_expand_compare_and_swap_loop): Update call of
14221 emit_cmp_and_jump_insns.
14223 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14226 * tree-inline.c (expand_call_inline): Combine profile statuses.
14228 2017-06-30 Andrew Pinski <apinski@cavium.com>
14230 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
14231 fold_stmt returned true.
14233 2017-06-30 Nathan Sidwell <nathan@acm.org>
14235 * ggc.h (empty_string): Delete.
14236 * cfgexpand.c (expand_asm_stmt): Use plain "".
14237 * optabs.c (expand_asm_memory_barrier): Likewise.
14238 * stringpool.c (empty_string): Delete.
14239 (digit_vector, digit_string): Delete.
14240 (ggc_alloc_string): Use plain "", don't optimize single digit
14241 strings. Use ggc_alloc_atomic.
14243 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
14245 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
14246 comparison set and one other set, use the cost of the non-comparison
14249 2017-06-30 Nathan Sidwell <nathan@acm.org>
14251 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
14254 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
14256 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
14257 loops. Remove now unneeded calls to gimple_switch_set_label() that
14258 just set removed labels to NULL_TREE.
14260 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
14262 * tree-ssanames.c (set_range_info_raw): Abstract from ...
14263 (set_range_info): ...here. Only call set_range_info_raw if domain
14265 (set_nonzero_bits): Call set_range_info_raw.
14266 * tree-ssanames.h (set_range_info_raw): New.
14268 2017-06-30 Jakub Jelinek <jakub@redhat.com>
14271 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
14272 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
14273 of nonimmediate_operand and <store_mask_constraint> instead of m
14274 for the input operand. For V8FI iterator, always split if input
14275 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
14276 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
14277 <store_mask_predicate> instead of register_operand and
14278 <store_mask_constraint> instead of v for the input operand. Make
14279 sure both operands aren't MEMs for if not <mask_applied>.
14281 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
14283 * lto-wrapper.c (copy_file) Close both file descriptors before
14286 2017-06-30 Martin Liska <mliska@suse.cz>
14289 * multiple_target.c (create_dispatcher_calls): Make ifunc
14290 also for function that don't have calls or are not referenced.
14292 2017-06-30 Richard Biener <rguenther@suse.de>
14294 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
14295 analyze the first scalar stmt. Move vector type computation
14296 for the BB case here from ...
14297 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
14298 live operation processing in the SLP case properly.
14300 2017-06-30 Richard Biener <rguenther@suse.de>
14302 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
14304 2017-06-30 Martin Liska <mliska@suse.cz>
14307 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
14308 before BUILT_IN_UNWIND_RESUME when ASAN is used.
14310 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
14312 * doc/invoke.texi (AArch64): Add missing options and remove redundant
14315 2017-06-30 Richard Biener <rguenther@suse.de>
14317 PR tree-optimization/81249
14318 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
14319 condition reduction result to original scalar type.
14321 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14323 * profile-count.h (enum profile_quality): Fix typos and whitespace
14326 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14328 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
14329 type for branch probabilities.
14331 2017-06-29 Julian Brown <julian@codesourcery.com>
14332 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14334 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
14335 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
14336 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
14337 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
14339 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14341 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
14342 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
14343 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
14344 CC usage from generic code to here.
14345 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
14346 CC usage into the target macros.
14348 2017-06-29 Maya Rashish <coypu@sdf.org>
14350 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
14353 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14355 * arm/arm-builtins.c: Include profile-count.h
14356 * except.c (sjlj_emit_function_enter): Use
14357 profile_probability::unlikely.
14359 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14361 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
14362 and tocrel_offset be pointer args rather than implicitly using
14364 (legitimate_constant_pool_address_p, rs6000_emit_move,
14365 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
14366 tocrel_offset and use in toc_relative_expr_p call.
14367 (print_operand, print_operand_address): Use static tocrel_base_oac
14368 and tocrel_offset_oac.
14369 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
14372 2017-06-29 Maya Rashish <coypu@sdf.org>
14374 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
14376 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
14378 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
14379 objects, take into account only the alignment of 'op0' and 'mode1' if
14382 2017-06-29 Steve Ellcey <sellcey@cavium.com>
14384 * ccmp.c (ccmp_tree_comparison_p): New function.
14385 (ccmp_candidate_p): Update to use above function.
14386 (get_compare_parts): New function.
14387 (expand_ccmp_next): Update to use new functions.
14388 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
14390 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
14391 take mode as argument.
14392 * ccmp.h (expand_ccmp_expr): Add mode as argument.
14393 * expr.c (expand_expr_real_1): Pass mode as argument.
14395 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
14397 * combine.c (combine_instructions): Print insns to dump_file, together
14400 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
14402 * asan.c (asan_emit_stack_protection): Update.
14403 (create_cond_insert_point): Update.
14404 * auto-profile.c (afdo_propagate_circuit): Update.
14405 * basic-block.h (struct edge_def): Turn probability to
14406 profile_probability.
14407 (EDGE_FREQUENCY): Update.
14408 * bb-reorder.c (find_traces_1_round): Update.
14409 (better_edge_p): Update.
14410 (sanitize_hot_paths): Update.
14411 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
14412 (make_single_succ_edge): Update.
14413 (check_bb_profile): Update.
14414 (dump_edge_info): Update.
14415 (update_bb_profile_for_threading): Update.
14416 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
14417 probabilitycount to 0.
14418 * cfgbuild.c (compute_outgoing_frequencies): Update.
14419 * cfgcleanup.c (try_forward_edges): Update.
14420 (outgoing_edges_match): Update.
14421 (try_crossjump_to_edge): Update.
14422 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
14423 (expand_gimple_tailcall): Update.
14424 (construct_init_block): Use make_single_succ_edge.
14425 (construct_exit_block): Use make_single_succ_edge.
14426 * cfghooks.c (verify_flow_info): Update.
14427 (redirect_edge_succ_nodup): Update.
14428 (split_edge): Update.
14429 (account_profile_record): Update.
14430 * cfgloopanal.c (single_likely_exit): Update.
14431 * cfgloopmanip.c (scale_loop_profile): Update.
14432 (set_zero_probability): Remove.
14433 (duplicate_loop_to_header_edge): Update.
14434 * cfgloopmanip.h (loop_version): Update prototype.
14435 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
14436 (force_nonfallthru_and_redirect): Update.
14437 (update_br_prob_note): Update.
14438 (rtl_verify_edges): Update.
14439 (purge_dead_edges): Update.
14440 (rtl_lv_add_condition_to_bb): Update.
14441 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
14442 * cgraphunit.c (init_lowered_empty_function): Update.
14443 (cgraph_node::expand_thunk): Update.
14444 * cilk-common.c: Include profile-count.h
14445 * dojump.c (inv): Remove.
14446 (jumpifnot): Update.
14447 (jumpifnot_1): Update.
14448 (do_jump_1): Update.
14450 (do_jump_by_parts_greater_rtx): Update.
14451 (do_compare_rtx_and_jump): Update.
14452 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
14453 do_jump_1. do_compare_rtx_and_jump): Update prototype.
14454 * dwarf2cfi.c: Include profile-count.h
14455 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
14456 (sjlj_emit_dispatch_table): Likewise.
14457 * explow.c: Include profile-count.h
14458 * expmed.c (emit_store_flag_force): Update.
14459 (do_cmp_and_jump): Update.
14460 * expr.c (compare_by_pieces_d::generate): Update.
14461 (compare_by_pieces_d::finish_mode): Update.
14462 (emit_block_move_via_loop): Update.
14463 (store_expr_with_bounds): Update.
14464 (store_constructor): Update.
14465 (expand_expr_real_2): Update.
14466 (expand_expr_real_1): Update.
14467 * expr.h (try_casesi, try_tablejump): Update prototypes.
14468 * gimple-pretty-print.c (dump_probability): Update.
14469 (dump_profile): New.
14470 (dump_gimple_label): Update.
14471 (dump_gimple_bb_header): Update.
14472 * graph.c (draw_cfg_node_succ_edges): Update.
14473 * hsa-gen.c (convert_switch_statements): Update.
14474 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
14475 (find_if_case_1): Update.
14476 (find_if_case_2): Update.
14477 * internal-fn.c (expand_arith_overflow_result_store): Update.
14478 (expand_addsub_overflow): Update.
14479 (expand_neg_overflow): Update.
14480 (expand_mul_overflow): Update.
14481 (expand_vector_ubsan_overflow): Update.
14482 * ipa-cp.c (good_cloning_opportunity_p): Update.
14483 * ipa-split.c (split_function): Use make_single_succ_edge.
14484 * ipa-utils.c (ipa_merge_profiles): Update.
14485 * loop-doloop.c (add_test): Update.
14486 (doloop_modify): Update.
14487 * loop-unroll.c (compare_and_jump_seq): Update.
14488 (unroll_loop_runtime_iterations): Update.
14489 * lra-constraints.c (lra_inheritance): Update.
14490 * lto-streamer-in.c (input_cfg): Update.
14491 * lto-streamer-out.c (output_cfg): Update.
14492 * mcf.c (adjust_cfg_counts): Update.
14493 * modulo-sched.c (sms_schedule): Update.
14494 * omp-expand.c (expand_omp_for_init_counts): Update.
14495 (extract_omp_for_update_vars): Update.
14496 (expand_omp_ordered_sink): Update.
14497 (expand_omp_for_ordered_loops): Update.
14498 (expand_omp_for_generic): Update.
14499 (expand_omp_for_static_nochunk): Update.
14500 (expand_omp_for_static_chunk): Update.
14501 (expand_cilk_for): Update.
14502 (expand_omp_simd): Update.
14503 (expand_omp_taskloop_for_outer): Update.
14504 (expand_omp_taskloop_for_inner): Update.
14505 * omp-simd-clone.c (simd_clone_adjust): Update.
14506 * optabs.c (expand_doubleword_shift): Update.
14507 (expand_abs): Update.
14508 (emit_cmp_and_jump_insn_1): Update.
14509 (expand_compare_and_swap_loop): Update.
14510 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
14511 * predict.c (predictable_edge_p): Update.
14512 (edge_probability_reliable_p): Update.
14513 (set_even_probabilities): Update.
14514 (combine_predictions_for_insn): Update.
14515 (combine_predictions_for_bb): Update.
14516 (propagate_freq): Update.
14517 (estimate_bb_frequencies): Update.
14518 (force_edge_cold): Update.
14519 * profile-count.c (profile_count::dump): Add missing space into dump.
14520 (profile_count::debug): Add newline.
14521 (profile_count::differs_from_p): Explicitly convert to unsigned.
14522 (profile_count::stream_in): Update.
14523 (profile_probability::dump): New member function.
14524 (profile_probability::debug): New member function.
14525 (profile_probability::differs_from_p): New member function.
14526 (profile_probability::differs_lot_from_p): New member function.
14527 (profile_probability::stream_in): New member function.
14528 (profile_probability::stream_out): New member function.
14529 * profile-count.h (profile_count_quality): Rename to ...
14530 (profile_quality): ... this one.
14531 (profile_probability): New.
14532 (profile_count): Update.
14533 * profile.c (compute_branch_probabilities): Update.
14534 * recog.c (peep2_attempt): Update.
14535 * sched-ebb.c (schedule_ebbs): Update.
14536 * sched-rgn.c (find_single_block_region): Update.
14537 (compute_dom_prob_ps): Update.
14538 (schedule_region): Update.
14539 * sel-sched-ir.c (compute_succs_info): Update.
14540 * stmt.c (struct case_node): Update.
14541 (do_jump_if_equal): Update.
14542 (get_outgoing_edge_probs): Update.
14543 (conditional_probability): Update.
14544 (emit_case_dispatch_table): Update.
14545 (expand_case): Update.
14546 (expand_sjlj_dispatch_table): Update.
14547 (emit_case_nodes): Update.
14548 * targhooks.c: Update.
14549 * tracer.c (better_p): Update.
14550 (find_best_successor): Update.
14551 * trans-mem.c (expand_transaction): Update.
14552 * tree-call-cdce.c: Update.
14553 * tree-cfg.c (gimple_split_edge): Upate.
14554 (move_sese_region_to_fn): Upate.
14555 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
14556 * tree-eh.c (lower_resx): Upate.
14557 (cleanup_empty_eh_move_lp): Upate.
14558 * tree-if-conv.c (version_loop_for_if_conversion): Update.
14559 * tree-inline.c (copy_edges_for_bb): Update.
14560 (copy_cfg_body): Update.
14561 * tree-parloops.c (gen_parallel_loop): Update.
14562 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
14563 (gimple_gen_time_profiler): Update.
14564 * tree-ssa-dce.c (remove_dead_stmt): Update.
14565 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
14566 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
14567 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
14568 (unloop_loops): Update.
14569 (try_peel_loop): Update.
14570 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
14571 * tree-ssa-loop-split.c (connect_loops): Update.
14572 (split_loop): Update.
14573 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
14574 (hoist_guard): Update.
14575 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
14576 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
14577 (value_replacement): Update.
14578 * tree-ssa-reassoc.c (branch_fixup): Update.
14579 * tree-ssa-tail-merge.c (replace_block_by): Update.
14580 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
14581 (create_edge_and_update_destination_phis): Update.
14582 (compute_path_counts): Update.
14583 (recompute_probabilities): Update.
14584 (update_joiner_offpath_counts): Update.
14585 (freqs_to_counts_path): Update.
14586 (duplicate_thread_path): Update.
14587 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
14588 (struct switch_conv_info): Update.
14589 (gen_inbound_check): Update.
14590 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
14591 (vect_do_peeling): Update.
14592 (vect_loop_versioning): Update.
14593 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
14594 (optimize_mask_stores): Update.
14595 * ubsan.c (ubsan_expand_null_ifn): Update.
14596 * value-prof.c (gimple_divmod_fixed_value): Update.
14597 (gimple_divmod_fixed_value_transform): Update.
14598 (gimple_mod_pow2): Update.
14599 (gimple_mod_pow2_value_transform): Update.
14600 (gimple_mod_subtract): Update.
14601 (gimple_mod_subtract_transform): Update.
14602 (gimple_ic): Update.
14603 (gimple_stringop_fixed_value): Update.
14604 (gimple_stringops_transform): Update.
14605 * value-prof.h: Update.
14607 2017-06-29 Carl Love <cel@us.ibm.com>
14609 * config/rs6000/rs6000-c.c: Add support for built-in functions
14610 vector signed int vec_signed (vector float);
14611 vector signed long long vec_signed (vector double);
14612 vector signed int vec_signed2 (vector double, vector double);
14613 vector signed int vec_signede (vector double);
14614 vector signed int vec_signedo (vector double);
14615 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
14616 instruction generator.
14617 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14618 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
14619 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
14621 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
14622 vunsignede_v2df): Add define_expands.
14623 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
14624 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
14625 VEC_UNSIGNEDO): Add definitions.
14626 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
14627 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
14628 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
14629 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
14630 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
14631 * config/rs6000/altivec.h (vec_signed, vec_signed2,
14632 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
14633 vec_unsignede, vec_unsignedo): Add builtin defines.
14634 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
14636 * doc/extend.texi: Update the built-in documentation file for the
14637 new built-in functions.
14639 2017-06-29 Richard Biener <rguenther@suse.de>
14641 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
14642 reduction chains to LOOP_VINFO_REDUCTIONS.
14643 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
14644 SLP reductions after processing reduction chains.
14646 2017-06-29 Nathan Sidwell <nathan@acm.org>
14648 * builtins.c (fold_builtin_FUNCTION): Use
14649 lang_hooks.decl_printable_name.
14651 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
14653 PR middle-end/81194
14654 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
14655 with only one label.
14656 * stmt.c (expand_case): Assert NCASES is greater than one.
14658 2017-06-29 Richard Biener <rguenther@suse.de>
14660 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
14662 (group_case_labels): Likewise.
14663 (find_taken_edge): Push sanity checking on val to workers...
14664 (find_taken_edge_cond_expr): ... here
14665 (find_taken_edge_switch_expr): ... and here, handle cases
14666 with just a default label.
14667 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
14668 (group_case_labels): Likewise.
14669 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
14670 group_case_labels does anything cleanup the CFG again.
14672 2017-06-29 Bin Cheng <bin.cheng@arm.com>
14674 PR tree-optimization/81196
14675 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
14676 exit condition comparing two IVs.
14678 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
14680 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
14681 profile to the dummy entry at the end of the list of architectures.
14682 * config/arm/arm-cpu-cdata.h: Regenerated.
14684 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14685 Michael Collison <michael.collison@arm.com>
14688 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
14690 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
14691 (*aarch64_reg_<mode>3_minus_mask): New pattern.
14692 (*aarch64_<optab>_reg_di3_mask2): New pattern.
14693 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
14694 of shift when the shift amount is masked with constant equal to
14695 the size of the mode.
14696 * config/aarch64/predicates.md (subreg_lowpart_operator): New
14699 2017-06-29 Martin Liska <mliska@suse.cz>
14701 * config/i386/i386.opt: Change range from [1,5] to [0,5].
14703 2017-06-29 Yury Gribov <tetra2005@gmail.com>
14706 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
14709 (edge_growth_cache_entry::edge_growth_cache_entry): New
14711 (reset_edge_growth_cache): Update to use constructor.
14713 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14715 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
14716 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14717 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
14719 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
14721 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
14722 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
14724 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
14726 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
14727 (*-linux-uclibc*): Add t-uclibc tmake_file.
14728 * config/t-musl: New.
14729 * config/t-uclibc: New.
14731 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
14733 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
14735 (gen_comm_data): Emit architectural setting of arch_prof.
14736 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
14738 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
14739 (armv8-m.base, armv8-m.main): Likewise.
14740 * arm-protos.h (arm_build_target): Add profile field.
14741 (arch_option): Likewise.
14742 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
14744 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
14745 arm_active_target.profile.
14747 2017-06-28 Richard Biener <rguenther@suse.de>
14749 PR middle-end/81227
14750 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
14751 TYPE_OVERFLOW_WRAPS.
14752 * match.pd (negate_expr_p): Likewise.
14753 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
14754 fold_build2, not fold_binary.
14756 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14758 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
14759 Convert memory address to Pmode.
14760 (aarch64_print_operand): Assert MEM operands are always Pmode.
14762 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
14765 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
14766 Remove redundant if.
14767 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
14768 * config/arm/aarch-common-protos.h
14769 (aarch_forward_to_shift_is_not_shifted_re): Remove.
14770 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
14772 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
14775 * multiple_target.c (create_dispatcher_calls): Set the default
14776 clone to be static, not public.
14778 2017-06-28 Richard Biener <rguenther@suse.de>
14780 * tree-vect-loop.c (vectorizable_reduction): Move special
14781 cond reduction IV var creation ...
14782 (vect_create_epilog_for_reduction): ... here. Remove induction_index
14783 parameter. Use STMT_VINFO_VECTYPE.
14784 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
14787 2017-06-28 Martin Liska <mliska@suse.cz>
14790 * ipa-visibility.c (non_local_p): Handle visibility.
14792 2017-06-28 Martin Liska <mliska@suse.cz>
14795 * common.opt: Add IntegerRange to various options.
14796 * opt-functions.awk (integer_range_info): New function.
14797 * optc-gen.awk: Add integer_range_info to cl_options struct.
14798 * opts-common.c (decode_cmdline_option): Handle
14799 CL_ERR_INT_RANGE_ARG.
14800 (cmdline_handle_error): Likewise.
14801 * opts.c (print_filtered_help): Show valid interval in
14802 when --help is provided.
14803 * opts.h (struct cl_option): Add range_min and range_max fields.
14804 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
14806 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
14808 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
14809 (x * C EQ/NE y * C): New transformation.
14811 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
14813 * genmultilib (combination_space): Accept '+' in option names.
14815 2017-06-28 Martin Liska <mliska@suse.cz>
14818 * asan.c (instrument_derefs): Bail out inner references
14819 that are hard register variables.
14821 2017-06-28 Jakub Jelinek <jakub@redhat.com>
14824 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
14825 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
14827 2017-06-28 Richard Biener <rguenther@suse.de>
14829 * tree-vectorizer.h (vect_get_vec_defs): Remove.
14830 (vect_get_slp_defs): Adjust.
14831 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
14833 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
14835 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
14836 get_initial_defs_for_reduction instead of vect_get_vec_defs.
14837 (vectorizable_reduction): Adjust.
14838 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
14840 (vect_get_slp_defs): Likewise.
14841 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
14842 (vectorizable_bswap): Adjust.
14843 (vectorizable_call): Likewise.
14844 (vectorizable_conversion): Likewise.
14845 (vectorizable_assignment): Likewise.
14846 (vectorizable_shift): Likewise.
14847 (vectorizable_operation): Likewise.
14848 (vectorizable_store): Likewise.
14849 (vectorizable_condition): Likewise.
14850 (vectorizable_comparison): Likewise.
14852 2017-06-28 Michael Collison <michael.collison@arm.com>
14855 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
14857 (arm_gen_movmemqi): Removed unused variable 'i'.
14858 Convert 'for' loop into 'while' loop.
14859 (arm_expand_prologue): Remove last unnecessary set of insn.
14860 (thumb_pop): Remove unused variable 'pushed_words'.
14861 (thumb_exit): Remove last unnecessary set of regs_to_pop.
14863 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14865 * config/s390/predicates.md: Use s390_rel_address_ok_p.
14866 * config/s390/s390-protos.h: Add prototype of
14867 s390_rel_address_ok_p.
14868 * config/s390/s390.c (s390_got_symbol): New function.
14869 (s390_rel_address_ok_p): New function.
14870 (legitimize_pic_address): Use s390_rel_address_ok_p.
14871 (s390_load_got): Use s390_got_symbol.
14872 (s390_option_override): Issue error if
14873 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
14874 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
14876 * config/s390/s390.opt: New option mpic-data-is-text-relative.
14878 2017-06-27 Andrew Pinski <apinski@cavium.com>
14880 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
14881 (X * copysign (1.0, X)): New pattern.
14882 (X * copysign (1.0, -X)): New pattern.
14883 (copysign (-1.0, CST)): New pattern.
14885 2017-06-27 Joseph Myers <joseph@codesourcery.com>
14887 * genmultilib (combination_space): Remove variable.
14888 Validate reuse rules against regular expression for any sequence
14889 of multilib options in any order.
14891 2017-06-27 Michael Collison <michael.collison@arm.com>
14893 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
14894 call aarch64_split_simd_combine.
14895 * (aarch64_combine_internal<mode>): Delete pattern.
14896 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
14897 Allow register and subreg operands.
14899 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14901 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
14902 specific need, just fallback on defaults.
14903 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
14905 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14906 Olivier Hainque <hainque@adacore.com>
14908 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
14910 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
14911 targets. Pick a default if no particular attempt applied.
14912 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
14915 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14917 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
14918 (x86_64-wrs-vxworks7): Likewise.
14920 2017-06-27 Marek Polacek <polacek@redhat.com>
14923 * ubsan.c (instrument_null): Check get_base_address's result for null.
14925 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14927 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
14929 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
14931 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
14932 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
14933 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
14934 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
14935 New function types.
14936 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
14937 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
14938 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
14939 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
14940 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
14941 BUILT_IN_FEUPDATEENV): New builtins.
14942 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
14943 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
14944 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
14945 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
14947 (builtin_structptr_types): Adjust size.
14948 * tree.c (builtin_structptr_types): Add four entries.
14950 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14951 Olivier Hainque <hainque@adacore.com>
14953 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
14954 (TLS_SYM): New local macro, forcing reference to __tls__ on
14955 link command lines for VxWorks 7 RTPs, triggering initialization
14957 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
14958 OS features TLS support, true for RTPs on VxWorks 7.
14959 * config/vxworks.c (vxworks_override_options): Setup emutls
14962 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14964 * predict.c (test_prediction_value_range): Use -1U instead of -1
14965 to avoid narrowing conversion warning.
14966 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
14967 to avoid narrowing conversion warning.
14968 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
14970 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
14972 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
14974 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
14975 64bit configurations.
14976 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
14977 (SIZE_TYPE): Likewise.
14978 * config/vxworks.c (vxworks_emutls_var_fields): Use
14979 long_unsigned_type_node instead of unsigned_type_node as the offset
14980 field type, which is "pointer" mode in emutls.c.
14982 2017-06-27 Jakub Jelinek <jakub@redhat.com>
14985 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
14987 PR middle-end/81207
14988 * gimple-fold.c (replace_call_with_call_and_fold): Handle
14989 gimple_vuse copying separately from gimple_vdef copying.
14991 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14993 * value-prof.c (free_hist): Remove call to memset and the enclosing if
14996 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
14997 Olivier Hainque <hainque@adacore.com>
14999 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
15000 for all vxworks7 targets.
15001 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
15002 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
15003 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
15004 variations for VX6/VX7 and 32/64bits later on in ...
15005 (VXWORKS_LIB_SPEC): Leverage new macros.
15006 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
15007 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
15009 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
15011 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
15012 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
15014 2017-06-26 Carl Love <cel@us.ibm.com>
15016 * config/rs6000/rs6000-c.c: Add support for built-in functions
15017 vector bool char vec_reve (vector bool char);
15018 vector signed char vec_reve (vector signed char);
15019 vector unsigned char vec_reve (vector unsigned char);
15020 vector bool int vec_reve (vector bool int);
15021 vector signed int vec_reve (vector signed int);
15022 vector unsigned int vec_reve (vector unsigned int);
15023 vector bool long long vec_reve (vector bool long long);
15024 vector signed long long vec_reve (vector signed long long);
15025 vector unsigned long long vec_reve (vector unsigned long long);
15026 vector bool short vec_reve (vector bool short);
15027 vector signed short vec_reve (vector signed short);
15028 vector double vec_reve (vector double);
15029 vector float vec_reve (vector float);
15030 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
15031 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
15032 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
15033 (altivec_vreve): New pattern.
15034 * config/rs6000/altivec.h (vec_reve): New define.
15035 * doc/extend.texi (vec_rev): Update the built-in documentation file
15036 for the new built-in functions.
15038 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15040 PR tree-optimization/71815
15041 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
15043 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
15045 (slsr_process_phi): Likewise.
15046 (replace_uncond_cands_and_profitable_phis): Don't replace a
15047 multiply candidate with a stride of 1 (copy or cast).
15048 (phi_incr_cost): Call uses_consumed_by_stmt rather than
15050 (lowest_cost_path): Likewise.
15051 (total_savings): Likewise.
15053 2017-06-26 Richard Biener <rguenther@suse.de>
15056 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
15057 Use def_builtin_pure for all gather builtins.
15059 2017-06-26 Richard Biener <rguenther@suse.de>
15061 PR tree-optimization/81203
15062 * tree-tailcall.c (find_tail_calls): Do not move stmts into
15063 non-dominating BBs.
15065 2017-06-26 Marek Polacek <polacek@redhat.com>
15068 * doc/invoke.texi: Document -Wmultistatement-macros.
15070 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
15072 * doc/sourcebuild.texi (ARM-specific attributes): Document new
15073 arm_neon_ok_no_float_abi effective target.
15075 2017-06-26 Richard Biener <rguenther@suse.de>
15077 PR tree-optimization/80928
15078 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
15079 (copy_bbs): Set BB_DUPLICATED flag early.
15080 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
15082 (execute_on_shrinking_pred): Likewise.
15083 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
15084 BB_DUPLICATED blocks.
15085 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
15086 iterate over all PHIs considering removal of *gsi.
15088 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
15090 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
15093 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15095 * config/rs6000/rs6000-string.c: (expand_block_clear,
15096 do_load_for_compare, select_block_compare_mode,
15097 compute_current_alignment, expand_block_compare,
15098 expand_strncmp_align_check, expand_strn_compare,
15099 expand_block_move, rs6000_output_load_multiple)
15100 Move functions related to string/block move/compare
15101 to a separate file.
15102 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
15103 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
15104 for this function which is now used in two files.
15105 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
15106 * config.gcc: Add rs6000-string.o to extra_objs for
15107 targets powerpc*-*-* and rs6000*-*-*.
15109 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
15112 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
15113 32-bit, since indexed is not valid for DImode.
15114 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
15115 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
15116 (define_peephole2 for Altivec d-form load): Add 32-bit support.
15117 (define_peephole2 for Altivec d-form store): Likewise.
15120 * multiple_target.c (create_dispatcher_calls): Only create the
15121 dispatcher call if the function is the default clone of a
15122 versioned function.
15124 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
15126 PR middle-end/80902
15127 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
15128 a call, force the call to not be a tail call.
15130 2017-06-23 Jeff Law <law@redhat.com>
15132 * doc/contrib.texi: Add entry for Steven Pemberton's work on
15135 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
15137 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
15138 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
15139 handling for early expansion of vector shifts (sl,sr,sra,rl).
15140 (builtin_function_type): Add vector shift right instructions
15141 to the unsigned argument list.
15143 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
15145 rtl-optimizatoin/79286
15146 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
15147 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
15148 trap. PIC register plus a const unspec without offset can never trap.
15150 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
15152 * tree.h (builtin_structptr_type): New type.
15153 (builtin_structptr_types): Declare new array.
15154 * tree.c (builtin_structptr_types): New array.
15155 (free_lang_data, build_common_tree_nodes): Use it.
15157 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
15160 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
15163 2017-06-22 Matt Turner <mattst88@gmail.com>
15165 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
15166 Lake models to skylake case. Assume skylake for unknown
15167 models with clflushopt.
15169 2017-06-22 Jeff Law <law@redhat.com>
15171 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
15172 frame sizes that do not satisfy aarch64_uimm12_shift.
15174 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
15176 * profile-count.h (apply_probability,
15177 apply_scale, probability_in): Fix checks for zero.
15179 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15181 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
15182 * doc/cppdiropts.texi (-I @var{dir}): Document it.
15184 2016-06-22 Richard Biener <rguenther@suse.de>
15186 * tree-vect-loop.c (vect_model_reduction_cost): Handle
15187 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
15188 REDUC_MAX_EXPR support.
15189 (vectorizable_reduction): Likewise.
15190 (vect_create_epilog_for_reduction): Likewise.
15192 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
15194 * match.pd (A / (1 << B) -> A >> B): New.
15195 * generic-match-head.c: Include optabs-tree.h.
15196 * gimple-match-head.c: Likewise.
15197 * optabs-tree.h (target_supports_op_p): New.
15198 * optabs-tree.c (target_supports_op_p): New.
15200 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
15202 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
15203 $gcc_cv_ld --help output.
15204 (gcc_cv_ld_demangle): Likewise.
15205 (gcc_cv_ld_eh_frame_hdr): Likewise.
15206 (gcc_cv_ld_pie): Likewise.
15207 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
15208 (gcc_cv_ld_buildid): Likewise.
15209 (gcc_cv_ld_sysroot): Likewise.
15210 (ld_bndplt_support): Likewise.
15211 (ld_pushpopstate_support): Likewise.
15212 * configure: Regenerate.
15213 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
15215 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15218 * config/i386/sse.md (round<mode>2): Renumber match_dup and
15219 operands indexes to avoid gap between operands and match_dups.
15221 2017-06-21 Andrew Pinski <apinski@cavium.com>
15223 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
15224 Increment Arith_shift and Arith_shift_reg by 1.
15225 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
15227 * config/aarch64/aarch64.c (thunderx_tunings): Enable
15228 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
15229 (aarch64_strip_extend): Add new argument and test for it.
15230 (aarch64_cheap_mult_shift_p): New function.
15231 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
15232 add a cost if it is true.
15233 Update calls to aarch64_strip_extend.
15234 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
15236 2017-06-21 Andrew Pinski <apinski@cavium.com>
15238 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
15240 (thunderxt88): Likewise.
15241 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
15242 (thunderx_prefetch_tune): New variable.
15243 (thunderx2t99_prefetch_tune): Update for the correct values.
15244 (thunderxt88_tunings): New variable.
15245 (thunderx_tunings): Use thunderx_prefetch_tune instead of
15246 generic_prefetch_tune.
15247 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
15249 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15251 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
15252 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
15253 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
15254 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
15255 (aarch64_atomic_cas<mode>, GPI): Likewise.
15257 2017-06-21 Martin Liska <mliska@suse.cz>
15259 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
15260 statements on cold and hot labels.
15261 * predict.c (tree_estimate_probability_bb): Remove the
15262 prediction from this place.
15264 2017-06-21 Martin Liska <mliska@suse.cz>
15266 PR tree-optimization/79489
15267 * gimplify.c (maybe_add_early_return_predict_stmt): New
15269 (gimplify_return_expr): Call the function.
15270 * predict.c (tree_estimate_probability_bb): Remove handling
15272 * predict.def: Update comment about early return predictor.
15273 * gimple-predict.h (is_gimple_predict): New function.
15274 * predict.def: Change default value of early return to 66.
15275 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
15277 * passes.def: Put pass_strip_predict_hints to the beginning of
15280 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
15282 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
15283 FUNCTION_DECL declarations.
15284 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
15286 (dwaf2out_decl): Likewise.
15287 * godump.c (go_early_global_decl): Skip call to the real debug hook
15288 for FUNCTION_DECL declarations.
15289 * passes.c (rest_of_decl_compilation): Skip call to the
15290 early_global_decl debug hook for FUNCTION_DECL declarations, unless
15291 -fdump-go-spec is passed.
15293 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15295 * config/i386/i386.c (struct builtin_isa): New field pure_p.
15296 Reorder for compactness.
15297 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
15298 (def_builtin_pure, def_builtin_pure2): New functions.
15299 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
15301 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
15303 * match.pd (nop_convert): New predicate.
15304 ((A +- CST1) +- CST2): Allow some NOP conversions.
15306 2017-06-21 Jakub Jelinek <jakub@redhat.com>
15309 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
15310 with ctors/dtors if GOVD_SHARED is set.
15312 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15314 * config/aarch64/aarch64.md (movti_aarch64):
15315 Emit mov rather than orr.
15316 (movtf_aarch64): Likewise.
15317 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
15318 Emit mov rather than orr.
15320 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15322 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
15323 Swap alternatives, make integer dup more expensive.
15325 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
15327 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
15328 Return true for non-tls symbols.
15330 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
15332 * config/aarch64/aarch64-cores.def (cortex-a55): New.
15333 (cortex-a75): Likewise.
15334 (cortex-a75.cortex-a55): Likewise.
15335 * config/aarch64/aarch64-tune.md: Regenerate.
15336 * doc/invoke.texi (-mtune): Document new values for -mtune.
15338 2017-06-21 Tom de Vries <tom@codesourcery.com>
15340 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
15341 stack_size feature.
15342 (Effective-Target Keywords, Other attributes): Suggest using
15343 dg-add-options stack_size feature to get stack limit in stack_size
15344 effective target documentation.
15346 2017-06-21 Julian Brown <julian@codesourcery.com>
15347 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15349 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
15350 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
15351 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
15353 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
15354 attribute type list for neon_multiply.
15355 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
15356 attribute type list for neon_multiply.
15357 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
15358 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
15359 attribute type list for neon_multiply.
15360 * config/arm/types.md (crypto_pmull): Add.
15361 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
15362 attribute type list.
15364 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
15366 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
15369 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15371 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
15372 to make sure not to dereference a NULL cost_classes_ptr pointer.
15374 2017-06-20 Carl Love <cel@us.ibm.com>
15376 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15377 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15378 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15379 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15380 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15381 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15382 VMULOSW): New enum "unspec" values.
15383 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15384 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15385 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15386 altivec_vmulosw): New patterns.
15387 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15388 VMULOSW): Add definitions.
15390 2017-06-20 Julia Koval <julia.koval@intel.com>
15392 * config/i386/i386.c: Fix rounding expand for new pattern.
15393 * config/i386/subst.md: Fix pattern (parallel -> unspec).
15395 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15397 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
15398 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
15400 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15402 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
15405 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
15407 * config/aarch64/aarch64-cores.def: Rearrange to sort by
15408 architecture, then by implementer ID.
15409 * config/aarch64/aarch64-tune.md: Regenerate.
15411 2017-06-20 Richard Biener <rguenther@suse.de>
15413 PR middle-end/81097
15414 * fold-const.c (split_tree): Fold to type before negating.
15416 2017-06-20 David Malcolm <dmalcolm@redhat.com>
15418 * diagnostic-show-locus.c
15419 (selftest::test_fixit_deletion_affecting_newline): New function.
15420 (selftest::diagnostic_show_locus_c_tests): Call it.
15422 2017-06-20 Andreas Schwab <schwab@suse.de>
15425 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
15428 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
15430 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
15431 __ARM_FEATURE_COPROC according to support.
15433 2017-06-20 Jakub Jelinek <jakub@redhat.com>
15435 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
15436 Rewritten to avoid overflow for > 32-bit pointers.
15439 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
15440 by removing enum keyword.
15441 (ubsan_type_descriptor): Likewise. Formatting fix.
15444 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
15445 splitter): Require TARGET_SSE2 in the condition.
15447 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
15450 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
15451 for doing vector set of SFmode on ISA 3.0.
15452 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
15453 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
15455 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
15456 SFmode value into a V4SF variable that was extracted from another
15457 V4SF variable without converting the element to double precision
15458 and back to single precision vector format.
15459 (vsx_insert_extract_v4sf_p9_2): Likewise.
15461 2017-06-19 Jakub Jelinek <jakub@redhat.com>
15463 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
15464 in UWHI to avoid undefined overflow.
15467 * ubsan.h (enum ubsan_encode_value_phase): New.
15468 (ubsan_encode_value): Change second argument to
15469 enum ubsan_encode_value_phase with default value of
15470 UBSAN_ENCODE_VALUE_GENERIC.
15471 * ubsan.c (ubsan_encode_value): Change second argument to
15472 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
15473 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
15474 create_tmp_var_raw instead of create_tmp_var and use a
15476 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
15477 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
15478 ubsan_encode_value callers.
15481 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
15482 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
15483 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
15485 2017-06-19 Richard Biener <rguenther@suse.de>
15487 PR middle-end/81118
15488 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
15489 estimates if we changed anything.
15491 2017-06-19 Richard Biener <rguenther@suse.de>
15493 PR tree-optimization/80887
15494 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
15495 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
15496 simplified lookups, then reset mprts_hook.
15497 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
15499 (try_to_simplify): Likewise.
15501 2017-06-19 Martin Liska <mliska@suse.cz>
15504 * gimplify.c (gimplify_switch_expr):
15505 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
15507 2017-06-19 Martin Liska <mliska@suse.cz>
15509 * doc/install.texi: Document that PGO runs in 4 stages.
15511 2017-06-19 Martin Liska <mliska@suse.cz>
15514 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
15515 to dispatcher function name.
15516 * multiple_target.c (replace_function_decl): New function.
15517 (create_dispatcher_calls): Redirect both edges and references.
15519 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
15521 * profile-count.c (profile_count::dump): Dump quality.
15522 (profile_count::differs_from_p): Update for unsigned val.
15523 * profile-count.h (profile_count_quality): New enum.
15524 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
15526 2017-06-19 Richard Biener <rguenther@suse.de>
15528 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
15529 struct function as arg.
15530 (estimate_numbers_of_iterations): Export overload with loop arg.
15531 (free_numbers_of_iterations_estimates_loop): Use an overload of
15532 free_numbers_of_iterations_estimates instead.
15533 * tree-cfg.c (remove_bb): Adjust.
15534 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
15535 * tree-parloops.c (gen_parallel_loop): Likewise.
15536 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
15538 (tree_unroll_loops_completely): Likewise.
15539 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
15540 Use an overload instead and export.
15541 (estimated_loop_iterations): Adjust.
15542 (max_loop_iterations): Likewise.
15543 (likely_max_loop_iterations): Likewise.
15544 (estimate_numbers_of_iterations): Take struct function as arg
15546 (loop_exits_before_overflow): Adjust.
15547 (free_numbers_of_iterations_estimates_loop): Use an overload.
15548 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
15549 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
15551 2017-06-19 Richard Biener <rguenther@suse.de>
15554 * ipa-prop.c (find_constructor_constant_at_offset): Handle
15555 RANGE_EXPR conservatively.
15557 2017-06-16 Carl Love <cel@us.ibm.com>
15559 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15560 definitions for vec_float, vec_float2, vec_floato,
15561 vec_floate built-ins.
15562 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
15563 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
15565 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
15566 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
15567 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
15568 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
15569 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
15570 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
15571 vec_floato): Add builtin defines.
15572 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
15573 Update the built-in documentation file for the new built-in
15576 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15578 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
15579 (mthumb): Mark as the negative of -marm.
15581 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15583 * doc/invoke.texi (ARM Options, -mcpu): Document supported
15585 (ARM Options, -mtune): Document that this accepts the same
15586 extension options as -mcpu.
15587 (ARM Options, -mfpu): Document addition of -mfpu=auto.
15589 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15591 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
15592 permitted extensions.
15594 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15596 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
15597 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
15598 (armv8-m.main): Add option +nodsp.
15599 * config/arm/arm-cpu-cdata.h: Regenerated.
15601 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15603 * config/arm/t-fuchsia: New file.
15604 * config.gcc (arm*-*-fuchsia*): Use it.
15606 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15608 * config/arm/t-symbian: Rewrite for new option infrastructure.
15610 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15612 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
15613 (MULTILIB_REQUIRED): Likewise.
15615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15617 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
15618 (MULTILIB_RESUE): Likewise.
15619 (MULTILIB_MATCHES): Likewise.
15620 (MULTLIB_REQUIRED): Likewise.
15622 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15624 * config/arm/t-rtems: Rewrite for new option framework.
15626 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15628 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
15629 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
15630 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
15631 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
15632 * config/arm/t-multilib: ... here.
15633 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
15634 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
15635 armv7-a and armv8*-a when A-profile libraries have not been built.
15636 * config/arm/t-rmprofile: Rewrite.
15638 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15640 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
15641 with a backslash. Remove the backslash after substituting unescaped
15643 * doc/fragments.texi (MULTILIB_REUSE): Document it.
15645 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15647 * config.gcc: (arm*-*-*): When building a-profile libraries, force
15648 the driver to pass through the default setting of -mfloat-abi.
15649 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
15651 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
15652 (all_feat_combs): New rule.
15653 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
15655 * config/arm/t-aprofile: Rewrite.
15657 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15659 * config/arm/arm.h (FPUTYPE_AUTO): Define.
15660 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
15661 fpu is not specified by the user/command-line.
15662 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
15663 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
15664 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
15665 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
15666 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
15667 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
15669 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15671 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
15672 * config/arm/t-arm-elf: Rewritten.
15674 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15676 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
15677 have some floating-point instructions.
15678 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
15679 (TARGET_MAYBE_HARD_FLOAT): New macro.
15680 * config/arm/arm-builtins.c (arm_init_builtins): Use
15681 TARGET_MAYBE_HARD_FLOAT.
15682 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
15684 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15686 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
15687 (configargs.h): Include it.
15688 (arm_print_hint_for_fpu_option): New function.
15689 (arm_parse_fpu_option): New function.
15690 (candidate_extension): New class.
15691 (arm_canon_for_multilib): New function.
15692 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
15693 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
15694 (ARCH_CANONICAL_SPECS): New macro.
15695 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
15697 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15699 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
15700 are set after handling multilib fragments. Set target_cpu_default2
15703 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15705 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
15707 (arm*-*-*): Set target_cpu_default2 to a quoted string.
15708 * config/arm/parsecpu.awk (check_cpu): Validate any extension
15710 (check_arch): Likewise.
15711 * config/arm/arm.c (arm_configure_build_target): Handle
15712 TARGET_CPU_DEFAULT being a string constant. Scan any feature
15713 options in the default.
15715 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15717 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
15718 when an option is an alias of another.
15719 * config/arm/parsecpu.awk (optalias): New parser token.
15720 (gen_comm_data): Mark non-alias options as such. Emit entries
15721 for extension aliases.
15722 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
15723 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
15724 (armv6kz, armv6zk, armv6t2): Likewise.
15725 (armv7): Make vfpv3-d16 an alias.
15726 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
15728 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
15729 Sort in canonical order.
15730 (armv8-a): Sort in canonical order.
15731 (armv8.1-a, armv8.2-a): Likewise.
15732 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
15734 (cortex-a9): Sort in canonical order.
15735 * config/arm/arm.c (selftests.h): Include it.
15736 (arm_test_cpu_arch_data): New function.
15737 (arm_run_self_tests): New function.
15738 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
15739 (targetm): Move declaration to the end of the file.
15740 * arm-cpu-cdata.h: Regenerated.
15742 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15744 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
15745 call to target_mode_check describing the type of option passed.
15746 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
15747 (arm_target_thumb_only): Use arm_parse_arch_option_name or
15748 arm_parse_cpu_option_name to match parameters against list of
15750 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
15751 arm_arch_core_flags data structure.
15752 * config/arm/arm-cpu_cdata.h: Regenerated.
15754 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15756 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
15758 (arm_print_hint_for_cpu_option): Likewise.
15759 (arm_print_hint_for_arch_option): Likewise.
15760 (arm_parse_cpu_option_name): Likewise.
15761 (arm_parse_arch_option_name): Likewise.
15762 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
15763 of entries in the all_fpus list.
15764 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
15765 (arm_parse_cpu_option_name): Declare.
15766 (arm_parse_arch_option_name): Declare.
15767 (arm_parse_option_features): Declare.
15768 (arm_intialize_isa): Declare.
15769 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
15771 (gen_comm_data): ... here. Make definitions non-static.
15772 * config/arm/arm-cpu-data.h: Regenerated.
15773 * config/arm/arm-cpu-cdata.h: Regenerated.
15775 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15777 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
15778 (cpu_arch_extension): New structure.
15779 (cpu_arch_option, arch_option, cpu_option): New structures.
15780 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
15781 architecture types.
15782 (gen_data): Generate new format data tables.
15783 * config/arm/arm.c (cpu_tune): New structure.
15784 (cpu_option, processors): Delete.
15785 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
15786 (arm_print_hint_for_cpu_option): ... this and ...
15787 (arm_print_hint_for_arch_option): ... this.
15788 (arm_parse_arch_cpu_name): Delete. Replace with ...
15789 (arm_parse_cpu_option_name): ... this and ...
15790 (arm_parse_arch_option_name): ... this.
15791 (arm_unrecognized_feature): Change type of target parameter to
15793 (arm_parse_arch_cpu_features): Delete. Replace with ...
15794 (arm_parse_option_features): ... this.
15795 (arm_configure_build_target): Rework to use new configuration data
15797 (arm_print_tune_info): Rework for new configuration data tables.
15798 * config/arm/arm-cpu-data.h: Regenerated.
15799 * config/arm/arm-cpu.h: Regenerated.
15801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15803 * Makefile.in (OBJS): Move sbitmap.o from here ...
15804 (OBJS-libcommon): ... to here.
15806 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15808 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
15809 (ISA_ALL_CRYPTO): New macro.
15810 (ISA_ALL_SIMD): New macro
15811 (ISA_ALL_FP): New macro.
15812 * config/arm/arm.c (fpu_bitlist): Update initializer.
15813 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
15815 (arm9e): Add fpu. Add option for nofp
15816 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
15817 (arm926ej-s, arm1026ej-s): Likewise.
15818 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
15819 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
15820 neon-fp16, neon-vfpv4, nofp and nosimd.
15821 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
15822 (cortex-a8): Add fpu. Add option for nofp.
15823 (cortex-a9): Add fpu. Add options for nosimd and nofp.
15824 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
15825 (cortex-r4f): Add fpu.
15826 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
15827 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
15829 (cortex-r8): Likewise.
15830 (cortex-m4): Add fpu. Add option for nofp.
15831 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
15832 (cortex-a17.cortex-a7): Likewise.
15833 (cortex-a32): Add fpu. Add options for crypto and nofp.
15834 (cortex-a35, cortex-a53): Likewise.
15835 (cortex-a57): Add fpu. Add option for crypto.
15836 (cortex-a72, cortex-a73): Likewise.
15837 (exynos-m1): Likewise.
15838 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
15839 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
15840 (cortex-m33): Add fpu. Add option for nofp.
15841 * config/arm/arm-cpu-cdata.h: Regenerated
15842 * config/arm/arm-cpu-data.h: Regenerated.
15844 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15846 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
15847 (armv5te, armv5tej): Likewise.
15848 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
15849 (armv7): Add options fp and vfpv3-d16.
15850 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
15851 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
15853 (armv7ve): Likewise.
15854 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
15855 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
15856 (armv8-a): Add nocrypto option.
15857 (armv8.1-a, armv8.2-a): Likewise.
15858 (armv8-m.main): add options fp, fp.dp and nofp.
15860 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15862 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
15864 (armv8-a+crc): Delete.
15865 (armv8.1-a): Add options simd, crypto and nofp.
15866 (armv8.2-a): Add options fp16, simd, crypto and nofp.
15867 (armv8.2-a+fp16): Delete.
15868 (armv8-m.main): Add option dsp.
15869 (armv8-m.main+dsp): Delete.
15870 (cortex-a8): Add fpu. Add nofp option.
15871 (cortex-a9): Add fpu. Add nofp and nosimd options.
15872 * config/arm/parsecpu.awk (gen_data): Generate option tables and
15873 link to main cpu and architecture data structures.
15874 (gen_comm_data): Only put isa attributes from the main architecture
15876 (option): New statement for architecture and CPU entries.
15877 * arm.c (struct cpu_option): New structure.
15878 (struct processors): Add entry for options.
15879 (arm_unrecognized_feature): New function.
15880 (arm_parse_arch_cpu_name): Ignore any characters after the first
15882 (arm_parse_arch_cpu_feature): New function.
15883 (arm_configure_build_target): Separate out any CPU and architecture
15884 features and parse separately. Don't error out if -mfpu=auto is
15885 used with only an architecture string.
15886 (arm_print_asm_arch_directives): New function.
15887 (arm_file_start): Call it.
15888 * config/arm/arm-cpu-cdata.h: Regenerated.
15889 * config/arm/arm-cpu-data.h: Likewise.
15890 * config/arm/arm-tables.opt: Likewise.
15892 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15894 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
15895 assembler when it is not -mfpu=auto.
15897 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15899 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
15900 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
15901 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
15902 (ASM_CPU_SPEC): Rewrite.
15903 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
15904 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
15905 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
15906 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
15907 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
15908 copied string is NUL-terminated. Also strip any characters prefixed
15910 (arm_rewrite_selected_arch): New function.
15911 (arm_rewrite_march): New function.
15913 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
15915 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
15916 (x_arm_cpu_string, x_arm_tune_string): Likewise.
15917 (march, mcpu, mtune): Convert to string-based options.
15918 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
15919 (arm_parse_arch_cpu_name): New function.
15920 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
15921 identify selected architecture or CPU.
15922 (arm_option_save): New function.
15923 (TARGET_OPTION_SAVE): Redefine.
15924 (arm_option_restore): Restore string options.
15925 (arm_option_print): Print string options.
15927 2017-06-16 Martin Sebor <msebor@redhat.com>
15929 PR tree-optimization/80933
15930 PR tree-optimization/80934
15931 * builtins.c (fold_builtin_3): Do not handle bcmp here.
15932 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
15933 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
15934 (gimple_fold_builtin): Call them.
15936 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15938 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
15939 as unlikely; update profile.
15941 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15943 * predict.c (force_edge_cold): Handle declaring edges impossible
15946 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15948 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
15950 (try_unroll_loop_completely): Fix reporting.
15952 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15954 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
15956 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
15959 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
15960 if given a non-constant argument for an intrinsic which requires a
15963 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15965 * profile.c (compare_freqs): New function.
15966 (branch_prob): Sort edge list.
15967 (find_spanning_tree): Assume that the list is priority sorted.
15969 2017-06-16 Richard Biener <rguenther@suse.de>
15971 PR tree-optimization/81090
15972 * passes.def (pass_record_bounds): Remove.
15973 * tree-pass.h (make_pass_record_bounds): Likewise.
15974 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
15975 make_pass_record_bounds): Likewise.
15976 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
15977 not free niter estimates at the beginning but at the end.
15978 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
15980 2017-06-16 Richard Biener <rguenther@suse.de>
15982 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
15983 initializer to workaround ICE in host GCC 4.8.
15985 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
15987 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
15989 (clone_inlined_nodes): Update.
15991 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15993 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
15994 prefetch settings, and enable prefetching by default at -O3.
15996 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
15998 * config/aarch64/aarch64.c (aarch64_override_options_internal):
15999 Set flag_prefetch_loop_arrays according to tuning data.
16001 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
16003 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
16004 New tune structure.
16005 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
16006 [Unrelated to main purpose of the patch] Place the pointer field last
16007 to enable type checking errors when tune structure are wrongly merged.
16008 * config/aarch64/aarch64.c (generic_prefetch_tune,)
16009 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
16010 (thunderx2t99_prefetch_tune): New tune constants.
16011 (tune_params *_tunings): Update all tunings (no functional change).
16012 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
16013 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
16014 from tunings structures.
16016 2017-06-16 Jakub Jelinek <jakub@redhat.com>
16019 * ubsan.c (instrument_null): Add T argument, use it instead
16020 of computing it based on IS_LHS.
16021 (instrument_object_size): Likewise.
16022 (pass_ubsan::execute): Adjust instrument_null and
16023 instrument_object_size callers to pass gimple_get_lhs or
16024 gimple_assign_rhs1 result to it. Use instrument_null instead of
16025 calling get_base_address and instrument_mem_ref. Handle
16026 aggregate call arguments for object-size sanitization.
16028 2017-06-16 Yury Gribov <tetra2005@gmail.com>
16030 PR tree-optimization/81089
16031 * tree-vrp.c (is_masked_range_test): Validate operands of
16034 2017-06-15 Martin Sebor <msebor@redhat.com>
16037 * dumpfile.c (dump_register): Avoid calling memset to initialize
16038 a class with a default ctor.
16039 * gcc.c (struct compiler): Remove const qualification.
16040 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
16041 * hash-table.h: Ditto.
16042 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
16044 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
16045 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
16047 * params.h (struct param_info): Make struct members non-const.
16048 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
16049 with default initialization.
16050 * vec.h (vec_copy_construct, vec_default_construct): New helper
16052 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
16053 with vec_copy_construct.
16054 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
16055 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
16056 * doc/invoke.texi (-Wclass-memaccess): Document.
16058 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16060 * emit-rtl.h (is_leaf): Update comment about local
16061 register allocator.
16063 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
16066 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
16067 for a variable to have a section before checking if the section has a
16069 Set section to.persistent if persistent attribute is set.
16070 Warn if .persistent attribute is used on an automatic variable.
16072 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
16074 PR rtl-optimization/80474
16075 * reorg.c (update_block): Do not ignore instructions in a delay slot.
16077 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
16079 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
16082 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
16084 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
16085 (casesi): Emit bounds checking as RTL.
16086 (casesi_internal_mips16_<mode>): Remove bounds checking.
16088 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
16090 * config/xtensa/xtensa.c (xtensa_option_override): Append
16091 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
16092 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
16093 xtensa_doloop_hooks): Define unconditionally.
16094 (xtensa_reorg_loops): Only call reorg_loops in the presence of
16096 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
16097 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
16098 for it in xtensa_option_override.
16099 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
16100 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
16102 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
16104 * doc/cppopts.texi: Document '-' special value to -MF.
16106 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
16108 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
16109 (cortex_a53_fconst): Likewise.
16110 (cortex_a53_fpmul): Likewise.
16111 (cortex_a53_f_load_64): Likewise.
16112 (cortex_a53_f_load_many): Likewise.
16113 (cortex_a53_advsimd_alu): Likewise.
16114 (cortex_a53_advsimd_alu_q): Likewise.
16115 (cortex_a53_advsimd_mul): Likewise.
16116 (cortex_a53_advsimd_mul_q): Likewise.
16117 (fpmac bypass): Add new bypass for fpmac-fpmac case.
16118 Add missing fmul, r2f_cvt and fconst cases.
16120 2017-06-14 Richard Biener <rguenther@suse.de>
16122 PR middle-end/81088
16123 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
16125 (fold_binary_loc): When associating do not treat pre-existing
16126 TREE_OVERFLOW on literal constants as a reason to allow
16127 TREE_OVERFLOW on associated literal constants.
16129 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16131 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
16132 (MASK_FEATURES): New macro.
16133 * config/sparc/sparc.c (sparc_option_override): Remove the special
16134 handling of -mfpu and generalize it to all MASK_FEATURES switches.
16136 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16138 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
16139 a division of 0 if non-call exceptions are enabled.
16141 2017-06-14 Andrew Pinski <apinski@cavium.com>
16142 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
16145 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
16146 Improve vector initialization code gen for only variable case.
16148 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
16150 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
16152 2017-06-14 Richard Biener <rguenther@suse.de>
16154 PR tree-optimization/81083
16155 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
16158 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16160 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
16161 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
16162 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
16163 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
16164 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
16165 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
16167 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16169 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
16170 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
16172 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16174 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
16176 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16178 * config/rs6000/t-rtems: Don't handle SPE.
16180 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16182 * config/rs6000/t-linux: Don't handle SPE.
16184 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16186 * config/rs6000/eabispe.h: Delete file.
16188 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16190 * config/rs6000/t-spe: Delete file.
16192 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16194 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
16195 (rs6000_legitimate_offset_address_p): Return false for anything in
16196 V2SImode or V2SFmode.
16198 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
16200 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
16201 except V2SF and V2SI. Rearrange the vector modes, and add comments.
16202 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
16204 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
16205 (rs6000_legitimate_offset_address_p): Ditto.
16206 (rs6000_emit_move): Ditto.
16207 (rs6000_init_builtins): Remove V4HI_type_node.
16209 2017-06-13 Martin Liska <mliska@suse.cz>
16212 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
16213 (gate_asan): Likewise.
16214 * asan.h (asan_no_sanitize_address_p): Remove the function.
16215 (sanitize_flags_p): New function.
16216 * builtins.def: Fix coding style.
16217 * common.opt: Use renamed enum value.
16218 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
16219 * doc/extend.texi: Document no_sanitize attribute.
16220 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
16221 to SANITIZE_UNDEFINED_NONDEFAULT.
16222 * gcc.c (sanitize_spec_function): Use the renamed enum value.
16223 * gimple-fold.c (optimize_atomic_compare_exchange_p):
16224 Use sanitize_flags_p.
16225 * gimplify.c (gimplify_function_tree): Likewise.
16226 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
16227 * opts.c (parse_no_sanitize_attribute): New function.
16228 (common_handle_option): Use renamed enum value.
16229 * opts.h (parse_no_sanitize_attribute): Declare.
16230 * tree.c (sanitize_flags_p): New function.
16231 * tree.h: Declared here.
16232 * tsan.c: Use sanitize_flags_p.
16233 * ubsan.c (ubsan_expand_null_ifn): Likewise.
16234 (instrument_mem_ref): Likewise.
16235 (instrument_bool_enum_load): Likewise.
16236 (do_ubsan_in_current_function): Remove the function.
16237 (pass_ubsan::execute): Use sanitize_flags_p.
16238 * ubsan.h: Remove do_ubsan_in_current_function
16239 * tree-cfg.c (print_no_sanitize_attr_value): New function.
16240 (dump_function_to_file): Use it here.
16242 2017-06-13 Martin Jambor <mjambor@suse.cz>
16244 PR tree-optimization/80803
16245 PR tree-optimization/81063
16246 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
16247 (propagate_subaccesses_across_link): Enqueue subtree whenever
16248 necessary instead of relying on the caller.
16250 2017-06-13 Martin Jambor <mjambor@suse.cz>
16252 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
16253 that have a first_link.
16254 (sort_and_splice_var_accesses): Do not check first_link before
16256 (subtree_mark_written_and_enqueue): Likewise.
16257 (propagate_all_subaccesses): Likewise and do not stop at first
16258 parent with a first_link.
16260 2017-06-13 Martin Jambor <mjambor@suse.cz>
16262 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
16265 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16267 * match.pd: New pattern.
16269 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16271 * tree-vrp.c (is_masked_range_test): New function.
16272 (register_edge_assert_for): Determine ranges for
16275 2017-06-13 Yury Gribov <tetra2005@gmail.com>
16277 PR tree-optimization/67328
16278 * fold-const.c (maskable_range_p): New function.
16279 (build_range_check): Generate bittests if possible.
16281 2017-06-13 Martin Liska <mliska@suse.cz>
16283 * gimple-pretty-print.c (dump_probability): Add new argument.
16284 (dump_edge_probability): Dump both probability and count.
16285 (dump_gimple_label): Likewise.
16286 (dump_gimple_bb_header): Likewise.
16288 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
16291 * config/avr/avr-devices.c: Fix indentation.
16292 * config/avr/gen-avr-mmcu-specs.c: Dito.
16294 2017-06-13 Richard Biener <rguenther@suse.de>
16296 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
16297 instead get vector type from stmt_info.
16298 (vectorizable_reduction): Adjust. Remove dead code.
16300 2017-06-13 Richard Biener <rguenther@suse.de>
16302 PR middle-end/81065
16303 * fold-const.c (extract_muldiv_1): Remove bogus distribution
16304 case of C * (x * C2 + C3).
16305 (fold_addr_of_array_ref_difference): Properly fold index difference.
16307 2017-06-12 David S. Miller <davem@davemloft.net>
16310 * config/sparc/sparc.md (return expander): Emit frame blockage if
16311 function uses alloca.
16313 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
16315 * combine.c (make_field_assignment): Check len rather than the mode
16316 precision when calling force_to_mode.
16318 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
16320 Support multilibs and devices that see flash in RAM address range.
16323 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
16324 (avr_mcu_t) <flash_pm_offset>: New field.
16325 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
16326 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
16327 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
16328 (AVR_TINY_PM_OFFSET): Remove macro.
16329 * config/avr/avr.opt (-mshort-calls): New option.
16330 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16331 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
16332 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
16333 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
16334 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
16335 instead of avr_arch->have_jmp_call.
16336 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
16337 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
16338 avr_arch->flash_pm_offset to define.
16339 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
16340 new field flash_pm_offset. Add entry for avrxmega3.
16341 (avr_texinfo): Add entry for avrxmega3.
16342 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
16343 attiny212, attiny214,
16344 attiny412, attiny414, attiny416, attiny417,
16345 attiny814, attiny816, attiny817,
16346 attiny1614, attiny1616, attiny1617,
16347 attiny3214, attiny3216, attiny3217.
16348 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
16349 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
16350 (avr_print_operand_address) [AVR_TINY]: Same.
16351 (avr_asm_init_sections) <readonly_data_section>: Only patch
16352 callback if avr_arch->flash_pm_offset = 0.
16353 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
16354 for rodata if avr_arch->flash_pm_offset != 0.
16355 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
16356 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
16357 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
16358 (m_options): Append opt_rcall.
16359 (m_dirnames): Append dir_rcall.
16360 * config/avr/t-multilib: Regenerate.
16362 * configure.ac [target=avr]: Check whether avrxmega3 default
16363 linker description file works as needed.
16364 * configure: Regenerate.
16365 * doc/avr-mmcu.texi: Regenerate.
16366 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
16367 <__AVR_ARCH__>: Document avrxmega3 and 103.
16368 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
16369 <__AVR_SHORT_CALLS__>: Document it.
16370 <__AVR_PM_BASE_ADDRESS__>: Document it.
16371 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
16372 (AVR Variable Attributes) <progmem>: Document this is
16373 not needed for avrxmega3.
16374 (AVR Named Address Spaces) <__flash>: Dito.
16376 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
16378 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
16380 2017-06-12 Doug Rupp <rupp@adacore.com>
16382 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
16383 Append vxworks-stdint.h to the tm_file list.
16384 * config/vxworks-stdint.h: New file.
16386 2017-06-12 Martin Liska <mliska@suse.cz>
16388 PR tree-optimization/81041
16389 * tree-profile.c (gimple_gen_ic_func_profiler):
16390 Create an extra BB in profile-generate
16391 (gimple_gen_time_profiler): Likewise.
16393 2017-06-12 Jakub Jelinek <jakub@redhat.com>
16395 PR tree-optimization/81003
16396 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
16397 (update_range_test): Use it instead of force_gimple_operand_gsi.
16399 2017-06-12 Richard Biener <rguenther@suse.de>
16401 PR tree-optimization/81053
16402 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
16403 with backedge value not defined in loop. Simplify def stmt
16406 2017-06-11 Tom de Vries <tom@codesourcery.com>
16409 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
16411 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
16412 nvptx_cannot_force_const_mem.
16414 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16416 * opts.c (finish_options): Move test for flag_split_stack after
16417 it has been initialized.
16419 2017-06-11 Jason Merrill <jason@redhat.com>
16421 * tree.h (id_equal): New.
16422 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
16423 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
16424 instead of strcmp of IDENTIFIER_POINTER.
16426 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16428 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
16429 (mark_all_inlined_calls_cdtor): Fix formating.
16430 (inline_transform): Rescale profile before inlining.
16432 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16434 * cgraph.h (cgraph_edge::clone): Update prototype.
16435 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
16436 (cgraph_node::create_clone): Update.
16437 (cgraph_node::create_version_clone): Update.
16438 * tree-inline.c (copy_bb): Update.
16439 (expand_call_inline): Update.
16441 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16443 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
16444 factored out from ...
16445 (rs6000_emit_prologue): ... here.
16447 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
16449 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
16450 factored out from ...
16451 (rs6000_emit_prologue): ... here.
16453 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16455 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
16457 (handle_missing_profiles): Fix computation of tp_first_run.
16458 (counts_to_freqs): Do not touch freqs when count is 0.
16460 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
16462 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
16465 2017-06-10 Tom de Vries <tom@codesourcery.com>
16467 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
16468 attributes): Document signal effective target.
16470 2017-06-10 Tom de Vries <tom@codesourcery.com>
16472 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
16473 Document effective target stack_size.
16475 2017-06-09 David Malcolm <dmalcolm@redhat.com>
16477 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
16478 to the edit_context if they can be auto-applied.
16480 2017-06-9 Ian Lance Taylor <iant@golang.org>
16482 * opts.c (finish_options): If -fsplit-stack, disable implicit
16483 -forder-blocks-and-partition.
16484 * doc/invoke.texi (Optimize Options): Document that when using
16485 -fsplit-stack -forder-blocks-and-partition is not implicitly
16488 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16490 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
16491 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
16492 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
16493 * builtins.def (abort, trap, unreachable): Declare cold.
16494 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
16495 * tree-core.h (ECF_COLD): New.
16496 * tree.c (set_call_expr_flags): Handle ECF_COLD.
16497 (build_common_builtin_nodes): Mark unreachable and abort as cold.
16499 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16501 * predict.c (unlikely_executed_stmt_p): Cleanup.
16503 2017-06-09 Richard Biener <rguenther@suse.de>
16505 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
16506 model if the ref is always written to.
16508 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16510 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
16512 2017-06-09 Tamar Christina <tamar.christina@arm.com>
16514 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
16517 2017-06-09 Tom de Vries <tom@codesourcery.com>
16520 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
16521 "target cannot support label values" when encountering LABEL_REF.
16523 2017-06-09 Martin Liska <mliska@suse.cz>
16525 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
16526 (gimple_gen_ic_func_profiler): Emit direct comparison
16527 of __gcov_indirect_call_callee with NULL.
16528 (gimple_gen_time_profiler): Change probability from
16529 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
16531 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16533 * profile.c (edge_gcov_counts): Turn to pointer.
16534 (compute_branch_probabilities, compute_branch_probabilities): Update.
16535 (branch_prob): Do not clear edge_gcov_count.
16536 * profile.h (edge_gcov_counts): Turn to pointer.
16537 (edge_gcov_count): Update.
16539 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
16541 * gimple.h (gimple_check_failed): Mark cold.
16543 2017-06-09 Richard Biener <rguenther@suse.de>
16545 PR tree-optimization/66623
16546 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
16547 refactor check_reduction into two parts, properly computing
16548 whether we have to check reduction validity for outer loop
16551 2017-06-09 Richard Biener <rguenther@suse.de>
16553 PR tree-optimization/79483
16554 * graphite-scop-detection.c (order): New global.
16555 (get_order): Compute bb to order mapping that satisfies code
16556 generation constraints.
16557 (cmp_pbbs): New helper.
16558 (build_scops): Start domwalk at entry block, sort generated
16561 2017-06-09 Richard Biener <rguenther@suse.de>
16563 PR middle-end/81007
16564 * ipa-polymorphic-call.c
16565 (ipa_polymorphic_call_context::restrict_to_inner_class):
16566 Skip FIELD_DECLs with error_mark_node type.
16567 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
16570 2017-06-09 Martin Liska <mliska@suse.cz>
16572 * predict.c (struct branch_predictor): New struct.
16573 (test_prediction_value_range): New test.
16574 (predict_c_tests): New function.
16575 * selftest-run-tests.c (selftest::run_tests): Run the function.
16576 * selftest.h: Declare new tests.
16578 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
16581 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
16582 gen_add3_insn did not fail.
16583 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
16584 r0, construct that number in a temporary reg and add that reg to r0.
16585 If asked to put the result in r0 as well, fail.
16587 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
16589 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
16590 for early expansion of vec_eqv.
16592 2017-06-08 Jakub Jelinek <jakub@redhat.com>
16594 PR middle-end/81005
16595 * ubsan.c (instrument_null): Avoid pointless code temporary.
16596 (pass_ubsan::execute): Instrument aggregate arguments of calls.
16598 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
16602 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
16605 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
16608 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16610 * predict.c (unlikely_executed_edge_p): Move ahead.
16611 (probably_never_executed_edge_p): Use it.
16613 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16615 PR middle-end/79988
16616 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
16617 gimple_call_builtin_p call.
16619 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16621 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
16622 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
16623 rtl_check_failed_type2, rtl_check_failed_code1,
16624 rtl_check_failed_code2, rtl_check_failed_code_mode,
16625 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
16626 rtvec_check_failed_bounds, rtl_check_failed_flag,
16627 _fatal_insn_not_found, _fatal_insn): Likewise.
16628 * tree.h (tree_contains_struct_check_failed,
16629 tree_check_failed, tree_not_check_failed,
16630 tree_class_check_failed, tree_range_check_failed,
16631 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
16632 tree_vec_elt_check_failed, phi_node_elt_check_failed,
16633 tree_operand_check_failed, omp_clause_check_failed,
16634 omp_clause_operand_check_failed, omp_clause_range_check_failed):
16637 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16639 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
16640 flag_branch_probabilities.
16641 * ipa-inline.c (edge_badness): Likewise.
16642 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16643 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
16644 * predict.c (maybe_hot_frequency_p): Likewise.
16645 (probably_never_executed): Likewise.
16646 * sched-ebb.c (schedule_ebbs): Likewise.
16647 * sched-rgn.c (find_single_block_region): Likewise.
16648 * tracer.c (tail_duplicate): Likewise.
16650 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16652 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
16653 longer requires x_flag_profile_use.
16655 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16657 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
16658 instead of flag_reorder_blocks_and_partition.
16659 * dbxout.c (dbxout_function_end): Likewise.
16660 * dwarf2out.c (gen_subprogram_die): Likewise.
16661 * haifa-sched.c (sched_create_recovery_edges): Likewise.
16662 * hw-doloop.c (reorg_loops): Likewise.
16663 * varasm.c (assemble_start_function,
16664 assemble_end_function): Likewise.
16665 (decide_function_section): Do not check for
16666 flag_reorder_blocks_and_partition.
16668 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
16670 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
16672 (chkp_get_hard_register_fake_addr_expr): Ditto.
16673 (chkp_build_addr_expr): Add check for hard reg case.
16674 (chkp_parse_array_and_component_ref): Ditto.
16675 (chkp_find_bounds_1): Ditto.
16676 (chkp_process_stmt): Don't generate bounds store for
16679 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16681 * predict.c (maybe_hot_bb_p): Do not check profile status.
16682 (maybe_hot_edge_p): Likewise.
16683 (probably_never_executed): Check for zero counts even if profile
16685 (unlikely_executed_edge_p): New function.
16686 (unlikely_executed_stmt_p): New function.
16687 (unlikely_executed_bb_p): New function.
16688 (set_even_probabilities): Use unlikely predicates.
16689 (combine_predictions_for_bb): Likewise.
16690 (predict_paths_for_bb): Likewise.
16691 (predict_paths_leading_to_edge): Likewise.
16692 (determine_unlikely_bbs): New function.
16693 (estimate_bb_frequencies): Use it.
16694 (compute_function_frequency): Use zero counts even if profile is
16696 * profile-count.h: Fix typo.
16698 2017-08-08 Julia Koval <julia.koval@intel.com>
16700 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
16701 _mm512_mask_cvtsepi16_storeu_epi8,
16702 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
16703 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
16704 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
16705 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
16706 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
16707 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
16708 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
16709 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
16710 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
16711 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
16712 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
16713 __builtin_ia32_pmovuswb256mem_mask,
16714 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
16715 __builtin_ia32_pmovwb512mem_mask): New builtins.
16717 2017-08-08 Julia Koval <julia.koval@intel.com>
16719 PR target/73350,80862
16720 * config/i386/subst.md (round): Fix round pattern.
16721 * config/i386/i386.c (ix86_erase_embedded_rounding):
16722 Fix erasing rounding for the fixed pattern.
16724 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
16726 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
16728 2017-06-08 Martin Liska <mliska@suse.cz>
16730 PR gcov-profile/80911
16731 * gcov.c (block_info::block_info): New constructor.
16733 2017-06-07 Carl Love <cel@us.ibm.com>
16735 * config/rs6000/rs6000-c: The return type of the following
16736 built-in functions was implemented as int not long long. Fix sign
16737 of return value for the unsigned version of vec_mulo and vec_mule.
16738 vector unsigned long long vec_bperm (vector unsigned long long,
16739 vector unsigned char)
16740 vector signed long long vec_mule (vector signed int,
16742 vector unsigned long long vec_mule (vector unsigned int,
16743 vector unsigned int)
16744 vector signed long long vec_mulo (vector signed int,
16746 vector unsigned long long vec_mulo (vector unsigned int,
16747 vector unsigned int)
16748 * doc/extend.texi: Fix the documentation for the built-in
16751 2017-06-07 Carl Love <cel@us.ibm.com>
16754 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
16757 2017-06-07 Carl Love <cel@us.ibm.com>
16759 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
16760 support, Generate doublehv for signed int/float for BE case only.
16762 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
16764 * doc/invoke.texi (mcx16): Rewrite.
16766 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16768 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
16769 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
16770 *mov<mode>_softfloat, and an anonymous splitter): Use
16771 nonimmediate_operand instead of rs6000_nonimmediate_operand.
16773 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16775 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
16777 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16778 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
16779 (rs6000_debug_reg_global): Adjust.
16780 (rs6000_init_hard_regno_mode_ok): Adjust.
16781 (rs6000_dbx_register_number): Adjust.
16782 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
16783 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16784 Remove SPE_ACC and SPEFSCR.
16785 (REG_ALLOC_ORDER): Ditto.
16786 (FRAME_POINTER_REGNUM): Change to 111.
16787 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
16788 (REG_CLASS_NAMES): Ditto.
16789 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
16790 (REGISTER_NAMES): Ditto.
16791 (ADDITIONAL_REG_NAMES): Ditto.
16792 (rs6000_reg_names): Ditto.
16793 * config/rs6000/rs6000.md: Renumber some register number
16796 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16798 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
16800 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
16801 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
16803 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
16804 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
16805 Delete the SPE high registers.
16806 (REG_ALLOC_ORDER): Ditto.
16807 (enum reg_class): Remove SPE_HIGH_REGS.
16808 (REG_CLASS_NAMES): Ditto.
16809 (REG_CLASS_CONTENTS): Delete the SPE high registers.
16810 (REGISTER_NAMES): Ditto.
16811 (rs6000_reg_names): Ditto.
16812 * doc/tm.texi.in: Remove SPE as example.
16813 * doc/tm.texi: Regenerate.
16815 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16817 * config/rs6000/8540.md (ppc8540_brinc): Delete.
16818 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
16819 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
16820 * config/rs6000/rs6000.md (type): Remove "brinc".
16822 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16824 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
16825 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
16826 * config/rs6000/linuxspe.h: Delete file.
16827 * config/rs6000/rs6000.md: Don't include spe.md.
16828 * config/rs6000/spe.h: Delete file.
16829 * config/rs6000/spe.md: Delete file.
16830 * config/rs6000/t-rs6000: Remove spe.md.
16832 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16834 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
16835 (reg_or_none500mem_operand): Delete.
16836 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
16837 instead of reg_or_none500mem_operand.
16839 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16841 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
16842 handling of SPE flags.
16843 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
16845 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16847 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
16849 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
16850 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
16851 paired_divv2sf3): Similar.
16852 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
16853 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
16854 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
16856 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
16857 Rename the paired_* instruction patterns.
16858 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
16860 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
16861 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
16862 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
16863 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
16864 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
16865 PAIRED_VECTOR_MODE.
16866 (struct machine_function): Delete field spe_insn_chain_scanned_p.
16867 (spe_func_has_64bit_regs_p): Delete.
16868 (spe_expand_predicate_builtin): Delete.
16869 (spe_expand_evsel_builtin): Delete.
16870 (TARGET_DWARF_REGISTER_SPAN): Do not define.
16871 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
16872 (invalid_e500_subreg): Delete.
16873 (rs6000_legitimize_address): Always force_reg op2 as well, for
16874 paired single memory accesses.
16875 (rs6000_member_type_forces_blk): Delete.
16876 (rs6000_spe_function_arg): Delete.
16877 (rs6000_expand_unop_builtin): Delete SPE handling.
16878 (rs6000_expand_binop_builtin): Ditto.
16879 (spe_expand_stv_builtin): Delete.
16880 (bdesc_2arg_spe): Delete.
16881 (spe_expand_builtin): Delete.
16882 (spe_expand_predicate_builtin): Delete.
16883 (spe_expand_evsel_builtin): Delete.
16884 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
16885 (spe_init_builtins): Delete.
16886 (spe_func_has_64bit_regs_p): Delete.
16887 (savres_routine_name): Delete "info" parameter. Adjust callers.
16888 (rs6000_emit_stack_reset): Ditto.
16889 (rs6000_dwarf_register_span): Delete.
16890 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
16891 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
16892 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
16894 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
16896 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
16897 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
16898 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
16899 mulv2sf3, divv2sf3): Delete expanders.
16901 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16903 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
16905 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16907 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
16908 * config/rs6000/rs6000.c: Ditto.
16910 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16912 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
16913 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
16914 comparison_operator.
16916 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16918 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
16919 * config/rs6000/rs6000.opt: Ditto.
16920 * config/rs6000/t-rtems: Ditto.
16922 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16924 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
16925 TARGET_E500_SINGLE by 0, simplify.
16926 * config/rs6000/rs6000.c: Ditto.
16927 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
16928 (spe_build_register_parallel): Delete.
16929 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
16930 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
16931 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
16932 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
16933 (E500_CONVERT): Delete.
16934 * config/rs6000/spe.md: Remove many patterns and all define_constants.
16936 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
16938 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
16939 * config/rs6000/dfp.md: Ditto.
16940 (negdd2, *negdd2_fpr): Merge.
16941 (absdd2, *absdd2_fpr): Merge.
16942 (negtd2, *negtd2_fpr): Merge.
16943 (abstd2, *abstd2_fpr): Merge.
16944 * config/rs6000/e500.h: Delete file.
16945 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
16946 TARGET_FPRS by 1 and simplify.
16947 * config/rs6000/rs6000-c.c: Ditto.
16948 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
16949 TARGET_DF_SPE by 0.
16950 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
16952 * config/rs6000/rs6000.md: Ditto.
16953 (floatdidf2, *floatdidf2_fpr): Merge.
16954 (move_from_CR_gt_bit): Delete.
16955 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
16956 (E500_CR_IOR_COMPARE): Delete.
16957 (All patterns that require !TARGET_FPRS): Delete.
16958 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
16960 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16962 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
16964 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16966 * graphds.c (add_edge): Intitialize edge's attached data.
16967 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
16968 pointer parameter. Call pointed function on each edge during
16969 graph traversing. Skip traversing the edge when the function
16971 (graphds_dfs, graphds_scc): Ditto.
16972 (for_each_edge): New parameter. Pass the new parameter to callback
16974 * graphds.h (skip_edge_callback): New function pointer type.
16975 (graphds_dfs, graphds_scc): New function pointer parameter.
16976 (graphds_edge_callback, for_each_edge): New parameter.
16978 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16980 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
16981 out code checking if runtime alias check is possible to below ...
16982 Call the new function.
16983 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
16984 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
16986 2017-06-07 Marek Polacek <polacek@redhat.com>
16989 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
16990 TYPE_OVERFLOW_WRAPS check.
16992 2017-06-07 Bin Cheng <bin.cheng@arm.com>
16994 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
16995 if versioning is required.
16996 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
16997 peeling with the check for versioning.
16999 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17001 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
17002 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
17003 Set true to new parameter if new ssa variable is defined.
17004 (vect_gen_vector_loop_niters): Refactor. Set range information
17005 for the new vector loop bound variable.
17006 (vect_do_peeling): Ditto.
17008 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17010 * tree-affine.c (ssa.h): Include header file.
17011 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
17012 has wrapping overflow behavior.
17014 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17016 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
17018 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17020 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
17021 (tree_to_aff_combination): ... here.
17023 2017-06-07 Bin Cheng <bin.cheng@arm.com>
17025 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
17026 reg_pressure model function.
17027 (ivopts_global_cost_for_size): Delete.
17028 (determine_set_costs, iv_ca_recount_cost): Call new model function
17029 ivopts_estimate_reg_pressure.
17031 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17033 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
17034 expensive than udiv. Remove floating point cases from mod.
17036 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17038 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
17039 Increase idiv cost.
17041 2017-06-07 Tamar Christina <tamar.christina@arm.com>
17043 * config/aarch64/aarch64.md
17044 (copysignsf3): Fix mask generation.
17046 2017-06-07 Jakub Jelinek <jakub@redhat.com>
17048 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
17050 (class dump_manager): Add register_dumps method.
17051 * dumpfile.c: Include langhooks.h.
17052 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
17053 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
17054 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
17055 (dump_manager::dump_register): Start with 512 entries instead of 32.
17056 (dump_manager::register_dumps): New method.
17057 * toplev.c (general_init): Instead of invoking register_dumps
17058 langhook, invoke register_dumps method on the dump manager.
17059 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
17062 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
17064 * doc/md.texi: Clarify the restrictions on a define_insn condition.
17065 Say that # requires an associated define_split to exist, and that
17066 the define_split must be suitable for use after register allocation.
17068 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17070 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
17071 (compute_outgoing_frequencies): Also initialize zero counts.
17072 (find_many_sub_basic_blocks): Do not produce uninitialized profile
17073 around loops; preserve more of profile when nothing changes.
17075 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
17077 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
17079 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
17080 * config/arm/arm-cpu-cdata.h: Regenerate.
17081 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
17082 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
17083 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
17084 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
17085 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
17087 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
17089 * config/arm/t-rmprofile: Likewise.
17090 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
17092 2017-06-06 David S. Miller <davem@davemloft.net>
17095 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
17096 blockage if function uses alloca.
17098 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17100 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
17101 New "uid" fields to hold pretty-print IDs of group and ref.
17102 Memory references are now identified as <group_id>:<ref_id>
17103 instead of using [random] addresses.
17104 (dump_mem_details): Simplify, no functional change.
17105 (dump_mem_ref): Simplify and make output more concise.
17106 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
17107 (find_or_create_group): Initialize group uid.
17108 (record_ref): Initialize ref uid. Improve debug output.
17109 (prune_group_by_reuse, should_issue_prefetch_p,)
17110 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
17111 (mark_nontemporal_store, determine_loop_nest_reuse):
17112 Improve debug output.
17114 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
17116 * dbgcnt.def (prefetch): New debug counter.
17117 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
17118 (schedule_prefetches): Stop issueing prefetches if debug counter
17121 2017-06-06 Tom de Vries <tom@codesourcery.com>
17123 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
17124 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
17126 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17128 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
17129 Use aarch64_reg_or_zero predicate for operand 4.
17130 (aarch64_compare_and_swap<mode> define_insn_and_split):
17131 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
17132 (aarch64_store_exclusive<mode>): Likewise for operand 2.
17134 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17136 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
17137 (arm_compute_save_core_reg_mask): This.
17138 (thumb1_compute_save_reg_mask): Rename into ...
17139 (thumb1_compute_save_core_reg_mask): This.
17140 (arm_compute_save_reg0_reg12_mask): Adapt comment.
17141 (arm_compute_frame_layout): Likewise.
17143 2017-06-06 Richard Biener <rguenther@suse.de>
17145 PR tree-optimization/80974
17146 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
17147 keep or clear leaders SSA info.
17149 2017-06-06 Tom de Vries <tom@codesourcery.com>
17151 * config/nvptx/nvptx.c (split_mode_p): New function.
17152 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
17154 2017-06-06 Tom de Vries <tom@codesourcery.com>
17156 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
17158 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17161 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
17164 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17166 * shrink-wrap.c (handle_simple_exit): Update profile.
17167 (try_shrink_wrapping): Upate profile.
17169 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17171 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
17172 (tree_guess_outgoing_edge_probabilities): New.
17173 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
17174 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
17176 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17178 * ipa-split.c (split_function): Initialize return bb profile.
17180 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
17182 * profile.c (compute_branch_probabilities): Also initialize
17183 EXIT_BLOCK profile.
17185 2017-06-06 Richard Biener <rguenther@suse.de>
17187 PR tree-optimization/80928
17188 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
17189 (vect_analyze_loop_operations): Properly guard analysis for
17191 (vect_transform_loop): Likewise.
17192 (vect_analyze_loop_2): Also reset SLP type on PHIs.
17193 (vect_model_induction_cost): Do not cost for pure SLP.
17194 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
17195 of induction in inner loop vectorization.
17196 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
17197 (vect_get_and_check_slp_defs): Handle vect_induction_def.
17198 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
17200 (vect_analyze_slp_cost_1): Cost induction.
17201 (vect_detect_hybrid_slp_stmts): Handle PHIs.
17202 (vect_get_slp_vect_defs): Likewise.
17203 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
17204 (vect_transform_stmt): Handle SLP reductions.
17205 * tree-vectorizer.h (vectorizable_induction): Adjust.
17207 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17209 * config/rs6000/rs6000.c (make_resolver_func): Update
17210 init_lowered_empty_function call.
17212 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
17214 * doc/invoke.texi: Document the -fprofile-abs-path option.
17215 * common.opt (fprofile-abs-path): New option.
17216 * gcov-io.h (gcov_write_filename): Declare.
17217 * gcov-io.c (gcov_write_filename): New function.
17218 * coverage.c (coverage_begin_function): Use gcov_write_filename.
17219 * profile.c (output_location): Likewise.
17221 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17223 * shring-wrap.c: Revert accidental commit.
17225 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
17227 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
17229 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17231 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
17233 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
17234 profile in callgraph edge.
17235 * profile-count.h (apply_probability): If THIS is 0, then result is 0
17236 (apply_scale): Likewise.
17237 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
17238 Also scale profile when inlining function with zero profile.
17239 (initialize_cfun): Update exit block profile even when it is zero.
17240 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
17241 when profile is read.
17243 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
17245 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
17246 (CLONE_*): New constants to define the processors we can generate
17247 code for with the target_clone attribute.
17248 (rs6000_clone_map): New array to identify which clone processors
17249 the current program is running on.
17250 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
17251 target_clone attribute.
17252 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
17253 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
17254 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
17255 (cpu_expand_builtin): Add support for target_clone attribute.
17256 (rs6000_valid_attribute_p): Allow "default" attribute.
17257 (get_decl_name): New debug function to simplify printing the
17258 current function name in debugging statements.
17259 (rs6000_clone_priority): New functions to support the target_clone
17260 attribute, and be able to generate code to switch between ISA 2.05
17261 through ISA 3.0 (power6 through power9).
17262 (rs6000_compare_version_priority): Likewise.
17263 (rs6000_get_function_versions_dispatcher): Likewise.
17264 (make_resolver_func): Likewise.
17265 (add_condition_to_bb): Likewise.
17266 (dispatch_function_versions): Likewise.
17267 (rs6000_generate_version_dispatcher_body): Likewise.
17268 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
17269 (fusion_gpr_load_p): Fix a spacing issue.
17270 * doc/extend.texi (Common Function Attributes): Document that the
17271 PowerPC supports the target_clone attribute.
17273 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
17275 * config/arm/arm.h: explain F symbol found in description of ARM
17276 register allocation in its legend.
17278 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17280 * config/mips/frame-header-opt.c: Include profile-count.h.
17281 * config/riscv/riscv.c: Include profile-count.h
17283 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
17285 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
17287 (sm_set_flag_if_changed): Add bbs field.
17288 (execute_sm_if_changed_flag_set): Pass BBS.
17289 (execute_sm): Update.
17291 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17293 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
17296 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17298 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
17299 (peephole2): New peephole2 to emit the above.
17300 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
17302 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17304 * config/aarch64/aarch64.c (define_peephole2 above
17305 *sub_<shift>_<mode>): New peephole.
17307 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
17309 * config/i386/i386.c (make_resolver_func): Update.
17310 * Makefile.in: Add profile-count.h and profile-count.o
17311 * auto-profile.c (afdo_indirect_call): Update to new API.
17312 (afdo_set_bb_count): Update.
17313 (afdo_propagate_edge): Update.
17314 (afdo_propagate_circuit): Update.
17315 (afdo_calculate_branch_prob): Update.
17316 (afdo_annotate_cfg): Update.
17317 * basic-block.h: Include profile-count.h
17318 (struct edge_def): Turn count to profile_count.
17319 (struct basic_block_def): Likewie.
17320 (REG_BR_PROB_BASE): Move to profile-count.h
17321 (RDIV): Move to profile-count.h
17322 * bb-reorder.c (max_entry_count): Turn to profile_count.
17323 (find_traces): Update.
17324 (rotate_loop):Update.
17325 (connect_traces):Update.
17326 (sanitize_hot_paths):Update.
17327 * bt-load.c (migrate_btr_defs): Update.
17328 * cfg.c (RDIV): Remove.
17329 (init_flow): Use alloc_block.
17330 (alloc_block): Uninitialize count.
17331 (unchecked_make_edge): Uninitialize count.
17332 (check_bb_profile): Update.
17333 (dump_edge_info): Update.
17334 (dump_bb_info): Update.
17335 (update_bb_profile_for_threading): Update.
17336 (scale_bbs_frequencies_int): Update.
17337 (scale_bbs_frequencies_gcov_type): Update.
17338 (scale_bbs_frequencies_profile_count): New.
17339 * cfg.h (update_bb_profile_for_threading): Update.
17340 (scale_bbs_frequencies_profile_count): Declare.
17341 * cfgbuild.c (compute_outgoing_frequencies): Update.
17342 (find_many_sub_basic_blocks): Update.
17343 * cfgcleanup.c (try_forward_edges): Update.
17344 (try_crossjump_to_edge): Update.
17345 * cfgexpand.c (expand_gimple_tailcall): Update.
17346 (construct_exit_block): Update.
17347 * cfghooks.c (verify_flow_info): Update.
17348 (dump_bb_for_graph): Update.
17349 (split_edge): Update.
17350 (make_forwarder_block): Update.
17351 (duplicate_block): Update.
17352 (account_profile_record): Update.
17353 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
17354 (get_estimated_loop_iterations): Update.
17355 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
17356 (single_likely_exit): Update.
17357 * cfgloopmanip.c (scale_loop_profile): Update.
17359 (set_zero_probability): Update.
17360 (lv_adjust_loop_entry_edge): Update.
17361 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
17362 (purge_dead_edges): Update.
17363 (rtl_account_profile_record): Update.
17364 * cgraph.c (cgraph_node::create): Uninitialize count.
17365 (symbol_table::create_edge): Uninitialize count.
17366 (cgraph_update_edges_for_call_stmt_node): Update.
17367 (cgraph_edge::dump_edge_flags): Update.
17368 (cgraph_node::dump): Update.
17369 (cgraph_edge::maybe_hot_p): Update.
17370 * cgraph.h: Include profile-count.h
17371 (create_clone), create_edge, create_indirect_edge): Update.
17372 (cgraph_node): Turn count to profile_count.
17373 (cgraph_edge0: Likewise.
17374 (make_speculative, clone): Update.
17375 (create_edge): Update.
17376 (init_lowered_empty_function): Update.
17377 * cgraphclones.c (cgraph_edge::clone): Update.
17378 (duplicate_thunk_for_node): Update.
17379 (cgraph_node::create_clone): Update.
17380 * cgraphunit.c (cgraph_node::analyze): Update.
17381 (cgraph_node::expand_thunk): Update.
17382 * final.c (dump_basic_block_info): Update.
17383 * gimple-streamer-in.c (input_bb): Update.
17384 * gimple-streamer-out.c (output_bb): Update.
17385 * graphite.c (print_global_statistics): Update.
17386 (print_graphite_scop_statistics): Update.
17387 * hsa-brig.c: Include basic-block.h.
17388 * hsa-dump.c: Include basic-block.h.
17389 * hsa-gen.c (T sum_slice): Update.
17390 (convert_switch_statements):Update.
17391 * hsa-regalloc.c: Include basic-block.h.
17392 * ipa-chkp.c (chkp_produce_thunks): Update.
17393 * ipa-cp.c (struct caller_statistics): Update.
17394 (init_caller_stats): Update.
17395 (gather_caller_stats): Update.
17396 (ipcp_cloning_candidate_p): Update.
17397 (good_cloning_opportunity_p): Update.
17398 (get_info_about_necessary_edges): Update.
17399 (dump_profile_updates): Update.
17400 (update_profiling_info): Update.
17401 (update_specialized_profile): Update.
17402 (perhaps_add_new_callers): Update.
17403 (decide_about_value): Update.
17404 (ipa_cp_c_finalize): Update.
17405 * ipa-devirt.c (struct odr_type_warn_count): Update.
17406 (struct decl_warn_count): Update.
17407 (struct final_warning_record): Update.
17408 (possible_polymorphic_call_targets): Update.
17409 (ipa_devirt): Update.
17410 * ipa-fnsummary.c (redirect_to_unreachable): Update.
17411 * ipa-icf.c (sem_function::merge): Update.
17412 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
17413 * ipa-inline.c (compute_uninlined_call_time): Update.
17414 (compute_inlined_call_time): Update.
17415 (want_inline_small_function_p): Update.
17416 (want_inline_self_recursive_call_p): Update.
17417 (edge_badness): Update.
17418 (lookup_recursive_calls): Update.
17419 (recursive_inlining): Update.
17420 (inline_small_functions): Update.
17421 (dump_overall_stats): Update.
17422 (dump_inline_stats): Update.
17423 * ipa-profile.c (ipa_profile_generate_summary): Update.
17424 (ipa_propagate_frequency): Update.
17425 (ipa_profile): Update.
17426 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
17427 * ipa-utils.c (ipa_merge_profiles): Update.
17428 * loop-doloop.c (doloop_modify): Update.
17429 * loop-unroll.c (report_unroll): Update.
17430 (unroll_loop_runtime_iterations): Update.
17431 * lto-cgraph.c (lto_output_edge): Update.
17432 (lto_output_node): Update.
17433 (input_node): Update.
17434 (input_edge): Update.
17435 (merge_profile_summaries): Update.
17436 * lto-streamer-in.c (input_cfg): Update.
17437 * lto-streamer-out.c (output_cfg): Update.
17438 * mcf.c (create_fixup_graph): Update.
17439 (adjust_cfg_counts): Update.
17440 (sum_edge_counts): Update.
17441 * modulo-sched.c (sms_schedule): Update.
17442 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
17443 * predict.c (maybe_hot_count_p): Update.
17444 (probably_never_executed): Update.
17445 (dump_prediction): Update.
17446 (combine_predictions_for_bb): Update.
17447 (propagate_freq): Update.
17448 (handle_missing_profiles): Update.
17449 (counts_to_freqs): Update.
17450 (rebuild_frequencies): Update.
17451 (force_edge_cold): Update.
17452 * predict.h: Include profile-count.h
17453 (maybe_hot_count_p, counts_to_freqs): UPdate.
17454 * print-rtl-function.c: Do not include cfg.h
17455 * print-rtl.c: Include basic-block.h
17456 * profile-count.c: New file.
17457 * profile-count.h: New file.
17458 * profile.c (is_edge_inconsistent): Update.
17459 (correct_negative_edge_counts): Update.
17460 (is_inconsistent): Update.
17461 (set_bb_counts): Update.
17462 (read_profile_edge_counts): Update.
17463 (compute_frequency_overlap): Update.
17464 (compute_branch_probabilities): Update; Initialize and deinitialize
17466 (branch_prob): Update.
17467 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
17468 (edge_gcov_count): New.
17469 (bb_gcov_count): New.
17470 * shrink-wrap.c (try_shrink_wrapping): Update.
17471 * tracer.c (better_p): Update.
17472 * trans-mem.c (expand_transaction): Update.
17473 (ipa_tm_insert_irr_call): Update.
17474 (ipa_tm_insert_gettmclone_call): Update.
17475 * tree-call-cdce.c: Update.
17476 * tree-cfg.c (gimple_duplicate_sese_region): Update.
17477 (gimple_duplicate_sese_tail): Update.
17478 (gimple_account_profile_record): Update.
17479 (execute_fixup_cfg): Update.
17480 * tree-inline.c (copy_bb): Update.
17481 (copy_edges_for_bb): Update.
17482 (initialize_cfun): Update.
17483 (freqs_to_counts): Update.
17484 (copy_cfg_body): Update.
17485 (expand_call_inline): Update.
17486 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17487 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
17488 (try_unroll_loop_completely): Update.
17489 (try_peel_loop): Update.
17490 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17491 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
17492 * tree-ssa-loop-split.c (connect_loops): Update.
17493 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
17494 * tree-ssa-reassoc.c (branch_fixup): Update.
17495 * tree-ssa-tail-merge.c (replace_block_by): Update.
17496 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
17497 (compute_path_counts): Update.
17498 (update_profile): Update.
17499 (recompute_probabilities): Update.
17500 (update_joiner_offpath_counts): Update.
17501 (estimated_freqs_path): Update.
17502 (freqs_to_counts_path): Update.
17503 (clear_counts_path): Update.
17504 (ssa_fix_duplicate_block_edges): Update.
17505 (duplicate_thread_path): Update.
17506 * tree-switch-conversion.c (case_bit_test_cmp): Update.
17507 (struct switch_conv_info): Update.
17508 * tree-tailcall.c (decrease_profile): Update.
17509 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17510 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17511 * value-prof.c (check_counter): Update.
17512 (gimple_divmod_fixed_value): Update.
17513 (gimple_mod_pow2): Update.
17514 (gimple_mod_subtract): Update.
17515 (gimple_ic_transform): Update.
17516 (gimple_stringop_fixed_value): Update.
17517 * value-prof.h (gimple_ic): Update.
17519 2017-06-02 Carl Love <cel@us.ibm.com>
17521 * config/rs6000/rs6000-c: Add support for built-in functions
17522 vector double vec_doublee (vector signed int);
17523 vector double vec_doublee (vector unsigned int);
17524 vector double vec_doublee (vector float);
17525 vector double vec_doubleh (vector signed int);
17526 vector double vec_doubleh (vector unsigned int);
17527 vector double vec_doubleh (vector float);
17528 vector double vec_doublel (vector signed int);
17529 vector double vec_doublel (vector unsigned int);
17530 vector double vec_doublel (vector float);
17531 vector double vec_doubleo (vector signed int);
17532 vector double vec_doubleo (vector unsigned int);
17533 vector double vec_doubleo (vector float);.
17534 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
17535 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
17537 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
17538 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
17539 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
17541 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
17542 vec_doublel, vec_doubleh.
17543 * doc/extend.texi: Update the built-in documentation file for the
17544 new built-in functions.
17546 2017-06-02 David Malcolm <dmalcolm@redhat.com>
17549 * ipa-inline-analysis.c (free_growth_caches): Set
17550 edge_removal_hook_holder to NULL after removing it.
17552 2017-06-02 Sudakshina Das <sudi.das@arm.com>
17554 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
17555 comparision with zero.
17557 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17558 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17559 for early expansion of vec_min and vec_max builtins.
17560 (builtin_function_type): Add min/max unsigned variants to those
17561 identified as having unsigned arguments.
17563 2017-06-02 Olivier Hainque <hainque@adacore.com>
17565 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
17567 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17569 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
17570 Use VALL_F16 iterator rather than VALL.
17572 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17574 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
17575 Emit CBNZ inside loop when doing a strong exchange and comparing
17576 against zero. Generate the CC flags after the loop.
17578 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
17580 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
17581 (dl_section_ref): New.
17582 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
17583 On AIX, append an expression to subtract the size of the
17584 section length to dl_section_ref.
17586 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
17588 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
17589 for early expansion of vector absolute builtins.
17591 2017-06-02 Richard Biener <rguenther@suse.de>
17593 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
17594 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
17596 2017-06-02 Richard Biener <rguenther@suse.de>
17598 PR tree-optimization/80948
17599 * tree-tailcall.c (find_tail_calls): Track stmts to move in
17600 stmt order as well.
17602 2017-06-02 Richard Biener <rguenther@suse.de>
17604 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
17606 * tree-vect-stmts.c (process_use): Do not mark backedge defs
17607 for inductions as relevant.
17609 2017-06-02 Richard Biener <rguenther@suse.de>
17611 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
17612 (vectorizable_induction): ... this. Remove dead code.
17614 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
17616 * builtins. (expand_builtin_alloca): Remove second parameter and
17617 infer its value from the first parameter instead.
17618 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
17620 2017-06-02 Jakub Jelinek <jakub@redhat.com>
17622 PR rtl-optimization/80903
17623 * loop-doloop.c (add_test): Unshare sequence.
17625 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
17627 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
17629 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
17631 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
17633 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
17634 xlogue_layout::get_instance, logue_layout::xlogue_layout,
17635 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
17636 (xlogue_layout::get_stub_rtx): Make static.
17637 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
17638 (xlogue_layout::compute_stub_managed_regs): Rename to...
17639 (xlogue_layout::count_stub_managed_regs): ...this.
17640 (xlogue_layout::is_stub_managed_reg): New function.
17641 (xlogue_layout::m_stub_names): Rename to...
17642 (xlogue_layout::s_stub_names): ...this, make static.
17643 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
17644 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
17645 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
17646 xlogue_layout::s_stub_names): Instantiate statics.
17647 (stub_managed_regs): Remove.
17648 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
17649 (disable_call_ms2sysv_xlogues): Rename to...
17650 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
17651 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
17653 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
17654 change after reload_completed.
17655 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
17657 (ix86_expand_prologue): Likewise.
17658 (ix86_expand_epilogue): Likewise.
17659 (ix86_expand_split_stack_prologue): Likewise.
17660 (ix86_compute_frame_layout): Remove frame parameter ...
17661 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
17662 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
17664 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
17665 (ix86_frame): Move from here ...
17666 * config/i386/i386.h (ix86_frame): ... to here.
17667 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
17668 complete ix86_frame data structure instead. Remove some_ld_name.
17670 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17672 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
17673 symbols that hold a DECL_VALUE_EXPR.
17675 2017-06-01 Martin Jambor <mjambor@suse.cz>
17677 PR tree-optimization/80898
17678 * tree-sra.c (process_subtree_disqualification): Removed.
17679 (disqualify_candidate): Do not acll
17680 process_subtree_disqualification.
17681 (subtree_mark_written_and_enqueue): New function.
17682 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
17683 RHS has been disqualified and re-queue LHS if necessary. Apart
17684 from that, ignore disqualified RHS.
17686 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17688 * config/s390/s390.c (s390_emit_epilogue): Disable early return
17689 address fetch for z10 or later.
17691 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17693 * config/arc/arc.md (tst_movb): Add guard when splitting.
17695 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17697 * config/arc/arc.c (arc_can_eliminate): Test against
17698 arc_frame_pointer_needed.
17700 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17702 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
17703 to prevent store reordering.
17704 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
17705 (type): Add block type.
17706 (stack_tie): Define special instruction to be used in
17709 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17711 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
17712 constraint. It is not valid for the pattern.
17713 (noncommutative_binary_comparison): Likewise.
17715 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17717 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
17720 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17722 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
17723 be used by the reg-alloc.
17725 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17727 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
17728 reg-alloc when having mul64 or mul32x16 instructions.
17729 (mulsidi3): Likewise.
17730 (umulsidi3): Likewise.
17731 (mulsi32x16): New pattern.
17732 (mulsi64): Likewise.
17733 (mulsidi64): Likewise.
17734 (umulsidi64): Likewise.
17735 (MUL32x16_REG): Define.
17736 (mul64_600): Use MUL32x16_REG.
17737 (mac64_600): Likewise.
17738 (umul64_600): Likewise.
17739 (umac64_600): Likewise.
17741 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
17743 * config/arc/arc.md (mulsi3_700): Make it commutative.
17745 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
17747 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
17749 (*sign_extendsidi2_insn): Likewise for movstosw.
17751 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
17753 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
17754 the type of the input discriminant value. Convert the
17755 discriminant value of signedness vary.
17757 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
17759 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
17760 Add to -Wall section.
17762 2017-06-01 Richard Biener <rguenther@suse.de>
17764 PR middle-end/66313
17765 * fold-const.c (fold_plusminus_mult_expr): If the factored
17766 factor may be zero use a wrapping type for the inner operation.
17767 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
17768 and handle moved defs.
17769 (process_assignment): Properly guard the unary op case. Return a
17770 tri-state indicating that moving the stmt before the call may allow
17771 to continue. Pass through to_move.
17772 (find_tail_calls): Handle moving unrelated defs before
17775 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
17778 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
17779 splitter result in the canonical way.
17781 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
17783 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
17784 also for 32bit target. Update insn attributes.
17785 (zero-extendsidi2 splitter): Allow all registers for operand 1.
17787 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
17789 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
17790 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
17791 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
17792 (_mm_maskz_min_ss): New intrinsics.
17794 2017-05-31 Martin Liska <mliska@suse.cz>
17796 * tree-vect-loop.c (vect_create_epilog_for_reduction):
17797 Change comment style to one we normally use.
17798 (vectorizable_reduction): Likewise.
17799 (vectorizable_induction): Likewise.
17800 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
17801 (vectorizable_call): Likewise.
17802 (vectorizable_simd_clone_call): Likewise.
17803 (vectorizable_conversion): Likewise.
17804 (vectorizable_assignment): Likewise.
17805 (vectorizable_shift): Likewise.
17806 (vectorizable_operation): Likewise.
17807 (vectorizable_store): Likewise.
17808 (vectorizable_load): Likewise.
17809 * tree-vectorizer.h: Likewise.
17811 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
17813 * passes.c (emergency_dump_function): New.
17814 * tree-pass.h (emergency_dump_function): Declare.
17815 * plugin.c (plugins_internal_error_function): Remove.
17816 * plugin.h (plugins_internal_error_function): Remove declaration.
17817 * toplev.c (internal_error_function): New static function. Use it...
17818 (general_init): ...here.
17820 2017-05-31 Graham Markall <graham.markall@embecosm.com>
17822 * config/arc/arc.c (arc_print_operand): Handle constant operands.
17823 (arc_rtx_costs): Add costs for new patterns.
17824 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
17825 * config/arc/predicates.md: Add _1_2_3_operand predicate.
17827 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17829 * tree-ssa-strlen.c (get_next_strinfo): New function.
17830 (get_stridx_plus_constant): Use it.
17831 (zero_length_string): Likewise.
17832 (adjust_related_strinfos): Likewise.
17833 (adjust_last_stmt): Likewise.
17835 2017-05-31 Richard Biener <rguenther@suse.de>
17838 * config/i386/i386.c (ix86_expand_builtin): Remove assert
17839 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
17841 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
17843 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
17844 loop_vinfo argument and use of dependence distance vectors.
17845 Check instead whether the two references differ only in their
17846 initial value and assume that they have the same alignment if the
17847 difference is a multiple of the vector alignment.
17848 (vect_analyze_data_refs_alignment): Update call accordingly.
17850 2017-05-31 Martin Liska <mliska@suse.cz>
17853 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
17855 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17857 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
17858 (create_intersect_range_checks): Move from ...
17859 * tree-data-ref.c (create_intersect_range_checks_index)
17860 (create_intersect_range_checks): ... to here.
17861 (create_runtime_alias_checks): New function factored from ...
17862 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
17863 here. Call above function.
17864 * tree-data-ref.h (create_runtime_alias_checks): New function.
17866 2017-05-31 Bin Cheng <bin.cheng@arm.com>
17868 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
17869 segment length for dr_b and compute it in wide_int.
17871 2017-05-31 Richard Biener <rguenther@suse.de>
17873 PR tree-optimization/80906
17874 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
17875 and pass through iv_map.
17876 (copy_bb_and_scalar_dependences): Adjust.
17877 (translate_pending_phi_nodes): Likewise.
17878 (copy_loop_close_phi_args): Handle code-generating IVs instead
17881 2017-05-30 David Malcolm <dmalcolm@redhat.com>
17883 * diagnostic-color.c (color_dict): Add "type-diff".
17884 (parse_gcc_colors): Update comment.
17885 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
17886 -fdiagnostics-show-template-tree and -fno-elide-type.
17887 (GCC_COLORS): Add type-diff to example.
17889 (-fdiagnostics-show-template-tree): New.
17890 (-fno-elide-type): New.
17891 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
17892 the pp_format_decoder callback. Call any m_format_postprocessor's
17894 (pretty_printer::pretty_printer): Initialize
17895 m_format_postprocessor.
17896 (pretty_printer::~pretty_printer): Delete any
17897 m_format_postprocessor.
17898 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
17899 (class format_postprocessor): New class.
17900 (struct pretty_printer::format_decoder): Document the new parameters.
17901 (struct pretty_printer::m_format_postprocessor): New field.
17902 * tree-diagnostic.c (default_tree_printer): Update for new
17903 bool and const char ** params.
17904 * tree-diagnostic.h (default_tree_printer): Likewise.
17906 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
17908 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
17909 (lwa_operand): Delete rs6000_gen_cell_microcode test.
17910 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
17911 rs6000_gen_cell_microcode code.
17912 (rs6000_final_prescan_insn): Delete.
17913 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
17914 "warn-cell-microcode" entries.
17915 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
17916 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
17917 throughout. Change cc_reg_not_micro_cr0_operand to
17918 cc_reg_not_cr0_operand throughout.
17919 (*extendhi<mode>2_noload): Delete.
17920 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
17921 (mwarn-cell-microcode): Delete.
17922 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
17923 -mgen-cell-microcode and -mwarn-cell-microcode.
17925 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
17928 * config/i386/constraints.md (Yd): New constraint.
17930 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
17931 and (?Yd, r) alternatives. Update insn attributes.
17932 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
17933 and (?*Yd, r) alternatives. Update insn attributes.
17934 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
17936 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
17938 * gimplify.c (gimplify_modify_expr): Don't create a
17939 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
17942 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
17944 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
17946 2017-05-30 Richard Biener <rguenther@suse.de>
17948 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
17949 and reduc_def fields.
17950 (STMT_VINFO_REDUC_TYPE): New define.
17951 (STMT_VINFO_REDUC_DEF): Likewise.
17952 (vect_force_simple_reduction): Adjust prototype.
17953 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
17954 (vect_is_simple_reduction): Remove check_reduction argument.
17955 (vect_force_simple_reduction): Adjust and set
17956 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17957 (vectorizable_reduction): Do not re-do reduction analysis
17958 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
17959 * tree-parloops.c (gather_scalar_reductions): Adjust.
17961 2017-05-30 Richard Biener <rguenther@suse.de>
17963 PR middle-end/80901
17964 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
17967 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
17969 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
17970 Introduce unknown_misalignment parameter and remove vf.
17971 (vect_peeling_hash_get_lowest_cost):
17972 Pass unknown_misalignment parameter.
17973 (vect_enhance_data_refs_alignment):
17974 Fix unsupportable data ref treatment.
17976 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17978 * tree-vect-data-refs.c (vect_get_data_access_cost):
17979 Workaround for SLP handling.
17980 (vect_enhance_data_refs_alignment):
17981 Compute costs for doing no peeling at all, compare to the best
17982 peeling costs so far and avoid peeling if cheaper.
17984 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17986 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
17987 Return peeling info and set costs to zero for unlimited cost
17989 (vect_enhance_data_refs_alignment): Also inspect all datarefs
17990 with unknown misalignment. Compute and costs for unknown
17991 misalignment, compare them to the costs for known misalignment
17992 and choose the cheapest for peeling.
17994 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
17996 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
17997 (vect_get_peeling_costs_all_drs): Create function.
17998 (vect_peeling_hash_get_lowest_cost):
17999 Use vect_get_peeling_costs_all_drs.
18000 (vect_peeling_supportable): Create function.
18001 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
18003 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
18005 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
18006 DR_HAS_NEGATIVE_STEP.
18007 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
18008 (vect_enhance_data_refs_alignment): Use.
18009 (vect_duplicate_ssa_name_ptr_info): Use.
18010 * tree-vectorizer.h (dr_misalignment): Use.
18011 (known_alignment_for_access_p): Use.
18013 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18016 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
18018 (has_section_name): New function.
18020 2017-05-30 Martin Liska <mliska@suse.cz>
18023 * auto-profile.c (get_function_decl_from_block): Fix
18026 2017-05-30 Richard Biener <rguenther@suse.de>
18028 PR middle-end/80876
18029 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
18031 2017-05-30 Martin Liska <mliska@suse.cz>
18033 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
18034 * dumpfile.h (struct dump_file_info): Remove ctors.
18036 2017-05-30 Martin Liska <mliska@suse.cz>
18038 * predict.def: Fix GNU coding style.
18040 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18042 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
18043 Mark 'to' argument with ATTRIBUTE_UNUSED.
18045 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
18047 * config/xtensa/xtensa.c (xtensa_emit_call): Use
18048 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
18049 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
18052 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
18054 * doc/install.texi (Options specification): Restore entry of
18055 --enable-sjlj-exceptions.
18057 2017-05-27 Michael Eager <eager@eagercon.com>
18060 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
18062 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
18064 * config/microblaze/microblaze.h
18065 (FIXED_REGISTERS): Update in macro.
18066 (CALL_USED_REGISTERS): Update in macro.
18068 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
18070 * doc/install.texi: Add links to macOS binary distributions.
18072 2017-05-27 Jakub Jelinek <jakub@redhat.com>
18076 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18078 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18080 2017-05-26 Martin Liska <mliska@suse.cz>
18082 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
18084 2017-05-26 Martin Liska <mliska@suse.cz>
18086 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
18087 always leading ';; '.
18088 (dump_bb_info): Likewise.
18089 (brief_dump_cfg): Likewise.
18090 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
18091 * dumpfile.c: Remove usage of TDF_VERBOSE.
18092 * dumpfile.h (enum dump_kind): Likewise.
18093 (dump_gimple_bb_header): Do not use TDF_COMMENT.
18094 * print-tree.c (debug_verbose): Remove.
18095 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
18096 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
18097 * tree-diagnostic.c (default_tree_printer): Replace
18098 TDF_DIAGNOSTIC with TDF_SLIM.
18100 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18102 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
18103 in parameter loop, rather than loop_vinfo.
18104 (create_intersect_range_checks): Ditto.
18105 (vect_create_cond_for_alias_checks): Update call to above functions.
18107 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18109 PR tree-optimization/80815
18110 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
18111 for merging runtime alias checks. Handle negative DR_STEPs.
18113 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18115 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
18117 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
18118 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
18119 out code pruning runtime alias checks.
18120 * tree-data-ref.c (prune_runtime_alias_test_list): New function
18121 factored out from above.
18122 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18124 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
18126 (prune_runtime_alias_test_list): New decalaration.
18128 2017-05-26 Bin Cheng <bin.cheng@arm.com>
18130 * tree-vect-data-refs.c (compare_tree): Rename and move ...
18131 * tree-data-ref.c (data_ref_compare_tree): ... to here.
18132 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
18133 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
18134 (operator==, comp_dr_with_seg_len_pair): Ditto.
18135 (vect_prune_runtime_alias_test_list): Ditto.
18137 2017-05-26 Martin Liska <mliska@suse.cz>
18140 * params.def: Bound partial-inlining-entry-probability param.
18142 2017-05-26 Marek Polacek <polacek@redhat.com>
18145 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
18148 2017-05-26 Richard Biener <rguenther@suse.de>
18150 PR tree-optimization/80842
18151 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
18154 2017-05-26 Richard Biener <rguenther@suse.de>
18156 PR tree-optimization/80844
18157 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
18159 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
18161 * doc/md.texi (Machine Constraints): Update x86 family
18162 machine constraints section to match 'config/i386/constraints.md'.
18164 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
18166 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
18168 2017-05-25 Nathan Sidwell <nathan@acm.org>
18170 * doc/invoke.texi (--enable-languages): Update documentation.
18172 2017-05-25 Martin Liska <mliska@suse.cz>
18174 * dumpfile.c: Add TDF_FOLDING.
18175 * dumpfile.h (enum dump_kind): Likewise.
18176 * genmatch.c (dt_simplify::gen_1): Use it.
18178 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18180 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
18182 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18184 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
18185 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
18187 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
18189 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
18190 * match.pd (X == C): Rewrite it here.
18191 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
18192 with_certain_nonzero_bits2): New predicates.
18193 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
18195 2017-05-24 Nathan Sidwell <nathan@acm.org>
18197 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
18200 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
18203 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18205 * config/powerpcspe: New port. Files are copied from the rs6000
18206 port, with "rs6000" in filenames replaced by "powerpcspe".
18208 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
18210 PR rtl-optimization/80754
18211 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
18213 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
18215 * config/sparc/sparc.md (length): Return the correct value for -mflat
18216 sibcalls to match output_sibcall.
18218 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
18222 * config/rs6000/rs6000.c (struct machine_function): Add new field
18224 (rs6000_get_separate_components): Init that field, use it.
18225 (rs6000_components_for_bb): Use the field.
18227 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
18229 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
18231 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
18233 PR middle-end/80823
18234 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
18236 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18239 * config/s390/s390.c (s390_check_qrst_address): Check incoming
18240 address against address_operand predicate.
18241 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
18243 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
18245 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
18246 parameters passed indirectly.
18248 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
18250 * config/i386/i386.md (*movdi_internal): Remove SSE4
18251 alternative 18 (?r, *v). Update insn attributes.
18252 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
18253 Update insn attributes.
18254 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
18255 Update insn attributes.
18256 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
18257 alternative 1 (r, v). Remove isa attribute.
18258 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
18259 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
18260 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
18262 2017-05-23 Tom de Vries <tom@codesourcery.com>
18264 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
18267 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18269 * cgraphunit.c (symbol_table::process_new_functions): Update.
18270 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
18271 (inline_generate_summary): Rename to ...
18272 (ipa_fn_summary_generate): ... this one.
18273 (inline_read_summary): Rename to ...
18274 (ipa_fn_summary_read): ... this one.
18275 (inline_write_summary): Rename to ...
18276 (ipa_fn_summary_write): ... this one.
18277 (inline_free_summary): Rename to ...
18278 (ipa_free_fn_summary): ... this one.
18279 (pass_data_local_fn_summary, pass_local_fn_summary,
18280 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
18281 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
18282 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
18283 make_pass_ipa_fn_summary): New.
18284 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
18285 inline_write_summary, inline_free_summary): Remove.
18286 (ipa_free_fn_summary) : New.
18287 * ipa-inline.c (ipa_inline): Update.
18288 (pass_ipa_inline): Do not generate summaries.
18289 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
18291 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
18292 and add pass_ipa_fn_summary.
18293 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
18295 (make_pass_inline_parameters): Remove.
18297 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
18299 * omp-low.c (struct omp_context): Remove "default_kind" member.
18302 * omp-offload.c (execute_oacc_device_lower): Remove the
18303 parallelism dimensions function attributes for unparallelized
18304 OpenACC kernels constructs.
18306 2017-05-23 Martin Liska <mliska@suse.cz>
18308 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
18310 (cgraph_edge::make_speculative): Likewise.
18311 (cgraph_edge::resolve_speculation): Likewise.
18312 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
18313 (cgraph_node::dump): Likewise.
18314 * cgraph.h: Likewise.
18315 * cgraphunit.c (analyze_functions): Likewise.
18316 (symbol_table::compile): Likewise.
18317 * ipa-cp.c (print_all_lattices): Likewise.
18318 (determine_versionability): Likewise.
18319 (initialize_node_lattices): Likewise.
18320 (ipcp_verify_propagated_values): Likewise.
18321 (estimate_local_effects): Likewise.
18322 (update_profiling_info): Likewise.
18323 (create_specialized_node): Likewise.
18324 (perhaps_add_new_callers): Likewise.
18325 (decide_about_value): Likewise.
18326 (decide_whether_version_node): Likewise.
18327 (identify_dead_nodes): Likewise.
18328 (ipcp_store_bits_results): Likewise.
18329 * ipa-devirt.c (dump_targets): Likewise.
18330 (ipa_devirt): Likewise.
18331 * ipa-icf.c (sem_item::dump): Likewise.
18332 (sem_function::equals): Likewise.
18333 (sem_variable::equals): Likewise.
18334 (sem_item_optimizer::read_section): Likewise.
18335 (sem_item_optimizer::execute): Likewise.
18336 (congruence_class::dump): Likewise.
18337 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
18338 (dump_inline_summary): Likewise.
18339 (estimate_node_size_and_time): Likewise.
18340 (inline_analyze_function): Likewise.
18341 * ipa-inline-transform.c (inline_call): Likewise.
18342 * ipa-inline.c (report_inline_failed_reason): Likewise.
18343 (want_early_inline_function_p): Likewise.
18344 (edge_badness): Likewise.
18345 (update_edge_key): Likewise.
18346 (inline_small_functions): Likewise.
18347 * ipa-profile.c (ipa_profile): Likewise.
18348 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
18349 (ipa_make_edge_direct_to_target): Likewise.
18350 (remove_described_reference): Likewise.
18351 (ipa_impossible_devirt_target): Likewise.
18352 (propagate_controlled_uses): Likewise.
18353 (ipa_print_node_params): Likewise.
18354 (ipcp_transform_function): Likewise.
18355 * ipa-pure-const.c (pure_const_read_summary): Likewise.
18356 (propagate_pure_const): Likewise.
18357 * ipa-reference.c (generate_summary): Likewise.
18358 (read_write_all_from_decl): Likewise.
18359 (propagate): Likewise.
18360 (ipa_reference_read_optimization_summary): Likewise.
18361 * ipa-utils.c (ipa_merge_profiles): Likewise.
18362 * ipa.c (walk_polymorphic_call_targets): Likewise.
18363 (symbol_table::remove_unreachable_nodes): Likewise.
18364 (ipa_single_use): Likewise.
18365 * passes.c (execute_todo): Likewise.
18366 * predict.c (drop_profile): Likewise.
18367 * symtab.c (symtab_node::get_dump_name): New function.
18368 (symtab_node::dump_name): Likewise.
18369 (symtab_node::dump_asm_name): Likewise.
18370 (symtab_node::dump_references): Likewise.
18371 (symtab_node::dump_referring): Likewise.
18372 (symtab_node::dump_base): Likewise.
18373 (symtab_node::debug_symtab): Likewise.
18374 * tree-sra.c (convert_callers_for_node): Likewise.
18375 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
18376 * value-prof.c (init_node_map): Likewise.
18378 2017-05-23 Martin Liska <mliska@suse.cz>
18380 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
18381 and symtab_node::debug_symtab to symbol_table::debug.
18382 * cgraphunit.c (analyze_functions): Use the renamed function.
18383 (symbol_table::compile): Likewise.
18384 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
18385 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
18386 * passes.c (execute_todo): Likewise.
18387 * symtab.c (symbol_table::dump): New function.
18388 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
18390 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18392 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
18393 that nonconst implies exec.
18395 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
18397 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18398 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18399 (inline_edge_summary_vec): Turn into ...
18400 (ipa_call_summaries): ... this one.
18401 (redirect_to_unreachable, edge_set_predicate,
18402 evaluate_properties_for_edge, inline_summary_alloc,
18403 reset_ipa_call_summary, reset_inline_summary,
18404 inline_summary_t::duplicate): Update.
18405 (inline_edge_duplication_hook): Turn to ...
18406 (ipa_call_summary_t::duplicate): ... this one.
18407 (inline_edge_removal_hook): Turn to ...
18408 (ipa_call_summary_t::remove): ... this one.
18409 (dump_inline_edge_summary): Turn to ...
18410 (dump_ipa_call_summary): ... this one.
18411 (estimate_function_body_sizes): Update.
18412 (inline_update_callee_summaries): Update.
18413 (remap_edge_change_prob): Update.
18414 (remap_edge_summaries): Update.
18415 (inline_merge_summary): Update.
18416 (do_estimate_edge_time): Update.
18417 (inline_generate_summary): Update.
18418 (inline_read_section): Update.
18419 (inline_read_summary): Update.
18420 (inline_free_summary): Update.
18421 * ipa-inline.c (can_inline_edge_p): Update.
18422 (compute_inlined_call_time): Update.
18423 (want_inline_small_function_p): Update.
18424 (edge_badness): Update.
18425 (early_inliner): Update.
18426 * ipa-inline.h (inline_edge_summary): Turn to ...
18427 (ipa_call_summary): ... this one.
18428 (ipa_call_summary_t): New class.
18429 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18430 (ipa_call_summaries): New.
18431 (inline_edge_summary): Remove.
18432 (estimate_edge_growth): Update.
18433 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18434 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18435 * ipa-split.c (execute_split_functions): Update.
18436 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18438 2017-05-23 Tom de Vries <tom@codesourcery.com>
18440 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18441 attributes): Document rdrand effective target.
18443 2017-05-23 Tom de Vries <tom@codesourcery.com>
18445 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
18446 attributes): Sort alphabetically.
18448 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
18450 * config/avr/genmultilib.awk: Use gsub instead of gensub.
18452 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
18455 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
18456 V2DF/V2DI splat into two separate patterns, one that handles
18457 registers, and the other that only handles memory. Drop support
18458 for splatting from a GPR on ISA 2.07 and then splitting the
18459 splat into direct move and splat.
18460 (vsx_splat_<mode>_reg): Likewise.
18461 (vsx_splat_<mode>_mem): Likewise.
18463 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
18465 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
18467 2017-05-22 Jakub Jelinek <jakub@redhat.com>
18469 PR middle-end/80809
18470 * omp-low.c (finish_taskreg_remap): New function.
18471 (finish_taskreg_scan): If unit size of ctx->record_type
18472 is non-constant, unshare the size expression and replace
18473 decls in it with possible outer var refs.
18475 PR middle-end/80809
18476 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
18477 GOVD_SHARED rather than GOVD_PRIVATE with it.
18478 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
18479 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
18481 PR middle-end/80853
18482 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
18483 as last argument to build_outer_var_ref for pointer bases of array
18484 section reductions.
18486 2017-05-19 Martin Sebor <msebor@redhat.com>
18488 * print-tree.c (print_node): Print DECL_READ_P flag.
18490 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18492 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
18493 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
18494 * cgraph.c: Likewise.
18495 * cgraphunit.c: Likewise.
18496 * gengtype.c: Likewise.
18497 * ipa-cp.c: Likewise.
18498 * ipa-devirt.c: Likewise.
18499 * ipa-icf.c: Likewise.
18500 * ipa-predicate.c: Likewise.
18501 * ipa-profile.c: Likewise.
18502 * ipa-prop.c: Likewise.
18503 * ipa-split.c: Likewise.
18505 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
18506 edge_predicate_pool, dump_inline_hints,
18507 inline_summary::account_size_time, redirect_to_unreachable,
18508 edge_set_predicate, set_hint_predicate,
18509 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
18510 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
18511 inline_summary_t::remove, remap_hint_predicate_after_duplication,
18512 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
18513 ipa_call_summary_t::remove, initialize_growth_caches,
18514 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
18515 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
18516 mark_modified, unmodified_parm_1, unmodified_parm,
18517 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
18518 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
18519 compute_bb_predicates, will_be_nonconstant_expr_predicate,
18520 will_be_nonconstant_predicate, record_modified_bb_info,
18521 get_minimal_bb, record_modified, param_change_prob,
18522 phi_result_unknown_predicate, predicate_for_phi_result,
18523 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
18524 estimate_function_body_sizes, compute_inline_parameters,
18525 compute_inline_parameters_for_curren, pass_data_inline_parameters,
18526 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
18527 inline_update_callee_summaries, remap_edge_change_prob,
18528 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
18529 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
18530 inline_analyze_function, inline_summary_t::insert,
18531 inline_generate_summary, read_ipa_call_summary, inline_read_section,
18532 inline_read_summary, write_ipa_call_summary, inline_write_summary,
18533 inline_free_summary): Move to ipa-fnsummary.h
18534 (predicate_t): Remove.
18535 * ipa-fnsummary.c: New file.
18536 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
18537 (enum inline_hints_vals, inline_hints, agg_position_info,
18538 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
18539 inline_summaries, ipa_call_summary, ipa_call_summary_t,
18540 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
18541 dump_inline_summary, dump_inline_hints, inline_generate_summary,
18542 inline_read_summary, inline_write_summary, inline_free_summary,
18543 inline_analyze_function, initialize_inline_failed,
18544 inline_merge_summary, inline_update_overall_summary,
18545 compute_inline_parameters): Move to ipa-fnsummary.h
18546 * ipa-fnsummary.h: New file.
18547 * ipa-inline-transform.h: Include ipa-inline.h.
18548 * ipa-inline.c: LIkewise.
18550 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18552 * ipa-inline.c (edge_badness): Use inlined_time instead of
18553 inline_summaries->get.
18555 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18557 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
18559 2017-05-22 Nathan Sidwell <nathan@acm.org>
18561 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
18562 (fdump-lang): Document 'raw' option.
18563 * dumpfile.h (TDI_tu): Delete.
18564 * dumpfile.c (dump_files): Remove translation-unit.
18565 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
18567 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
18569 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
18570 command option from $(AWK) call.
18571 * config/avr/genmultilib.awk: Simplify and rewrite so that it
18572 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
18573 [FORMAT]: Remove handling of variable.
18574 * config/avr/t-multilib: Regenerate.
18576 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18578 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
18580 (dump_inline_summary): Do not print self_time.
18581 (estimate_function_body_sizes): Do not set self_time.
18582 (compute_inline_parameters): Likewise.
18583 (inline_read_section, inline_write_summary): Do not stream self_time.
18584 * ipa-inline.h (inline_summary): Drop self_time.
18586 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
18588 * ipa-inline-analysis.c (account_size_time): Rename to ...
18589 (inline_summary::account_size_time): ... this one.
18590 (reset_ipa_call_summary): Turn to ...
18591 (ipa_call_summary::reset): ... this one.
18592 (reset_inline_summary): Turn to ...
18593 (inline_summary::reset): ... this one.
18594 (inline_summary_t::remove): Update.
18595 (inline_summary_t::duplicate): Update.
18596 (ipa_call_summary_t::remove): Update.
18597 (dump_inline_summary): Update.
18598 (estimate_function_body_sizes): Update.
18599 (compute_inline_parameters): Update.
18600 (estimate_node_size_and_time): Update.
18601 (inline_merge_summary): Update.
18602 (inline_update_overall_summary): Update.
18603 (inline_read_section): Update.
18604 (inline_write_summary): Update.
18605 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
18606 add account_size_time and reset member functions.
18607 (ipa_call_summary): Add reset function.
18608 * ipa-predicate.h (predicate::operator &): Constify.
18610 2017-05-22 Richard Biener <rguenther@suse.de>
18612 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
18614 2017-05-19 Jason Merrill <jason@redhat.com>
18616 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
18618 2017-05-19 Marek Polacek <polacek@redhat.com>
18621 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
18622 TYPE_OVERFLOW_WRAPS checks.
18624 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
18626 * tree-core.h (enum omp_clause_default_kind): Add
18627 "OMP_CLAUSE_DEFAULT_PRESENT".
18628 * tree-pretty-print.c (dump_omp_clause): Handle it.
18629 * gimplify.c (enum gimplify_omp_var_data): Add
18630 "GOVD_MAP_FORCE_PRESENT".
18631 (gimplify_adjust_omp_clauses_1): Map it to
18632 "GOMP_MAP_FORCE_PRESENT".
18633 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
18635 * gimplify.c (oacc_default_clause): Clarify.
18637 2017-05-19 Nathan Sidwell <nathan@acm.org>
18639 LANG_HOOK_REGISTER_DUMPS
18640 * toplev.c (general_init): Call register dump lang hook.
18641 * doc/invoke.texi: Document -fdump-lang option family.
18642 * dumpfile.c (dump_files): Remove class dump here.
18643 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
18644 * dumpfile.h (tree_dump_index): Remove TDI_class.
18645 * langhooks-def.h (lhd_register_dumps): Declare.
18646 (LANG_HOOKS_REGISTER_DUMPS): Define.
18647 (LANG_HOOKS_INITIALIZER): Add it.
18648 * langhooks.c (lhd_register_dumps): Define.
18649 * langhooks.h (struct lang_hooks): Add register_dumps.
18651 2017-05-19 Nathan Sidwell <nathan@acm.org>
18653 * context.h (context::set_passes): New.
18654 * context.c (context::context): Do not create pass manager.
18655 * toplev.c (general_init): Create pass manager here.
18657 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
18659 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
18660 use this splitter if two add or or instructions would also work for
18661 the constant we want to generate.
18663 2017-05-19 Richard Biener <rguenther@suse.de>
18666 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
18667 predicate evaluation.
18669 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18671 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
18673 * ipa-inline.c (want_inline_small_function_p): Do not cast to
18676 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
18678 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
18679 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
18680 (inline_edge_summary_vec): Turn into ...
18681 (ipa_call_summaries): ... this one.
18682 (redirect_to_unreachable, edge_set_predicate,
18683 evaluate_properties_for_edge, inline_summary_alloc,
18684 reset_ipa_call_summary, reset_inline_summary,
18685 inline_summary_t::duplicate): Update.
18686 (inline_edge_duplication_hook): Turn to ...
18687 (ipa_call_summary_t::duplicate): ... this one.
18688 (inline_edge_removal_hook): Turn to ...
18689 (ipa_call_summary_t::remove): ... this one.
18690 (dump_inline_edge_summary): Turn to ...
18691 (dump_ipa_call_summary): ... this one.
18692 (estimate_function_body_sizes): Update.
18693 (inline_update_callee_summaries): Update.
18694 (remap_edge_change_prob): Update.
18695 (remap_edge_summaries): Update.
18696 (inline_merge_summary): Update.
18697 (do_estimate_edge_time): Update.
18698 (inline_generate_summary): Update.
18699 (inline_read_section): Update.
18700 (inline_read_summary): Update.
18701 (inline_free_summary): Update.
18702 * ipa-inline.c (can_inline_edge_p): Update.
18703 (compute_inlined_call_time): Update.
18704 (want_inline_small_function_p): Update.
18705 (edge_badness): Update.
18706 (early_inliner): Update.
18707 * ipa-inline.h (inline_edge_summary): Turn to ...
18708 (ipa_call_summary): ... this one.
18709 (ipa_call_summary_t): New class.
18710 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
18711 (ipa_call_summaries): New.
18712 (inline_edge_summary): Remove.
18713 (estimate_edge_growth): Update.
18714 * ipa-profile.c (ipa_propagate_frequency_1): Update.
18715 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
18716 * ipa-split.c (execute_split_functions): Update.
18717 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
18719 2017-05-19 Richard Biener <rguenther@suse.de>
18721 PR middle-end/80764
18722 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
18724 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
18726 * config/rs6000/rs6000.c (struct machine_function): Add field
18727 fpr_is_wrapped_separately.
18728 (rs6000_get_separate_components): Use 64 components. Handle the
18729 new FPR components.
18730 (rs6000_components_for_bb): Handle the FPR components.
18731 (rs6000_emit_prologue_components): Handle the FPR components.
18732 (rs6000_emit_epilogue_components): Handle the FPR components.
18733 (rs6000_set_handled_components): Handle the FPR components.
18734 (rs6000_emit_prologue): Don't output prologue code for those FPRs
18735 that are already separately shrink-wrapped.
18736 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
18737 that are already separately shrink-wrapped.
18739 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
18742 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
18745 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
18746 (define_peephole2 for Altivec d-form load): Add peepholes to catch
18747 cases where the register allocator uses a move and an offsettable
18748 memory operation to/from a FPR register on ISA 2.06/2.07.
18749 (define_peephole2 for Altivec d-form store): Likewise.
18751 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
18754 * config/i386/mmx.md (*mov<mode>_internal): Enable
18755 alternatives 11, 12, 13 and 14 also for 32bit targets.
18756 Remove alternatives 15, 16, 17 and 18.
18757 * config/i386/sse.md (vec_concatv2di): Change
18758 alternative (!x, *y) to (x, ?!*Yn).
18760 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
18762 * dumpfile.h (enum dump_kind): Remove stray comma.
18764 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18766 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
18767 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
18768 predicate::num_conditions
18769 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
18770 (CHANGED): turn into predicate::changed.
18771 (agg_position_info): Move to ipa-predicate.h
18772 (add_condition, predicate::add_clause, predicate::operator &=,
18773 predicate::or_with, predicate::evaluate, predicate::probability,
18774 dump_condition, dump_clause, predicate::dump,
18775 predicate::remap_after_duplication, predicate::remap_after_inlining,
18776 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
18777 (evaluate_conditions_for_known_args): Update.
18778 (set_cond_stmt_execution_predicate): Update.
18779 * ipa-inline.h: Include ipa-predicate.h
18780 (condition, inline_param_summary, conditions, agg_position_info,
18781 predicate): Move to ipa-predicate.h
18782 * ipa-predicate.c: New file.
18783 * ipa-predicate.h: New file.
18785 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
18787 * final.c (leaf_function_p): Check we are not in a sequence.
18789 2017-05-18 Martin Liska <mliska@suse.cz>
18791 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
18792 * dumpfile.c (dump_register): Use new enum dump_kind.
18793 (get_dump_file_name): Likewise.
18794 (dump_enable_all): Likewise.
18795 (dump_switch_p_1): Likewise.
18796 (enable_rtl_dump_file): Remove usage of TDF_RTL.
18797 * dumpfile.h (enum dump_kind): New enum type.
18798 (struct dump_file_info): Create constructor and
18799 format fields and comments.
18800 * passes.c (pass_manager::register_one_dump_file):
18802 * statistics.c (statistics_early_init): Likewise.
18803 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
18804 TDF_TREE with TDF_SLIM.
18805 (gather_memory_references_ref): Likewise.
18807 2017-05-18 Martin Liska <mliska@suse.cz>
18809 * vec.h (struct vnull): Use it.
18811 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
18813 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
18814 (true_predicate, false_predicate, true_predicate_p,
18815 false_predicate_p): Remove.
18816 (single_cond_predicate, not_inlined_predicate): Turn to member function
18818 (add_condition): Update.
18819 (add_clause): Turn to...
18820 (predicate::add_clause): ... this one; update; allow passing NULL
18822 (and_predicates): Turn to ...
18823 (predicate::operator &=): ... this one.
18824 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
18825 (or_predicates): Turn to ...
18826 (predicate::or_with): ... this one.
18827 (evaluate_predicate): Turn to ...
18828 (predicate::evaluate): ... this one.
18829 (predicate_probability): Turn to ...
18830 (predicate::probability): ... this one.
18831 (dump_condition): Update.
18832 (dump_predicate): Turn to ...
18833 (predicate::dump): ... this one.
18834 (account_size_time): Update.
18835 (edge_set_predicate): Update.
18836 (set_hint_predicate): UPdate.
18837 (evaluate_conditions_for_known_args): Update.
18838 (evaluate_properties_for_edge): Update.
18839 (remap_predicate_after_duplication): Turn to...
18840 (predicate::remap_after_duplication): ... this one.
18841 (remap_hint_predicate_after_duplication): Update.
18842 (inline_summary_t::duplicate): UPdate.
18843 (dump_inline_edge_summary): Update.
18844 (dump_inline_summary): Update.
18845 (set_cond_stmt_execution_predicate): Update.
18846 (set_switch_stmt_execution_predicate): Update.
18847 (compute_bb_predicates): Update.
18848 (will_be_nonconstant_expr_predicate): Update.
18849 (will_be_nonconstant_predicate): Update.
18850 (phi_result_unknown_predicate): Update.
18851 (predicate_for_phi_result): Update.
18852 (array_index_predicate): Update.
18853 (estimate_function_body_sizes): Update.
18854 (estimate_node_size_and_time): Update.
18855 (estimate_ipcp_clone_size_and_time): Update.
18856 (remap_predicate): Rename to ...
18857 (predicate::remap_after_inlining): ... this one.
18858 (remap_hint_predicate): Update.
18859 (inline_merge_summary): Update.
18860 (inline_update_overall_summary): Update.
18861 (estimate_size_after_inlining): Update.
18862 (read_predicate): Rename to ...
18863 (predicate::stream_in): ... this one.
18864 (read_inline_edge_summary): Update.
18865 (write_predicate): Rename to ...
18866 (predicate::stream_out): ... this one.
18867 (write_inline_edge_summary): Update.
18868 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
18869 (clause_t): Turn to uint32_t
18870 (predicate): Turn to class; implement constructor and operators
18872 (size_time_entry): Update.
18873 (inline_summary): Update.
18874 (inline_edge_summary): Update.
18876 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
18878 * fold-const.c (fold_binary_loc): Move transformation...
18879 * match.pd (C - X CMP X): ... here.
18881 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
18883 * config/sparc/sparc.c (sparc_option_override): Set function
18884 alignment for -mcpu=niagara7 to 64 to match the I$ line.
18885 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
18887 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
18889 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
18891 2017-05-18 Marek Polacek <polacek@redhat.com>
18894 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
18895 (pass_ubsan::execute): Call gimple_assign_single_p instead of
18896 gimple_assign_load_p.
18898 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
18900 PR middle-end/80692
18901 * real.c (do_compare): Give decimal_do_compare preference over
18902 comparing just the signs.
18904 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
18906 * doc/md.texi (Canonicalization of Instructions): Describe the
18907 canonical form of instructions that inherently set a condition
18910 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
18912 PR middle-end/80775
18913 * tree-cfg.c: Move deletion of unreachable case statements to after
18914 the merging of consecutive case labels.
18916 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
18918 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
18919 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
18920 restoring of callee-saved registers.
18922 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
18924 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
18925 * config/visium/visium.c (single_set_and_flags): Likewise.
18926 * config/visium/visium.md (Substitutions): Likewise.
18928 2017-05-17 Martin Liska <mliska@suse.cz>
18930 * cfg.c: Introduce dump_flags_t type and
18931 use it instead of int type.
18933 * cfghooks.c: Likewise.
18934 * cfghooks.h (struct cfg_hooks): Likewise.
18935 * cfgrtl.c: Likewise.
18936 * cfgrtl.h: Likewise.
18937 * cgraph.c (cgraph_node::get_body): Likewise.
18938 * coretypes.h: Likewise.
18939 * domwalk.c: Likewise.
18940 * domwalk.h: Likewise.
18941 * dumpfile.c (struct dump_option_value_info): Likewise.
18942 (dump_enable_all): Likewise.
18943 (dump_switch_p_1): Likewise.
18944 (opt_info_switch_p): Likewise.
18945 * dumpfile.h (enum tree_dump_index): Likewise.
18946 (struct dump_file_info): Likewise.
18947 * genemit.c: Likewise.
18948 * generic-match-head.c: Likewise.
18949 * gengtype.c (open_base_files): Likewise.
18950 * gimple-pretty-print.c: Likewise.
18951 * gimple-pretty-print.h: Likewise.
18952 * graph.c (print_graph_cfg): Likewise.
18953 * graphite-scop-detection.c (dot_all_sese): Likewise.
18954 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
18955 * loop-unroll.c (report_unroll): Likewise.
18956 * passes.c (pass_manager::register_one_dump_file): Likewise.
18957 * print-tree.c: Likewise.
18958 * statistics.c: Likewise.
18959 * tree-cfg.c: Likewise.
18960 * tree-cfg.h: Likewise.
18961 * tree-dfa.c: Likewise.
18962 * tree-dfa.h: Likewise.
18963 * tree-dump.c (dump_function): Likewise.
18964 * tree-dump.h (struct dump_info): Likewise.
18965 * tree-pretty-print.c: Likewise.
18966 * tree-pretty-print.h: Likewise.
18967 * tree-ssa-live.c: Likewise.
18968 * tree-ssa-live.h: Likewise.
18969 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
18970 * tree-vect-loop.c: Likewise.
18971 * tree-vect-slp.c: Likewise.
18973 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
18974 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18976 PR tree-optimization/80457
18977 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
18978 of all arguments to a statement as scalar_to_vec operations.
18979 (vectorizable_call): Adjust call to vect_model_simple_cost for
18981 (vectorizable_conversion): Likewise.
18982 (vectorizable_assignment): Likewise.
18983 (vectorizable_shift): Likewise.
18984 (vectorizable_operation): Likewise.
18985 (vectorizable_comparison): Likewise.
18986 (vect_is_simple_cond): Record the def types for operands.
18987 (vectorizable_condition): Likewise, call vect_model_simple_cost.
18988 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
18989 for statement argument count.
18991 2017-05-16 Carl Love <cel@us.ibm.com>
18993 * config/rs6000/rs6000-c: Add support for built-in functions
18994 vector unsigned long long vec_bperm (vector unsigned long long,
18995 vector unsigned char)
18996 vector signed long long vec_mule (vector signed int,
18998 vector unsigned long long vec_mule (vector unsigned int,
18999 vector unsigned int)
19000 vector signed long long vec_mulo (vector signed int,
19002 vector unsigned long long vec_mulo (vector unsigned int,
19003 vector unsigned int)
19004 vector signed char vec_sldw (vector signed char,
19005 vector signed char,
19007 vector unsigned char vec_sldw (vector unsigned char,
19008 vector unsigned char,
19010 vector signed short vec_sldw (vector signed short,
19011 vector signed short,
19013 vector unsigned short vec_sldw (vector unsigned short,
19014 vector unsigned short,
19016 vector signed int vec_sldw (vector signed int,
19019 vector unsigned int vec_sldw (vector unsigned int,
19020 vector unsigned int,
19022 vector signed long long vec_sldw (vector signed long long,
19023 vector signed long long,
19025 vector unsigned long long vec_sldw (vector unsigned long long,
19026 vector unsigned long long,
19028 * config/rs6000/rs6000-c: Add support for built-in functions
19029 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
19030 * config/rs6000/altivec.h: Add defintion for vec_sldw.
19031 * doc/extend.texi: Update the built-in documentation for the
19032 new built-in functions.
19034 2017-05-16 Marek Polacek <polacek@redhat.com>
19038 * tree.c (save_expr): Don't fold the expression.
19040 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
19042 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
19043 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
19044 and (?*y,m). Update insn attributes.
19046 2017-05-16 Martin Liska <mliska@suse.cz>
19048 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
19049 flags argument of print_gimple_stmt, print_gimple_expr,
19050 print_generic_stmt and print_generic_expr.
19051 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
19052 * coretypes.h: Likewise.
19053 * except.c (dump_eh_tree): Likewise.
19054 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
19055 * gimple-pretty-print.h: Likewise.
19056 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
19057 (backprop::push_to_worklist): Likewise.
19058 (backprop::pop_from_worklist): Likewise.
19059 (backprop::process_use): Likewise.
19060 (backprop::intersect_uses): Likewise.
19061 (note_replacement): Likewise.
19062 * gimple-ssa-store-merging.c
19063 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
19064 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
19065 (pass_store_merging::execute): Likewise.
19066 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
19067 (ssa_base_cand_dump_callback): Likewise.
19068 (dump_incr_vec): Likewise.
19069 (replace_refs): Likewise.
19070 (replace_mult_candidate): Likewise.
19071 (create_add_on_incoming_edge): Likewise.
19072 (create_phi_basis): Likewise.
19073 (insert_initializers): Likewise.
19074 (all_phi_incrs_profitable): Likewise.
19075 (introduce_cast_before_cand): Likewise.
19076 (replace_one_candidate): Likewise.
19077 * gimplify.c (gimplify_expr): Likewise.
19078 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
19079 (set_rename): Likewise.
19080 (rename_uses): Likewise.
19081 (copy_loop_phi_nodes): Likewise.
19082 (add_close_phis_to_merge_points): Likewise.
19083 (copy_loop_close_phi_args): Likewise.
19084 (copy_cond_phi_args): Likewise.
19085 (graphite_copy_stmts_from_block): Likewise.
19086 (translate_pending_phi_nodes): Likewise.
19087 * graphite-poly.c (print_pdr): Likewise.
19088 (dump_gbb_cases): Likewise.
19089 (dump_gbb_conditions): Likewise.
19090 (print_scop_params): Likewise.
19091 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
19092 (build_cross_bb_scalars_use): Likewise.
19093 (gather_bbs::before_dom_children): Likewise.
19094 * hsa-dump.c (dump_hsa_immed): Likewise.
19095 * ipa-cp.c (print_ipcp_constant_value): Likewise.
19096 (get_replacement_map): Likewise.
19097 * ipa-inline-analysis.c (dump_condition): Likewise.
19098 (estimate_function_body_sizes): Likewise.
19099 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
19100 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
19101 * ipa-prop.c (ipa_dump_param): Likewise.
19102 (ipa_print_node_jump_functions_for_edge): Likewise.
19103 (ipa_modify_call_arguments): Likewise.
19104 (ipa_modify_expr): Likewise.
19105 (ipa_dump_param_adjustments): Likewise.
19106 (ipa_dump_agg_replacement_values): Likewise.
19107 (ipcp_modif_dom_walker::before_dom_children): Likewise.
19108 * ipa-pure-const.c (check_stmt): Likewise.
19109 (pass_nothrow::execute): Likewise.
19110 * ipa-split.c (execute_split_functions): Likewise.
19111 * omp-offload.c (dump_oacc_loop_part): Likewise.
19112 (dump_oacc_loop): Likewise.
19113 * trans-mem.c (tm_log_emit): Likewise.
19114 (tm_memopt_accumulate_memops): Likewise.
19115 (dump_tm_memopt_set): Likewise.
19116 (dump_tm_memopt_transform): Likewise.
19117 * tree-cfg.c (gimple_verify_flow_info): Likewise.
19118 (print_loop): Likewise.
19119 * tree-chkp-opt.c (chkp_print_addr): Likewise.
19120 (chkp_gather_checks_info): Likewise.
19121 (chkp_get_check_result): Likewise.
19122 (chkp_remove_check_if_pass): Likewise.
19123 (chkp_use_outer_bounds_if_possible): Likewise.
19124 (chkp_reduce_bounds_lifetime): Likewise.
19125 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
19126 (chkp_mark_completed_bounds): Likewise.
19127 (chkp_register_incomplete_bounds): Likewise.
19128 (chkp_mark_invalid_bounds): Likewise.
19129 (chkp_maybe_copy_and_register_bounds): Likewise.
19130 (chkp_build_returned_bound): Likewise.
19131 (chkp_get_bound_for_parm): Likewise.
19132 (chkp_build_bndldx): Likewise.
19133 (chkp_get_bounds_by_definition): Likewise.
19134 (chkp_generate_extern_var_bounds): Likewise.
19135 (chkp_get_bounds_for_decl_addr): Likewise.
19136 * tree-chrec.c (chrec_apply): Likewise.
19137 * tree-data-ref.c (dump_data_reference): Likewise.
19138 (dump_subscript): Likewise.
19139 (dump_data_dependence_relation): Likewise.
19140 (analyze_overlapping_iterations): Likewise.
19141 * tree-inline.c (expand_call_inline): Likewise.
19142 (tree_function_versioning): Likewise.
19143 * tree-into-ssa.c (dump_defs_stack): Likewise.
19144 (dump_currdefs): Likewise.
19145 (dump_names_replaced_by): Likewise.
19146 (dump_update_ssa): Likewise.
19147 (update_ssa): Likewise.
19148 * tree-object-size.c (pass_object_sizes::execute): Likewise.
19149 * tree-parloops.c (build_new_reduction): Likewise.
19150 (try_create_reduction_list): Likewise.
19151 (ref_conflicts_with_region): Likewise.
19152 (oacc_entry_exit_ok_1): Likewise.
19153 (oacc_entry_exit_single_gang): Likewise.
19154 * tree-pretty-print.h: Likewise.
19155 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
19156 (get_scalar_evolution): Likewise.
19157 (add_to_evolution): Likewise.
19158 (get_loop_exit_condition): Likewise.
19159 (analyze_evolution_in_loop): Likewise.
19160 (analyze_initial_condition): Likewise.
19161 (analyze_scalar_evolution): Likewise.
19162 (instantiate_scev): Likewise.
19163 (number_of_latch_executions): Likewise.
19164 (gather_chrec_stats): Likewise.
19165 (final_value_replacement_loop): Likewise.
19166 (scev_const_prop): Likewise.
19167 * tree-sra.c (dump_access): Likewise.
19168 (disqualify_candidate): Likewise.
19169 (create_access): Likewise.
19170 (reject): Likewise.
19171 (maybe_add_sra_candidate): Likewise.
19172 (create_access_replacement): Likewise.
19173 (analyze_access_subtree): Likewise.
19174 (analyze_all_variable_accesses): Likewise.
19175 (sra_modify_assign): Likewise.
19176 (initialize_constant_pool_replacements): Likewise.
19177 (find_param_candidates): Likewise.
19178 (decide_one_param_reduction): Likewise.
19179 (replace_removed_params_ssa_names): Likewise.
19180 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
19181 * tree-ssa-copy.c (dump_copy_of): Likewise.
19182 (copy_prop_visit_cond_stmt): Likewise.
19183 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
19184 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
19185 (record_equivalences_from_stmt): Likewise.
19186 * tree-ssa-dse.c (compute_trims): Likewise.
19187 (delete_dead_call): Likewise.
19188 (delete_dead_assignment): Likewise.
19189 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
19190 (forward_propagate_into_cond): Likewise.
19191 (pass_forwprop::execute): Likewise.
19192 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
19193 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
19195 (move_computations_worker): Likewise.
19196 (execute_sm): Likewise.
19197 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
19198 (remove_exits_and_undefined_stmts): Likewise.
19199 (remove_redundant_iv_tests): Likewise.
19200 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
19201 (adjust_iv_update_pos): Likewise.
19202 * tree-ssa-math-opts.c (bswap_replace): Likewise.
19203 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
19204 (value_replacement): Likewise.
19205 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
19206 * tree-ssa-pre.c (print_pre_expr): Likewise.
19207 (get_representative_for): Likewise.
19208 (create_expression_by_pieces): Likewise.
19209 (insert_into_preds_of_block): Likewise.
19210 (eliminate_insert): Likewise.
19211 (eliminate_dom_walker::before_dom_children): Likewise.
19212 (eliminate): Likewise.
19213 (remove_dead_inserted_code): Likewise.
19214 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
19215 * tree-ssa-reassoc.c (get_rank): Likewise.
19216 (eliminate_duplicate_pair): Likewise.
19217 (eliminate_plus_minus_pair): Likewise.
19218 (eliminate_not_pairs): Likewise.
19219 (undistribute_ops_list): Likewise.
19220 (eliminate_redundant_comparison): Likewise.
19221 (update_range_test): Likewise.
19222 (optimize_range_tests_var_bound): Likewise.
19223 (optimize_vec_cond_expr): Likewise.
19224 (rewrite_expr_tree): Likewise.
19225 (rewrite_expr_tree_parallel): Likewise.
19226 (linearize_expr): Likewise.
19227 (break_up_subtract): Likewise.
19228 (linearize_expr_tree): Likewise.
19229 (attempt_builtin_powi): Likewise.
19230 (attempt_builtin_copysign): Likewise.
19231 (transform_stmt_to_copy): Likewise.
19232 (transform_stmt_to_multiply): Likewise.
19233 (dump_ops_vector): Likewise.
19234 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
19235 (print_scc): Likewise.
19236 (set_ssa_val_to): Likewise.
19237 (visit_reference_op_store): Likewise.
19238 (visit_use): Likewise.
19239 (sccvn_dom_walker::before_dom_children): Likewise.
19240 (run_scc_vn): Likewise.
19241 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
19243 (expr_hash_elt::print): Likewise.
19244 (const_and_copies::pop_to_marker): Likewise.
19245 (const_and_copies::record_const_or_copy_raw): Likewise.
19246 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
19247 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
19248 (dump_predicates): Likewise.
19249 (find_uninit_use): Likewise.
19250 (warn_uninitialized_phi): Likewise.
19251 (pass_late_warn_uninitialized::execute): Likewise.
19252 * tree-ssa.c (verify_vssa): Likewise.
19253 (verify_ssa): Likewise.
19254 (maybe_optimize_var): Likewise.
19255 * tree-vrp.c (dump_value_range): Likewise.
19256 (dump_all_value_ranges): Likewise.
19257 (dump_asserts_for): Likewise.
19258 (register_edge_assert_for_2): Likewise.
19259 (vrp_visit_cond_stmt): Likewise.
19260 (vrp_visit_switch_stmt): Likewise.
19261 (vrp_visit_stmt): Likewise.
19262 (vrp_visit_phi_node): Likewise.
19263 (simplify_cond_using_ranges_1): Likewise.
19264 (fold_predicate_in): Likewise.
19265 (evrp_dom_walker::before_dom_children): Likewise.
19266 (evrp_dom_walker::push_value_range): Likewise.
19267 (evrp_dom_walker::pop_value_range): Likewise.
19268 (execute_early_vrp): Likewise.
19270 2017-05-16 Richard Biener <rguenther@suse.de>
19272 * dwarf2out.c (loc_list_from_tree_1): Do not create
19273 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
19275 2017-05-16 Richard Biener <rguenther@suse.de>
19277 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
19279 (note_variable_value_in_expr): If we resolved the decl ref
19280 do not push to the stack.
19282 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
19284 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
19285 operations in fast-math mode.
19286 (vaddq_f16): Likewise.
19287 (vmul_f16): Likewise.
19288 (vmulq_f16): Likewise.
19289 (vsub_f16): Likewise.
19290 (vsubq_f16): Likewise.
19291 * config/arm/neon.md (add<mode>3): New.
19293 (fma:<VH:mode>3): New. Also remove outdated comment.
19296 2017-05-16 Martin Liska <mliska@suse.cz>
19300 * ipa-devirt.c (warn_types_mismatch): Fix typo.
19301 (odr_types_equivalent_p): Likewise.
19303 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
19305 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
19307 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
19310 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
19311 non-interunit SSE move alternatives with '?'.
19312 (zero-extendsidi peephole2): New peephole to skip intermediate
19313 general register in SSE zero-extend sequence.
19315 2017-05-15 Jeff Law <law@redhat.com>
19317 * reorg.c (relax_delay_slots): Create a new variable to hold
19318 the temporary target rather than clobbering TARGET_LABEL.
19320 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
19321 missing argument to extract_bit_field call.
19322 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
19324 2017-05-15 Martin Liska <mliska@suse.cz>
19327 * gcc.c (process_command): Do not allow empty argument of -o option.
19329 2017-05-15 Renlin Li <renlin.li@arm.com>
19331 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
19332 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
19333 * config/aarch64/constraints.md (Usf): Add long call check.
19334 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
19335 (call_value): Likewise.
19336 (sibcall): Likewise.
19337 (sibcall_value): Likewise.
19339 (call_value_insn): New.
19340 (sibcall_insn): Update rtx pattern.
19341 (sibcall_value_insn): Likewise.
19342 (call_internal): Remove.
19343 (call_value_internal): Likewise.
19344 (sibcall_internal): Likewise.
19345 (sibcall_value_internal): Likewise.
19346 (call_reg): Likewise.
19347 (call_symbol): Likewise.
19348 (call_value_reg): Likewise.
19349 (call_value_symbol): Likewise.
19351 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
19354 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
19356 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
19358 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
19359 compatible with CCGOCmode and with CCZmode.
19361 2017-05-14 Martin Sebor <msebor@redhat.com>
19363 PR middle-end/77671
19364 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
19365 (gimple_fold_builtin_snprintf): Same.
19366 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
19367 (gimple_fold_builtin_snprintf): Same.
19368 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
19369 of character types.
19370 (is_call_safe): New function.
19371 (try_substitute_return_value): Call it.
19372 (try_simplify_call): New function.
19373 (pass_sprintf_length::handle_gimple_call): Call it.
19375 2017-05-14 Martin Sebor <msebor@redhat.com>
19377 PR middle-end/80669
19378 * builtins.c (expand_builtin_stpncpy): Simplify.
19380 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
19382 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
19383 * config/i386/i386.h
19384 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
19385 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
19386 (struct machine_function): Add new members call_ms2sysv,
19387 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
19388 (struct machine_frame_state): New fields sp_realigned and
19389 sp_realigned_offset.
19390 * config/i386/i386.c
19391 (enum xlogue_stub): New enum.
19392 (enum xlogue_stub_sets): New enum.
19393 (class xlogue_layout): New class.
19394 (struct ix86_frame): New fields stack_realign_allocate_offset,
19395 stack_realign_offset and outlined_save_offset. Modify comments to
19396 detail stack layout when using out-of-line stubs.
19397 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
19398 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
19399 -mcall-ms2sysv-xlogues.
19400 (stub_managed_regs): New static variable.
19401 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
19402 registers managed by out-of-line stub.
19403 (disable_call_ms2sysv_xlogues): New function.
19404 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
19405 m->call_ms2sysv when appropriate and compute frame layout for
19407 (sp_valid_at, fp_valid_at): New inline functions.
19408 (choose_basereg): New function.
19409 (choose_baseaddr): Add align parameter, use choose_basereg and modify
19411 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
19412 Use align parameter of choose_baseaddr to generated aligned SSE movs
19414 (pro_epilogue_adjust_stack): Modify to track
19415 machine_frame_state::sp_realigned.
19416 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
19417 (ix86_nsaved_sseregs): Likewise.
19418 (ix86_emit_save_regs): Likewise.
19419 (ix86_emit_save_regs_using_mov): Likewise.
19420 (ix86_emit_save_sse_regs_using_mov): Likewise.
19421 (get_scratch_register_on_entry): Likewise.
19422 (gen_frame_set): New function.
19423 (gen_frame_load): Likewise.
19424 (gen_frame_store): Likewise.
19425 (emit_outlined_ms2sysv_save): Likewise.
19426 (emit_outlined_ms2sysv_restore): Likewise.
19427 (ix86_expand_prologue): Modify stack re-alignment code and call
19428 emit_outlined_ms2sysv_save when appropriate.
19429 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
19430 parameter rtx_insn *insn, which allows the function to be used to only
19431 generate the notes.
19432 (ix86_expand_epilogue): Modify validity checks of frame and stack
19433 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
19434 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
19435 * config/i386/predicates.md
19436 (save_multiple): New predicate.
19437 (restore_multiple): Likewise.
19438 * config/i386/sse.md
19439 (save_multiple<mode>): New pattern.
19440 (save_multiple_realign<mode>): Likewise.
19441 (restore_multiple<mode>): Likewise.
19442 (restore_multiple_and_return<mode>): Likewise.
19443 (restore_multiple_leave_return<mode>): Likewise.
19444 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
19446 2017-05-14 Julia Koval <julia.koval@intel.com>
19448 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
19449 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
19450 (__builtin_ia32_xsetbv): New builtins.
19451 * config/i386/i386.c (ix86_expand_special_args_builtin):
19453 (ix86_expand_builtin): Special expand for new intrinsics.
19454 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
19455 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
19456 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
19458 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19460 * cfganal.c (inverted_post_order_compute): Change argument type
19462 * cfganal.h (inverted_post_order_compute): Adjust prototype.
19463 * df-core.c (rest_of_handle_df_initialize): Adjust.
19464 (rest_of_handle_df_finish): Likewise.
19465 (df_analyze_1): Likewise.
19466 (df_analyze): Likewise.
19467 (loop_inverted_post_order_compute): Change argument to be a vec *.
19468 (df_analyze_loop): Adjust.
19469 (df_get_n_blocks): Likewise.
19470 (df_get_postorder): Likewise.
19471 * df.h (struct df_d): Change field to be a vec.
19472 * lcm.c (compute_laterin): Adjust.
19473 (compute_available): Likewise.
19474 * lra-lives.c (lra_create_live_ranges_1): Likewise.
19475 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
19476 * tree-ssa-pre.c (compute_antic): Likewise.
19478 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19480 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
19481 (depth_first_search::depth_first_search): Change structure init
19482 function to this constructor.
19483 (depth_first_search::add_bb): Rename function to this member.
19484 (depth_first_search::execute): Likewise.
19485 (flow_dfs_compute_reverse_finish): Adjust.
19487 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19489 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
19490 (longest_simple_path): Likewise.
19491 * shrink-wrap.c (spread_components): Likewise.
19492 (disqualify_problematic_components): Likewise.
19493 (emit_common_heads_for_components): Likewise.
19494 (emit_common_tails_for_components): Likewise.
19495 (insert_prologue_epilogue_for_components): Likewise.
19497 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19499 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
19502 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19504 * df-core.c (df_set_blocks): Start using auto_bitmap.
19505 (df_compact_blocks): Likewise.
19506 * df-problems.c (df_rd_confluence_n): Likewise.
19507 * df-scan.c (df_insn_rescan_all): Likewise.
19508 (df_process_deferred_rescans): Likewise.
19509 (df_update_entry_block_defs): Likewise.
19510 (df_update_exit_block_uses): Likewise.
19511 (df_entry_block_bitmap_verify): Likewise.
19512 (df_exit_block_bitmap_verify): Likewise.
19513 (df_scan_verify): Likewise.
19514 * lra-constraints.c (lra_constraints): Likewise.
19515 (undo_optional_reloads): Likewise.
19516 (lra_undo_inheritance): Likewise.
19517 * lra-remat.c (calculate_gen_cands): Likewise.
19518 (do_remat): Likewise.
19519 * lra-spills.c (assign_spill_hard_regs): Likewise.
19520 (spill_pseudos): Likewise.
19521 * tree-ssa-pre.c (bitmap_set_and): Likewise.
19522 (bitmap_set_subtract_values): Likewise.
19524 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19526 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
19527 management with auto_bitmap.
19528 (fix_inter_tick): Likewise.
19529 (fix_recovery_deps): Likewise.
19530 * ira.c (add_store_equivs): Likewise.
19531 (find_moveable_pseudos): Likewise.
19532 (split_live_ranges_for_shrink_wrap): Likewise.
19533 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
19534 (rtx_reuse_manager::seen_def_p): Likewise.
19535 (rtx_reuse_manager::set_seen_def): Likewise.
19536 * print-rtl.h (class rtx_reuse_manager): Likewise.
19538 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19540 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
19542 (migrate_btr_def): Likewise.
19543 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
19544 * df-core.c (loop_post_order_compute): Likewise.
19545 (loop_inverted_post_order_compute): Likewise.
19546 * hsa-common.h: Likewise.
19547 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
19548 * init-regs.c (initialize_uninitialized_regs): Likewise.
19549 * ipa-inline.c (resolve_noninline_speculation): Likewise.
19550 (inline_small_functions): Likewise.
19551 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
19552 * ira.c (combine_and_move_insns): Likewise.
19553 (build_insn_chain): Likewise.
19554 * loop-invariant.c (find_invariants): Likewise.
19555 * lower-subreg.c (propagate_pseudo_copies): Likewise.
19556 * predict.c (tree_predict_by_opcode): Likewise.
19557 (predict_paths_leading_to): Likewise.
19558 (predict_paths_leading_to_edge): Likewise.
19559 (estimate_loops_at_level): Likewise.
19560 (estimate_loops): Likewise.
19561 * shrink-wrap.c (try_shrink_wrapping): Likewise.
19562 (spread_components): Likewise.
19563 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
19564 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
19565 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
19566 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
19567 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
19568 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
19569 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
19570 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
19571 (mark_threaded_blocks): Likewise.
19572 (thread_through_all_blocks): Likewise.
19573 * tree-ssa.c (verify_ssa): Likewise.
19574 (execute_update_addresses_taken): Likewise.
19575 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
19577 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19579 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
19581 (post_order_compute): Likewise.
19582 (inverted_post_order_compute): Likewise.
19583 (pre_and_rev_post_order_compute_fn): Likewise.
19585 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19587 * genrecog.c (int_set::int_set): Explicitly construct our
19588 auto_vec base class.
19589 * vec.h (auto_vec::auto_vec): New constructor.
19591 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19593 * bitmap.h (class auto_bitmap): New constructor taking
19594 bitmap_obstack * argument.
19596 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
19598 * bitmap.h (class auto_bitmap): Change type of m_bits to
19599 bitmap_head, and adjust ctor / dtor and member operators.
19601 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
19603 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
19604 when returned register mode doesn't match original mode.
19606 2017-05-12 Jeff Law <law@redhat.com>
19607 Jakub Jelinek <jakub@redhat.com>
19609 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
19610 we look for cc setter after the compare-elim changes.
19611 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
19612 within the vector to match what compare-elim now expects.
19613 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
19614 (xorsi3_flags, one_cmplsi2_flags): Likewise.
19616 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
19617 after the compare-elim changes.
19618 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
19619 the vector to match what compare-elim now expects.
19620 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
19621 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
19622 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
19623 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
19624 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
19626 * config/visium/visium.c (single_set_and_flags): Fix where
19627 we look for cc setter after the compare-elim changes.
19628 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
19629 with the vector to match what compare-elim now expects.
19630 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
19631 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
19632 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
19633 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
19634 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
19635 (neg<mode>2_insn_set_overflow): Likewise.
19637 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
19639 PR middle-end/79794
19640 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
19641 maybe_expand_insn call, set ops[0].target. If still set after call,
19642 set alt_rtl. Add extra arg to recursive calls.
19643 (extract_bit_field): Add alt_rtl argument. Pass to
19645 * expmed.h (extract_bit_field): Fix prototype.
19646 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
19647 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
19648 to extract_bit_field_calls.
19649 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
19650 Pass alt_rtl to extract_bit_field calls.
19651 * calls.c (store_unaligned_arguments_into_psuedos)
19652 load_register_parameters): Pass extra NULL to extract_bit_field calls.
19653 * optabs.c (maybe_legitimize_operand): Clear op->target when call
19655 * optabs.h (struct expand_operand): Add target bitfield.
19657 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19659 * compare-elim.c (try_eliminate_compare): Canonicalize
19660 operation with embedded compare to
19661 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
19662 (set (reg) (operation)].
19664 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
19666 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
19669 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
19670 cost of adding a carry flag for ADC instruction.
19671 [case MINUS]: Ignore the cost of subtracting a carry flag
19672 for SBB instruction.
19674 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
19676 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
19678 * config/rs6000/bmiintrin.h: New file.
19679 * config/rs6000/bmi2intrin.h: New file.
19680 * config/rs6000/x86intrin.h: New file.
19682 2017-05-12 Jeff Law <law@redhat.com>
19684 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
19687 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
19689 PR middle-end/80707
19690 * tree-cfg.c: Remove cfg edges of unreachable case statements.
19692 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19694 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19695 early expansion of vector divide builtins.
19696 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
19697 builtins identified as having unsigned arguments.
19699 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19701 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
19702 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
19703 expansion of vector logical operations (and, andc, or, xor,
19706 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19708 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
19709 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
19711 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
19713 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
19714 early GIMPLE expansion of vector multiplies.
19716 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19718 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
19719 TARGET_HAVE_MOVT conditional.
19720 (movt splitter): Likewise.
19722 2017-05-12 Richard Biener <rguenther@suse.de>
19724 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
19725 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
19726 Fold all stmts not inplace.
19728 2017-05-12 Richard Biener <rguenther@suse.de>
19730 PR tree-optimization/80713
19731 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
19732 inserted_exprs bit for not removed stmts.
19734 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
19736 PR middle-end/69921
19737 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
19738 parallelized" attribute for parallelized OpenACC kernels.
19739 * omp-offload.c (execute_oacc_device_lower): Use it.
19741 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
19742 Set "oacc kernels" attribute.
19743 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
19744 parameter. Adjust all users.
19745 (oacc_fn_attrib_kernels_p): Remove function.
19746 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
19747 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
19748 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
19749 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
19750 assert "oacc kernels" attribute is set.
19752 2017-05-11 Carl Love <cel@us.ibm.com>
19754 * config/rs6000/rs6000-c: Add support for built-in functions
19755 vector unsigned char vec_popcnt (vector signed char)
19756 vector unsigned char vec_popcnt (vector unsigned char)
19757 vector unsigned short vec_popcnt (vector signed short)
19758 vector unsigned short vec_popcnt (vector unsigned short)
19759 vector unsigned int vec_popcnt (vector signed int)
19760 vector unsigned int vec_popcnt (vector unsigned int)
19761 vector unsigned long long vec_popcnt (vector signed long long)
19762 vector unsigned long long vec_popcnt (vector unsigned long long)
19763 vector signed long long vec_slo (vector signed long long,
19764 vector signed char)
19765 vector signed long long vec_slo (vector signed long long,
19766 vector unsigned char)
19767 vector unsigned long long vec_slo (vector unsigned long long,
19768 vector signed char)
19769 vector unsigned long long vec_slo (vector unsigned long long,
19770 vector unsigned char)
19771 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
19772 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
19773 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
19774 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
19775 * doc/extend.texi: Update the built-in documentation file for the
19776 new built-in functions.
19778 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
19780 * attribs.h (sorted_attr_string): Move machine independent
19781 functions for target clone support from the i386 port to common
19782 code. Rename ix86_function_versions to common_function_versions.
19783 Rename make_name to make_unique_name.
19784 (common_function_versions): Likewise.
19785 (make_unique_name): Likewise.
19786 (make_dispatcher_decl): Likewise.
19787 (is_function_default_version): Likewise.
19788 * attribs.c (attr_strcmp): Likewise.
19789 (sorted_attr_string): Likewise.
19790 (common_function_versions): Likewise.
19791 (make_unique_name): Likewise.
19792 (make_dispatcher_decl): Likewise.
19793 (is_function_default_version): Likewise.
19794 * config/i386/i386.c (attr_strcmp): Likewise.
19795 (sorted_attr_string): Likewise.
19796 (ix86_function_versions): Likewise.
19797 (make_name): Likewise.
19798 (make_dispatcher_decl): Likewise.
19799 (is_function_default_version): Likewise.
19800 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
19802 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19805 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
19806 Account for direct move costs for vec_construct of integer
19809 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
19812 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
19813 (UNSPEC_STX_ATOMIC): Ditto.
19814 (loaddi_via_sse): New insn.
19815 (storedi_via_sse): Ditto.
19816 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
19817 Update corresponding peephole2 patterns.
19818 (atomic_storedi_fpu): Ditto.
19820 2017-05-11 Julia Koval <julia.koval@intel.com>
19822 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
19823 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
19825 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
19826 (__builtin_ia32_rsqrt14ss_mask): New builtins.
19827 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
19829 2017-05-11 Nathan Sidwell <nathan@acm.org>
19831 * graphite-poly.c: Include dumpfile.h.
19833 * dumpfle.h (dump_function): Declare here ...
19834 * tree-dump.h (dump_function): ... not here.
19835 * dumpfile.c: #include tree-cfg.h.
19836 (dump_function): Move here from ...
19837 * tree-dump.c (dump_function): ... here.
19838 * gimplify.c: #include splay-tree.h, not tree-dump.h.
19839 * graphite-poly.c: Don't include tree-dump.h.
19840 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
19841 * print-tree.c: Likewise.
19842 * stor-layout.c: Likewise.
19843 * tree-nested.c: Likewise.
19845 * dumpfile.c (dump_start): Use TDF_FLAGS.
19846 (dump_enable_all): Fix TDF_KIND check thinko.
19848 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
19850 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19851 array entries to represent two legal parameterizations of the
19852 overloaded __builtin_cmpb function, as represented by the
19853 P6_OV_BUILTIN_CMPB constant.
19854 (altivec_resolve_overloaded_builtin): Add special case handling
19855 for the __builtin_cmpb function, as represented by the
19856 P6_OV_BUILTIN_CMPB constant.
19857 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
19858 (BU_P6_64BIT_2): New macro.
19859 (BU_P6_OVERLOAD_2): New macro
19860 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
19861 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
19862 (CMPB): Add overload support to represent both 32-bit and 64-bit
19863 compare-bytes function.
19864 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
19865 support for TARGET_CMPB.
19866 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
19867 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
19868 documentation of the __builtin_cmpb overloaded built-in function.
19870 2017-05-11 Richard Biener <rguenther@suse.de>
19872 PR tree-optimization/80705
19873 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
19874 bases are not vectorizable.
19876 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19878 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
19879 when counting register pressure.
19881 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19883 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
19884 (iv_ca_more_deps): Renamed to ...
19885 (iv_ca_compare_deps): ... this.
19886 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
19888 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19890 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
19892 (determine_group_iv_costs): ... here.
19893 (find_inv_vars_cb): Record inv var if it's not recorded before.
19895 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19897 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
19898 (get_shiftadd_cost): Ditto.
19900 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19902 * tree-ssa-address.c: Include header file.
19903 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
19905 (add_to_parts): Refactor.
19906 (addr_to_parts): New parameter. Update use of move_hint_to_base.
19907 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
19910 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19912 PR tree-optimization/53090
19913 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
19915 (extract_cond_operands): Detect condition with IV on both sides
19916 and return COMP_IV_EXPR_2.
19917 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
19918 (rewrite_use_compare): Simplify by removing call to function
19919 extract_cond_operands.
19921 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19923 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
19924 (extract_cond_operands): Detect condition comparing against non-
19925 invariant bound and return appropriate enum value.
19926 (find_interesting_uses_cond): Update use of extract_cond_operands.
19927 Handle its return value accordingly.
19928 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
19930 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19932 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
19933 nonlinear iv_use computation in loop invariant sensitive way.
19935 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19937 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
19938 (find_iv_candidates): Call relate_compare_use_with_all_cands.
19940 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19942 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
19943 (dump_cand): Support iv_cand.inv_exprs.
19944 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
19946 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
19949 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19951 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
19953 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
19954 as local function. Include necessary header files.
19955 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
19957 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19959 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
19961 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19963 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
19964 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
19965 RSHIFT_EXPR and BIT_NOT_EXPR.
19967 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19969 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
19970 (adjust_setup_cost): New parameter supporting round up adjustment.
19971 (struct address_cost_data): Delete.
19972 (force_expr_to_var_cost): Don't bound cost with spill_cost.
19973 (split_address_cost, ptr_difference_cost): Delete.
19974 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
19975 (struct ainc_cost_data): New struct.
19976 (get_address_cost_ainc): New function.
19977 (get_address_cost, get_computation_cost): Reimplement.
19978 (determine_group_iv_cost_address): Record inv_expr for all uses of
19980 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
19981 (iv_ca_has_deps): Reimplemented to ...
19982 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
19984 (iv_ca_extend): Call iv_ca_more_deps.
19986 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19988 * tree-ssa-address.c (struct mem_address): Move to header file.
19989 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
19990 * tree-ssa-address.h (struct mem_address): Move from C file.
19991 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
19993 2017-05-11 Bin Cheng <bin.cheng@arm.com>
19995 * tree-affine.h (aff_combination_type): New interface.
19996 (aff_combination_zero_p): Remove static.
19997 (aff_combination_const_p): New interface.
19998 (aff_combination_singleton_var_p): New interfaces.
20000 2017-05-11 Richard Biener <rguenther@suse.de>
20002 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
20003 Skip unreachable blocks and destinations.
20004 (eliminate): Move stmt removal and fixup ...
20005 (fini_eliminate): ... here. Skip inserted exprs.
20006 (pass_pre::execute): Move fini_pre after fini_eliminate.
20007 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
20008 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
20009 PRE to get rid of dead code that has invalid SSA form and
20010 split critical edges again.
20012 2017-05-11 Bin Cheng <bin.cheng@arm.com>
20014 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20016 2017-05-11 Richard Biener <rguenther@suse.de>
20018 * passes.c (execute_function_todo): Verify loops if they are
20019 said to be up-to-date.
20020 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
20021 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
20023 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
20026 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
20027 handle calling assemble_external ourself.
20030 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
20031 modes with zero size. Enhance comment.
20033 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20035 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
20036 built-ins for vec_xl and vec_xst with short and char pointer
20039 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20041 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
20042 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
20043 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
20044 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
20045 (_mm_maskz_min_round_ss): New intrinsics.
20046 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
20047 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
20048 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
20049 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
20050 (__builtin_ia32_minss_mask_round): New builtins.
20051 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20052 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20053 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
20055 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
20056 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
20058 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
20061 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
20063 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
20064 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
20065 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
20066 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
20067 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
20068 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
20069 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
20070 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20071 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
20072 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
20073 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
20074 (__builtin_ia32_mulss_mask_round): New builtins.
20075 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
20076 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
20077 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
20079 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
20080 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
20082 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
20085 2017-05-10 Julia Koval <julia.koval@intel.com>
20087 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
20088 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
20089 (_mm256_setr_m128i): New intrinsics.
20091 2017-05-10 Julia Koval <julia.koval@intel.com>
20093 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
20094 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
20095 (_mm_maskz_rcp14_ss): New intrinsics.
20096 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
20097 (__builtin_ia32_rcp14ss_mask): New builtins.
20098 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
20100 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
20102 PR tree-optimization/51513
20103 * tree-cfg.c (gimple_seq_unreachable_p): New function.
20104 (assert_unreachable_fallthru_edge_p): Use it.
20105 (group_case_labels_stmt): Likewise.
20106 * tree-cfg.h: Prototype it.
20107 * stmt.c: Include cfghooks.h and tree-cfg.h.
20108 (emit_case_dispatch_table) <gap_label>: New local variable.
20109 Use it to fill dispatch table gaps.
20110 Test for default_label before updating probabilities.
20111 (expand_case) <default_label>: Remove unneeded initialization.
20112 Test for unreachable default case statement and remove its edge.
20113 Set default_label accordingly.
20114 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
20116 2017-05-10 Carl Love <cel@us.ibm.com>
20118 * config/rs6000/rs6000-c: Add support for built-in functions
20119 vector signed char vec_neg (vector signed char)
20120 vector signed short int vec_neg (vector short int)
20121 vector signed int vec_neg (vector signed int)
20122 vector signed long long vec_neg (vector signed long long)
20123 vector float vec_neg (vector float)
20124 vector double vec_neg (vector double)
20125 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
20127 * config/rs6000/altivec.h: Add define for vec_neg
20128 * doc/extend.texi: Update the built-in documentation for the
20129 new built-in functions.
20131 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20133 PR tree-optimization/77644
20134 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
20136 2017-05-10 Nathan Sidwell <nathan@acm.org>
20138 * dumpfile.h (TDI_lang_all): New.
20139 (TDF_KIND): New. Renumber others
20140 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
20142 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
20144 (get_dump_file_name): Adjust suffix generation.
20145 (dump_enable_all): Use TDF_KIND.
20146 * doc/invoke.texi (-fdump-lang-all): Document.
20148 * dumpfile.h: Tabify.
20150 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
20153 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
20154 Move member access before delete.
20156 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
20158 * tree-inline.c (expand_call_inline): Split block at stmt
20161 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
20164 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
20165 are now unused after splitting mov{sf,sd}_hardfloat.
20166 (f32_lr2): Likewise.
20167 (f32_lm): Likewise.
20168 (f32_lm2): Likewise.
20169 (f32_li): Likewise.
20170 (f32_li2): Likewise.
20171 (f32_lv): Likewise.
20172 (f32_sr): Likewise.
20173 (f32_sr2): Likewise.
20174 (f32_sm): Likewise.
20175 (f32_sm2): Likewise.
20176 (f32_si): Likewise.
20177 (f32_si2): Likewise.
20178 (f32_sv): Likewise.
20179 (f32_dm): Likewise.
20180 (f32_vsx): Likewise.
20181 (f32_av): Likewise.
20182 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
20183 For movsf, order stores so the VSX stores occur before the GPR
20184 store which encourages the register allocator to use a traditional
20185 FPR instead of a GPR. For movsd, order the stores so that the GPR
20186 store comes before the VSX stores to allow the power6 to work.
20187 This is due to the power6 not having a 32-bit integer store
20188 instruction from a FPR.
20189 (movsf_hardfloat): Likewise.
20190 (movsd_hardfloat): Likewise.
20192 2017-05-09 Martin Sebor <msebor@redhat.com>
20194 PR translation/80280
20195 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
20198 PR translation/80280
20199 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
20200 data member added in r247778.
20201 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
20203 2017-05-09 Nathan Sidwell <nathan@acm.org>
20205 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
20207 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
20210 2017-05-09 Marek Polacek <polacek@redhat.com>
20212 * doc/invoke.texi: Fix typo.
20214 2017-05-09 Richard Biener <rguenther@suse.de>
20216 * tree-vrp.c (vrp_val_is_max): Adjust comment.
20217 (vrp_val_is_min): Likewise.
20218 (set_value_range_to_value): Likewise.
20219 (set_value_range_to_nonnegative): Likewise.
20220 (gimple_assign_nonzero_p): Likewise.
20221 (gimple_stmt_nonzero_p): Likewise.
20222 (vrp_int_const_binop): Likewise. Remove unreachable case.
20223 (adjust_range_with_scev): Adjust comments.
20224 (compare_range_with_value): Likewise.
20225 (extract_range_from_phi_node): Likewise.
20226 (test_for_singularity): Likewise.
20228 2017-05-09 Richard Biener <rguenther@suse.de>
20230 * tree-vrp.c (get_single_symbol): Add assert that we don't
20231 get overflowed constants as invariant part.
20232 (compare_values_warnv): Add comment before the TREE_NO_WARNING
20233 checks. Use wi::cmp instead of recursing for integer constants.
20234 (compare_values): Just ignore whether we assumed undefined
20235 overflow instead of failing the compare.
20236 (extract_range_for_var_from_comparison_expr): Add comment before the
20237 TREE_NO_WARNING sets.
20238 (test_for_singularity): Likewise.
20239 (extract_range_from_comparison): Do not disable optimization
20240 when we assumed undefined overflow.
20241 (extract_range_basic): Remove init of unused var.
20243 2017-05-09 Richard Biener <rguenther@suse.de>
20245 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
20246 (extract_range_from_multiplicative_op_1): Adjust.
20247 (extract_range_from_binary_expr_1): Use int_const_binop.
20249 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
20252 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
20253 rs6000_store_data_bypass_p in seven define_bypass directives and
20254 in several comments.
20255 * config/rs6000/rs6000-protos.h: Add prototype for
20256 rs6000_store_data_bypass_p function.
20257 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
20258 function implements slightly different (rs6000-specific) semantics
20259 than store_data_bypass_p, returning false rather than aborting
20260 with assertion error when arguments do not satisfy the
20261 requirements of store data bypass.
20262 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
20263 rs6000_store_data_bypass_p.
20265 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
20267 * config/xtensa/xtensa-protos.h
20268 (xtensa_initial_elimination_offset): New declaration.
20269 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20270 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
20271 macro definition, add case for FRAME_POINTER_REGNUM when
20272 FRAME_GROWS_DOWNWARD.
20273 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
20274 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
20275 xtensa_initial_elimination_offset.
20277 2017-05-08 Nathan Sidwell <nathan@acm.org>
20279 * doc/invoke.texi: Alphabetize -fdump options.
20281 2017-05-08 Martin Sebor <msebor@redhat.com>
20283 PR translation/80280
20284 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
20286 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
20288 * target.def (compute_frame_layout): New optional target hook.
20289 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
20290 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
20291 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
20293 * reload1.c (verify_initial_elim_offsets): Likewise.
20294 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
20295 (use_simple_return_p): Call arm_compute_frame_layout if needed.
20296 (arm_get_frame_offsets): Split up into this ...
20297 (arm_compute_frame_layout): ... and this function.
20299 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
20301 * config/aarch64/constraints.md (Usa): New constraint.
20302 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
20304 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
20306 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
20307 with_multilib_list after it has been checked.
20309 2017-05-08 Richard Biener <rguenther@suse.de>
20311 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
20312 (bitmap_set_subtract_values): Likewise.
20314 2017-05-08 Richard Biener <rguenther@suse.de>
20316 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
20317 (gimple_assign_nonzero): ... this and remove strict_overflow_p
20319 (gimple_stmt_nonzero_warnv_p): Rename to ...
20320 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
20322 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
20323 (extract_range_basic): Adjust, do not disable propagation on
20324 strict overflow sensitive simplification.
20325 (vrp_visit_cond_stmt): Likewise.
20327 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
20329 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
20330 body size unconditionally.
20332 2017-05-07 Jeff Law <law@redhat.com>
20335 2017-05-06 Jeff Law <law@redhat.com>
20336 PR tree-optimization/78496
20337 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20340 PR tree-optimization/78496
20341 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20342 (simplify_stmt_using_ranges): Call it.
20343 (vrp_dom_walker::before_dom_children): Extract equivalences
20344 from an ASSERT_EXPR with an equality comparison against a
20347 2017-05-06 Jeff Law <law@redhat.com>
20349 PR tree-optimization/78496
20350 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
20353 PR tree-optimization/78496
20354 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
20355 (simplify_stmt_using_ranges): Call it.
20356 (vrp_dom_walker::before_dom_children): Extract equivalences
20357 from an ASSERT_EXPR with an equality comparison against a
20360 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20362 * lra-constraints.c (lra_copy_reg_equiv): New function.
20363 (split_reg): Use it to copy equivalence information from the
20364 original register to the spill register.
20366 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
20368 PR rtl-optimization/75964
20369 * simplify-rtx.c (simplify_const_relational_operation): Remove
20370 invalid handling of comparisons of integer ABS.
20372 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
20374 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
20375 initialize to zero.
20376 (init_regs): Remove declaration.
20377 (function_arg_advance_32): Initialize error_p as boolean variable.
20379 2017-05-05 Nathan Sidwell <nathan@acm.org>
20381 * store-motion.c (remove_reachable_equiv_notes): Reformat long
20382 lines. Use for (;;).
20384 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20386 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
20387 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
20388 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
20389 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
20390 VF=2 that require versioning.
20392 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20394 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
20397 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20399 * diagnostic.h (diagnostic_override_option_index): Convert from
20400 macro to inline function.
20402 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20404 * diagnostic.c (last_module_changed_p): New function.
20405 (set_last_module): New function.
20406 (diagnostic_report_current_module): Convert macro usage to
20407 the above functions.
20408 * diagnostic.h (diagnostic_context::last_module): Strengthen
20409 from const line_map * to const line_map_ordinary *.
20410 (diagnostic_last_module_changed): Delete macro.
20411 (diagnostic_set_last_module): Delete macro.
20413 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20415 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
20416 with diagnostic_report_diagnostic.
20417 (diagnostic_n_impl_richloc): Likewise.
20418 * diagnostic.h (report_diagnostic): Delete macro.
20419 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
20420 with diagnostic_report_diagnostic.
20421 * substring-locations.c (format_warning_va): Likewise.
20423 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20425 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
20426 save/restor of format_spec. Move option-printing code to...
20427 (print_option_information): ...this new function, and
20428 reimplement by simply printing to the pretty_printer,
20429 rather than appending to the format string.
20431 2017-05-05 David Malcolm <dmalcolm@redhat.com>
20433 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
20434 handling logic into...
20435 (update_effective_level_from_pragmas): ...this new function.
20437 2017-05-04 Andrew Waterman <andrew@sifive.com>
20439 * config/riscv/riscv.opt (mstrict-align): New option.
20440 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
20441 (SLOW_UNALIGNED_ACCESS): Define.
20442 (riscv_slow_unaligned_access): Declare.
20443 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
20445 (riscv_slow_unaligned_access): New variable.
20446 (rocket_tune_info): Set slow_unaligned_access to true.
20447 (optimize_size_tune_info): Set slow_unaligned_access to false.
20448 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
20449 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
20450 (riscv_option_override): Set riscv_slow_unaligned_access.
20451 * doc/invoke.texi: Add -mstrict-align to RISC-V.
20453 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
20455 * config/riscv/riscv.md: Unify indentation.
20457 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20462 * config/rs6000/rs6000.md (u code attribute): Add FIX and
20464 (extendsi<mode>2): Add support for doing sign extension via
20465 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
20466 don't have ISA 3.0 instructions.
20467 (extendsi<mode>2 splitter): Likewise.
20468 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
20469 generate the normal insns since SImode can now go in vector
20470 registers. Disallow the special UNSPECs needed for previous
20471 machines to hide SImode being used. Add new insns
20472 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
20473 (fix_trunc<mode>si2_stfiwx): Likewise.
20474 (fix_trunc<mode>si2_internal): Likewise.
20475 (fixuns_trunc<mode>si2): Likewise.
20476 (fixuns_trunc<mode>si2_stfiwx): Likewise.
20477 (fctiw<u>z_<mode>_smallint): Likewise.
20478 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
20479 of floating point to 32-bit integer from doing a direct move to
20480 the GPR registers to do a store.
20481 (fctiwz_<mode>): Break long line.
20483 2017-05-05 Bin Cheng <bin.cheng@arm.com>
20485 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
20486 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
20487 (addr_list, addr_offset_valid_p): New.
20488 (split_address_groups): Check offset validity with above function.
20489 (gt-tree-ssa-loop-ivopts.h): Include header file.
20491 2017-05-05 Nathan Sidwell <nathan@acm.org>
20493 * config.gcc (arm*-*-*): Add missing 'fi'.
20495 2017-05-05 Steve Ellcey <sellcey@cavium.com>
20497 * doc/invoke.texi (-fopt-info): Explicitly say order of options
20498 included in -fopt-info does not matter.
20499 * doc/optinfo.texi (-fopt-info): Fix description of default
20500 behavour. Explicitly say order of options included in -fopt-info
20503 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20505 * config.gcc: Allow combinations of aprofile and rmprofile values for
20506 --with-multilib-list.
20507 * config/arm/t-multilib: New file.
20508 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
20509 variables. Remove setting of ISA and floating-point ABI in
20510 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
20511 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
20512 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
20513 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
20515 * config/arm/t-rmprofile: Likewise except for the matches changes.
20516 * doc/install.texi (--with-multilib-list): Document the combination of
20517 aprofile and rmprofile values and warn about pitfalls in doing that.
20519 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20521 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
20522 (movdi_aarch64): Likewise.
20524 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20526 PR tree-optimization/80632
20527 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
20529 (build_arrays): Initialize it for virtual phis.
20530 (fix_phi_nodes): Use it for virtual phis.
20532 PR tree-optimization/80558
20533 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
20534 [x, y] op z into [x op, y op z] for op & or | if conditions
20537 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
20538 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20541 * config/arm/arm.md (use_literal_pool): Remove.
20542 (64-bit immediate split): No longer takes cost into consideration
20543 if arm_disable_literal_pool is enabled.
20544 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
20545 used when arm_disable_literal_pool is enabled.
20546 (arm_max_const_double_inline_cost): Remove use of
20547 arm_disable_literal_pool.
20548 (push_minipool_fix): Add assert.
20549 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
20550 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
20551 (no_literal_pool_sf_immediate): New.
20553 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20555 PR tree-optimization/80613
20556 * tree-ssa-dce.c (propagate_necessity): Remove cases for
20557 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
20559 2017-05-05 Richard Biener <rguenther@suse.de>
20561 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
20563 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
20565 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
20566 of this flag from insn conditions due to removal from r247495.
20568 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
20570 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
20572 (arm_early_store_addr_dep_ptr): Likewise.
20573 * config/arm/aarch-common-protos.h
20574 (arm_early_load_addr_dep_ptr): Add prototype.
20575 (arm_early_store_addr_dep_ptr): Likewise.
20576 * config/arm/cortex-a53.md: Add new bypasses.
20578 2017-05-05 Jakub Jelinek <jakub@redhat.com>
20580 * tree.c (next_type_uid): Change type to unsigned.
20581 (type_hash_canon): Decrement back next_type_uid if
20582 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
20583 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
20586 2017-05-04 Martin Sebor <msebor@redhat.com>
20588 * builtins.c: Fix a trivial typo in a comment.
20590 PR middle-end/79234
20591 * builtins.c (check_sizes): Adjust to handle reading past the end.
20592 Avoid printing excessive upper bound of ranges. Use %E to print
20593 tree nodes instead of converting them to %wu.
20594 (expand_builtin_memchr): New function.
20595 (compute_dest_size): Rename...
20596 (compute_objsize): ...to this.
20597 (expand_builtin_memcpy): Adjust.
20598 (expand_builtin_mempcpy): Adjust.
20599 (expand_builtin_strcat): Adjust.
20600 (expand_builtin_strcpy): Adjust.
20601 (check_strncat_sizes): Adjust.
20602 (expand_builtin_strncat): Adjust.
20603 (expand_builtin_strncpy): Adjust and simplify.
20604 (expand_builtin_memset): Adjust.
20605 (expand_builtin_bzero): Adjust.
20606 (expand_builtin_memcmp): Adjust.
20607 (expand_builtin): Handle memcmp.
20608 (maybe_emit_chk_warning): Check strncat just once.
20610 2017-05-04 Martin Sebor <msebor@redhat.com>
20612 PR preprocessor/79214
20613 PR middle-end/79222
20614 PR middle-end/79223
20615 * builtins.c (check_sizes): Add inlining context and issue
20616 warnings even when -Wno-system-headers is set.
20617 (check_strncat_sizes): Same.
20618 (expand_builtin_strncat): Same.
20619 (expand_builtin_memmove): New function.
20620 (expand_builtin_stpncpy): Same.
20621 (expand_builtin): Handle memmove and stpncpy.
20623 2017-05-04 Bin Cheng <bin.cheng@arm.com>
20625 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
20626 which is not used any more.
20628 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20630 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
20632 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20634 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
20635 (cortexa53_tunings): Likewise.
20636 (cortexa57_tunings): Likewise.
20637 (cortexa72_tunings): Likewise.
20638 (cortexa73_tunings): Likewise.
20640 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20642 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
20643 Set loop alignment to 8.
20645 2017-05-04 Martin Sebor <msebor@redhat.com>
20647 PR translation/80280
20648 * builtins.c (expand_builtin_object_size): Add missing quoting to
20649 %D and like directives.
20650 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
20651 (hsa_type_for_tree_type): Same.
20652 (verify_function_arguments): Same.
20653 * symtab.c (symbol_table::change_decl_assembler_name): Same.
20654 * varasm.c (get_section): Same.
20657 2017-05-04 Martin Sebor <msebor@redhat.com>
20659 PR translation/80280
20660 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
20662 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
20664 * config/aarch64/aarch64.c (generic_addrcost_table):
20665 Change HI/TI mode setting.
20667 2017-05-04 Martin Jambor <mjambor@suse.cz>
20669 PR tree-optimization/80622
20670 * tree-sra.c (comes_initialized_p): New function.
20671 (build_accesses_from_assign): Only set write lazily when
20672 comes_initialized_p is false.
20673 (analyze_access_subtree): Use comes_initialized_p.
20674 (propagate_subaccesses_across_link): Assert !comes_initialized_p
20675 instead of testing for PARM_DECL.
20677 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20679 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
20680 constraint on operand 0 to allow more general addressing modes.
20681 Adjust output template.
20682 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
20684 * config/aarch64/aarch64-protos.h
20685 (aarch64_address_valid_for_prefetch_p): Declare prototype.
20686 * config/aarch64/constraints.md (Dp): New address constraint.
20687 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
20690 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
20692 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
20693 update use of estimate_ipcp_clone_size_and_time.
20694 (estimate_local_effects): Update use of
20695 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
20696 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
20697 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
20698 Return nonspecialized time.
20700 2017-05-04 Richard Biener <rguenther@suse.de>
20702 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
20703 for the last VUSE which def dominates the PHI. Directly call
20705 (get_continuation_for_phi_1): Remove.
20707 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
20709 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
20710 to explain the use of truncating division. Cap the number of
20711 iterations to the maximum given by nb_iterations_upper_bound,
20714 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
20716 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
20717 * configure: Regenerate.
20718 * config.in: Regenerate.
20719 * config/i386/driver-mingw32.c: new file.
20720 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
20721 * config.host: Link driver-mingw32.o on MinGW host.
20722 * doc/install.texi: Document new --enable-mingw-wildcard configure
20725 2017-05-04 Marek Polacek <polacek@redhat.com>
20727 PR tree-optimization/80612
20728 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
20730 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20731 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
20733 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
20734 (movt splitter): Likewise.
20735 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
20736 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
20737 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
20738 block for Thumb-1 with MOVT.
20739 (thumb2_legitimate_address_p): Move code block ...
20740 (can_avoid_literal_pool_for_label_p): ... into this new function.
20741 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
20743 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
20744 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
20745 "M-profile targets with the MOVT instruction".
20747 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
20749 * config/arm/arm-builtins.c (arm_init_builtins): Rename
20750 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
20751 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
20753 2017-05-04 Martin Liska <mliska@suse.cz>
20755 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
20756 variable cond_code.
20758 2017-05-04 Richard Biener <rguenther@suse.de>
20760 * tree.c (array_at_struct_end_p): Handle arrays at struct
20761 end with flexarrays more conservatively. Refactor and treat
20762 arrays of arrays or aggregates more strict. Fix
20763 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
20764 * tree.c (array_at_struct_end_p): Adjust prototype.
20765 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
20766 * gimple-fold.c (get_range_strlen): Likewise.
20767 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
20769 2017-05-04 Richard Biener <rguenther@suse.de>
20771 PR tree-optimization/31130
20772 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
20774 (supports_overflow_infinity): Likewise.
20775 (is_negative_overflow_infinity): Likewise.
20776 (is_positive_overflow_infinity): Likewise.
20777 (is_overflow_infinity): Likewise.
20778 (stmt_overflow_infinity): Likewise.
20779 (overflow_infinity_range_p): Likewise.
20780 (usable_range_p): Remove as always returning true.
20781 (make_overflow_infinity): Remove.
20782 (negative_overflow_infinity): Likewise.
20783 (positive_overflow_infinity): Likewise.
20784 (avoid_overflow_infinity): Likewise.
20785 (set_value_range): Adjust accordingly.
20786 (set_value_range_to_nonnegative): Likewise, remove now unused
20787 overflow_infinity arg.
20788 (vrp_operand_equal_p): Adjust.
20789 (update_value_range): Likewise.
20790 (range_int_cst_singleton_p): Likewise.
20791 (operand_less_p): Likewise.
20792 (compare_values_warnv): Likewise.
20793 (extract_range_for_var_from_comparison_expr): Likewise.
20794 (vrp_int_const_binop): Likewise.
20795 (zero_nonzero_bits_from_vr): Likewise.
20796 (extract_range_from_multiplicative_op_1): Likewise.
20797 (extract_range_from_binary_expr_1): Likewise.
20798 (extract_range_from_unary_expr): Likewise.
20799 (extract_range_from_comparison): Likewise.
20800 (extract_range_basic): Likewise.
20801 (adjust_range_with_scev): Likewise.
20802 (compare_ranges): Likewise.
20803 (compare_range_with_value): Likewise.
20804 (dump_value_range): Likewise.
20805 (test_for_singularity): Likewise, remove strict_overflow_p parameter
20807 (simplify_cond_using_ranges): Adjust.
20809 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
20811 * brig-builtins.def: Added a builtin for class_f64.
20812 * builtin-types.def: Added a builtin type needed by class_f64.
20814 2017-05-03 Jason Merrill <jason@redhat.com>
20816 * timevar.def: Add TV_CONSTEXPR.
20818 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20820 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
20822 2017-05-03 Martin Jambor <mjambor@suse.cz>
20824 * ipa-prop.c (ipa_update_after_lto_read): Removed.
20825 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
20826 * ipa-cp.c (ipcp_propagate_stage): Do not call
20827 ipa_update_after_lto_read.
20828 * ipa-inline.c (ipa_inline): Likewise.
20830 2017-05-03 Martin Jambor <mjambor@suse.cz>
20832 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
20833 tag. Added a default constructor and a destructor.
20834 (ipa_edge_args_sum_t): New class;
20835 (ipa_edge_args_sum): Declare.
20836 (ipa_edge_args_vector): Remove declaration.
20837 (IPA_EDGE_REF): Use ipa_edge_args_sum.
20838 (ipa_free_edge_args_substructures): Remove declaration.
20839 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
20840 (ipa_edge_args_info_available_for_edge_p): Likewise.
20841 * ipa-prop.c (ipa_edge_args_vector): Removed.
20842 (edge_removal_hook_holder): Likewise.
20843 (edge_duplication_hook_holder): Likewise.
20844 (ipa_edge_args_sum): New variable.
20845 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
20846 ipa_edge_args_vector.
20847 (ipa_free_edge_args_substructures): Likewise.
20848 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
20849 ipa_edge_args_vector.
20850 (ipa_edge_removal_hook): Turned into method
20851 ipa_edge_args_sum_t::remove.
20852 (ipa_edge_duplication_hook): Turned into method
20853 ipa_edge_args_sum_t::duplicate.
20854 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
20855 registering edge hooks.
20856 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
20857 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
20858 ipa_edge_args_sum instead of ipa_edge_args_vector.
20859 * ipa-profile.c (ipa_profile): Likewise.
20861 2017-05-03 Martin Jambor <mjambor@suse.cz>
20863 * symbol-summary.h (function_summary): New method exists.
20864 (function_summary::symtab_removal): Deallocate through release.
20865 (call_summary): New class.
20866 (gt_ggc_mx): New overload.
20867 (gt_pch_nx): Likewise.
20868 (gt_pch_nx): Likewise.
20870 2017-05-03 Jeff Law <law@redhat.com>
20872 PR tree-optimization/78496
20873 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
20874 from simplify_cond_using_ranges. Split off code to walk
20875 backwards through casts into ...
20876 (simplify_cond_using_ranges_2): New function.
20877 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
20878 (execute_vrp): After identifying jump threads, call
20879 simplify_cond_using_ranges_2.
20881 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
20884 * ipa-inline.h (inline_summary): Add ctor.
20885 (create_ggc): Do not use ggc_cleared_alloc.
20887 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
20888 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20890 * gcc.c (handle_braces): Support escaping in switch matching
20892 * doc/invoke.texi (Spec Files): Document it.
20893 Remove superfluous @code markup in items.
20895 2017-05-03 David Malcolm <dmalcolm@redhat.com>
20897 * diagnostic-show-locus.c (struct column_range): New struct.
20898 (get_affected_columns): New function.
20899 (get_printed_columns): New function.
20900 (struct correction): New struct.
20901 (correction::ensure_capacity): New function.
20902 (correction::ensure_terminated): New function.
20903 (struct line_corrections): New struct.
20904 (line_corrections::~line_corrections): New dtor.
20905 (line_corrections::add_hint): New function.
20906 (layout::print_trailing_fixits): Reimplement in terms of the new
20908 (selftest::test_overlapped_fixit_printing): New function.
20909 (selftest::diagnostic_show_locus_c_tests): Call it.
20911 2017-05-03 Nathan Sidwell <nathan@acm.org>
20913 Canonicalize canonical type hashing
20914 * tree.h (type_hash_canon_hash): Declare.
20915 * tree.c (type_hash_list, attribute_hash_list): Move into
20916 type_hash_canon_hash.
20917 (build_type_attribute_qual_variant): Break out hash code calc into
20918 type_hash_canon_hash.
20919 (type_hash_canon_hash): New. Generic type hash computation.
20920 (build_range_type_1, build_array_type_1, build_function_type,
20921 build_method_type_directly, build_offset_type, build_complex_type,
20922 make_vector_type): Call it.
20924 2017-05-03 Richard Biener <rguenther@suse.de>
20926 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
20927 When all DRs have unknown misaligned do not always peel
20928 when there is a store but apply the same costing model as if
20929 there were only loads.
20931 2017-05-03 Richard Biener <rguenther@suse.de>
20934 PR tree-optimization/80492
20935 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
20936 compare_base_decls returning dont-know properly.
20938 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
20940 * config/arm/iterators.md (CCSI): New mode iterator.
20941 (arch): New mode attribute.
20942 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
20943 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
20944 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
20945 code iterator for success result mode.
20946 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
20947 the corresponding new insn generators.
20949 2017-05-03 Bin Cheng <bin.cheng@arm.com>
20952 2017-05-02 Bin Cheng <bin.cheng@arm.com>
20953 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
20955 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
20957 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
20958 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
20959 (DDR_A): Wrap DDR argument in brackets.
20960 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
20961 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
20962 (DDR_REVERSED_P): Likewise.
20964 2017-05-03 Jakub Jelinek <jakub@redhat.com>
20966 PR tree-optimization/79472
20967 * tree-switch-conversion.c (struct switch_conv_info): Add
20968 contiguous_range and default_case_nonstandard fields.
20969 (collect_switch_conv_info): Compute contiguous_range and
20970 default_case_nonstandard fields, don't clear final_bb if
20971 contiguous_range and only the default case doesn't have the required
20973 (check_all_empty_except_final): Set default_case_nonstandard instead
20974 of failing if contiguous_range and the default case doesn't have empty
20976 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
20977 and only the default case doesn't have the required constants. Skip
20979 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
20980 if default_case_nonstandard.
20981 (build_constructors): Build constant 1 just once. Assert that default
20982 values aren't inserted in between cases if contiguous_range. Skip
20984 (build_arrays): Skip virtual phis.
20985 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
20986 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
20987 Handle virtual phis.
20988 (gen_inbound_check): Handle default_case_nonstandard case.
20989 (process_switch): Adjust check_final_bb caller. Call
20990 gather_default_values with the first non-default case instead of
20991 default case if default_case_nonstandard.
20993 2017-05-02 Nathan Sidwell <nathan@acm.org>
20995 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
20996 check. Fix formatting.
20998 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
21000 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
21001 errors when comparing specialized and unspecialized times.
21003 2017-05-02 David Malcolm <dmalcolm@redhat.com>
21005 * diagnostic-show-locus.c
21006 (layout::should_print_annotation_line_p): Make private.
21007 (layout::print_annotation_line): Make private.
21008 (layout::annotation_line_showed_range_p): Make private.
21009 (layout::show_ruler): Make private.
21010 (layout::print_source_line): Make private. Pass in line and
21011 line_width, rather than calling location_get_source_line. Drop
21013 (layout::print_leading_fixits): New method.
21014 (layout::print_any_fixits): Rename to...
21015 (layout::print_trailing_fixits): ...this, and make private.
21016 Don't print newline fixits.
21017 (diagnostic_show_locus): Move logic for printing one row into...
21018 (layout::print_line): ...this new function. Move the
21019 location_get_source_line call and error-handling from
21020 print_source_line to here. Call print_leading_fixits, and rename
21021 print_any_fixits to print_trailing_fixits.
21022 (selftest::test_fixit_insert_containing_newline): Update now that
21023 newlines are partially supported.
21024 (selftest::test_fixit_insert_containing_newline_2): New test.
21025 (selftest::test_fixit_replace_containing_newline): Update comments.
21026 (selftest::diagnostic_show_locus_c_tests): Call the new test.
21027 * edit-context.c (class added_line): New class.
21028 (class edited_line): Describe newline handling in comment.
21029 (edited_line::actually_edited_p): New method.
21030 (edited_line::print_content): Delete redundant decl.
21031 (edited_line::m_predecessors): New field.
21032 (edited_file::print_content): Call edited_line::print_content.
21033 (edited_file::print_diff): Update to support newlines.
21034 (edited_file::print_diff_hunk): Likewise.
21035 (edited_file::print_run_of_changed_lines): New function.
21036 (edited_file::print_diff_line): Convert to...
21037 (print_diff_line): ...this.
21038 (edited_file::get_effective_line_count): New function.
21039 (edited_line::edited_line): Initialize new field m_predecessors.
21040 (edited_line::~edited_line): Clean up m_predecessors.
21041 (edited_line::apply_fixit): Handle newlines.
21042 (edited_line::get_effective_line_count): New function.
21043 (edited_line::print_content): New function.
21044 (edited_line::print_diff_lines): New function.
21045 (selftest::test_applying_fixits_insert_containing_newline): New
21047 (selftest::test_applying_fixits_replace_containing_newline): New
21049 (selftest::insert_line): New function.
21050 (selftest::test_applying_fixits_multiple_lines): Add example of
21052 (selftest::edit_context_c_tests): Call the new tests.
21054 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21056 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
21057 parameter cand. Update dump information.
21058 (get_computation_cost): Update uses.
21060 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21062 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
21063 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
21064 (get_computation_at, rewrite_use_address): Update use of
21065 get_computation_aff.
21067 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21069 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
21070 (get_computation): Delete.
21071 (get_computation_cost): Implement like get_computation_cost_at.
21072 Use get_computation_at.
21073 (get_computation_cost_at): Delete.
21074 (rewrite_use_nonlinear_expr): Use get_computation_at.
21075 (rewrite_use_compare, remove_unused_ivs): Ditto.
21077 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21079 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
21081 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21083 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
21084 (ivopts_global_cost_for_size): Rename parameter and update uses.
21085 (iv_ca_recount_cost): Update uses.
21086 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
21087 candidates seperately in n_invs and n_cands.
21088 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
21090 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21092 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
21093 (find_inv_vars_cb): New.
21094 (find_depends): Renamed to ...
21095 (find_inv_vars): ... this.
21096 (add_candidate_1, force_var_cost): Call find_inv_vars.
21097 (split_address_cost, determine_group_iv_cost_cond): Ditto.
21099 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21101 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
21102 inv_vars. Add inv_exprs.
21103 (struct iv_cand): Rename depends_on to inv_vars.
21104 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
21105 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
21106 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
21107 (dump_cand): Dump inv_vars.
21108 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
21109 (record_invariant, find_depends, add_candidate_1): Ditto.
21110 (set_group_iv_cost, force_var_cost): Ditto.
21111 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
21112 (get_computation_cost_at, get_computation_cost): Ditto.
21113 (determine_group_iv_cost_generic): Ditto.
21114 (determine_group_iv_cost_address): Ditto.
21115 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
21116 (determine_group_iv_costs): Ditto.
21117 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
21118 (iv_ca_set_remove_invariants): Renamed to ...
21119 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
21120 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
21121 (iv_ca_set_add_invariants): Renamed to ...
21122 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
21123 (iv_ca_set_cp): Use iv_ca_set_add_invs.
21124 (iv_ca_has_deps): Support inv_vars and inv_exprs.
21125 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
21126 (create_new_ivs): Remove useless dump.
21128 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21130 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
21132 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
21133 (iv_ca_set_no_cp, create_new_iv): Ditto.
21135 2017-05-02 Bin Cheng <bin.cheng@arm.com>
21137 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
21139 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21141 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
21142 function tree_check2.
21144 2017-05-02 Martin Liska <mliska@suse.cz>
21146 * doc/gcov.texi: Add missing preposition.
21147 * gcov.c (function_info::function_info): Properly fill up
21148 all member variables.
21150 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21152 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
21154 2017-05-02 Tamar Christina <tamar.christina@arm.com>
21156 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
21158 2017-05-02 Martin Liska <mliska@suse.cz>
21161 * lto-streamer-in.c (lto_read_tree_1): Remove
21162 LTO_STREAMER_DEBUG.
21163 * lto-streamer.c (struct tree_hash_entry): Likewise.
21164 (struct tree_entry_hasher): Likewise.
21165 (tree_entry_hasher::hash): Likewise.
21166 (tree_entry_hasher::equal): Likewise.
21167 (lto_streamer_init): Likewise.
21168 (lto_orig_address_map): Likewise.
21169 (lto_orig_address_get): Likewise.
21170 (lto_orig_address_remove): Likewise.
21171 * lto-streamer.h: Likewise.
21172 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
21173 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
21175 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
21177 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
21178 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
21179 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
21180 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
21181 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
21182 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
21183 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
21184 (mm_maskz_sub_ss): New intrinsics.
21185 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21186 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
21187 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
21188 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
21189 (__builtin_ia32_subss_mask_round): New builtins.
21190 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
21191 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
21192 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
21194 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
21195 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
21197 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
21200 2017-05-02 Martin Jambor <mjambor@suse.cz>
21202 PR tree-optimization/78687
21203 * tree-sra.c (access): New field parent.
21204 (process_subtree_disqualification): New function.
21205 (disqualify_candidate): Call it.
21206 (build_accesses_from_assign): Reset write flag if creating an
21208 (build_access_subtree): Fill in parent field and also prpagate
21209 down grp_write flag.
21210 (create_artificial_child_access): New parameter set_grp_write, set
21211 grp_write to its value.
21212 (propagate_subaccesses_across_link): Also propagate grp_write flag
21214 (propagate_all_subaccesses): Push the closest parent back to work
21215 queue if add_access_to_work_queue returned true.
21217 2017-05-02 Richard Biener <rguenther@suse.de>
21219 * common.opt (fstrict-overflow): Alias negative to fwrapv.
21220 * doc/invoke.texi (fstrict-overflow): Remove all traces of
21221 -fstrict-overflow documentation.
21222 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
21223 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
21224 flag_strict_overflow.
21225 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
21226 * lto-opts.c (lto_write_options): Do not stream it.
21227 * lto-wrapper.c (merge_and_complain): Do not handle it.
21228 * opts.c (default_options_table): Do not set -fstrict-overflow.
21229 (finish_options): Likewise do not clear it when sanitizing.
21230 * simplify-rtx.c (simplify_const_relational_operation): Do not
21231 test flag_strict_overflow.
21233 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21235 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
21236 using enabled attribute.
21237 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
21238 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
21239 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
21240 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
21241 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
21242 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
21243 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
21244 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
21245 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
21246 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
21248 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
21250 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
21252 2017-05-02 Richard Biener <rguenther@suse.de>
21254 PR tree-optimization/80591
21256 2017-04-10 Richard Biener <rguenther@suse.de>
21258 * tree-ssa-structalias.c (find_func_aliases): Properly handle
21261 2017-05-02 Richard Biener <rguenther@suse.de>
21263 PR tree-optimization/80549
21264 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
21265 (cleanup_tree_cfg_noloop): Create forwarders to known loop
21266 headers if they do not have a preheader.
21268 2017-05-02 Martin Liska <mliska@suse.cz>
21271 * common.opt: Fix typo.
21272 * doc/invoke.texi: Likewise.
21274 2017-05-01 Jan Beulich <jbeulich@suse.com>
21276 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
21277 swapping, add (x,x,m,x,n) alternative.
21279 2017-05-01 Nathan Sidwell <nathan@acm.org>
21281 * calls.c (combine_pending_stack_adjustment_and_call): Remove
21282 unnecessary unadjusted_alignment check.
21284 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
21287 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
21289 * gimplify.c (gimplify_cilk_detach): New function.
21290 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
21291 * tree-core.h: Document EXPR_CILK_SPAWN.
21292 * tree.h (EXPR_CILK_SPAWN): Define.
21294 2017-05-01 David Malcolm <dmalcolm@redhat.com>
21296 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
21297 to use new fixit_hint representation, using the "replace" logic.
21298 (get_line_span_for_fixit_hint): Likewise.
21299 (layout::print_any_fixits): Likewise.
21300 (selftest::test_one_liner_many_fixits): Rename to...
21301 (selftest::test_one_liner_many_fixits_1): ...this, and update
21302 comment and expected output to reflect that the multiple fix-it
21303 hints are now consolidated into one insertion.
21304 (selftest::test_one_liner_many_fixits_2): New test.
21305 (selftest::test_diagnostic_show_locus_one_liner): Update for
21307 (selftest::test_fixit_consolidation): Update for fix-it API
21309 * diagnostic.c (print_parseable_fixits): Likewise.
21310 * edit-context.c (edited_line::m_line_events): Convert from
21311 auto_vec <line_event *> to auto_vec <line_event>.
21312 (class line_event): Convert from abstract base class to a concrete
21313 class, taking over the role of replace_event.
21314 (class insert_event): Delete.
21315 (class replace_event): Rename to class line_event. Convert to
21317 (edit_context::add_fixits): Reimplement.
21318 (edit_context::apply_insert): Delete.
21319 (edit_context::apply_replace): Rename to...
21320 (edit_context::apply_fixit): ...this. Convert to half-open range.
21321 (edited_file::apply_insert): Delete.
21322 (edited_file::apply_replace): Rename to...
21323 (edited_file::apply_fixit): ...this.
21324 (edited_line::~edited_line): Drop deletion of events.
21325 (edited_line::apply_insert): Delete.
21326 (edited_line::apply_replace): Rename to...
21327 (edited_line::apply_fixit): ...this. Convert to half-open range.
21328 Update for change to type of m_line_events.
21329 * edit-context.h (edit_context::apply_insert): Delete.
21330 (edit_context::apply_replace): Rename to...
21331 (edit_context::apply_fixit): ...this.
21333 2017-05-01 Martin Sebor <msebor@redhat.com>
21335 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
21338 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
21341 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
21342 __get_cpuid_max returns 0.
21343 (__get_cpuid_count): Ditto.
21345 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
21347 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
21348 replacement expression is another instance of one of its arguments.
21350 2017-05-01 Jakub Jelinek <jakub@redhat.com>
21353 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
21354 check for stack push/pop autoinc.
21355 * config/i386/i386.c (ix86_agi_dependent): Return false
21356 if the only reason why modified_in_p returned true is that
21357 addr is SP based and set_insn is a push or pop.
21359 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21361 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
21364 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
21367 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
21368 (account_size_time): Use two predicates - exec_pred and
21370 (evaluate_conditions_for_known_args): Compute both clause and
21372 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
21373 (inline_summary_t::duplicate): Update.
21374 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
21376 (compute_inline_parameters): Likewise.
21377 (estimate_edge_size_and_time): Update caluclation of time.
21378 (estimate_node_size_and_time): Compute both time and nonspecialized
21380 (estimate_ipcp_clone_size_and_time): Update.
21381 (inline_merge_summary): Update.
21382 (do_estimate_edge_time): Update.
21383 (do_estimate_edge_size): Update.
21384 (do_estimate_edge_hints): Update.
21385 (inline_read_section, inline_write_summary): Stream both new predicates.
21386 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
21388 (compute_inlined_call_time): Cleanup.
21389 (big_speedup_p): Update.
21390 (edge_badness): Update.
21391 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
21392 (size_time_entry): Replace predicate by exec_predicate and
21393 nonconst_predicate.
21394 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
21395 (estimate_edge_time): Return also nonspec_time.
21396 (reset_edge_growth_cache): Update.
21398 2017-04-29 Jakub Jelinek <jakub@redhat.com>
21400 PR rtl-optimization/80491
21401 * ifcvt.c (noce_process_if_block): When looking for x setter
21402 with missing else_bb, don't check only the insn right before
21403 cond_earliest, but look for the last insn that x is modified in
21404 within the same bb.
21406 PR rtl-optimization/80491
21407 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
21409 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
21411 PR tree-optimization/80487
21412 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
21414 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21416 PR tree-optimization/79697
21417 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
21418 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
21419 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
21421 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
21422 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
21424 2017-04-28 Martin Sebor <msebor@redhat.com>
21426 PR tree-optimization/80523
21427 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
21428 (init_target_to_host_charmap, target_to_host, target_strtol10): New
21430 (maybe_warn, format_directive, parse_directive): Use new functions.
21431 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
21433 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
21435 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
21437 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
21439 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
21440 target_header_dir): Set correctly.
21441 * configure: Regenerated.
21442 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
21443 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
21444 instead of SYSTEM_HEADER_DIR.
21446 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
21448 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
21449 (estimate_local_effects): Likewise.
21450 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
21451 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
21452 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
21453 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
21454 do_estimate_edge_time, estimate_edge_time): Likewise.
21455 * ipa-inline-analysis.c (estimate_node_size_and_time,
21456 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
21457 (estimate_time_after_inlining): Remove.
21459 2017-04-28 Martin Liska <mliska@suse.cz>
21461 * doc/gcov.texi: Enhance documentation of gcov.
21463 2017-04-28 Martin Liska <mliska@suse.cz>
21465 * doc/gcov.texi: Sort options in alphabetic order.
21466 * doc/gcov-dump.texi: Likewise.
21467 * doc/gcov-tool.texi: Likewise.
21468 * gcov.c (print_usage): Likewise.
21469 * gcov-dump.c (print_usage): Likewise.
21470 * gcov-tool.c (print_merge_usage_message): Likewise.
21471 (print_rewrite_usage_message): Likewise.
21472 (print_overlap_usage_message): Likewise.
21474 2017-04-28 Martin Liska <mliska@suse.cz>
21476 PR gcov-profile/53915
21477 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
21479 2017-04-28 Martin Liska <mliska@suse.cz>
21481 PR gcov-profile/79891
21482 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
21483 is marked by compiler as living on a line.
21484 (get_cycles_count): Remove usage of the union.
21485 (output_intermediate_file): Likewise.
21486 (find_source): Fix GNU coding style.
21487 (accumulate_line_counts): Remove old non-all block mode.
21488 (output_lines): Remove usage of the union.
21489 * profile.c (output_location): Include all BBs, even if
21490 belonging to a same line (and file) as a previous BB.
21492 2017-04-28 Martin Liska <mliska@suse.cz>
21494 * gcov.c (process_args): Handle new argument 'w'.
21495 (read_graph_file): Assign ID to BBs.
21496 (output_branch_count): Display BB # if verbose flag is set.
21497 (output_lines): Likewise for arcs.
21498 (print_usage): Add '--verbose' option help.
21499 * doc/gcov.texi: Document --verbose (-w) option.
21501 2017-04-28 Martin Liska <mliska@suse.cz>
21503 * gcov.c (struct block_location_info): New struct.
21504 (process_file): Fill up the new structure.
21505 (read_graph_file): Replace usage of encoding by the newly added
21507 (add_line_counts): Likewise.
21508 (accumulate_line_counts): Remove usage of the union.
21509 (function_info::function_info): New function.
21510 (function_info::~function_info): Likewise.
21511 (process_file): Call delete instead of release_function.
21512 (release_function): Release the function.
21513 (release_structures): Call delete instead of release_function.
21514 (solve_flow_graph): Replace usage of num_blocks.
21515 (find_exception_blocks): Likewise.
21516 (output_lines): Fix GNU coding style.
21518 2017-04-28 Martin Liska <mliska@suse.cz>
21521 * coverage.c (coverage_remove_note_file): New function.
21522 * coverage.h: Declare the function.
21523 * toplev.c (finalize): Clean if an error has been seen.
21525 2017-04-28 Martin Liska <mliska@suse.cz>
21527 PR gcov-profile/80031
21528 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
21529 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
21530 * gcov.c (read_graph_file): Read just number of blocks.
21531 * profile.c (branch_prob): Do not stream 0 flags per a basic
21534 2017-04-28 Martin Liska <mliska@suse.cz>
21536 * gcov-dump.c (tag_*): Add new argument to declarations.
21537 (dump_gcov_file): Likewise.
21538 (tag_blocks): Add and use new argument depth.
21539 (tag_arcs): Likewise.
21540 (tag_lines): Likewise.
21541 (tag_counters): Likewise.
21542 (tag_summary): Likewise.
21543 (dump_working_sets): Use depth to do a proper indentation.
21545 2017-04-28 Jakub Jelinek <jakub@redhat.com>
21548 * cgraph.h (symtab_node::debug_symtab): No longer inline.
21549 * symtab.c (symtab_node::debug_symtab): Move definition here.
21551 2017-04-28 Richard Biener <rguenther@suse.de>
21553 * lto-streamer.h (LTO_major_version): Bump to 7.
21555 2017-04-28 Richard Biener <rguenther@suse.de>
21557 * tree-vrp.c (assert_info): New struct.
21558 (add_assert_info): New helper.
21559 (register_edge_assert_for_2): Refactor to add asserts to a vector
21561 (register_edge_assert_for_1): Likewise.
21562 (register_edge_assert_for): Likewise.
21563 (finish_register_edge_assert_for): New helper actually registering
21564 asserts where live on edge.
21565 (find_conditional_asserts): Adjust.
21566 (find_switch_asserts): Likewise.
21567 (evrp_dom_walker::try_find_new_range): Generalize.
21568 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
21570 2017-04-27 Marek Polacek <polacek@redhat.com>
21573 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
21574 arg10 and arg11 to itype.
21576 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
21578 * doc/extend.texi (Object Size Checking): Improve grammar.
21580 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
21583 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
21584 that the logic for permitting reciprocal estimates matches that
21587 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21590 * tree.c (type_cache_hasher::equal): Only compare
21591 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
21592 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
21593 non-aggregate element types.
21594 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
21595 about the flag on ARRAY_TYPEs in the comment, formatting fix.
21597 2017-04-27 Richard Biener <rguenther@suse.de>
21599 PR middle-end/80533
21600 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
21601 stripping ARRAY_REFs from MEM_EXPR make sure we're not
21602 keeping a reference to a trailing array.
21604 2017-04-27 Richard Biener <rguenther@suse.de>
21606 PR middle-end/80539
21607 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
21608 being in loop-closed SSA form conservatively.
21609 (chrec_fold_multiply_poly_poly): Likewise.
21611 2017-04-27 Tamar Christina <tamar.christina@arm.com>
21613 PR middle-end/79665
21614 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
21615 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
21617 2017-04-27 Jakub Jelinek <jakub@redhat.com>
21620 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
21621 (aarch64_function_arg_alignment): Return unsigned int again, but still
21622 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
21623 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21624 Don't emit -Wpsabi note.
21625 (aarch64_function_arg_boundary): Likewise.
21626 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21629 2017-04-26 Nathan Sidwell <nathan@acm.org>
21631 * tree.h (crc32_unsigned_n): Declare.
21632 (crc32_unsigned, crc32_unsigned): Make inline.
21633 * tree.c (crc32_unsigned_bits): Replace with ...
21634 (crc32_unsigned_n): ... this.
21635 (crc32_unsigned, crc32_byte): Remove.
21636 (crc32_string): Remove unnecessary braces.
21638 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21640 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
21641 * ipa-inline-analysis.c (MAX_TIME): Remove.
21642 (account_size_time): Use sreal for time.
21643 (dump_inline_summary): Update.
21644 (estimate_function_body_sizes): Update.
21645 (estimate_edge_size_and_time): Update.
21646 (estimate_calls_size_and_time): Update.
21647 (estimate_node_size_and_time): Update.
21648 (inline_merge_summary): Update.
21649 (inline_update_overall_summary): Update.
21650 (estimate_time_after_inlining): Update.
21651 (inline_read_section): Update.
21652 (inline_write_summary): Update.
21653 * ipa-inline.c (compute_uninlined_call_time): Update.
21654 (compute_inlined_call_time): Update.
21655 (recursive_inlining): Update.
21656 (inline_small_functions): Update.
21657 (dump_overall_stats): Update.
21658 * ipa-inline.h: Include sreal.h.
21659 (size_time_entry): Turn time to sreal.
21660 (inline_summary): Turn self_time nad time to sreal.
21662 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
21664 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
21666 (sreal::stream_out, sreal::stream_in): New.
21667 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
21669 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21671 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
21674 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
21677 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
21679 (dimode_scalar_chain::compute_convert_gain): Ditto.
21680 (dimode_scalar_chain::make_vector_copies): Ditto.
21681 (dimode_scalar_chain::convert_reg): Ditto.
21682 (dimode_scalar_chain::convert_insn): Ditto.
21683 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
21684 (VI248_AVX512BW_1): New mode iterator.
21685 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
21686 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
21689 2017-04-25 Martin Sebor <msebor@redhat.com>
21691 PR tree-optimization/80497
21692 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
21693 constants are representable in HOST_WIDE_INT.
21694 (parse_directive): Ditto.
21696 2017-04-25 Martin Sebor <msebor@redhat.com>
21699 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
21700 (new_zero_array): Adjust signature.
21701 (dom_info::dom_init): Used unsigned rather that size_t.
21702 (dom_info::dom_info): Same.
21704 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21705 Jakub Jelinek <jakub@redhat.com>
21708 * config/arm/arm.c: Include gimple.h.
21709 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21710 returns negative, increment ncrn only if it returned positive.
21711 (arm_needs_doubleword_align): Return int instead of bool,
21712 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
21713 members, but if there is any such non-FIELD_DECL
21714 > PARM_BOUNDARY aligned decl, return -1 instead of false.
21715 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
21716 returns negative, increment nregs only if it returned positive.
21717 (arm_setup_incoming_varargs): Likewise.
21718 (arm_function_arg_boundary): Emit -Wpsabi note if
21719 arm_needs_doubleword_align returns negative, return
21720 DOUBLEWORD_ALIGNMENT only if it returned positive.
21722 2017-04-25 Marek Polacek <polacek@redhat.com>
21725 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
21726 first argument to type.
21728 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
21731 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
21732 type checks to test for compatibility instead of equality.
21734 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
21735 Jakub Jelinek <jakub@redhat.com>
21738 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
21740 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
21741 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
21742 the alignment computation, but return their maximum in warn_alignment.
21743 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
21744 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
21746 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
21747 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
21750 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21752 * config/arc/simdext.md (dmpyh): Fix typo.
21754 2017-04-25 Richard Biener <rguenther@suse.de>
21756 PR tree-optimization/80492
21757 * alias.c (compare_base_decls): Handle registers with asm
21758 specification conservatively.
21759 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
21760 compare_base_decls returning dont-know properly.
21762 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21764 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
21765 (legitimate_offset_address_p): New function.
21766 (arc_legitimate_address_p): Use above function.
21768 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21770 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
21772 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21774 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
21775 ACCH registers whenever they are available.
21777 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21779 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
21780 double regs fix when not used.
21782 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21784 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
21786 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
21787 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
21789 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21791 * config/arc/arc.c (arc_output_addsi): Check for h-register class
21792 when emitting short ADD instructions.
21794 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
21796 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
21798 (cmpsi_cc_c_insn): Likewise.
21799 (cbranchsi4_scratch): Compute proper instruction length using
21800 compact_hreg_operand.
21801 * config/arc/predicates.md (compact_hreg_operand): New predicate.
21803 2017-04-25 Richard Biener <rguenther@suse.de>
21805 PR middle-end/80509
21806 * passes.c (pass_manager::pass_manager): Initialize
21807 m_name_to_pass_map.
21809 2017-04-25 Richard Biener <rguenther@suse.de>
21811 PR tree-optimization/79201
21812 * tree-ssa-sink.c (statement_sink_location): Handle calls.
21814 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21817 * config/s390/vector.md: Split MEM->GPR vector moves for
21818 non-s_operand addresses.
21820 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21823 * config/s390/predicates.md (reload_const_wide_int_operand): New
21825 * config/s390/s390.md ("movti"): Remove d/P alternative.
21826 ("movti_bigconst"): New pattern definition.
21828 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21831 * s390-protos.h (s390_expand_cs_hqi): Removed.
21832 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
21833 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
21834 modes as well as CCZ1mode and CCZmode.
21835 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
21836 signature of s390_emit_compare_and_swap.
21837 (s390_expand_cs_hqi): Likewise, make static.
21838 (s390_expand_cs_tdsi): Generate an explicit compare before trying
21839 compare-and-swap, in some cases.
21840 (s390_expand_cs): Wrapper function.
21841 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
21843 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
21844 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
21845 patterns for small and large integers. Forbid symref memory operands.
21846 Move expander to s390.c. Require cc register.
21847 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
21848 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
21849 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
21850 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
21851 symref memory operands. Remove CC mode and call s390_match_ccmode
21853 ("atomic_exchange<mode>"): Allow and implement all integer modes.
21855 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21857 * config/s390/s390.md (define_peephole2): New peephole to help
21858 combining the load-and-test pattern with volatile memory.
21860 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
21862 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
21863 with CCZmode for TARGET_Z196.
21865 2017-04-25 Jakub Jelinek <jakub@redhat.com>
21867 PR rtl-optimization/80501
21868 * combine.c (make_compound_operation_int): Set subreg_code to SET
21869 even for AND with mask of the sign bit of mode.
21871 PR rtl-optimization/80500
21872 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
21873 sum's initial value.
21875 2017-04-25 Julian Brown <julian@codesourcery.com>
21876 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21878 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
21880 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
21882 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
21884 2017-04-25 Julian Brown <julian@codesourcery.com>
21885 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21887 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
21888 (thunderx2t99_sha): New Reservation.
21890 2017-04-25 Julian Brown <julian@codesourcery.com>
21891 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
21893 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
21894 type for 1-element load.
21896 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
21898 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
21900 2017-04-24 Martin Jambor <mjambor@suse.cz>
21902 PR tree-optimization/80293
21903 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
21904 char arrays not totally scalarizable if it is false.
21905 (analyze_all_variable_accesses): Pass correct value in the new
21906 parameter. Add a statistics counter.
21908 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
21910 PR middle-end/79931
21911 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
21913 2017-04-24 Richard Biener <rguenther@suse.de>
21915 PR tree-optimization/80494
21916 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
21917 out for complex types.
21919 2017-04-24 Richard Biener <rguenther@suse.de>
21921 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
21922 * tree-ssa-sccvn.c (print_scc): Print SCC size.
21923 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
21924 (DFS): Adjust and never fail.
21925 (sccvn_dom_walker::fail): Remove.
21926 (sccvn_dom_walker::before_dom_children): Adjust.
21927 (run_scc_vn): Likewise and never fail.
21928 * tree-ssa-pre.c (pass_pre::execute): Adjust.
21929 (pass_fre::execute): Likewise.
21931 2017-04-24 Richard Biener <rguenther@suse.de>
21933 PR tree-optimization/79725
21934 * tree-ssa-sink.c (statement_sink_location): Return whether
21935 failure reason was zero uses. Move that check later.
21936 (sink_code_in_bb): Deal with zero uses by removing the stmt
21939 2017-04-24 Richard Biener <rguenther@suse.de>
21942 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
21943 pointer-based references.
21945 2017-04-24 Richard Biener <rguenther@suse.de>
21948 * pass_manager.h (pass_manager::operator new): Remove.
21949 (pass_manager::operator delete): Likewise.
21950 * passes.c (pass_manager::operator new): Remove.
21951 (pass_manager::operator delete): Likewise.
21952 (pass_manager::pass_manager): Zero individual pass members.
21954 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
21957 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
21958 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
21959 Check "XEXP (src, 1)" operand here.
21960 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
21961 Check "XEXP (src, 1)" operand here.
21962 (dimode_scalar_chain::make_vector_copies): Detect count register
21963 of a shift instruction. Zero extend count register from QImode
21964 to DImode to satisfy vector shift pattern count operand predicate.
21965 Substitute vector shift count operand with a DImode copy.
21966 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
21969 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
21971 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
21972 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21973 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
21974 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
21975 (UNSPEC_NOREX_MEM): Remove definition.
21977 2017-04-21 Richard Biener <rguenther@suse.de>
21979 PR tree-optimization/79547
21980 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
21981 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
21982 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
21983 without any constraints.
21985 2017-04-21 Richard Biener <rguenther@suse.de>
21987 PR tree-optimization/78847
21988 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
21990 2017-04-21 Richard Biener <rguenther@suse.de>
21992 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
21993 (build_distinct_type_copy): Likewise.
21994 (build_variant_type_copy): Likewise.
21995 * tree.c (build_qualified_type): Pass down mem-stat info.
21996 (build_distinct_type_copy): Likewise.
21997 (build_variant_type_copy): Likewise.
21999 2017-04-21 Richard Biener <rguenther@suse.de>
22001 PR tree-optimization/80237
22002 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
22004 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
22005 for a simplified result.
22007 2016-04-21 Richard Biener <rguenther@suse.de>
22009 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
22010 sth as strict as a simple_iv but a chrec without symbols and an
22011 operand defined in the loop we are peeling (and not some subloop).
22012 (propagate_constants_for_unrolling): Propagate all constants.
22014 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22017 * config/i386/i386.c (print_reg): Remove assert for disalowed
22018 regno values, call output_operand_lossage instead.
22020 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
22023 * config/i386/constraints.md (Yc): New register constraint.
22024 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
22025 Use Yc constraint for alternative 2 of operand 0. Remove
22026 preferred_for_speed attribute.
22028 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22030 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
22031 lastprivate clauses in SIMT case.
22033 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
22035 * doc/invoke.texi (-Wextra-semi): Document new warning option.
22037 2017-04-20 Richard Biener <rguenther@suse.de>
22039 PR tree-optimization/57796
22040 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
22041 as N scalar stores.
22042 (vect_model_load_cost): Cost gathers as N scalar loads.
22044 2017-04-20 Richard Biener <rguenther@suse.de>
22046 * ggc-page.c (ggc_allocated_p): Rename to ...
22047 (safe_lookup_page_table_entry): ... this and return the lookup
22049 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
22051 2017-04-20 Richard Biener <rguenther@suse.de>
22053 PR tree-optimization/80453
22054 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
22055 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
22056 from the conditions.
22057 (vn_phi_eq): Pass them down.
22058 (vn_phi_lookup): Record them.
22059 (vn_phi_insert): Likewise.
22061 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
22063 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
22064 uninitialized variable warning to avoid buffer overrun.
22066 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
22069 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
22070 is suppressed for '{ 0 }' in C.
22072 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22074 * BASE-VER: Set to 8.0.0.
22076 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
22078 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
22079 priority .init_array and .fini_array section with SECTION_NOTYPE
22082 2017-04-20 Jakub Jelinek <jakub@redhat.com>
22084 PR middle-end/80423
22085 * tree.h (build_array_type): Add typeless_storage default argument.
22086 * tree.c (type_cache_hasher::equal): Also compare
22087 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
22088 (build_array_type): Add typeless_storage argument, set
22089 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
22091 (build_nonshared_array_type): Adjust build_array_type_1 caller.
22092 (build_array_type): Likewise. Add typeless_storage argument.
22094 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22095 Jakub Jelinek <jakub@redhat.com>
22097 PR tree-optimization/80426
22098 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
22099 operation on symbolic operands, also compute the overflow for the
22100 invariant part when the operation degenerates into a negation.
22102 2017-04-19 Jakub Jelinek <jakub@redhat.com>
22105 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
22106 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
22109 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
22111 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
22114 * config/avr/avr.c (tree.h): Include it.
22115 (cgraph.h): Include it.
22116 (avr_encode_section_info): Don't warn for uninitialized progmem
22117 variable if it's just an alias.
22119 2017-04-19 Richard Biener <rguenther@suse.de>
22122 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
22123 when needed by AutoPGO.
22125 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
22128 * doc/lto.texi: Remove an extra 'that'.
22130 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
22132 PR rtl-optimization/80429
22133 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
22134 are only used in debug insns.
22136 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
22137 Vladimir Makarov <vmakarov@redhat.com>
22139 * config/sparc/predicates.md (input_operand): Add comment. Return
22140 true for any memory operand when LRA is in progress.
22141 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
22143 2017-04-18 Jeff Law <law@redhat.com>
22146 * mips.md ({return,simple_return}_internal): Do not overwrite
22149 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22151 PR tree-optimization/80443
22152 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
22153 instead of adding 1, subtract -1 and similarly instead of subtracting
22156 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
22158 PR rtl-optimization/80357
22159 * haifa-sched.c (tmp_bitmap): New variable.
22160 (model_recompute): Handle duplicate use records.
22161 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
22162 (free_global_sched_pressure_data): Free it.
22164 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
22167 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
22168 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
22169 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
22170 instead of SYSTEM_HEADER_DIR.
22172 2017-04-18 Jeff Law <law@redhat.com>
22174 PR middle-end/80422
22175 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
22176 predecessors after walking up the insn chain.
22178 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22181 * dwarf2out.c (modified_type_die): Try harder not to emit internal
22182 sizetype type into debug info.
22184 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22187 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
22188 unneeded test for TARGET_UPPER_REGS_SF.
22189 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
22191 2017-04-18 Jakub Jelinek <jakub@redhat.com>
22194 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
22195 instead of gsi_after_labels.
22197 2017-04-18 Jeff Law <law@redhat.com>
22199 * regcprop.c (maybe_mode_change): Avoid creating copies of the
22203 2017-04-13 Jeff Law <law@redhat.com>
22204 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22205 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22207 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
22210 * config/avr/avr.c (intl.h): Include it.
22211 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
22213 2017-04-18 Martin Liska <mliska@suse.cz>
22215 PR gcov-profile/78783
22216 * gcov-tool.c (gcov_output_files): Validate that destination
22217 file is either removed by the tool or by a user.
22219 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22220 Guy Benyei <guybe@mellanox.com>
22222 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
22223 block, and do not negate it, the stored id is already negative.
22225 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
22227 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
22229 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
22232 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
22233 masks of options that should be turned off if the VSX vector
22234 options are turned off.
22235 (OTHER_P8_VECTOR_MASKS): Likewise.
22236 (OTHER_VSX_VECTOR_MASKS): Likewise.
22237 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
22238 rs6000_disable_incompatible_switches to validate no type switches
22240 (rs6000_incompatible_switch): New function to disallow turning on
22241 other vector options if -mno-vsx, -mno-power8-vector, or
22242 -mno-power9-vector are specified.
22244 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22246 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
22248 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22250 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
22251 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
22252 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
22253 (ARG_POINTER_CFA_OFFSET): Likewise.
22255 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
22257 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
22258 conditions to take advantage of various optimizations.
22260 2017-04-13 Jeff Law <law@redhat.com>
22262 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
22263 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
22264 (zero_extendsidi2_dext): Likewise.
22266 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22269 * fold-const.c (fold_ternary_loc): Revert
22270 use op0 instead of fold_convert_loc (loc, type, arg0) part of
22273 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
22275 PR rtl-optimization/80343
22276 * lra-remat.c (update_scratch_ops): Assign original hard reg to
22277 new scratch pseudo.
22279 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
22282 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
22283 to ubsan_encode_value.
22285 2017-04-13 Jeff Law <law@redhat.com>
22287 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
22288 appearing in DEBUG_INSNs.
22290 2017-04-13 Martin Liska <mliska@suse.cz>
22292 PR gcov-profile/80413
22293 * gcov-io.c (gcov_write_string): Copy to buffer just when
22294 allocated size is greater than zero.
22296 2017-04-13 Jakub Jelinek <jakub@redhat.com>
22299 * dwarf2out.c (decls_for_scope): Ignore declarations of
22300 current_function_decl in BLOCK_NONLOCALIZED_VARS.
22302 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
22305 * ipa-visibility.c (non_local_p): Fix typos.
22306 (localize_node): When localizing symbol in same comdat group,
22307 dissolve the group only when we know external symbols are going
22309 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
22311 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22313 PR tree-optimization/79390
22314 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
22315 order does not result in usable sequence, retry with reversed operand
22321 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
22322 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
22323 op0 instead of fold_convert_loc (loc, type, arg0).
22325 2017-04-12 Jeff Law <law@redhat.com>
22327 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
22328 has a delay slot in the generated code.
22330 * config/cris/cris.md (cris_preferred_reload_class): Return
22331 GENNONACR_REGS rather than GENERAL_REGS.
22333 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22336 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
22337 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
22338 signedness of the result type.
22340 2017-04-12 Richard Biener <rguenther@suse.de>
22341 Jeff Law <law@redhat.com>
22343 PR tree-optimization/80359
22344 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
22345 trim stores to TARGET_MEM_REFs.
22347 2017-04-12 Richard Biener <rguenther@suse.de>
22349 PR tree-optimization/79390
22350 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
22351 threading case even more.
22353 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
22356 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
22357 for quad_address_p for TImode, instead of just not indexed_address.
22359 2017-04-12 Richard Biener <rguenther@suse.de>
22360 Bernd Edlinger <bernd.edlinger@hotmail.de>
22362 PR middle-end/79671
22363 * alias.c (component_uses_parent_alias_set_from): Handle
22364 TYPE_TYPELESS_STORAGE.
22365 (get_alias_set): Likewise.
22366 * tree-core.h (tree_type_common): Add typeless_storage flag.
22367 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
22368 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
22369 for types containing members with TYPE_TYPELESS_STORAGE.
22370 (place_field): Likewise.
22371 (layout_type): Likewise for ARRAY_TYPE.
22372 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
22373 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
22374 TYPE_TYPELESS_STORAGE.
22375 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
22377 2017-04-12 Jakub Jelinek <jakub@redhat.com>
22380 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
22381 first argument to type.
22383 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22387 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
22388 CONST0_RTX (mode) rather than const0_rtx where appropriate.
22389 (rs6000_expand_binop_builtin): Likewise.
22390 (rs6000_expand_ternop_builtin): Likewise; also add missing
22391 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
22392 vshasigma built-ins.
22393 * doc/extend.texi: Document that vec_xxpermdi's third argument
22394 must be a constant.
22396 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
22398 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
22399 Use shift_const cost parameter when calculating gain of STV shifts.
22401 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
22403 PR rtl-optimization/70478
22404 * lra-constraints.c (process_alt_operands): Check memory for
22405 disfavoring memory insn operand.
22407 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22409 PR middle-end/80100
22410 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
22411 left shift in unsigned HOST_WIDE_INT type.
22413 PR rtl-optimization/80385
22414 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
22415 (not (neg X)) into (plus X -1) for complex or non-integral modes.
22418 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
22419 if they have any depend clauses.
22421 2017-04-11 Martin Liska <mliska@suse.cz>
22424 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
22425 * ipa-split.c (split_function): Create a local comdat symbol
22426 if caller is in a comdat group.
22428 2017-04-11 Martin Liska <mliska@suse.cz>
22431 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
22434 2017-04-11 Martin Sebor <msebor@redhat.com>
22436 PR middle-end/80364
22437 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
22438 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
22440 (directive::set_width, directive::set_precision, format_character):
22442 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
22445 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
22448 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
22449 conflict, set target->arch_name instead of target->cpu_name.
22451 2017-04-11 Richard Biener <rguenther@suse.de>
22453 PR tree-optimization/80374
22454 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
22455 build_zero_cst, remove fold_convertible_p check again.
22457 2017-04-11 Martin Liska <mliska@suse.cz>
22460 * ubsan.c (instrument_object_size): Do not instrument register
22463 2017-04-11 Jakub Jelinek <jakub@redhat.com>
22466 * config/i386/i386-builtin-types.def
22467 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
22468 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
22469 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
22470 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
22471 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
22472 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
22473 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
22474 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
22475 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
22476 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
22477 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
22478 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
22479 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
22480 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
22481 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
22482 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
22483 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
22484 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
22485 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
22486 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
22487 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
22488 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
22489 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
22490 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
22491 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
22492 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
22493 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
22494 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
22495 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
22496 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
22497 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
22498 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
22499 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
22500 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
22501 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
22502 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
22503 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
22504 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
22505 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
22506 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
22507 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
22508 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
22509 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
22510 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
22511 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
22512 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
22514 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
22515 flag to second_arg_count, handle 4 argument function type _COUNT
22516 aliases, handle second_arg_count on second argument rather than last.
22518 2017-04-10 Jeff Law <law@redhat.com>
22520 PR tree-optimization/80374
22521 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
22522 record anything if we can not convert integer_zero_node to the
22525 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
22528 * config/rs6000/rs6000.c (rs6000_option_override_internal):
22529 Enhance special handling given to the TARGET_P9_MINMAX option in
22530 relation to certain other options.
22532 2017-04-10 Bin Cheng <bin.cheng@arm.com>
22534 PR tree-optimization/80153
22535 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
22536 remove POINTER_PLUS_EXPR's base part directly, rather than through
22539 2017-04-10 Richard Biener <rguenther@suse.de>
22540 Bin Cheng <bin.cheng@arm.com>
22542 PR tree-optimization/80153
22543 * tree-affine.c (aff_combination_to_tree): Get base pointer from
22544 the first element of pointer type aff_tree. Build result expr in
22546 (add_elt_to_tree): Convert to type unconditionally. Remove other
22547 fold_convert calls.
22548 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
22549 (rewrite_use_nonlinear_expr): Check invariant using iv information.
22551 2017-04-10 Richard Biener <rguenther@suse.de>
22553 * tree-ssa-structalias.c (find_func_aliases): Properly handle
22556 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
22558 PR rtl-optimization/70478
22559 * lra-constraints.c (curr_small_class_check): New.
22560 (update_and_check_small_class_inputs): New.
22561 (process_alt_operands): Update curr_small_class_check. Disfavor
22562 alternative insn memory operands. Check available regs for small
22565 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
22568 * config/mips/mips.opt (-mvirt): Update description.
22569 * doc/invoke.texi (-mvirt): Likewise.
22571 2017-04-10 Richard Biener <rguenther@suse.de>
22573 PR middle-end/80362
22574 * fold-const.c (fold_binary_loc): Look at unstripped ops when
22575 looking for NEGATE_EXPR in -A / -B to A / B folding.
22577 2017-04-10 Martin Liska <mliska@suse.cz>
22579 PR gcov-profile/80224
22580 * gcov.c (print_usage): Fix usage string.
22581 (get_gcov_intermediate_filename): Remove.
22582 (output_gcov_file): Use both for normal and intermediate format.
22583 (generate_results): Do not initialize special file for
22584 intermediate format.
22586 2017-04-10 Richard Biener <rguenther@suse.de>
22588 PR tree-optimization/80304
22589 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
22592 2017-04-10 Nathan Sidwell <nathan@acm.org>
22595 * config/rs6000/rs6000.c (rs6000_vector_type): New.
22596 (rs6000_init_builtins): Use it.
22598 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22600 * config/arm/arm.md (<mrc>): Add mode to SET source.
22601 (<mrrc>): Likewise.
22603 2017-04-10 Richard Biener <rguenther@suse.de>
22605 PR middle-end/80344
22606 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
22608 2017-04-10 Jakub Jelinek <jakub@redhat.com>
22611 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
22612 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
22613 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
22614 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
22615 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
22616 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
22617 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
22618 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
22619 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
22620 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
22621 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
22622 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
22623 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
22624 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
22625 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
22626 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
22627 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
22628 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
22629 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
22630 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
22631 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
22632 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
22633 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
22635 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
22637 PR rtl-optimization/70478
22638 * lra-constraints.c: Reverse the last patch.
22640 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
22642 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
22643 Add comment for WCHAR_T.
22645 2017-04-08 Martin Liska <mliska@suse.cz>
22648 2017-04-07 Martin Liska <mliska@suse.cz>
22651 * ipa-split.c (split_function): Add function part to a same comdat
22654 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
22657 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
22659 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
22661 * rs6000/rs6000.c (vec_load_pendulum): Rename...
22662 (vec_pairing): ...to this.
22663 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
22664 (rs6000_sched_init): Adjust for name change.
22665 (struct rs6000_sched_context): Likewise.
22666 (rs6000_init_sched_context): Likewise.
22667 (rs6000_set_sched_context): Likewise.
22669 2017-04-07 Jakub Jelinek <jakub@redhat.com>
22675 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
22677 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
22678 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
22679 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
22681 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
22683 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
22685 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22687 PR rtl-optimization/70703
22688 * ira-color.c (update_conflict_hard_regno_costs): Use
22689 int64_t instead of HOST_WIDE_INT.
22691 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
22693 PR rtl-optimization/70478
22694 * lra-constraints.c (process_alt_operands): Disfavor alternative
22695 insn memory operands.
22697 2017-04-07 Jeff Law <law@redhat.com>
22699 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
22700 CALL and NOTE_INSN_CALL_ARG_LOCATION.
22702 2017-04-07 Martin Liska <mliska@suse.cz>
22705 * config/aarch64/aarch64.c (aarch64_process_target_attr):
22706 Show error message instead of an ICE.
22708 2017-04-07 Martin Liska <mliska@suse.cz>
22711 * ipa-split.c (split_function): Add function part to a same comdat
22714 2017-04-07 Richard Biener <rguenther@suse.de>
22716 PR middle-end/80341
22717 * tree.c (get_unwidened): Also handle ! for_type case for
22719 * convert.c (do_narrow): Split out from ...
22720 (convert_to_integer_1): ... here. Do not pass final truncation
22721 type to get_unwidened for TRUNC_DIV_EXPR.
22723 2017-04-07 Richard Biener <rguenther@suse.de>
22725 * tree-affine.c (wide_int_ext_for_comb): Take type rather
22727 (aff_combination_const): Adjust.
22728 (aff_combination_scale): Likewise.
22729 (aff_combination_add_elt): Likewise.
22730 (aff_combination_add_cst): Likewise.
22731 (aff_combination_convert): Likewise.
22732 (add_elt_to_tree): Likewise. Remove unused argument.
22733 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
22735 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
22737 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
22739 * config/arm/arm.c (arm_default_short_enums): Use
22740 ARM_DEFAULT_SHORT_ENUMS.
22741 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
22743 2017-04-06 Jakub Jelinek <jakub@redhat.com>
22746 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
22747 members with redundant out-of-class redeclaration.
22749 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22752 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
22753 * config/i386/i386.md (*zero_extendsidi2):
22754 Add (?*x,*x) and (?*v,*v) alternatives.
22756 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
22759 * config/i386/i386.c (ix86_expand_builtin)
22760 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
22761 mode from insn data. Convert operands to insn operand mode.
22762 Copy operands that don't satisfy insn predicate to a register.
22764 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
22766 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
22769 2017-04-06 Richard Biener <rguenther@suse.de>
22771 PR tree-optimization/80334
22772 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
22773 preserve alignment of accesses.
22775 2017-04-06 Richard Biener <rguenther@suse.de>
22777 PR tree-optimization/80262
22778 * tree-sra.c (build_ref_for_offset): Preserve address-space
22780 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
22781 Drop useless address-space information on MEM_REF offsets.
22783 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
22785 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
22787 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22789 PR rtl-optimization/70703
22790 * ira-color.c (update_conflict_hard_regno_costs): Use
22791 HOST_WIDE_INT instead of long.
22793 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
22796 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
22797 not defined for x86_64 target. Add -mmmx target option when __SSE2__
22799 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
22800 for x86_64 target. Handle -m3dnowa option.
22802 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
22804 PR rtl-optimization/70703
22805 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
22806 (update_conflict_hard_regno_costs): Use long instead of unsigned
22807 arithmetic for cost calculation.
22809 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22810 Bernd Edlinger <bernd.edlinger@hotmail.de>
22813 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
22816 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
22819 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
22820 ptr_mode with Pmode throughout.
22821 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
22822 into probe_stack_range and use DImode.
22824 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
22827 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
22828 call_eh_return is true.
22830 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22832 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
22833 Initialize last_match_fntype_index.
22835 2017-04-05 Jakub Jelinek <jakub@redhat.com>
22838 * tree-nvr.c: Include internal-fn.h.
22839 (pass_return_slot::execute): Ignore internal calls without
22842 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22843 Richard Biener <rguenther@suse.de>
22846 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
22847 captures used multiple times, except for the last use.
22848 * generic-match-head.c: Include gimplify.h.
22850 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22852 PR tree-optimization/79390
22853 * target.h (struct noce_if_info): Declare.
22854 * targhooks.h (default_noce_conversion_profitable_p): Declare.
22855 * target.def (noce_conversion_profitable_p): New target hook.
22856 * ifcvt.h (struct noce_if_info): New type, moved from ...
22857 * ifcvt.c (struct noce_if_info): ... here.
22858 (noce_conversion_profitable_p): Renamed to ...
22859 (default_noce_conversion_profitable_p): ... this. No longer
22861 (noce_try_store_flag_constants, noce_try_addcc,
22862 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
22863 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
22864 instead of noce_conversion_profitable_p.
22865 * config/i386/i386.c: Include ifcvt.h.
22866 (ix86_option_override_internal): Don't override
22867 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
22868 (ix86_noce_conversion_profitable_p): New function.
22869 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
22870 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
22871 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
22872 * doc/tm.texi: Regenerated.
22874 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22876 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
22879 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
22882 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
22883 instructions for small multiply cores.
22885 2017-04-04 Jeff Law <law@redhat.com>
22887 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
22889 (mips_expand_vec_perm_const): Initialize elements in orig_perm
22890 that are not set by the loop over the elements.
22892 2017-04-04 Jakub Jelinek <jakub@redhat.com>
22895 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
22896 int mode, convert_modes it to mode as unsigned, otherwise use
22897 lowpart_subreg to mode rather than SImode.
22898 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
22899 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
22900 Use DImode instead of SImode for the shift count operand.
22901 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
22904 2017-04-04 Richard Biener <rguenther@suse.de>
22906 PR middle-end/80281
22907 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
22908 arithmetic done for the negate or the plus. Simplify.
22909 (A - (-B) -> A + B): Likewise.
22910 * fold-const.c (split_tree): Make sure to not negate pointers.
22912 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
22914 PR rtl-optimization/60818
22915 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
22916 a compare of comparisons with the thing compared if this results
22917 in a different machine mode.
22919 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
22921 * alias.c (base_alias_check): Fix typo in comment.
22922 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
22923 * cgraphunit.c (symbol_table::compile): Likewise.
22924 * collect2.c (maybe_run_lto_and_relink): Likewise.
22925 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
22926 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
22927 * config/avr/avr.c (avr_map_op_t): Likewise.
22928 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
22929 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
22930 * config/epiphany/epiphany.md (movcc): Likewise.
22931 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
22932 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
22934 * config/mips/mips.c (mips_save_restore_reg): Likewise.
22935 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
22936 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
22937 * config/sh/sh.c (sh_rtx_costs): Likewise.
22938 * fold-const.c (fold_truth_andor): Likewise.
22939 * genautomata.c (collapse_flag): Likewise.
22940 * gengtype.h (struct type::u::s): Likewise.
22941 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
22942 * input.c (FORMAT_AMOUNT): Likewise.
22943 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
22944 (known_aggs_to_agg_replacement_list): Likewise.
22945 * ipa-inline-analysis.c: Likewise.
22946 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
22947 * ipa-polymorphic-call.c
22948 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
22949 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
22950 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
22952 * modulo-sched.c (apply_reg_moves): Likewise.
22953 * omp-expand.c (build_omp_regions_1): Likewise.
22954 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
22955 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
22956 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
22957 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
22958 * value-prof.c: Likewise.
22959 * var-tracking.c (val_reset): Likewise.
22961 2017-04-03 Richard Biener <rguenther@suse.de>
22963 PR tree-optimization/80275
22964 * fold-const.c (split_address_to_core_and_offset): Handle
22967 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
22969 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
22970 descriptors is at least equal to that of functions.
22972 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22974 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
22976 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
22979 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
22980 (mov<IMOD4:mode>): New expander.
22981 (*mov<IMOD4:mode>_internal): New insn and split pattern.
22983 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
22985 PR rtl-optimization/79405
22986 * fwprop.c (propagations_left): New variable.
22987 (forward_propagate_into): Decrement it.
22988 (fwprop_init): Initialize it.
22989 (fw_prop): If the variable has reached zero, stop propagating.
22990 (fwprop_addr): Ditto.
22992 2017-03-31 Jakub Jelinek <jakub@redhat.com>
22995 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
22996 a FUNCTION_DECL, pass it as decl instead of origin to
22999 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
23001 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
23004 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
23007 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
23008 TARGET_VSX_SMALL_INTEGER.
23010 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23012 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23013 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
23015 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
23017 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
23018 extraction from odd-numbered MSA register.
23020 2017-03-31 Jakub Jelinek <jakub@redhat.com>
23022 PR middle-end/80173
23023 * expmed.c (store_bit_field_1): Don't attempt to create
23024 a word subreg out of hard registers wider than word if they
23025 have HARD_REGNO_NREGS of 1 for their mode.
23027 PR middle-end/80163
23028 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
23029 conversions to integer types wider than word and pointer.
23032 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
23033 (rtx_equal_for_cselib_p): Pass 0 to it.
23034 * cselib.c (cselib_hasher::equal): Likewise.
23035 (rtx_equal_for_cselib_1): Add depth argument. If depth
23036 is 128, don't look up VALUE locs and punt. Increment
23037 depth in recursive calls when walking VALUE locs.
23039 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
23041 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
23042 (make_gcov_file_name): Use the canonical path name for generating
23044 (read_line): Fix handling of files with ascii null bytes.
23046 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
23048 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
23049 to initialise a vector register instead
23050 of using a const_int.
23052 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23054 PR translation/80189
23055 * gimplify.c (omp_default_clause): Use %qs instead of %s in
23056 diagnostic messages.
23058 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
23061 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
23062 (dfp_diex_<mode>): Update mode of operand 1.
23063 * doc/extend.texi (dxex, dxexq): Document change to return type.
23064 (diex, diexq): Document change to argument type.
23066 2017-03-30 Martin Jambor <mjambor@suse.cz>
23069 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
23070 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
23071 it reflects the signature changes performed at the callee side.
23072 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
23073 to cgraph_build_function_type_skip_args.
23074 (build_function_decl_skip_args): Adjust call to the above function.
23076 2017-03-30 Jakub Jelinek <jakub@redhat.com>
23079 * config/i386/sse.md
23080 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
23081 register as dest whenever it is a MEM not rtx_equal_p to the
23082 corresponding dup operand, and when forcing into reg move the
23083 reg into the memory afterwards.
23084 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
23085 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
23086 for the force_reg mode.
23087 (avx512vl_vextractf128<mode>): Use register as dest either
23088 always when a MEM, or when it is a MEM not rtx_equal_p to the
23089 corresponding dup operand, or even not when it is a CONST_VECTOR
23090 depending on the mode and lo vs. hi.
23091 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
23093 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
23094 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
23095 Likewise. Require that operands[2] is even.
23096 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
23097 Remove extraneous parens. Require that operands[2] is a multiple
23099 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
23100 operands[0] is a MEM if <mask_applied>, the predicates/constraints
23101 disallow memory then.
23103 2017-03-30 Richard Biener <rguenther@suse.de>
23105 PR tree-optimization/77498
23106 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
23107 to non-constants over backedges.
23109 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
23111 PR rtl-optimization/80233
23112 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
23113 as last_combined_insn. Do not test for BARRIER_P separately.
23115 2017-03-29 Andreas Schwab <schwab@suse.de>
23118 * calls.c (prepare_call_address): Convert funexp to Pmode before
23119 copying to temp reg.
23121 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23123 PR tree-optimization/80158
23124 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
23125 Handle possible future case of more than one alternate
23127 (replace_rhs_if_not_dup): Likewise.
23128 (replace_one_candidate): Likewise.
23130 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
23132 PR rtl-optimization/80193
23133 * ira.c (ira): Do not check allocation for LRA.
23135 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
23137 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
23138 (nvptx_output_simt_exit): Declare.
23139 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
23140 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
23141 (init_softstack_frame): Move initialization of crtl->is_leaf to...
23142 (nvptx_declare_function_name): ...here. Emit declaration of local
23143 memory space buffer for omp_simt_enter insn.
23144 (nvptx_output_unisimt_switch): New.
23145 (nvptx_output_softstack_switch): New.
23146 (nvptx_output_simt_enter): New.
23147 (nvptx_output_simt_exit): New.
23148 * config/nvptx/nvptx.h (struct machine_function): New fields
23149 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
23150 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
23151 (UNSPECV_SIMT_EXIT): Ditto.
23152 (omp_simt_enter_insn): New insn.
23153 (omp_simt_enter): New expansion.
23154 (omp_simt_exit): New insn.
23155 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
23157 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
23158 (expand_GOMP_SIMT_ENTER_ALLOC): New.
23159 (expand_GOMP_SIMT_EXIT): New.
23160 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
23161 (GOMP_SIMT_ENTER_ALLOC): Ditto.
23162 (GOMP_SIMT_EXIT): Ditto.
23163 * target-insns.def (omp_simt_enter): New insn.
23164 (omp_simt_exit): Ditto.
23165 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
23167 (lower_rec_simd_input_clauses): Implement SIMT privatization.
23168 (lower_rec_input_clauses): Likewise.
23169 (lower_lastprivate_clauses): Handle SIMT privatization.
23171 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
23172 (ompdevlow_adjust_simt_enter): New.
23173 (find_simtpriv_var_op): New.
23174 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
23175 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
23177 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
23178 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
23179 (copy_decl_for_dup_finish): Ditto.
23181 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
23183 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
23186 * config/i386/i386.c (ix86_option_override_internal): Always
23187 allow -mpreferred-stack-boundary=3 for 64-bit targets.
23189 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23191 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
23193 2017-03-28 Bin Cheng <bin.cheng@arm.com>
23195 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
23196 mark new edge's irreducible flag accordign to it.
23197 (vect_do_peeling): Check loop preheader edge's irreducible flag
23198 and pass it to function slpeel_add_loop_guard.
23200 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
23202 PR tree-optimization/80218
23203 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
23204 Update block frequencies and counts.
23206 2017-03-28 Richard Biener <rguenther@suse.de>
23208 PR tree-optimization/78644
23209 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
23210 of a simplification result we may not use it at all.
23212 2017-03-28 Richard Biener <rguenther@suse.de>
23215 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
23216 without arguments, generate default definition of a SSA name.
23218 2017-03-28 Richard Biener <rguenther@suse.de>
23220 PR middle-end/80222
23221 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
23222 TYPE_REF_CAN_ALIAS_ALL references.
23223 * fold-const.c (fold_indirect_ref_1): Likewise.
23225 2017-03-28 Martin Liska <mliska@suse.cz>
23228 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
23229 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
23231 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23232 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
23234 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
23235 (EXTRA_SPECS): Define.
23236 (SUBTARGET_EXTRA_SPECS): Likewise.
23237 (SUBTARGET_CPP_SPEC): Likewise.
23238 * config/arc/elf.h (EXTRA_SPECS): Renamed to
23239 SUBTARGET_EXTRA_SPECS.
23240 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
23242 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
23244 * config/arc/simdext.md (vst64_insn): Update pattern.
23245 (vld32wh_insn): Likewise.
23246 (vld32wl_insn): Likewise.
23247 (vld64_insn): Likewise.
23248 (vld32_insn): Likewise.
23250 2017-03-28 Marek Polacek <polacek@redhat.com>
23253 * fold-const.c (fold_comparison): Use protected_set_expr_location
23254 instead of SET_EXPR_LOCATION.
23256 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
23258 * tree.c (add_expr): Avoid name lookup warning.
23260 2017-03-27 Jeff Law <law@redhat.com>
23262 PR tree-optimization/80216
23263 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
23264 function name. Limit recursion depth.
23265 (record_temporary_equivalences): Corresponding changes.
23267 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
23269 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
23272 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23275 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
23277 * cfgcleanup.c (reg_note_cfa_p): New array.
23278 (insns_have_identical_cfa_notes): New function.
23279 (old_insns_match_p): Don't cross-jump in between /f
23280 and non-/f instructions. If both i1 and i2 are frame related,
23281 verify all CFA notes, their order and content.
23283 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
23286 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
23287 HImode and SImode with zero extend to DImode to one insn.
23288 (bswap<mode>2_extenddi): Likewise.
23289 (bswapsi2_extenddi): Likewise.
23290 (bswaphi2_extendsi): Likewise.
23291 (bswaphi2): Combine bswap HImode and SImode into one insn.
23292 Separate memory insns from swapping register.
23293 (bswapsi2): Likewise.
23294 (bswap<mode>2): Likewise.
23295 (bswaphi2_internal): Delete, no longer used.
23296 (bswapsi2_internal): Likewise.
23297 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
23298 store, and gpr<-gpr swap insns.
23299 (bswap<mode>2_store): Likewise.
23300 (bswaphi2_reg): Register only splitter, combine with the splitter.
23301 (bswaphi2 splitter): Likewise.
23302 (bswapsi2_reg): Likewise.
23303 (bswapsi2 splitter): Likewise.
23304 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
23305 the insns into load, store, and register/register insns.
23306 (bswapdi2_ldbrx): Likewise.
23307 (bswapdi2_load): Likewise.
23308 (bswapdi2_store): Likewise.
23309 (bswapdi2_reg): Likewise.
23311 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
23313 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
23314 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
23316 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
23319 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
23321 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23322 special handling for target option conflicts between dform
23323 options (-mpower9-dform, -mpower9-dform-vector,
23324 -mpower9-dform-scalar) and -mno-direct-move.
23326 2017-03-27 Richard Biener <rguenther@suse.de>
23328 PR tree-optimization/80181
23329 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
23331 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23333 * config/arc/predicates.md (move_double_src_operand): Replace the
23334 call to move_double_src_operand with a call to address_operand.
23336 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23338 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
23339 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
23340 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
23342 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23344 * config/arc/predicates.md (long_immediate_loadstore_operand):
23345 Consider scaled addresses cases.
23347 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
23349 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
23350 restored when in interrupt.
23351 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
23352 doesn't have delay slot.
23354 2017-03-27 Richard Biener <rguenther@suse.de>
23357 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
23358 inlined thunk clones.
23360 2017-03-27 Jakub Jelinek <jakub@redhat.com>
23363 * asan.c (instrument_derefs): Copy over last operand from
23364 original COMPONENT_REF to the new COMPONENT_REF with
23365 DECL_BIT_FIELD_REPRESENTATIVE.
23366 * ubsan.c (instrument_object_size): Likewise.
23368 2017-03-27 Richard Biener <rguenther@suse.de>
23370 PR tree-optimization/80170
23371 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
23372 sure DR/SCEV didnt fold in constants we do not see when looking
23373 at the reference base alignment.
23375 2017-03-27 Richard Biener <rguenther@suse.de>
23377 PR middle-end/80171
23378 * gimple-fold.c (fold_ctor_reference): Properly guard against
23379 NULL return value from canonicalize_constructor_val.
23381 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
23384 * config/i386/i386.c (ix86_expand_builtin)
23385 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
23386 flags reg setting and flags reg using instructions.
23387 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
23388 clobbering instructions to zero extend op2.
23390 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
23392 * doc/install.texi (Configuration) <--with-aix-soname>:
23393 Update link to AIX ld.
23395 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
23397 PR rtl-optimization/80160
23398 PR rtl-optimization/80159
23399 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
23400 reg_alternate_class into account.
23402 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
23405 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
23406 to consider in curr_insn_transform.
23408 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23410 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
23411 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
23412 and emit_mode_inner.
23414 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23416 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
23417 argument to the overloaded builtin variants. Use the new flag to
23418 deprecate certain builtin variants.
23419 * config/s390/s390-builtin-types.def: Add new builtin types.
23420 * config/s390/s390-builtins.h: Support new flags field for
23421 overloaded builtins.
23422 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
23423 (s390_macro_to_expand): Enable vector float data type.
23424 (s390_cpu_cpp_builtins_internal): Indicate support of the new
23425 builtins by incrementing the __VEC__ version number.
23426 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
23428 (s390_resolve_overloaded_builtin): Emit error messages depending
23429 on the builtin flags.
23430 * config/s390/s390.c (s390_expand_builtin): Support additional
23431 flags argument. Change error message to match the messages
23432 emitted in s390-c.c.
23433 * config/s390/s390.md: New UNSPEC_* constants.
23434 (op_type): Add new instruction types.
23435 * config/s390/vecintrin.h: Add new builtins and test data class
23437 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
23438 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
23439 (VEC_INEXACT, VEC_NOINEXACT): New constants.
23440 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
23441 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
23442 ("vec_mergel<mode>"): V_HW -> VEC_HW.
23444 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
23445 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
23446 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
23447 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
23449 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
23450 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
23451 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
23452 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
23454 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
23455 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
23456 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
23457 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
23458 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
23459 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
23460 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
23462 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
23463 ("vec_scatter_element<V_HW_4:mode>_DI")
23464 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
23465 ("vec_fpint<mode>", "vflls")
23466 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
23467 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
23468 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
23469 ("*vec_cmphe<mode>_cc"): ... these.
23471 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
23472 mode constant instead of magic value.
23474 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23476 * config/s390/s390.c (s390_expand_vec_compare): Support other
23477 vector floating point modes than just V2DF.
23478 (s390_expand_vcond): Likewise.
23479 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
23480 (s390_cannot_change_mode_class): Prevent mode changes between TF
23481 and V1TF in vector registers.
23482 * config/s390/s390.md (DF, SF): New mode attributes.
23483 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
23484 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
23485 SFmode support for VRs.
23486 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
23488 (VFT, VF_HW): New mode iterators.
23489 (vw, sdx): New mode attributes.
23490 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
23491 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
23492 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
23493 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
23494 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
23495 also the new vector floating point modes. Renaming to ...
23497 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
23498 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
23499 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
23500 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
23501 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
23502 ("vec_unordered<mode>"): ... these.
23504 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
23505 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
23506 ("*vec_extendv2df"): New insn definitions.
23508 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23510 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
23511 ("mulditi3_2", "*muldi3_sign"): New patterns.
23512 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
23513 rename the pattern definition.
23515 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23517 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
23519 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
23521 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23523 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
23524 instruction if possible.
23525 * config/s390/vector.md (vec_halfnumelts): New mode
23527 ("*vec_vllezlf<mode>"): New pattern.
23529 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23531 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
23532 ("popcountv4si2", "popcountv2di2"): Rename to ...
23533 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
23534 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
23536 ("popcount<mode>2_vxe"): New pattern.
23538 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23540 * common/config/s390/s390-common.c (processor_flags_table): Add
23542 * config.gcc: Add arch12.
23543 * config/s390/driver-native.c (s390_host_detect_local_cpu):
23544 Default to arch12 for unknown CPU model numbers.
23545 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
23546 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
23547 PROCESSOR_max sanity check.
23548 * config/s390/s390-opts.h (enum processor_type): Add
23550 * config/s390/s390.c (processor_table): Add arch12.
23551 (s390_expand_builtin): Add check for B_VXE flag.
23552 (s390_issue_rate): Add PROCESSOR_ARCH12.
23553 (s390_get_sched_attrmask): Likewise.
23554 (s390_get_unit_mask): Likewise.
23555 (s390_sched_score): Enable z13 scheduling for arch12.
23556 (s390_sched_reorder): Likewise.
23557 (s390_sched_variable_issue): Likewise.
23558 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
23560 (s390_tune_attr): Use z13 scheduling also for arch12.
23561 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
23562 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
23563 (TARGET_VXE_P): New macros.
23564 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
23565 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
23566 * config/s390/s390.opt: Add arch12 as processor_type.
23568 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23570 * config/s390/s390.md
23571 ("fixuns_truncdddi2", "fixuns_trunctddi2")
23572 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
23573 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
23575 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
23576 Rename expanders to ...
23578 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
23579 ("fixuns_truncdddi2_emu"): ... these.
23581 ("fixuns_trunc<mode>si2_emu"): New expander.
23583 ("*fixuns_truncdfdi2_z13"): Rename to ...
23584 ("*fixuns_truncdfdi2_vx"): ... this.
23586 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23588 * config/s390/2964.md: Remove the single element vector compare
23589 instructions which are no longer used.
23590 * config/s390/s390.c (s390_select_ccmode): Remove handling of
23592 (s390_canonicalize_comparison): Remove handling of DFmode
23594 (s390_expand_vec_compare_scalar): Remove function.
23595 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
23596 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
23598 ("*cmp<mode>_ccs"): Add wfcdb instruction.
23600 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23602 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
23604 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
23605 will anyway by matched by mov<mode>_64dfp.
23607 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23609 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
23610 vlef/vstef. Add missing operand to vleif.
23612 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23614 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
23615 pair for all vector types with 64 bit elements.
23616 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
23617 * config/s390/vector.md (V_HW_64): ... here.
23618 (V_128_NOSINGLE): New mode iterator.
23619 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
23620 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
23621 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
23622 ("*vec_load_pairv2di"): Change to ...
23623 ("*vec_load_pair<mode>"): ... this one.
23625 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23627 * config/s390/constraints.md: Add comments.
23628 (jKK): Reject element sizes > 8 bytes.
23629 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
23631 * config/s390/s390.md: Add the s_operand checks formerly in
23632 s390_split_ok_p to various splitters where they are still
23634 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
23635 for 128 bit vectors. Plus two splitters.
23637 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23639 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
23642 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23645 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
23646 error if the boundary argument is not constant.
23648 2017-03-24 Jakub Jelinek <jakub@redhat.com>
23650 PR rtl-optimization/80112
23651 * loop-doloop.c (doloop_condition_get): Don't check condition
23652 if cmp isn't SET with IF_THEN_ELSE src.
23654 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23656 PR tree-optimization/80158
23657 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
23658 replacing a candidate statement, also replace it for the
23659 candidate's alternate interpretation.
23660 (replace_rhs_if_not_dup): Likewise.
23661 (replace_one_candidate): Likewise.
23663 2017-03-24 Richard Biener <rguenther@suse.de>
23665 PR tree-optimization/80167
23666 * graphite-isl-ast-to-gimple.c
23667 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
23669 (translate_isl_ast_to_gimple::get_rename): Likewise.
23671 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
23673 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
23674 handling of certain combinations of target options, including the
23675 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
23676 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
23678 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23681 * config/arm/arm.md (*load_multiple): Add reload_completed to
23682 matching condition.
23684 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23685 Richard Biener <rguenth@suse.de>
23687 PR tree-optimization/79908
23688 PR tree-optimization/80136
23689 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23690 been cast away, gimplify_and_add suffices.
23692 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
23694 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
23696 2017-03-23 Richard Biener <rguenther@suse.de>
23698 PR tree-optimization/80032
23699 * gimplify.c (gimple_push_cleanup): Forced unconditional
23700 cleanups still have to go to the conditional_cleanups
23703 2017-03-22 Jakub Jelinek <jakub@redhat.com>
23705 PR tree-optimization/80072
23706 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
23708 (next_operand_entry_id): Change type to unsigned int.
23709 (sort_by_operand_rank): Make sure to return the right return value
23710 even if unsigned fields are bigger than INT_MAX.
23711 (struct oecount): Change cnt and id type to unsigned int.
23712 (oecount_hasher::equal): Formatting fix.
23713 (oecount_cmp): Make sure to return the right return value
23714 even if unsigned fields are bigger than INT_MAX.
23715 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
23718 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
23719 TREE_READONLY on result if writing it more than once.
23722 * doc/invoke.texi (-fsanitize=thread): Document that with
23723 -fnon-call-exceptions atomics are not able to throw
23727 * tsan.c: Include tree-eh.h.
23728 (instrument_builtin_call): Call maybe_clean_eh_stmt or
23729 maybe_clean_or_replace_eh_stmt where needed.
23730 (instrument_memory_accesses): Add cfg_changed argument.
23731 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
23732 if it returned true.
23733 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
23735 PR rtl-optimization/63191
23736 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
23737 wrapper function, moved the whole old content into ...
23738 (ix86_delegitimize_address_1): ... this. New inline function.
23739 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
23740 true as last argument instead of ix86_delegitimize_address.
23742 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23744 * config/aarch64/aarch64.c (generic_branch_cost): Copy
23745 cortexa57_branch_cost.
23747 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
23749 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
23751 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23754 * doc/md.texi (Constraints): Document wA constraint.
23755 * config/rs6000/constraints.md (wA): New.
23756 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
23757 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
23758 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
23759 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
23761 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
23764 * gimplify.c (is_oacc_declared): New function.
23765 (oacc_default_clause): Use it to set default flags for acc declared
23766 variables inside parallel regions.
23767 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
23768 declared variables.
23769 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
23770 declare attribute to any decl as necessary.
23772 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
23775 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
23776 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
23777 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
23778 (arm_arch_lpae): This.
23779 * config/arm/arm.c (arm_arch7ve): Rename into ...
23780 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
23781 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
23784 2017-03-22 Martin Liska <mliska@suse.cz>
23787 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
23788 error message instead of an ICE.
23790 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23792 * doc/extend.texi (6.11 Additional Floating Types): Revise.
23794 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
23796 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
23798 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
23801 2017-03-21 Martin Sebor <msebor@redhat.com>
23803 * doc/extend.texi: Use "cannot" instead of "can't."
23804 * doc/hostconfig.texi: Same.
23805 * doc/install.texi: Same.
23806 * doc/invoke.texi: Same.
23807 * doc/loop.texi: Same.
23808 * doc/md.texi: Same.
23809 * doc/objc.texi: Same.
23810 * doc/rtl.texi: Same.
23811 * doc/tm.texi: Same.
23812 * doc/tm.texi.in: Same.
23813 * doc/trouble.texi: Same.
23815 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
23818 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
23819 (collect_checksum_attributes): Set it.
23820 (die_checksum_ordered): Use it.
23822 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23824 PR tree-optimization/79908
23825 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
23826 change: For a VA_ARG whose LHS has been cast away, use
23827 force_gimple_operand to construct the side effects.
23829 2017-03-21 David Malcolm <dmalcolm@redhat.com>
23831 PR translation/80001
23832 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
23833 more amenable to translation.
23834 (oacc_loop_auto_partitions): Likewise.
23836 2017-03-21 Marek Polacek <polacek@redhat.com>
23837 Martin Sebor <msebor@redhat.com>
23839 PR tree-optimization/80109
23840 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
23841 on INTEGRAL_TYPE_P.
23843 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23844 Segher Boessenkool <segher@kernel.crashing.org>
23847 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
23848 check reg_used_between_p between insn and one of succ or succ2
23849 depending on if succ is artificial insn not inserted into insn
23852 2017-03-21 Martin Liska <mliska@suse.cz>
23854 PR gcov-profile/80081
23855 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
23856 * doc/gcc.texi: Include gcov-dump stuff.
23857 * doc/gcov-dump.texi: New file.
23859 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
23861 PR rtl-optimization/79150
23862 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
23863 conditional jump, if the jump is the last insn of the loop.
23865 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23866 Richard Biener <rguenth@suse.de>
23868 PR tree-optimization/79908
23869 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
23870 been cast away, use force_gimple_operand to construct the side
23873 2017-03-21 Martin Liska <mliska@suse.cz>
23875 PR libfortran/79956
23876 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
23879 2017-03-21 Brad Spengler <spender@grsecurity.net>
23882 * plugin.c (htab_hash_plugin): New function.
23883 (add_new_plugin): Use it and adjust.
23884 (parse_plugin_arg_opt): Adjust.
23885 (init_one_plugin): Likewise.
23887 2017-03-21 Richard Biener <rguenther@suse.de>
23889 PR tree-optimization/80032
23890 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
23891 if set force the cleanup to happen unconditionally.
23892 (gimplify_target_expr): Push inserted clobbers with force_uncond
23893 to avoid them being removed by control-dependent DCE.
23895 2017-03-21 Richard Biener <rguenther@suse.de>
23897 PR tree-optimization/80122
23898 * tree-inline.c (copy_bb): Do not expans va-arg packs or
23899 va_arg_pack_len when the inlined call stmt requires pack
23901 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
23903 2017-03-21 Jakub Jelinek <jakub@redhat.com>
23906 * tsan.c (instrument_builtin_call): If the memory model argument
23907 is not a constant, assume it is valid.
23910 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
23913 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23915 PR rtl-optimization/79910
23916 * combine.c (can_combine_p): Do not allow combining an I0 or I1
23917 if its dest is used by an insn before I2 (other than the combined
23918 insns themselves, which are properly handled already).
23920 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
23923 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
23925 * combine.c (record_used_regs): New static function.
23926 (try_combine): Handle situations where there is an additional
23927 instruction between I2 and I3 which needs to have a LOG_LINK
23931 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
23933 * combine.c (try_combine): Delete redundant i1 test. Call
23934 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
23936 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23939 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
23940 alternatives 13/14.
23942 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23944 PR tree-optimization/80054
23945 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
23946 the optimization if a PHI or any of its arguments is not dominated
23947 by the candidate's basis. Use gphi* rather than gimple* as
23949 (replace_profitable_candidates): Clean up a gimple* variable that
23950 should be a gphi* variable.
23952 2017-03-20 Martin Sebor <msebor@redhat.com>
23955 * doc/extend.texi (attribute constructor): Document present limitation.
23957 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
23960 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
23961 __POWER9_VECTOR__ #ifdef control, change template definition to
23962 use Power9-specific built-in function.
23963 (vec_any_eq): Likewise.
23964 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
23965 to control outcomes from this test.
23966 (vector_ae_<mode>p): For VEC_F modes, likewise.
23968 2017-03-20 Ian Lance Taylor <iant@google.com>
23970 * config/i386/i386.c (ix86_function_regparm): Save an extra
23971 register for -fsplit-stack with DECL_STATIC_CHAIN.
23973 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23976 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
23977 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
23979 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
23981 * config/riscv/riscv.c (riscv_print_operand): Use "fence
23983 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
23986 2017-03-20 Marek Polacek <polacek@redhat.com>
23989 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
23991 2017-03-20 Richard Biener <rguenther@suse.de>
23993 PR tree-optimization/80113
23994 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
23995 allocate extra SSA name for PHI def.
23996 (add_close_phis_to_outer_loops): Likewise.
23997 (add_close_phis_to_merge_points): Likewise.
23998 (copy_loop_close_phi_args): Likewise.
23999 (copy_cond_phi_nodes): Likewise.
24001 2017-03-20 Martin Liska <mliska@suse.cz>
24003 PR middle-end/79753
24004 * tree-chkp.c (chkp_build_returned_bound): Do not build
24005 returned bounds for a LHS that's not a BOUNDED_P type.
24007 2017-03-20 Martin Liska <mliska@suse.cz>
24011 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
24012 COMPLEX_CST and VECTOR_CST.
24014 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24017 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
24018 target operand. A new splitter adds the clobber statement in case
24019 the target operand is dead anyway.
24021 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
24023 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
24024 to age-old versions of binutils and glibc.
24026 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
24028 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
24030 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24032 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
24034 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
24036 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
24037 requirement for binutils 2.13.
24039 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
24041 * combine.c (try_combine): Delete redundant i1 test. Call
24042 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
24044 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
24046 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
24047 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
24049 <riscv64-*-elf>: Re-arrange section
24050 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
24051 <riscv32-*-linux>: Likewise.
24052 <riscv64-*-elf>: Likewise
24053 <riscv64-*-linux>: Likewise.
24055 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
24058 * aarch64.opt(verbose-cost-dump): Fix typo.
24060 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
24063 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
24064 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
24066 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
24068 * reload.c (find_reloads): When reloading a nonoffsettable address,
24069 use RELOAD_OTHER for it and its address reloads.
24071 PR rtl-optimization/79910
24072 * combine.c (record_used_regs): New static function.
24073 (try_combine): Handle situations where there is an additional
24074 instruction between I2 and I3 which needs to have a LOG_LINK
24077 2017-03-17 Jeff Law <law@redhat.com>
24079 PR tree-optimization/71437
24080 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
24081 conditional in the hash table first.
24082 (vrp_dom_walker::before_dom_children): Extract condition from
24083 ASSERT_EXPR. Record condition, its inverion and any implied
24084 conditions as well.
24086 2017-03-17 Marek Polacek <polacek@redhat.com>
24087 Markus Trippelsdorf <markus@trippelsdorf.de>
24089 PR tree-optimization/80079
24090 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
24093 2017-03-17 Richard Biener <rguenther@suse.de>
24095 PR middle-end/80075
24096 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
24097 Properly verify the LHS before the RHS possibly claims to be
24099 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
24102 2017-03-17 Martin Jambor <mjambor@suse.cz>
24104 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
24105 (List of -O2 options): Likewise.
24106 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
24109 2017-03-17 Tom de Vries <tom@codesourcery.com>
24111 * gcov-dump.c (print_usage): Print bug_report_url.
24113 2017-03-17 Richard Biener <rguenther@suse.de>
24115 PR middle-end/80050
24116 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
24117 (parser::peek): Likewise.
24119 2017-03-17 Richard Biener <rguenther@suse.de>
24121 PR tree-optimization/80048
24122 * sese.c (free_sese_info): Properly release rename_map and
24123 copied_bb_map elements.
24125 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
24127 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
24128 Add linked-list forward and backlinks. Insert on
24129 construction, remove on destruction.
24130 (class pass_store_merging): Add m_stores_head field.
24131 (pass_store_merging::terminate_and_process_all_chains):
24132 Iterate over m_stores_head list.
24133 (pass_store_merging::terminate_all_aliasing_chains):
24135 (pass_store_merging::execute): Check for debug stmts first.
24136 Push new chains onto the m_stores_head stack.
24138 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
24141 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
24142 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
24143 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
24145 2017-03-16 Jeff Law <law@redhat.com>
24147 PR tree-optimization/71437
24148 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
24149 member function. Implementation moved into after_dom_children
24150 member function and into the threader's thread_outgoing_edges
24152 (dom_opt_dom_walker::after_dom_children): Simplify by moving
24153 some code into new thread_outgoing_edges.
24154 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
24155 definition. Simplify marker handling (do it here). Assume we always
24156 have the available expression and the const/copies tables.
24157 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
24159 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
24160 * tree-vrp.c (equiv_stack): No longer file scoped.
24161 (vrp_dom_walker): New class.
24162 (vrp_dom_walker::before_dom_children): New member function.
24163 (vrp_dom_walker::after_dom_children): Likewise.
24164 (identify_jump_threads): Setup domwalker. Use it rather than
24165 walking edges in a random order by hand. Simplify setup/finalization.
24166 (finalize_jump_threads): Remove.
24167 (vrp_finalize): Do not call identify_jump_threads here.
24168 (execute_vrp): Do it here instead and call thread_through_all_blocks
24171 PR tree-optimization/71437
24172 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
24174 (simplify_stmt_for_jump_threading): Add basic_block argument. All
24176 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
24177 (dom_opt_dom_walker::thread_across_edge): Remove
24178 handle_dominating_asserts argument. All callers changed.
24179 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
24180 changes. Remove calls to lhs_of_dominating_assert. Other
24181 uses of handle_dominating_asserts turn into unconditional code
24182 (simplify_control_stmt_condition_1): Likewise.
24183 (simplify_control_stmt_condition): Likewise.
24184 (thread_through_normal_block, thread_across_edge): Likewise.
24185 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
24186 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
24187 object if it is not an SSA_NAME.
24188 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
24189 before calling into the VRP specific simplifiers.
24190 (identify_jump_threads): Remove handle_dominating_asserts
24193 2017-03-16 Jakub Jelinek <jakub@redhat.com>
24196 * tree-diagnostic.c (default_tree_printer): No longer static.
24197 * tree-diagnostic.h (default_tree_printer): New prototype.
24199 2017-03-16 Tamar Christina <tamar.christina@arm.com>
24201 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
24202 Change ins into fmov.
24204 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24206 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
24207 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
24208 Use h_con constraint for operand 1.
24209 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
24210 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
24212 2017-03-15 Jeff Law <law@redhat.com>
24214 PR tree-optimization/71437
24215 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
24216 (record_temporary_equivalences): Use it.
24218 PR tree-optimization/71437
24219 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
24220 tree-ssa-scopedtables.
24221 (lookup_avail_expr, build_and_record_new_cond): Likewise.
24222 (record_conditions, record_cond, vuse_eq): Likewise.
24223 (record_edge_info): Adjust to API tweak of record_conditions.
24224 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
24225 (record_temporary_equivalences, optimize_stmt): Likewise.
24226 (eliminate_redundant_computations): Likewise.
24227 (record_equivalences_from_stmt): Likewise.
24228 * tree-ssa-scopedtables.c: Include options.h and params.h.
24229 (vuse_eq): New function, moved from tree-ssa-dom.c
24230 (build_and_record_new_cond): Likewise.
24231 (record_conditions): Likewise. Accept vector of conditions rather
24232 than edge_equivalence structure for first argument.
24233 for the first argument.
24234 (avail_exprs_stack::lookup_avail_expr): New member function, moved
24235 from tree-ssa-dom.c.
24236 (avail_exprs_stack::record_cond): Likewise.
24237 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
24238 from tree-ssa-dom.c.
24239 (avail_exprs_stack): Add new member functions lookup_avail_expr
24241 (record_conditions): Declare.
24243 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
24246 * lra-constraints.c (process_alt_operands): Increase reject for
24247 reloading an input/output operand.
24249 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
24252 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
24253 insns to convert from signed/unsigned char/short to IEEE 128-bit
24255 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
24257 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
24260 * config/i386/i386.c (ix86_vector_duplicate_value): Create
24261 subreg of inner mode for values already in registers.
24263 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
24265 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
24266 iteration reg is used after the loop.
24268 2017-03-14 Martin Sebor <msebor@redhat.com>
24270 PR tree-optimization/79800
24271 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
24272 precision in negative-positive range.
24273 (format_floating): Call non-const overload with adjusted precision.
24275 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24278 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
24281 2017-03-14 Martin Sebor <msebor@redhat.com>
24283 PR middle-end/80020
24284 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
24285 * builtins.def (aligned_alloc): Use it.
24288 * Makefile.in (GTFILES): Add calls.c.
24289 * calls.c: Include "gt-calls.h".
24291 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
24293 PR rtl-optimization/79728
24294 * regs.h (struct target_regs): New field
24295 x_contains_allocatable_regs_of_mode.
24296 (contains_allocatable_regs_of_mode): New macro.
24297 * reginfo.c (init_reg_sets_1): Initialize it, and change
24298 contains_reg_of_mode so it includes global regs as well.
24299 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
24300 rather than contains_regs_of_mode.
24302 2017-03-14 Martin Liska <mliska@suse.cz>
24304 * doc/invoke.texi: Document options that can't be combined with
24305 -fcheck-pointer-bounds.
24307 2017-03-14 Martin Liska <mliska@suse.cz>
24309 PR middle-end/79831
24310 * doc/invoke.texi (-Wchkp): Document the option.
24312 2017-03-14 Martin Liska <mliska@suse.cz>
24314 * Makefile.in: Install gcov-dump.
24316 2017-03-14 Martin Liska <mliska@suse.cz>
24318 * multiple_target.c (expand_target_clones): Bail out for
24319 an invalid attribute.
24321 2017-03-14 Richard Biener <rguenther@suse.de>
24323 * alias.c (struct alias_set_entry): Pack properly.
24324 * cfgloop.h (struct loop): Likewise.
24325 * cse.c (struct set): Likewise.
24326 * ipa-utils.c (struct searchc_env): Likewise.
24327 * loop-invariant.c (struct invariant): Likewise.
24328 * lra-remat.c (struct cand): Likewise.
24329 * recog.c (struct change_t): Likewise.
24330 * rtl.h (struct address_info): Likewise.
24331 * symbol-summary.h (function_summary): Likewise.
24332 * tree-loop-distribution.c (struct partition): Likewise.
24333 * tree-object-size.c (struct object_size_info): Likewise.
24334 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
24335 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
24336 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
24337 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
24338 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
24339 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
24340 (struct _stmt_vec_info): Likewise.
24342 2017-03-14 Martin Liska <mliska@suse.cz>
24345 * multiple_target.c (create_dispatcher_calls): Check that
24346 a target can create a function dispatcher.
24348 2017-03-14 Martin Liska <mliska@suse.cz>
24351 * multiple_target.c (expand_target_clones): Drop local.local
24352 flag for default implementation.
24354 2017-03-14 Richard Biener <rguenther@suse.de>
24356 PR tree-optimization/80030
24357 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
24359 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
24361 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
24362 gcc_fallthrough() instead of __attribute__((fallthrough));
24364 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24366 * doc/gcc.texi: Remove "up" link to (DIR).
24367 * doc/gccint.texi: Ditto.
24369 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
24371 * doc/install.texi (Specific) <avr>: Remove reference to
24374 2017-03-13 Jeff Law <law@redhat.com>
24376 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
24377 attribute rather than comments.
24379 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
24380 match_scratch operand is highest.
24382 2017-03-13 Martin Liska <mliska@suse.cz>
24384 PR middle-end/78339
24385 * ipa-pure-const.c (warn_function_noreturn): If the declarations
24386 is a CHKP clone, use original declaration.
24388 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24390 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
24391 (arc_conditional_register_usage): Use a different allocation order
24392 when optimizing for size.
24393 * common/config/arc/arc-common.c (arc_option_optimization_table):
24394 Section anchors default on when optimizing for size.
24396 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24398 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
24400 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24402 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
24403 * config/arc/arc.md (cpu_facility): Add cd variant.
24404 (*movqi_insn): Add code density variant.
24405 (*movhi_insn): Likewise.
24406 (*movqi_insn): Likewise.
24407 (*addsi3_mixed): Likewise.
24408 (subsi3_insn): Likewise.
24410 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24412 * config/arc/arc.md (movsi_cond_exec): Update constraint.
24414 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
24416 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
24417 expressions with MINUS and UNARY ops.
24419 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24422 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
24424 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
24425 between vec_select and vector argument.
24426 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
24427 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
24428 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
24429 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
24430 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
24431 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
24433 2017-03-13 Richard Biener <rguenther@suse.de>
24436 * params.def (vect-max-peeling-for-alignment): Fix typo.
24438 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24440 * doc/install.texi (Specific) <mips-*-*>: Remove description of
24441 issue that only occurred with binutils below 2.18.
24443 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24445 * doc/install.texi (Specific) <cris-axis-elf>: No longer
24446 refer to binutils 2.11/2.12 minimum.
24448 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
24450 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
24451 ftp.kernel.org and simplify binutils requirement.
24453 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
24455 * doc/invoke.texi (Warning Options): Fix spelling of link-time
24457 (Optimize Options): Ditto. Also remove redundancy.
24459 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24461 PR translation/79848
24462 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
24464 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
24465 to G_ to avoid double translation.
24467 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24469 PR translation/79923
24470 * auto-profile.c (get_combined_location): Convert leading
24471 character of diagnostics to lower case and remove trailing period.
24472 (read_profile): Likewise for various diagnostics.
24473 * config/arm/arm.c (arm_option_override): Remove trailing period
24474 from various diagnostics.
24475 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
24476 (msp430_expand_delay_cycles): Likewise.
24478 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24481 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
24482 full command-line argument, rather than just "str".
24483 (aarch64_validate_march): Likewise.
24484 (aarch64_validate_mtune): Likewise.
24486 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
24488 PR rtl-optimization/78911
24489 * lra-assigns.c (must_not_spill_p): New function.
24490 (spill_for): Use it.
24492 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24494 PR tree-optimization/79981
24495 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
24496 ATOMIC_COMPARE_EXCHANGE ifn result.
24497 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
24498 IFN_ATOMIC_COMPARE_EXCHANGE.
24500 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24503 * opts.c (parse_sanitizer_options): Add missing question mark to
24504 "did you mean" message.
24506 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24508 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
24510 (VMULEUH_UNS): Likewise.
24511 (VMULOUB_UNS): Likewise.
24512 (VMULOUH_UNS): Likewise.
24513 * config/rs6000/rs6000.c (builtin_function_type): Remove
24514 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
24516 2017-03-10 David Malcolm <dmalcolm@redhat.com>
24519 * read-rtl-function.c (function_reader::read_rtx_operand): Update
24520 x with result of extra_parsing_for_operand_code_0.
24521 (function_reader::extra_parsing_for_operand_code_0): Convert
24522 return type from void to rtx, returning x. When reading
24523 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
24524 larger size containing struct block_symbol.
24526 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
24528 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
24529 -mfloat128-hardware without -m64.
24531 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
24534 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
24535 entries to the case statement that marks unsigned arguments to
24536 overloaded functions.
24538 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
24540 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
24541 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
24543 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
24546 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
24547 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
24549 2017-03-10 Martin Liska <mliska@suse.cz>
24553 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
24554 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
24557 2017-03-10 Olivier Hainque <hainque@adacore.com>
24559 * tree-switch-conversion (array_value_type): Start by resetting
24560 candidate type to it's main variant.
24562 2017-03-10 Jakub Jelinek <jakub@redhat.com>
24564 PR rtl-optimization/79909
24565 * combine.c (try_combine): Use simplify_replace_rtx on individual
24566 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
24567 of the whole CALL_INSN_FUNCTION_USAGE.
24569 PR tree-optimization/79972
24570 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
24571 get_range_info on SSA_NAMEs. Formatting fixes.
24573 2017-03-10 Richard Biener <rguenther@suse.de>
24574 Jakub Jelinek <jakub@redhat.com>
24576 PR tree-optimization/77975
24577 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
24578 edge to be constant.
24579 (get_val_for): For constant x return it. Formatting fix.
24580 (loop_niter_by_eval): Avoid pointless looping if the next iteration
24581 would use the same bases as the current one.
24583 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24585 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
24586 instead of vec_select for V1TImode.
24587 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
24589 (VSX_LE_128): Add V1TI to this mode iterator.
24590 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
24591 (*vsx_le_perm_store_<mode>): Likewise.
24592 (pre-reload splitter for VSX stores): Likewise.
24593 (post-reload splitter for VSX stores): Likewise.
24594 (*vsx_xxpermdi2_le_<mode>): Likewise.
24595 (*vsx_lxvd2x2_le_<mode>): Likewise.
24596 (*vsx_stxvd2x2_le_<mode>): Likewise.
24598 2017-03-09 Michael Eager <eager@eagercon.com>
24600 Correct failures with --enable-checking=yes,rtl.
24602 * config/microblaze/microblaze.c (microblaze_expand_shift):
24603 Replace GET_CODE test with CONST_INT_P and INTVAL test with
24604 test for const0_rtx.
24605 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
24606 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
24608 2017-03-09 Richard Biener <rguenther@suse.de>
24610 PR tree-optimization/79977
24611 * graphite-scop-detection.c (scop_detection::merge_sese):
24612 Handle the case of extra exits to blocks dominating the entry.
24614 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
24616 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
24619 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
24621 PR rtl-optimization/79949
24622 * lra-constraints.c (process_alt_operands): Check memory when
24623 trying to predict a cycle. Print about the overall increase.
24625 2017-03-09 Richard Biener <rguenther@suse.de>
24627 PR middle-end/79971
24628 * gimple-expr.c (useless_type_conversion_p): Preserve
24629 TYPE_SATURATING for fixed-point types.
24631 2017-03-09 Richard Biener <rguenther@suse.de>
24634 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
24635 alignment of BLKmode params.
24637 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24640 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
24641 (VALL_NO_V2Q): Likewise.
24643 * config/aarch64/aarch64-simd.md
24644 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
24646 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
24647 VALL_NO_V2Q mode iterator.
24648 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
24650 2017-03-09 Martin Liska <mliska@suse.cz>
24652 PR tree-optimization/79631
24653 * tree-chkp-opt.c (chkp_is_constant_addr): Call
24654 tree_int_cst_sign_bit just for INTEGER constants.
24656 2017-03-09 Martin Liska <mliska@suse.cz>
24660 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
24663 2017-03-09 Marek Polacek <polacek@redhat.com>
24666 * tree.c (inchash::add_expr): Handle TREE_VEC.
24668 2017-03-09 Martin Liska <mliska@suse.cz>
24671 (chkp_narrow_size_and_offset): New function.
24672 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
24673 (void chkp_parse_bit_field_ref): New function.
24674 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
24675 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
24677 2017-03-09 Martin Liska <mliska@suse.cz>
24680 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
24681 (chkp_find_bounds_1): Remove gcc_unreachable.
24683 2017-03-09 Jakub Jelinek <jakub@redhat.com>
24686 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
24687 BUILT_IN_SYNC*, determine the access type from the size suffix and
24688 always build a MEM_REF with that type. Handle forgotten
24689 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
24692 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
24693 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
24694 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
24695 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
24696 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
24697 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
24698 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
24699 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
24700 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
24701 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
24702 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
24703 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
24704 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
24705 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
24706 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
24707 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
24708 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
24709 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
24710 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
24711 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
24712 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
24713 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
24714 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
24715 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
24716 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
24717 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
24718 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
24719 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
24720 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
24721 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
24722 definitions outside of __OPTIMIZE__ guarded section.
24725 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
24726 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
24727 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
24728 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
24731 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24733 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
24734 ("vfenez<mode>"): Add missing constraints.
24736 2017-03-08 Martin Sebor <msebor@redhat.com>
24739 * config/nds32/nds32.c (nds32_option_override):
24740 Fix misspelled diagnostic.
24742 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24745 * gimplify.c (gimplify_omp_for): Replace index var in outer
24746 taskloop statement with an artificial variable and add
24747 OMP_CLAUSE_PRIVATE clause for it.
24749 2017-03-08 Richard Biener <rguenther@suse.de>
24751 PR tree-optimization/79955
24752 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
24753 for accesses that are completely outside of the variable.
24755 2017-03-08 Andrew Haley <aph@redhat.com>
24757 PR tree-optimization/79943
24758 * tree-ssa-loop-split.c (compute_new_first_bound): When
24759 calculating the new upper bound, (END-BEG) should be added, not
24762 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24764 * config/avr/avr.md (setmemhi): Make sure match_dup
24765 operand number comes before match_scratch.
24767 2017-03-08 Richard Biener <rguenther@suse.de>
24769 PR tree-optimization/79920
24770 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
24771 with ncopies == 1 to ...
24772 (vect_transform_slp_perm_load): ... here. Properly compute
24773 all element loads by iterating VF times over the group. Do
24774 not handle ncopies (computed in a broken way) in
24775 vect_create_mask_and_perm.
24777 2017-03-08 Jakub Jelinek <jakub@redhat.com>
24780 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
24781 is a uniform vector, use uniform_vector_p return value instead of
24782 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
24784 2017-03-07 Marek Polacek <polacek@redhat.com>
24786 PR middle-end/79809
24787 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
24788 (alloca_call_type): Likewise.
24790 2017-03-07 Martin Liska <mliska@suse.cz>
24792 * gcov.c (process_args): Put comment to correct location.
24794 2017-03-07 Martin Liska <mliska@suse.cz>
24796 PR middle-end/68270
24797 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
24798 Use array_at_struct_end_p instead of DECL_CHAIN (field).
24799 (chkp_narrow_bounds_for_field): Likewise.
24800 (chkp_parse_array_and_component_ref): Pass one more argument to
24803 2017-03-07 Richard Biener <rguenther@suse.de>
24805 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
24808 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
24810 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
24811 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
24813 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24816 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
24817 to end of description.
24818 (PARAM_MAX_STORES_TO_MERGE): Likewise.
24820 2017-03-07 Jakub Jelinek <jakub@redhat.com>
24822 PR rtl-optimization/79901
24823 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
24825 (*avx512f_<code><mode>3<mask_name>): ... this.
24826 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
24827 iterator instead of VI8_AVX2_AVX512BW.
24829 PR rtl-optimization/79901
24830 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
24831 min/max expander, expand it using expand_vec_cond_expr.
24834 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
24837 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24840 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
24841 to void * for PCH reasons.
24842 * dwarf2out.c (output_loc_operands, output_die): Cast
24843 v.val_vec.array to unsigned char *.
24845 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
24848 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
24851 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
24853 PR rtl-optimization/79571
24854 * lra-constraints.c (process_alt_operands): Calculate static
24855 reject and subtract it from overall when only addresses will be
24858 2017-03-06 Julia Koval <julia.koval@intel.com>
24861 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
24862 incoming stack boundary to 128 for 64-bit targets.
24864 2017-03-06 Richard Biener <rguenther@suse.de>
24866 PR tree-optimization/79894
24867 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
24868 to NULL after folding it.
24870 2017-03-06 Richard Biener <rguenther@suse.de>
24872 PR tree-optimization/79824
24873 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
24874 check disabling peeling for gaps.
24876 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
24878 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
24879 attributes): Document gettimeofday.
24881 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24883 * config/s390/s390.c (s390_option_override_internal): Set
24884 PARAM_MIN_VECT_LOOP_BOUND
24886 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
24888 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
24889 * config/s390/s390.md: Likewise.
24891 2017-03-06 Jakub Jelinek <jakub@redhat.com>
24894 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
24895 (<avx2_avx512>_perm<mode>): Rename to ...
24896 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
24898 (<avx512>_perm<mode>_mask): Rename to ...
24899 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
24901 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
24902 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
24903 instead of VI8F_256_512.
24904 (avx512f_perm<mode>): New define_expand.
24905 (avx512f_perm<mode>_mask): Likewise.
24906 (avx512f_perm<mode>_1<mask_name>): New define_insn.
24907 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
24909 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24911 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
24912 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
24914 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
24916 2017-03-06 Martin Liska <mliska@suse.cz>
24919 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
24920 when having a SSA NAME w/o VAR_DECL assigned to it.
24922 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24924 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
24925 msa_dpsub_<su>_d): Fix MODE for vec_select.
24927 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
24929 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
24931 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
24933 2017-03-06 Richard Biener <rguenther@suse.de>
24935 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
24936 * plugin.c (register_plugin_info): Likewise.
24937 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
24939 2017-03-05 Jakub Jelinek <jakub@redhat.com>
24941 * config/i386/sse.md (sse_storehps, sse_storelps,
24942 avx_<castmode><avxsizesuffix>_<castmode>,
24943 avx512f_<castmode><avxsizesuffix>_<castmode>,
24944 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
24945 in condition that at least one operand is not a MEM.
24947 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24949 PR middle-end/79805
24950 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
24951 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
24953 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
24954 gimple_call_nothrow_p flag based on whether original builtin can throw.
24955 If it can, emit following stmts on the fallthrough edge.
24956 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
24957 don't create new bb if inserting just debug stmts on the edge, try to
24958 insert them on the fallthru bb or just reset debug stmts.
24960 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
24963 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
24964 restore recog_data (including the operand rtxes inside it) around
24965 the call to get_insn_template.
24967 2017-03-03 Martin Sebor <msebor@redhat.com>
24969 PR tree-optimization/79699
24970 * context.c (context::~context): Free MPFR caches to avoid
24971 a memory leak on program exit.
24973 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24975 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
24976 Use wide_int::ulow () instead of .elt (0).
24978 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
24980 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
24981 (*pushxf): Limit oF constraint to 32bit targets and add oC
24982 constraint for 64bit targets.
24983 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
24984 (*pushdf): Change rmF constraint to rmC.
24986 2017-03-03 Martin Liska <mliska@suse.cz>
24988 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
24989 Remove unused variable.
24991 2017-03-03 Jakub Jelinek <jakub@redhat.com>
24994 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
24995 is a memory operand, increase num_memory.
24996 (ix86_expand_args_builtin): Likewise.
24998 2017-03-03 Jan Hubicka <jh@suse.cz>
25001 * ipa-devirt.c (maybe_record_node): Properly handle
25002 __cxa_pure_virtual visibility.
25004 2017-03-03 Martin Liska <mliska@suse.cz>
25006 PR tree-optimization/79803
25007 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
25009 (pass_loop_prefetch::execute): Disabled optimization if an
25010 assumption about L1 cache size is not met.
25012 2017-03-03 Martin Liska <mliska@suse.cz>
25014 PR rtl-optimization/79574
25015 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
25016 (hash_scan_set): Likewise.
25017 (dump_hash_table): Likewise.
25018 (hoist_code): Likewise.
25020 2017-03-03 Richard Biener <rguenther@suse.de>
25022 * fixed-value.c (fixed_from_string): Restore use of elt (1)
25023 in place of uhigh ().
25024 (fixed_convert_from_real): Likewise.
25026 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
25029 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
25031 2017-03-03 Richard Biener <rguenther@suse.de>
25033 PR middle-end/79818
25034 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
25035 TYPE_OVERFLOW_UNDEFINED check.
25037 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25039 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
25041 (vector_ae_<mode>_p): Likewise.
25042 (vector_nez_<mode>_p): Likewise.
25043 (vector_ne_v2di_p): Likewise.
25044 (vector_ae_v2di_p): Likewise.
25045 (vector_ne_<mode>_p): Likewise.
25046 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
25048 (vsx_tsqrt<mode>2_fe): Likewise.
25050 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
25053 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
25055 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25057 PR rtl-optimization/79780
25058 * cprop.c (one_cprop_pass): When second and further conditional trap
25059 in a single basic block is turned into an unconditional trap, turn it
25060 into a deleted note to avoid RTL verification failures.
25062 2017-03-02 Richard Biener <rguenther@suse.de>
25064 * fold-const.c (const_binop): Use ulow () instead of elt (0).
25066 2017-03-02 Richard Biener <rguenther@suse.de>
25068 PR tree-optimization/79345
25070 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
25071 param and abort the walk, returning -1 if it is hit.
25072 (walk_aliased_vdefs): Take a limit param and pass it on.
25073 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
25074 defaulting to 0 and return a signed int.
25075 * tree-ssa-uninit.c (struct check_defs_data): New struct.
25076 (check_defs): New helper.
25077 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
25078 about uninitialized memory.
25079 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
25080 bogus uninitialized warning.
25081 (fixed_convert_from_real): Likewise.
25083 2017-03-02 Bin Cheng <bin.cheng@arm.com>
25085 PR tree-optimization/66768
25086 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
25087 iv_use if base object can't be determined.
25089 2017-03-02 Jakub Jelinek <jakub@redhat.com>
25091 PR tree-optimization/79345
25092 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
25093 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
25094 (get_pattern_stats): Initialize it.
25095 * genemit.c (gen_expand): Verify match_scratch numbers come after
25096 match_operand/match_dup numbers.
25097 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
25098 match_scratch numbers.
25099 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
25101 * config/s390/s390.md (trunctdsd2): Likewise.
25103 2017-03-02 Richard Biener <rguenther@suse.de>
25105 * wide-int.h (wide_int_storage::operator=): Implement in terms
25108 2017-03-02 Richard Biener <rguenther@suse.de>
25110 PR tree-optimization/79777
25111 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
25112 the to insert expression to sth existing.
25114 2017-03-01 Martin Sebor <msebor@redhat.com>
25116 PR middle-end/79692
25117 * gimple-ssa-sprintf.c
25118 (directive::known_width_and_precision): New function.
25119 (format_integer): Use it.
25120 (get_mpfr_format_length): Consider the full range of precision
25121 when computing %g output with the # flag. Set the likely byte
25122 count to 3 rather than 1 when precision is indeterminate.
25123 (format_floating): Correct the lower bound of precision.
25125 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25127 * doc/invoke.texi: Document default code model for 64-bit Linux.
25129 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25132 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
25133 udiv rather than div since input pattern is unsigned.
25135 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
25137 * config/i386/i386.c (print_reg): Warn for values of
25138 unsupported size in integer register.
25140 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
25143 * config/rs6000/predicates.md (current_file_function_operand): Do
25144 not allow self calls to be local if the function is replaceable.
25146 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
25149 * config/rs6000/altivec.h (vec_ctz and others): Change the
25150 preprocessor macro that controls conditional compilation from
25151 _ARCH_PWR9 to __POWER9_VECTOR__.
25152 (vec_all_ne): Change parameterization of __altivec_scalar_pred
25153 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
25154 control (instead of _ARCH_PWR9 control) so that template
25155 definition uses power9-specific function.
25156 (vec_any_eq): Likewise.
25157 (vec_all_ne): Change macro definition to use a power9-specific
25158 expansion under #ifdef __POWER9_VECTOR__ control (instead of
25159 _ARCH_PWR9 control).
25160 (vec_any_eq) Likewise.
25161 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
25162 expansion for CMPNEF to remove support for xvcmpnesp instruction.
25163 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
25164 support for xvcmpnedp instruction.
25165 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
25166 macro expansion so that Power9 implementation of vec_all_ne does
25167 not use the AltiVec predicate framework.
25168 (VCMPNEH_P): Likewise.
25169 (VCMPNEW_P): Likewise.
25170 (VCMPNED_P): Likewise.
25171 (VCMPNEFP_P): Likewise.
25172 (VCMPNEDP_P): Likewise.
25173 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
25174 implementation of vec_any_eq to not use AltiVec predicate
25176 (VCMPAEH_P): Likewise.
25177 (VCMPAEW_P): Likewise.
25178 (VCMPAED_P): Likewise.
25179 (VCMPAEFP_P): Likewise.
25180 (VCMPAEDP_P): Likewise.
25181 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
25182 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
25183 not use the AltiVec predicate framework.
25184 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
25185 of vec_any_eq to not use AltiVec predicate framework.
25186 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
25187 support for predefined __POWER9_VECTOR__ macro to indicate that
25188 Power9 instruction selection is enabled.
25189 (altivec_overloaded_builtins): Remove extraneous
25190 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
25191 function argument types RS6000_BTI_bool_V16QI and
25192 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
25193 entry for overloaded function argument types RS6000_BTI_bool_V4SI
25194 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
25195 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
25196 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
25197 Power9 for implementations of vec_cmpne. Change the signature for
25198 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
25199 (representing vec_all_ne) to remove the previously described first
25200 argument of type RS6000_BTI_INTSI, as this was an artifact of
25201 reliance on the AltiVec predicate framework, which is no longer
25202 used in the implementation of these functions. Add
25203 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
25204 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
25205 since, unlike the AltiVec predicate framework implementation, we
25206 do not share function descriptors between vec_alle and vec_anyeq.
25207 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
25208 set of modes that receive special treatment even when
25209 TARGET_P9_VECTOR is true. The special treatment emits code that
25210 does not depend on Power9 instructions.
25211 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
25212 define_expand to not rely on AltiVec predicate framework.
25213 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25215 (vector_ne_v2di_p): Change this define_expand to not rely on
25216 AltiVec predicate framework.
25217 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
25219 (vector_ne_<mode>_p): Change this define_expand to not rely on
25220 AltiVec predicate framework.
25221 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
25223 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
25224 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
25225 define_insn pattern.
25226 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
25227 define_insn pattern because the xvcmpne<VSs>. instruction is not
25229 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
25230 instruction is not supported.
25232 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25234 * config/nvptx/nvptx.c: Include intl.h.
25236 2017-03-01 Martin Jambor <mjambor@suse.cz>
25239 * ipa-prop.h (ipa_bits): Removed field known.
25240 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
25241 to pointers. Adjusted their comments to warn about their sharing.
25242 (ipcp_transformation_summary): Change bits to a vector of pointers.
25243 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
25244 (ipa_get_ipa_bits_for_value): Declare.
25245 * tree-vrp.h (value_range): Mark as GTY((for_user)).
25246 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
25247 (ipa_bits_hash_table): Likewise.
25248 (ipa_vr_ggc_hash_traits): Likewise.
25249 (ipa_vr_hash_table): Likewise.
25250 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
25251 being pointers and vr_known being removed.
25252 (ipa_set_jf_unknown): Likewise.
25253 (ipa_get_ipa_bits_for_value): New function.
25254 (ipa_set_jfunc_bits): Likewise.
25255 (ipa_get_value_range): New overloaded functions.
25256 (ipa_set_jfunc_vr): Likewise.
25257 (ipa_compute_jump_functions_for_edge): Use the above functions to
25258 construct bits and vr parts of jump functions.
25259 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
25260 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25262 (ipcp_grow_transformations_if_necessary): Also allocate
25263 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
25265 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
25266 them. Fix too long lines.
25267 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
25268 vr_known being removed.
25269 (ipa_read_jump_function): Use new setter functions to construct bits
25270 and vr parts of jump functions or set them to NULL.
25271 (write_ipcp_transformation_info): Adjust for bits being pointers.
25272 (read_ipcp_transformation_info): Likewise.
25273 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
25275 Include gt-ipa-prop.h.
25276 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
25278 (ipcp_store_bits_results): Likewise.
25279 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
25280 Do not write to existing jump functions but use a temporary instead.
25282 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25285 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
25286 attempt to use its first operand as BIT_FIELD_REF base.
25288 2017-03-01 Richard Biener <rguenther@suse.de>
25290 PR middle-end/79721
25291 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
25292 interpolating formula in wrapping arithmetic.
25293 (chrec_apply): Convert chrec_evaluate return value to wanted type.
25295 2017-03-01 Jakub Jelinek <jakub@redhat.com>
25297 PR tree-optimization/79734
25298 * tree-vect-generic.c (expand_vector_condition): Optimize
25299 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
25300 Handle VEC_COND_EXPR where comparison has different inner width from
25301 type's inner width.
25303 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
25305 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
25306 markup, and similar issues. Remove @opindex entries for things
25307 that aren't options. Add missing -mmpy-option entries.
25309 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25311 PR tree-optimization/79737
25312 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
25313 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
25314 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
25315 instead of byte_size. Formatting fix.
25316 (shift_bytes_in_array_right): Formatting fix.
25318 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
25321 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
25322 condition on optimize for the leaf function test.
25324 2017-02-28 Martin Liska <mliska@suse.cz>
25327 * read-rtl-function.c (function_reader::handle_unknown_directive):
25328 Bail out when one uses -flto.
25330 2017-02-28 Martin Liska <mliska@suse.cz>
25332 * common.opt: Replace space with tabular for options of <number>
25334 * config/i386/i386.opt: Show <number> value for
25335 -mlarge-data-threshold.
25336 * opts.c (print_filtered_help): Do not display number in hexadecimal
25339 2017-02-28 Martin Liska <mliska@suse.cz>
25341 * common.opt: Fix --help=option -Q for options which are of
25344 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
25346 * config/i386/i386.c (print_reg): Error out for values
25347 of 8-bit size in invalid integer register.
25349 2017-02-28 Martin Sebor <msebor@redhat.com>
25351 PR tree-optimization/79691
25352 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
25354 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25357 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
25358 gcc_unreachable with output_operand_lossage.
25360 2017-02-28 Richard Biener <rguenther@suse.de>
25362 PR tree-optimization/79740
25363 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
25365 (visit_nary_op): Insert the nary into the hashtable if we
25366 pattern-matched sth.
25367 * tree-ssa-pre.c (eliminate_insert): Robustify.
25369 2017-02-28 Richard Biener <rguenther@suse.de>
25371 PR middle-end/79731
25372 * fold-const.c (decode_field_reference): Reject out-of-bound
25375 2017-02-28 Jakub Jelinek <jakub@redhat.com>
25377 * config/i386/i386.c: Include intl.h.
25378 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
25379 instead of just cond ? "..." : "...".
25380 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
25381 * coverage.c (read_counts_file): Likewise.
25382 * omp-offload.c: Include intl.h.
25383 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
25384 of just cond ? "..." : "...".
25385 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
25386 of just cond ? "..." : "...".
25388 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
25391 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
25393 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
25394 'tune for' CPU name.
25395 * config/arm/arm-cpu-data.h: Regenerated.
25397 2017-02-28 Richard Biener <rguenther@suse.de>
25399 PR tree-optimization/79732
25400 * tree-inline.c (expand_call_inline): Do not shadow var.
25402 2017-02-28 Richard Biener <rguenther@suse.de>
25404 PR tree-optimization/79723
25405 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
25406 address-space properly.
25408 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
25410 * doc/optinfo.texi (Optimization groups): Fix option used for
25412 * doc/invoke.texi (-fopt-info): Document "omp".
25413 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
25414 (OPTGROUP_ALL): Add OPTGROUP_OMP.
25415 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
25416 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
25417 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
25419 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
25421 * dumpfile.c (optgroup_options): Instead of "openmp", associate
25422 OPTGROUP_OMP with "omp".
25424 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
25427 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
25428 for arithmetic shift of unsigned V2DI.
25430 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
25432 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
25433 arc/linux.h headers.
25434 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
25435 (LINK_SPEC): Likewise.
25436 (ARC_TLS_EXTRA_START_SPEC): Likewise.
25437 (EXTRA_SPECS): Likewise.
25438 (STARTFILE_SPEC): Likewise.
25439 (ENDFILE_SPEC): Likewise.
25440 (LIB_SPEC): Likewise.
25441 (TARGET_SDATA_DEFAULT): Likewise.
25442 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
25443 (MULTILIB_DEFAULTS): Likewise.
25444 (DWARF2_UNWIND_INFO): Likewise.
25445 * config/arc/big.h: New file.
25446 * config/arc/elf.h: Likewise.
25447 * config/arc/linux.h: Likewise.
25448 * config/arc/t-uClibc: Remove.
25450 2017-02-27 Bin Cheng <bin.cheng@arm.com>
25452 PR tree-optimization/77536
25453 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
25454 (tree_transform_and_unroll_loop): Use above function to compute the
25455 estimated niter of unrolled loop and use it when scaling profile.
25456 Also use count info rather than frequency if it's non-zero.
25457 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
25458 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
25459 (vect_transform_loop): Call above function.
25461 2017-02-27 Richard Biener <rguenther@suse.de>
25463 PR tree-optimization/45397
25464 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
25465 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
25466 (visit_nary_op): Add pattern matching for CSEing sign-changed
25467 or truncated operations with wider ones.
25469 2017-02-27 Richard Biener <rguenther@suse.de>
25471 PR tree-optimization/79690
25472 * tree-vect-stmts.c (vectorizable_store): Use vector type
25473 built from the DR with address-space.
25475 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
25477 * doc/invoke.texi (Optimize Options): Refine the description
25478 of asan-use-after-return.
25480 2017-02-25 Alan Modra <amodra@gmail.com>
25482 PR rtl-optimization/79584
25483 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
25484 base, not ad->base_term, the reg within base. Remove assertion
25485 that ad->base == ad->base_term. Replace gen_int_mode using
25486 bogus mode with const0_rtx.
25488 2017-02-25 Jakub Jelinek <jakub@redhat.com>
25490 PR middle-end/79396
25491 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
25492 FMA_EXPR like tcc_binary or tcc_unary.
25494 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
25497 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
25499 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
25500 (output_loc_operands): Handle DW_OP_call_ref and
25501 DW_OP_GNU_variable_value.
25502 (struct variable_value_struct): New type.
25503 (struct variable_value_hasher): Likewise.
25504 (variable_value_hash): New variable.
25505 (string_types): Remove.
25506 (copy_loc_descr): New function.
25507 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
25508 (prepend_loc_descr_to_each): New function.
25509 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
25510 instead of add_loc_descr_to_each if the first argument is single
25511 location list and the second has multiple.
25512 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
25513 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
25514 when looking for variable value which doesn't have other location info.
25515 (loc_list_from_tree): Formatting fix.
25516 (gen_array_type_die): Simplify DW_AT_string_length handling.
25517 (adjust_string_types): Remove.
25518 (gen_subprogram_die): Don't call adjust_string_types nor test/set
25519 string_types. Call resolve_variable_values.
25520 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
25521 (resolve_addr_in_expr): Likewise. Add A argument.
25522 (copy_deref_exprloc): Remove deref argument. Adjust for the
25523 original expression being DW_OP_GNU_variable_value with optionally
25524 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
25525 optionally after it.
25526 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
25527 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
25528 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
25529 (variable_value_hasher::hash, variable_value_hasher::equal): New
25531 (resolve_variable_value_in_expr, resolve_variable_value,
25532 resolve_variable_values, note_variable_value_in_expr,
25533 note_variable_value): New functions.
25534 (dwarf2out_early_finish): Call note_variable_value on all toplevel
25537 2017-02-24 Jakub Jelinek <jakub@redhat.com>
25540 * opts.h (handle_generated_option): Add GENERATED_P argument.
25541 * opts-common.c (handle_option): Adjust function comment.
25542 (handle_generated_option): Add GENERATED_P argument, pass it to
25544 (control_warning_option): Pass false to handle_generated_option
25546 * opts.c (maybe_default_option): Pass true to handle_generated_option
25548 * optc-gen.awk: Likewise.
25550 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25552 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
25553 a REG, look at the REG it is a SUBREG of.
25554 (splitter for cmpeqsi_t): Ditto.
25556 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25558 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
25559 the special USEs with the pattern of the insn, not the insn itself.
25561 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
25564 * doc/invoke.texi: Document -mload-store-pairs.
25566 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25567 Sandra Loosemore <sandra@codesourcery.com>
25569 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
25570 argument isn't a CONST_INT.
25571 (nios2_alternate_compare_const): Assert op is a CONST_INT.
25572 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
25573 (nios2_validate_compare): Bypass alternate compare logic if *op2
25574 is not a CONST_INT.
25575 (ldstwm_operation_p): Return false if first_base is not a REG or
25576 if first_offset is not a CONST_INT.
25578 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25580 * config/cris/cris.md: Use correct operand in a define_peephole2.
25582 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25584 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
25586 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
25588 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
25589 this_insn if it is an INSN or JUMP_INSN.
25590 (force_offsettable): Look at base, not at addr.
25591 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
25592 on things that aren't necessarily CONST_INTs.
25594 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
25596 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
25597 -mfpmath=sse is the default also for x86-32 targets with SSE2
25598 instruction set when @option{-ffast-math} is enabled
25600 2017-02-24 Jeff Law <law@redhat.com>
25602 PR rtl-optimizatoin/79286
25603 * ira.c (update_equiv_regs): Drop may_trap_p exception to
25606 2017-02-24 Richard Biener <rguenther@suse.de>
25608 PR tree-optimization/79389
25609 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
25612 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
25614 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
25615 function comment to reflect reality.
25616 (loop_exits_before_overflow): Fix typo in function description.
25618 2017-02-24 Richard Biener <rguenther@suse.de>
25620 PR tree-optimization/79389
25621 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
25622 properly that a threading opportunity exists. Detect conditional
25623 copy/constant propagation opportunities.
25625 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
25627 * config/visium/visium.md (type): Add trap.
25628 (b): New mode attribute.
25629 (*btst): Rename into...
25630 (*btst<mode>): ...this and adjust.
25631 (*cbranchsi4_btst_insn): Rename into...
25632 (*cbranch<mode>4_btst_insn): ...this and adjust.
25633 (trap): New define_insn.
25635 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25637 PR tree-optimization/79389
25638 * ifcvt.c (struct noce_if_info): Add rev_cond field.
25639 (noce_reversed_cond_code): New function.
25640 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
25641 reversed_comparison_code. Formatting fix.
25642 (noce_try_store_flag): Test rev_cond != NULL in addition to
25643 reversed_comparison_code.
25644 (noce_try_store_flag_constants): Likewise.
25645 (noce_try_store_flag_mask): Likewise.
25646 (noce_try_addcc): Use rev_cond if non-NULL instead of
25647 reversed_comparison_code.
25648 (noce_try_cmove_arith): Likewise. Formatting fixes.
25649 (noce_try_minmax, noce_try_abs): Clear rev_cond.
25650 (noce_find_if_block): Initialize rev_cond.
25651 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
25652 instead of false as last argument never attempt to reverse it
25655 2017-02-23 Bin Cheng <bin.cheng@arm.com>
25657 PR tree-optimization/79663
25658 * tree-predcom.c (combine_chains): Process refs in reverse order
25659 only for ZERO length chains, and add explaining comment.
25661 2017-02-23 Jeff Law <law@redhat.com>
25663 PR tree-optimization/79578
25664 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
25665 in call to operand_equal_p.
25667 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
25670 * config/i386/cpuid.h: Fix another undefined behavior.
25672 2017-02-23 Richard Biener <rguenther@suse.de>
25674 PR tree-optimization/79683
25675 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
25676 vector types for data-refs.
25678 2017-02-23 Martin Liska <mliska@suse.cz>
25680 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
25682 2017-02-23 Jakub Jelinek <jakub@redhat.com>
25684 PR middle-end/79665
25685 * internal-fn.c (get_range_pos_neg): Moved to ...
25686 * tree.c (get_range_pos_neg): ... here. No longer static.
25687 * tree.h (get_range_pos_neg): New prototype.
25688 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
25689 are known to be in between 0 and signed maximum inclusive, try to
25690 expand both unsigned and signed divmod and use the cheaper one from
25693 2017-02-22 Jeff Law <law@redhat.com>
25695 PR tree-optimization/79578
25696 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
25697 to compare base operands.
25699 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
25702 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
25703 gpc_reg_operand instead of fpr_reg_operand.
25705 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
25707 * config/mips/mips.c (mips_return_in_memory): Force FP
25708 vector types to be returned in memory for o32 ABI.
25710 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25712 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
25713 instead of DW_TAG_member for static data member declarations and don't
25714 set no_linkage_name for static inline data members.
25715 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
25718 2017-02-22 Martin Liska <mliska@suse.cz>
25720 * doc/invoke.texi: Replace inequality signs with square brackets
25723 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25725 PR tree-optimization/68644
25726 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
25728 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
25731 * lra-constraints.c (simplify_operand_subreg): Handle
25732 WORD_REGISTER_OPERATIONS targets.
25734 2017-02-22 Jakub Jelinek <jakub@redhat.com>
25737 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
25738 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
25739 elimination by swapping fld*.
25741 2017-02-22 Richard Biener <rguenther@suse.de>
25743 PR tree-optimization/79673
25744 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
25745 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
25746 irrelevant address-space qualifiers and avoiding a
25747 ADDR_SPACE_CONVERT_EXPR from fold_convert.
25749 2017-02-22 Richard Biener <rguenther@suse.de>
25751 PR tree-optimization/79666
25752 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
25753 to not symbolically negate if that may introduce undefined
25756 2017-02-22 Martin Liska <mliska@suse.cz>
25759 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
25760 * data-streamer-out.c (streamer_write_gcov_count_stream):
25762 * value-prof.c (stream_out_histogram_value): Make assert more
25763 precise based on type of counter.
25765 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
25768 * config/i386/i386.md (standard_x87sse_constant_load splitter):
25769 Use nonimmediate_operand instead of memory_operand for operand 1.
25770 (float-extend standard_x87sse_constant_load splitter): Ditto.
25772 2017-02-21 Jeff Law <law@redhat.com>
25774 PR tree-optimization/79621
25775 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
25776 blocks with edges to themselves.
25778 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25781 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
25782 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
25783 Use gimple_call_builtin_p.
25786 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
25787 on temporarily removed DEBUG_INSNs.
25789 PR tree-optimization/79649
25790 * tree-loop-distribution.c (classify_partition): Give up on
25791 non-generic address space loads/stores.
25793 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
25795 * doc/loop.texi (Loop manipulation): Remove nonexistent
25796 tree_ssa_loop_version from the documentation.
25797 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
25799 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25802 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
25803 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
25804 * config/rs6000/rs6000.c: Include except.h.
25805 (rs6000_expand_split_stack_prologue): Call
25806 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
25808 2017-02-21 Martin Jambor <mjambor@suse.cz>
25811 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
25812 have been analyzed.
25814 2017-02-21 Martin Jambor <mjambor@suse.cz>
25816 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
25817 for backward compatibility only.
25818 * doc/invoke.texi (Option Summary): Remove all references to
25819 -fipa-cp-alignment.
25821 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
25825 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25827 * lra-constraints.c (curr_insn_transform): Handle
25828 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25830 2017-02-21 Martin Liska <mliska@suse.cz>
25832 * config/i386/i386.opt: Replace -masm-dialect with -masm.
25834 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
25836 PR translation/79638
25837 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
25839 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
25842 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
25843 (arm_function_ok_for_sibcall): Return false for an indirect call by
25844 descriptor if all the argument registers are used.
25845 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
25846 alignment of the function.
25848 2017-02-21 Jakub Jelinek <jakub@redhat.com>
25850 PR tree-optimization/61441
25851 * simplify-rtx.c (simplify_const_unary_operation): For
25852 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
25853 the sNaN unmodified.
25855 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
25857 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25858 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25859 instead of SYSTEM_HEADER_DIR.
25861 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
25862 Martin Liška <mliska@suse.cz>
25864 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
25865 Fix typos and grammar, use active voice, and clarify.
25867 2017-02-20 Marek Polacek <polacek@redhat.com>
25869 PR middle-end/79537
25870 * gimplify.c (gimplify_expr): Handle unused *&&L;.
25873 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
25875 2017-02-20 Jakub Jelinek <jakub@redhat.com>
25878 * config/i386/i386.c (ix86_expand_builtin): Handle
25879 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
25880 ix86_builtins_isa[fcode].isa as a requirement of those
25881 flags and any other flag in the bitmask.
25882 (ix86_init_mmx_sse_builtins): Use 0 instead of
25883 ~OPTION_MASK_ISA_64BIT as mask.
25884 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
25885 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
25886 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
25887 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
25889 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25892 * lra-constraints.c (split_reg): Check requested split mode
25893 is supported by the register.
25895 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25897 * lra-constraints.c (simplify_operand_subreg): Remove early
25900 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25903 * lra-constraints.c (curr_insn_transform): Tighten condition
25904 for converting SUBREG reloads from OP_OUT to OP_INOUT.
25906 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
25909 * lra-constraints.c (curr_insn_transform): Handle
25910 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
25912 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
25915 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
25917 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
25919 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
25922 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
25925 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25927 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
25930 2017-02-18 Jakub Jelinek <jakub@redhat.com>
25932 * final.c (last_columnnum, override_columnnum): New variables.
25933 (final_start_function): Set last_columnnum, pass it to begin_prologue
25934 hook and pass 0 to dwarf2out_begin_prologue.
25935 (final_scan_insn): Update override_columnnum. Pass last_columnnum
25936 to source_line debug hook.
25937 (notice_source_line): Compute last_columnnum and for debug_column_info
25938 return true on column changes.
25939 * debug.h (struct gcc_debug_hooks): Add column argument to
25940 source_line and begin_prologue hooks.
25941 (debug_nothing_int_charstar_int_bool): Remove prototype.
25942 (debug_nothing_int_int_charstar,
25943 debug_nothing_int_int_charstar_int_bool): New prototypes.
25944 (dwarf2out_begin_prologue): Add column argument.
25945 * debug.c (do_nothing_debug_hooks): Adjust source_line and
25946 begin_prologue hooks.
25947 (debug_nothing_int_charstar_int_bool): Remove.
25948 (debug_nothing_int_int_charstar,
25949 debug_nothing_int_int_charstar_int_bool): New functions.
25950 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
25951 through to dwarf2out_source_line.
25952 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
25953 (dwarf2out_source_line): Add column argument, emit it if requested.
25954 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
25956 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25957 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
25958 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
25959 through to dwarf2out_begin_prologue.
25960 (vmsdbgout_source_line): Add column argument, pass it through to
25961 dwarf2out_source_line.
25962 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
25963 dbxout_source_line caller.
25964 (dbxout_source_line): Add column argument.
25966 * common.opt (gno-column-info, gcolumn-info): New options.
25967 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
25968 (check_die): Also test for multiple DW_AT_decl_column attributes.
25969 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
25970 DW_AT_decl_column if requested.
25971 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
25973 (gen_variable_die): Likewise.
25974 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
25975 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
25978 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
25979 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
25980 (ix86_handle_option): Handle OPT_m3dnowa.
25981 * doc/invoke.texi (-m3dnowa): Document.
25982 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
25983 -m3dnowa instead of -m3dnow -march=athlon.
25986 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
25987 instead of gcc_assert for K, r and R code checks. Formatting fixes.
25989 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25992 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
25993 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
25994 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
25995 generator for vsx_xxpermdi_<mode>_be.
25996 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
25997 force big-endian semantics.
25998 (vsx_xxpermdi_<mode>_be): New define_expand with same
25999 implementation as previous version of vsx_xxpermdi_<mode>.
26001 2017-02-17 Jakub Jelinek <jakub@redhat.com>
26003 PR tree-optimization/79327
26004 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
26005 variable, its initialization and use.
26007 2017-02-17 Julia Koval <julia.koval@intel.com>
26009 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
26010 (OPTION_MASK_ISA_PKU_UNSET): New.
26011 (ix86_handle_option): Handle -mrdpid.
26012 * config/i386/cpuid.h (bit_RDPID): New.
26013 * config/i386/driver-i386.c (host_detect_local_cpu):
26014 Detect RDPID feature.
26015 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
26016 * config/i386/i386-c.c (ix86_target_macros_internal):
26018 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
26019 (ix86_valid_target_attribute_inner_p): Add "rdpid".
26020 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
26021 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
26022 * config/i386/i386.md (define_insn "rdpid"): New.
26023 * config/i386/i386.opt Add -mrdpid.
26024 * config/i386/immintrin.h (_rdpid_u32): New.
26026 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
26028 PR rtl-optimization/79541
26029 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
26030 instead of transforming it into USE.
26032 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
26034 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
26035 If HONOR_SNANS (SFmode) force the input to a register.
26036 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
26037 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
26038 an frsp or similar insn.
26040 2017-02-17 Martin Liska <mliska@suse.cz>
26042 PR rtl-optimization/79577
26043 * params.def (selsched-max-sched-times): Increase minimum to 1.
26045 2017-02-17 Martin Liska <mliska@suse.cz>
26047 PR rtl-optimization/79574
26048 * gcse.c (want_to_gcse_p): Prevent integer overflow.
26050 2017-02-17 Martin Liska <mliska@suse.cz>
26052 PR tree-optimization/79529
26053 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
26054 ssa_defined_default_def_p to handle cases which are implicitly
26056 * tree-ssa.c (ssa_defined_default_def_p): New function.
26057 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
26058 which are implicitly defined.
26059 * tree-ssa.h (ssa_defined_default_def_p): Declare.
26061 2017-02-17 Richard Biener <rguenther@suse.de>
26063 PR middle-end/79576
26064 * params.def (max-ssa-name-query-depth): Limit to 10.
26066 2017-02-17 Richard Biener <rguenther@suse.de>
26068 PR tree-optimization/79552
26069 * tree-ssa-structalias.c (visit_loadstore): Properly verify
26072 2017-02-17 Richard Biener <rguenther@suse.de>
26075 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
26077 2017-02-17 Marek Polacek <polacek@redhat.com>
26079 PR middle-end/79536
26080 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
26081 (fold_negate_expr): New wrapper.
26083 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
26085 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
26086 Correct terminology and de-emphasize pre-standard behavior.
26088 2017-02-16 Alan Modra <amodra@gmail.com>
26090 PR rtl-optimization/79286
26091 * ira.c (def_dominates_uses): New function.
26092 (update_equiv_regs): Don't create an equivalence for insns that
26093 may trap where the register def does not dominate the use.
26095 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
26097 PR rtl-optimization/78127
26098 * lra.c (lra): Call lra_eliminate before finish the loop after
26101 2017-02-16 Richard Biener <rguenther@suse.de>
26103 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
26105 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
26106 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
26107 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
26108 (isl_val_int_from_wi): New function.
26109 (extract_affine_gmp): Rename to ...
26110 (extract_affine_wi): ... this, take a widest_int.
26111 (extract_affine_int): Just wrap extract_affine_wi.
26112 (add_param_constraints): Use isl_val_int_from_wi.
26113 (add_loop_constraints): Likewise, and extract_affine_wi.
26115 2017-02-15 Jeff Law <law@redhat.com>
26117 PR middle-end/79521
26118 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
26119 ira_init_register_move_cost_if_necessary.
26121 2017-02-15 Martin Sebor <msebor@redhat.com>
26123 PR middle-end/32003
26124 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
26125 removed in a prior commit.
26127 2017-02-15 Bin Cheng <bin.cheng@arm.com>
26129 PR tree-optimization/79347
26130 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
26131 counters during peeling.
26133 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
26135 * Makefile.in (site.exp): Remove "set ISLVER".
26137 2017-02-15 Jakub Jelinek <jakub@redhat.com>
26140 * real.c (real_from_integer): Call real_convert even for decimal.
26142 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
26145 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
26147 2017-02-14 Andrew Pinski <apinski@cavium.com>
26149 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
26150 cores and change the partno/implementer to be correct.
26151 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
26152 the 'B" as the implementer.
26153 * config/aarch64/aarch64-tune.md: Regenerate.
26155 2017-02-14 Carl Love <cel@us.ibm.com>
26157 * config/rs6000/rs6000.c: Add case statement entry to make the
26158 xvcvuxdsp built-in argument unsigned.
26159 * config/rs6000/vsx.md: Fix the source and return operand types so they
26160 match the instruction definitions from the ISA document. Fix typo
26161 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
26164 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
26167 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
26168 member early_clobber_alts.
26169 * lra-lives.c (reg_early_clobber_p): New.
26170 (process_bb_lives): Use it.
26171 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
26172 (debug_operand_data): Initialize early_clobber_alts.
26173 (setup_operand_alternative): Set up early_clobber_alts.
26174 (collect_non_operand_hard_regs): Ditto. Pass early clobber
26175 alternatives to new_insn_reg.
26176 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
26178 (lra_update_insn_regno_info): Pass the new arg.
26180 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26182 PR middle-end/79505
26183 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
26184 (new_oacc_loop_raw): Don't clear already cleared fields.
26187 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
26188 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
26189 _mm512_prefetch_i64gather_ps): New inline functions and macros.
26191 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
26194 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
26196 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
26199 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
26200 the extra instruction to the right place to store 128-bit constant
26203 2017-02-14 Martin Sebor <msebor@redhat.com>
26205 PR middle-end/79448
26206 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
26207 warning for strings of unknown length.
26209 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26211 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
26213 2017-02-14 Jeff Law <law@redhat.com>
26216 * ira-costs.c (scan_one_insn): Initialize register move costs
26217 for pseudos seen in USE/CLOBBER insns.
26219 PR tree-optimization/79095
26220 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
26221 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
26222 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
26223 if the operands are known to be not equal, then the resulting range
26225 (intersect_ranges): If the new range is ~[0,0] and the old range is
26226 wide, then prefer ~[0,0].
26227 * tree-vrp.c (overflow_comparison_p_1): New function.
26228 (overflow_comparison_p): New function.
26229 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
26230 if NAME is used in an overflow test.
26231 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
26232 overflow check that can be expressed as an equality test, then adjust
26233 ops to be that equality test.
26235 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26237 * config/s390/s390-builtin-types.def: Remove flags argument.
26238 * config/s390/s390.c (s390_init_builtins): Likewise.
26240 2017-02-14 Martin Liska <mliska@suse.cz>
26242 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
26243 vector. Fix trailing white spaces.
26245 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
26247 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
26250 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26252 PR rtl-optimization/68664
26253 * config/arm/arm.c (arm_sched_can_speculate_insn):
26254 New function. Declare prototype.
26255 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26257 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26259 PR rtl-optimization/68664
26260 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
26262 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
26264 2017-02-14 Amit Pawar <amit.pawar@amd.com>
26266 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
26267 max skip bytes for function, loop and jump.
26269 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
26271 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
26272 ABS_EXPR for gimple dump.
26274 2017-02-14 Jakub Jelinek <jakub@redhat.com>
26277 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
26279 PR tree-optimization/79408
26280 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
26281 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
26282 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
26283 also if rhs1 is INTEGER_CST.
26285 2017-02-14 Richard Biener <rguenther@suse.de>
26287 PR middle-end/79432
26288 * tree-into-ssa.c (insert_phi_nodes): When the function can
26289 have abnormal edges rewrite SSA names with broken use-def
26290 dominance out of SSA and register them for PHI insertion.
26292 2017-02-13 Martin Sebor <msebor@redhat.com>
26294 PR middle-end/79496
26295 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
26296 clearing info.nowrite flag when snprintf size argument is a range.
26298 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26300 * cprop.c (cprop_jump): Add missing space in string literal.
26301 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
26302 (get_constraint_for_component_ref): Likewise.
26303 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
26304 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
26305 * lra-constraints.c (process_alt_operands): Likewise.
26306 * ipa-inline.c (inline_small_functions): Likewise.
26307 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
26308 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
26309 * trans-mem.c (diagnose_tm_1_op): Likewise.
26310 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
26311 (grid_parallel_clauses_gridifiable): Likewise.
26313 * config/nvptx/mkoffload.c (process): Add space in between
26316 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
26317 "MOD4_SSE_REGS" and "ALL_REGS".
26319 * spellcheck.c (test_data): Add , in between "foo" and "food".
26321 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26324 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
26325 boundary crossing check and subsequent code generation agree.
26327 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26329 * config/aarch64/aarch64.c (has_memory_op): Delete.
26330 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
26333 2017-02-13 Jakub Jelinek <jakub@redhat.com>
26335 PR rtl-optimization/79388
26336 PR rtl-optimization/79450
26337 * combine.c (distribute_notes): When removing TEM_INSN for which
26338 corresponding dest has last value recorded, invalidate that last
26341 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26343 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
26344 of explicit '@'. Add missing assembly comment marker on branch costs
26347 2017-02-13 Nathan Sidwell <nathan@acm.org>
26349 * gengtype-lex.l (<in_struct>): Add '/'.
26351 2017-02-13 Martin Liska <mliska@suse.cz>
26354 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
26356 2017-02-13 Richard Biener <rguenther@suse.de>
26358 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
26360 * configure: Re-generate.
26361 * config.in: Likewise.
26362 * graphite-dependences.c: Simplify as if
26363 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
26364 * graphite-isl-ast-to-gimple.c: Likewise.
26365 * graphite-optimize-isl.c: Likewise.
26366 * graphite-poly.c: Likewise.
26367 * graphite-sese-to-poly.c: Likewise.
26368 * graphite.h: Likewise.
26369 * toplev.c: Include isl/version.h and use isl_version () for
26370 printing the ISL version.
26371 * doc/install.texi: Update ISL requirement.
26373 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26375 * doc/standards.texi (Standards): Update reference to
26378 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
26380 * doc/extend.texi (Named Address Spaces): sourceware.org now
26382 * doc/install.texi (Binaries): Ditto.
26385 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26387 * doc/cpp.texi: Replace "stringify"/"stringification" with C
26388 standard terminology "stringize"/"stringizing" throughout.
26389 * doc/cppinternals.texi: Likewise.
26391 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
26393 * doc/extend.texi: Fix some spelling mistakes and typos.
26394 * doc/invoke.texi: Likewise.
26396 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26399 * params.def (inline-min-speedup) Change from 10 to 8.
26401 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26403 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
26406 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
26409 * ipa-inline-analysis.c (get_minimal_bb): New function.
26410 (record_modified): Use it.
26411 (remap_edge_change_prob): Handle also ancestor functions.
26413 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
26415 * doc/contrib.texi (Contributors): Remove broken link into
26416 the Mauve CVS repository.
26418 2017-02-11 Jakub Jelinek <jakub@redhat.com>
26420 PR middle-end/79454
26421 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
26422 result computation whenever lhs doesn't have vector mode, not
26423 just when it has BLKmode.
26425 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26427 * doc/makefile.texi (profiledbootstrap): Refer to the
26428 installation instructions only in textual form.
26430 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26433 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
26435 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
26437 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
26438 (Specific): Update mingw-w64 reference.
26440 (Specific): Remove broken link to Renesas RX processor.
26442 2017-02-10 Richard Biener <rguenther@suse.de>
26444 * toplev.c (process_options): Do not mention obsolete graphite
26445 options when printing sorry message about missing graphite support.
26446 Mention -floop-nest-optimize.
26448 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
26450 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
26451 (vtst_p16): Likewise.
26452 (vtstq_p8): Likewise.
26453 (vtstq_p16): Likewise.
26455 (vtstq_p64): Likewise.
26456 * config/arm/arm_neon.h (vgetq_lane_p64): New.
26457 (vset_lane_p64): New.
26458 (vsetq_lane_p64): New.
26460 2017-02-10 Jakub Jelinek <jakub@redhat.com>
26462 PR tree-optimization/79411
26463 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
26464 stmt operands are SSA_NAMEs used in abnormal phis.
26465 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
26468 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26471 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
26474 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
26476 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
26478 2017-02-09 Jakub Jelinek <jakub@redhat.com>
26480 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
26484 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
26485 not arbitrary TREE_CONSTANT.
26488 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
26489 "omp declare target link" attribute unless is_global_var.
26490 * omp-offload.c (find_link_var_op): Likewise.
26492 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
26493 Chung-Lin Tang <cltang@codesourcery.com>
26495 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
26497 (gimplify_adjust_omp_clauses): Don't delete TILE.
26498 (gimplify_omp_for): Deal with TILE.
26499 * internal-fn.c (expand_GOACC_TILE): New function.
26500 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
26502 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
26503 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
26505 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
26506 avoid DIV for outermost collapse var.
26507 (expand_oacc_for): Insert tile element loop as needed. Adjust.
26508 Remove out of date comments, fix whitespace.
26509 * omp-general.c (omp_extract_for_data): Deal with tiling.
26510 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
26511 adjust OLF_DIM_BASE value.
26512 (struct omp_for_data): Add tiling field.
26513 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
26514 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
26515 for auto loops. Remove default auto determining, moved to
26516 oacc_loop_fixed_partitions.
26517 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
26518 stmts, add e_mask field.
26519 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
26520 (oacc_thread_numbers): Use oacc_dim_call.
26521 (oacc_xform_tile): New.
26522 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
26523 (finish_oacc_loop): Adjust for ifns vector.
26524 (oacc_loop_discover_walk): Append loop abstraction sites to list,
26525 add case for GOACC_TILE fns.
26526 (oacc_loop_xform_loop): Delete.
26527 (oacc_loop_process): Iterate over call list directly, and add
26528 handling for GOACC_TILE fns.
26529 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
26531 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
26532 vector partitioning to outer loops. Assign 2 partitions to loops
26533 when available. Add TILE handling.
26534 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
26535 (execite_oacc_device_lower): Process GOACC_TILE fns,
26536 ignore unknown specs.
26537 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
26538 * tree.c (omp_clause_num_ops): Adjust TILE ops.
26539 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
26541 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
26543 * configure.ac (ACX_BUGURL): Update.
26544 * configure: Regenerate.
26546 2017-02-09 Richard Biener <rguenther@suse.de>
26548 PR tree-optimization/69823
26549 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
26550 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
26552 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26554 * config/arc/arc-c.def: Add __NPS400__ definition.
26555 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
26556 (TARGET_NPS400): Define.
26558 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
26560 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
26562 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
26563 pointer, arch_info.
26564 (arc_cpu_types): Fill the arch_info field with a pointer into the
26565 arc_arch_types table.
26566 (arc_selected_cpu): Declare.
26567 * config/arc/arc.c (arc_selected_cpu): Make global.
26568 (arc_selected_arch): Delete.
26569 (arc_base_cpu): Delete.
26570 (arc_override_options): Remove references to deleted variables,
26571 update access to arch information.
26572 (ARC_OPT): Update access to arch information.
26573 (ARC_OPTX): Likewise.
26574 * config/arc/arc.h (arc_base_cpu): Remove declaration.
26575 (TARGET_ARC600): Update access to arch information.
26576 (TARGET_ARC601): Likewise.
26577 (TARGET_ARC700): Likewise.
26578 (TARGET_EM): Likewise.
26579 (TARGET_HS): Likewise.
26580 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
26583 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
26586 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
26587 condition/operands for integer GE/LE/GEU/LEU operations.
26589 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
26591 PR translation/79397
26592 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
26595 2017-02-08 Martin Jambor <mjambor@suse.cz>
26598 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
26599 whether allocation happened.
26600 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
26601 nothing was allocated.
26603 2017-02-08 Jakub Jelinek <jakub@redhat.com>
26605 PR tree-optimization/79408
26606 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
26607 constant, but SSA_NAME with a known integer range, use the minimum
26608 of that range instead of op1 to determine if modulo can be replaced
26609 with its first operand.
26611 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26613 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
26615 2017-02-08 Richard Biener <rguenther@suse.de>
26617 PR tree-optimization/71824
26618 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
26619 Check all loops contained in the merged region.
26621 2017-02-07 Andrew Pinski <apinski@cavium.com>
26623 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
26625 2017-02-07 Andrew Pinski <apinski@cavium.com>
26627 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
26628 (thunderxt88): Likewise.
26629 (thunderxt81): Disable LSE and change v8.1 to v8.
26630 (thunderxt83): Likewise.
26632 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26633 Richard Biener <rguenther@suse.de>
26635 PR middle-end/79399
26636 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
26637 type from int to size_t.
26638 * ira-costs.c (struct_costs_size): Change type from int to size_t.
26640 2017-02-07 Jakub Jelinek <jakub@redhat.com>
26642 PR rtl-optimization/79386
26643 * cprop.c (bypass_conditional_jumps): Initialize
26644 bypass_last_basic_block already before splitting bbs after
26645 unconditional traps...
26646 (bypass_conditional_jumps): ... rather than here.
26649 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
26650 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
26651 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
26652 fix -masm=intel patterns.
26654 2017-02-07 Richard Biener <rguenther@suse.de>
26656 PR tree-optimization/79256
26657 PR middle-end/79278
26658 * builtins.c (get_object_alignment_2): Use min_align_of_type
26659 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
26660 and ADJUST_FIELD_ALIGN.
26662 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
26664 * doc/tm.texi: Regenerate.
26665 * stor-layout.c (layout_decl): Adjust.
26666 (update_alignment_for_field): Likewise.
26667 (place_field): Likewise.
26668 (min_align_of_type): Likewise.
26669 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
26670 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
26671 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
26672 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
26673 * config/frv/frv.c (frv_adjust_field_align): Likewise.
26674 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
26675 * config/i386/i386.c (x86_field_alignment): Likewise.
26676 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
26677 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
26678 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
26679 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
26680 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
26681 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
26685 2017-01-30 Richard Biener <rguenther@suse.de>
26687 PR tree-optimization/79256
26688 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
26689 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
26692 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
26694 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
26695 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
26696 builtins to SImode and emit a zero-extend, if necessary.
26698 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26700 * docs/invoke.texi (RISC-V Options): Alphabetize.
26702 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26704 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
26707 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
26709 * config/riscv/riscv.c: New file.
26710 * common/config/riscv/riscv-common.c: Likewise.
26711 * config.gcc: Likewise.
26712 * config/riscv/constraints.md: Likewise.
26713 * config/riscv/elf.h: Likewise.
26714 * config/riscv/generic.md: Likewise.
26715 * config/riscv/linux.h: Likewise.
26716 * config/riscv/multilib-generator: Likewise.
26717 * config/riscv/peephole.md: Likewise.
26718 * config/riscv/pic.md: Likewise.
26719 * config/riscv/predicates.md: Likewise.
26720 * config/riscv/riscv-builtins.c: Likewise.
26721 * config/riscv/riscv-c.c: Likewise.
26722 * config/riscv/riscv-ftypes.def: Likewise.
26723 * config/riscv/riscv-modes.def: Likewise.
26724 * config/riscv/riscv-opts.h: Likewise.
26725 * config/riscv/riscv-protos.h: Likewise.
26726 * config/riscv/riscv.h: Likewise.
26727 * config/riscv/riscv.md: Likewise.
26728 * config/riscv/riscv.opt: Likewise.
26729 * config/riscv/sync.md: Likewise.
26730 * config/riscv/t-elf-multilib: Likewise.
26731 * config/riscv/t-linux: Likewise.
26732 * config/riscv/t-linux-multilib: Likewise.
26733 * config/riscv/t-riscv: Likewise.
26734 * configure.ac: Likewise.
26735 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
26736 Waterman as RISC-V maintainers.
26737 * doc/install.texi: Add RISC-V entries.
26738 * doc/invoke.texi: Add RISC-V options section.
26739 * doc/md.texi: Add RISC-V constraints section.
26740 * configure: Regenerated.
26742 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
26745 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
26746 false values to be constant vectors with all 0 or all 1 bits set.
26747 (vcondu<mode><mode>): Likewise.
26748 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
26750 (fpmask_comparison_operator): Update comment.
26751 (vecint_comparison_operator): New predicate.
26752 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
26753 vector conditionals when the true and false values are constant
26754 vectors with all 0 bits or all 1 bits set.
26756 2017-02-06 Martin Sebor <msebor@redhat.com>
26758 PR tree-optimization/79376
26759 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
26761 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
26763 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
26764 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
26765 to simplify split condition.
26767 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26769 * omp-expand.c (oxpand_omp_atomic_fetch_op,
26770 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
26773 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
26775 PR rtl-optimization/68664
26776 * target.def (can_speculate_insn): New hook.
26777 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
26778 * doc/tm.texi: Regenerate.
26779 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
26780 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
26781 (rs6000_sched_can_speculate_insn): New function.
26783 2017-02-06 Jakub Jelinek <jakub@redhat.com>
26785 PR tree-optimization/79284
26786 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
26787 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
26788 vectorizable_mask_load_store, vectorizable_operation,
26789 vect_is_simple_cond, get_same_sized_vectype): Use it instead
26790 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
26791 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
26792 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
26793 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
26794 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
26795 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
26796 is_gimple_assign (stmt). Replace another such test with
26797 is_gimple_assign (stmt).
26799 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26802 * config/avr/avr.c (rtl-iter.h): Include it.
26803 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
26804 (avr_legitimate_combined_insn): ...and implementation.
26806 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
26808 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
26809 * config/s390/s390.c (s390_const_operand_ok)
26810 (s390_canonicalize_comparison, s390_extract_part)
26811 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
26812 (s390_contiguous_bitmask_p, s390_rtx_costs)
26813 (legitimize_pic_address): Likewise.
26814 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
26815 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
26816 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
26817 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
26818 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
26820 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
26822 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
26823 REGNO($0) == REGNO($1).
26825 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26827 * config/s390/linux.h(SIZE_TYPE): Add comment.
26829 2017-02-06 Julian Brown <julian@codesourcery.com>
26830 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
26831 Virendra Pathak <virendra.pathak@broadcom.com>
26833 * config/aarch64/aarch64-cores.def: Change the scheduler
26835 * config/aarch64/aarch64.md: Include thunderx2t99.md.
26836 * config/aarch64/thunderx2t99.md: New file.
26838 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26840 * doc/standards.texi (Go Language): Update link to language
26843 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26845 * tree-eh.c (lower_resx): Sanitize profile.
26846 (cleanup_empty_eh_move_lp): Likewise.
26848 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
26851 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
26853 * cfgloopmanip.h (loop_version): Update prototype.
26854 * modulo-sched.c (sms_schedule): Update call of loop_version.
26855 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
26856 * tree-parloops.c (gen_parallel_loop): Likewise.
26857 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
26858 * tree-ssa-loop-split.c (split_loop): Likewise.
26859 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
26860 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
26862 2017-02-05 Martin Liska <mliska@suse.cz>
26865 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
26867 (print_operand_address): Initialize a struct to zero.
26869 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26871 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
26872 garbage collector only in textual form.
26874 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
26876 * doc/extend.texi (x86 specific memory model extensions for
26877 transactional memory): Simplify a phrase.
26879 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
26882 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
26883 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
26884 (atomic_storedi_1): Likewise.
26886 2017-02-04 Jakub Jelinek <jakub@redhat.com>
26888 PR tree-optimization/79338
26889 * tree-parloops.c (gather_scalar_reductions): Don't call
26890 vect_analyze_loop_form for loop->inner before destroying loop's
26893 2017-02-03 Martin Sebor <msebor@redhat.com>
26895 PR tree-optimization/79327
26896 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
26897 when precision has resulted in leading zeros.
26898 (format_integer): Adjust the likely counter to assume an unknown
26899 argument that may be zero is non-zero.
26901 2017-02-03 Jason Merrill <jason@redhat.com>
26904 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
26905 avoid copying non-taken branch.
26907 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26909 PR tree-optimization/79340
26910 * tree-vect-loop.c (vectorizable_reduction): Release
26911 vec_defs elements after safe_splicing them into other vectors.
26914 PR tree-optimization/79327
26915 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
26916 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
26918 (format_integer): Use wide_int_to_tree instead of build_int_cst
26919 + to_?hwi. If argmin is NULL, just set argmin and argmax to
26920 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
26921 of shortest and longest sequence.
26923 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
26925 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
26926 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
26928 2017-02-03 Walter Lee <walt@tilera.com>
26931 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
26932 after initial stackframe link reg save.
26933 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
26935 2017-02-03 Jakub Jelinek <jakub@redhat.com>
26938 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
26939 wu for stxssp alternative.
26941 2017-02-03 Martin Sebor <msebor@redhat.com>
26943 PR tree-optimization/79352
26944 * gimple-fold.c (get_range_strlen): Add argument.
26945 (get_range_strlen): Change return type to bool.
26946 (get_maxval_strlen): Pass in a dummy argument.
26947 * gimple-fold.h (get_range_strlen): Change return type to bool.
26948 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
26949 * tree.h (array_at_struct_end_p): Add argument.
26950 * tree.c (array_at_struct_end_p): Handle it.
26952 2017-02-03 Martin Liska <mliska@suse.cz>
26955 * multiple_target.c (create_dispatcher_calls): Redirect edge
26956 from a caller of a dispatcher.
26957 (expand_target_clones): Make the clones local.
26958 (ipa_target_clone): Do both target clones and resolvers.
26959 (ipa_dispatcher_calls): Remove the pass.
26960 (pass_dispatcher_calls::gate): Likewise.
26961 (make_pass_dispatcher_calls): Likewise.
26962 * passes.def (pass_target_clone): Put as very first IPA early
26965 2017-02-03 Martin Liska <mliska@suse.cz>
26967 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
26968 in case of a function with ifunc attribute.
26970 2017-02-03 Martin Liska <mliska@suse.cz>
26972 * cgraph.c (cgraph_node::dump): Dump function version info.
26973 * symtab.c (symtab_node::dump_base): Add missing new line.
26975 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
26977 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
26978 (ifcombine_ifandif): Use it.
26980 2017-02-03 Martin Liska <mliska@suse.cz>
26982 * doc/invoke.texi: Document default value for
26983 use-after-scope-direct-emission-threshold.
26985 2017-02-03 Martin Liska <mliska@suse.cz>
26987 PR tree-optimization/79339
26988 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
26989 (format_floating): Likewise.
26991 2017-02-03 Martin Liska <mliska@suse.cz>
26994 * ipa-prop.c (ipa_node_params_t::insert): Remove current
26996 (ipa_node_params_t::remove): Likewise.
26997 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
26998 initialization from removed ipa_node_params_t::insert.
26999 (ipa_node_params::~ipa_node_params): Move from removed
27000 ipa_node_params_t::release.
27001 * symbol-summary.h (symbol_summary::m_released): New member.
27002 Do not release a summary twice. Do not allow to call finalizer
27003 for types of a summary that live in GGC memory.
27005 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
27007 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
27010 2017-02-02 Martin Sebor <msebor@redhat.com>
27012 PR middle-end/79275
27013 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
27014 (format_string): Tighten up the range of output for non-constant
27015 strings and correct the expected range for wide non-constant strings.
27017 2017-02-02 Martin Sebor <msebor@redhat.com>
27019 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
27021 PR middle-end/32003
27022 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
27024 (-fdump-tree-@var): Add to index and document how to come up
27025 with pass-specific option and dump file names.
27026 (-fdump-passes): Clarify where to look for output.
27028 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
27030 PR middle-end/77445
27031 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
27032 statistics of the analyzed path; allow threading for speed when
27033 any of BBs along the path are optimized for speed.
27035 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
27037 PR middle-end/78468
27038 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
27039 settings of the virtual registers.
27042 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
27044 * explow.c (get_dynamic_stack_size): Take known alignment of stack
27045 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
27048 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27050 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
27051 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
27053 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27055 * config/s390/s390.md: Add missing comments with the expanded
27057 * config/s390/vector.md: Likewise.
27058 * config/s390/vx-builtins.md: Likewise.
27060 2017-02-02 Jakub Jelinek <jakub@redhat.com>
27063 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
27064 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
27065 conditions on a single line.
27067 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27069 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27070 __S390_VX__ to __VX__.
27072 2017-02-01 Andrew Pinski <apinski@cavium.com>
27074 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
27075 stmt_info to record_stmt_cost.
27076 (vect_get_known_peeling_cost): Pass stmt_info if known to
27078 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
27079 cpu_vector_cost field into
27080 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
27081 field into vec_int_stmt_cost and vec_fp_stmt_cost.
27082 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
27083 splitting of scalar_stmt_cost and vec_stmt_cost.
27084 (thunderx_vector_cost): Likewise.
27085 (cortexa57_vector_cost): LIkewise.
27086 (exynosm1_vector_cost): Likewise.
27087 (xgene1_vector_cost): Likewise.
27088 (thunderx2t99_vector_cost): Improve after the splitting of the two
27090 (aarch64_builtin_vectorization_cost): Update for the splitting of
27091 scalar_stmt_cost and vec_stmt_cost.
27093 2017-02-01 Torvald Riegel <triegel@redhat.com>
27094 Richard Henderson <rth@redhat.com>
27096 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
27097 conditional on existance of a fast atomic load.
27098 * optabs-query.c (can_atomic_load_p): New function.
27099 * optabs-query.h (can_atomic_load_p): Declare it.
27100 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
27101 no fast atomic load is available for the particular size of access.
27102 (expand_atomic_compare_and_swap): Likewise.
27103 (expand_atomic_load): Likewise.
27104 (expand_atomic_store): Likewise.
27105 (expand_atomic_fetch_op): Likewise.
27106 * testsuite/lib/target-supports.exp
27107 (check_effective_target_sync_int_128): Remove x86 because it provides
27108 no fast atomic load.
27109 (check_effective_target_sync_int_128_runtime): Likewise.
27111 2017-02-01 Richard Biener <rguenther@suse.de>
27113 * graphite.c: Include tree-vectorizer.h for find_loop_location.
27114 (graphite_transform_loops): Provide opt-info for optimized nests.
27115 * tree-parloop.c (parallelize_loops): Provide opt-info for
27116 parallelized loops.
27118 2017-02-01 Richard Biener <rguenther@suse.de>
27120 PR middle-end/79315
27121 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
27122 was not set before.
27124 2017-02-01 Richard Biener <rguenther@suse.de>
27126 PR tree-optimization/71824
27127 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
27128 Verify the loops are valid in the merged SESE region.
27129 (scop_detection::can_represent_loop_1): Check analyzing the
27130 evolution of the number of iterations in the region succeeds.
27132 2017-01-31 Ian Lance Taylor <iant@golang.org>
27134 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
27135 REG_ARGS_SIZE note to 32-bit push insns and call insn.
27137 2017-01-31 David Malcolm <dmalcolm@redhat.com>
27139 PR preprocessor/79210
27140 * input.c (get_substring_ranges_for_loc): Replace line_width
27141 assertion with error-handling.
27143 2017-01-31 Richard Biener <rguenther@suse.de>
27145 PR tree-optimization/77318
27146 * graphite-sese-to-poly.c (extract_affine): Fix assert.
27147 (create_pw_aff_from_tree): Take loop parameter.
27148 (add_condition_to_pbb): Pass loop of the condition to
27149 create_pw_aff_from_tree.
27151 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27153 * config/s390/s390.c (s390_asan_shadow_offset): New function.
27154 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
27156 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
27160 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
27162 (convert_int_to_float128): Likewise.
27163 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
27164 (convert_int_to_float128): Likewise.
27165 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
27166 (UNSPEC_IEEE128_CONVERT): Likewise.
27167 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
27168 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
27169 Use local variables for IBM extended format.
27170 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
27171 (fix_trunc<mode>si2_fprs): Likewise.
27172 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
27173 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
27174 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
27175 to know that we can now have integers of all sizes in vector
27177 (fix<uns>_<mode>di2_hw): Likewise.
27178 (float<uns>_<mode>si2_hw): Likewise.
27179 (fix_<mode>si2_hw): Likewise.
27180 (fixuns_<mode>si2_hw): Likewise.
27181 (float<uns>_<mode>di2_hw): Likewise.
27182 (float_<mode>di2_hw): Likewise.
27183 (float_<mode>si2_hw): Likewise.
27184 (floatuns_<mode>di2_hw): Likewise.
27185 (floatuns_<mode>si2_hw): Likewise.
27186 (xscvqp<su>wz_<mode>): Delete, no longer used.
27187 (xscvqp<su>dz_<mode>): Likewise.
27188 (xscv<su>dqp_<mode>): Likewise.
27189 (ieee128_mfvsrd_64bit): Likewise.
27190 (ieee128_mfvsrd_32bit): Likewise.
27191 (ieee128_mfvsrwz): Likewise.
27192 (ieee128_mtvsrw): Likewise.
27193 (ieee128_mtvsrd_64bit): Likewise.
27194 (ieee128_mtvsrd_32bit): Likewise.
27196 2017-01-31 Martin Liska <mliska@suse.cz>
27199 * ipa-prop.c (ipa_free_all_node_params): Call release method
27200 instead of ~sumbol_summary to not to trigger double times
27203 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
27205 PR tree-optimization/71691
27206 * bitmap.h (class auto_bitmap): New.
27207 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
27208 is_maybe_undefined instead of ssa_undefined_value_p.
27210 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27212 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
27213 __S390_ARCH_LEVEL__ to __ARCH__.
27215 2017-01-31 Jakub Jelinek <jakub@redhat.com>
27217 PR tree-optimization/79267
27218 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
27219 if should_remove_lhs_p is true.
27221 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
27224 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
27225 (add_alignment_attribute): New.
27226 (base_type_die): Add alignment attribute.
27227 (subrange_type_die): Likewise.
27228 (modified_type_die): Likewise.
27229 (gen_array_type_die): Likewise.
27230 (gen_descr_array_type_die: Likewise.
27231 (gen_enumeration_type_die): Likewise.
27232 (gen_subprogram_die): Likewise.
27233 (gen_variable_die): Likewise.
27234 (gen_field_die): Likewise.
27235 (gen_ptr_to_mbr_type_die): Likewise.
27236 (gen_struct_or_union_type_die): Likewise.
27237 (gen_subroutine_type_die): Likewise.
27238 (gen_typedef_die): Likewise.
27239 (base_type_cmp): Compare alignment attribute.
27241 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27244 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
27245 (setb_unsigned) New pattern for setb with CCUNS.
27246 * config/rs6000/rs6000.c (expand_block_compare): Use a different
27247 subfc./subfe sequence to avoid overflow problems. Generate a
27248 shorter sequence with cmpld/setb for power9.
27249 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
27250 for generating subfc. instruction.
27251 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
27252 now uses this instruction.
27254 2017-01-30 Ian Lance Taylor <iant@google.com>
27257 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
27258 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
27260 2017-01-30 Martin Sebor <msebor@redhat.com>
27262 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
27263 Move constant to the right of a relational operator.
27264 (get_mpfr_format_length, format_character, format_string): Ditto.
27265 (should_warn_p, maybe_warn): Same.
27267 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
27269 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
27272 * asan.c (get_translation_unit_decl): Remove function.
27273 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
27275 2017-01-30 Martin Liska <mliska@suse.cz>
27277 PR gcov-profile/79259
27278 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
27279 -fprofile-generate.
27281 2017-01-30 Martin Liska <mliska@suse.cz>
27284 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
27285 Initialize variables with NULL value.
27287 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
27290 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
27292 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
27294 2017-01-30 Richard Biener <rguenther@suse.de>
27296 PR tree-optimization/79276
27297 * tree-vrp.c (process_assert_insertions): Properly adjust common
27298 when removing a duplicate.
27300 * gcc.dg/torture/pr79276.c: New testcase.
27302 2017-01-30 Richard Biener <rguenther@suse.de>
27304 PR tree-optimization/79256
27305 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
27306 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
27308 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
27310 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
27313 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
27314 ("*r<noxa>sbg_<mode>_sll_bitmask")
27315 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
27316 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
27317 Use contiguous_bitmask_nowrap_operand.
27319 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27322 * config/rs6000/altivec.h (vec_xl): Revise #define.
27323 (vec_xst): Likewise.
27325 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
27327 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
27329 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
27331 PR rtl-optimization/79194
27332 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
27333 traps before call to bypass_conditional_jumps.
27335 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27337 PR tree-optimization/71374
27338 * lra-constraints.c (check_conflict_input_operands): New.
27339 (match_reload): Use it.
27341 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
27344 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
27345 account to calculate conflict_set.
27347 2017-01-27 Bin Cheng <bin.cheng@arm.com>
27349 PR rtl-optimization/78559
27350 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
27351 other_insn in combine.
27353 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
27355 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
27356 uint16_type_node for BT_UINT16.
27358 2017-01-27 David Malcolm <dmalcolm@redhat.com>
27360 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
27361 "RTL Tests" to menu.
27362 (GIMPLE Tests): New node.
27363 (RTL Tests): New node.
27365 2017-01-27 Richard Biener <rguenther@suse.de>
27367 PR tree-optimization/79245
27368 * tree-loop-distribution.c (distribute_loop): Apply cost
27369 modeling also to detected patterns.
27371 2017-01-27 Richard Biener <rguenther@suse.de>
27373 PR tree-optimization/71433
27374 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
27375 (compare_assert_loc): New function.
27376 (process_assert_insertions): Sort and optimize assert locations
27377 to remove duplicates and push down identical assertions on
27378 edges to their destination block.
27380 2017-01-27 Richard Biener <rguenther@suse.de>
27382 PR tree-optimization/79244
27383 * tree-vrp.c (remove_range_assertions): Forcefully propagate
27384 out SSA names even if abnormal.
27386 2017-01-27 Jakub Jelinek <jakub@redhat.com>
27388 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
27389 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
27390 instead of MPFR_RNDN.
27392 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
27395 * arm.c (arm_option_override): Don't call build_target_option_node
27396 until after doing all option overrides.
27397 (arm_valid_target_attribute_tree): Likewise.
27399 2017-01-27 Martin Liska <mliska@suse.cz>
27401 * doc/invoke.texi (-fprofile-arcs): Document profiling support
27402 for {cd}tors and C++ {cd}tors.
27404 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
27406 * config/s390/s390.md ("*setmem_long_and")
27407 ("*setmem_long_and_31z"): Use zero_extend instead of and.
27409 2017-01-26 Martin Sebor <msebor@redhat.com>
27411 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
27414 2017-01-26 Martin Sebor <msebor@redhat.com>
27416 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
27417 HAVE_DFmode before using XFmode or DFmode.
27418 (parse_directive): Avoid using the z length modifier to avoid
27419 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
27421 PR middle-end/78703
27422 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
27423 to accept adjustment as an array.
27424 (get_int_range): New function.
27425 (struct directive): Make width and prec arrays.
27426 (directive::set_width, directive::set_precision): Call get_int_range.
27427 (format_integer, format_floating): Handle width and precision ranges.
27428 (format_string, parse_directive): Same.
27430 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27433 * dwarf2out.c (generate_skeleton_bottom_up): For children with
27434 comdat_type_p set, just clone them, but keep the children in the
27438 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
27439 which have direct callers with -fvar-tracking-assignments enabled
27441 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
27442 inside of type units.
27444 2017-01-26 Martin Sebor <msebor@redhat.com>
27446 PR middle-end/78703
27447 * gimple-ssa-sprintf.c (struct result_range): Add likely and
27449 (struct format_result): Replace number_chars, number_chars_min,
27450 and number_chars_max with a single member of struct result_range.
27452 (format_result::operator+=): Adjust.
27453 (struct fmtresult): Remove bounded. Handle likely and unlikely
27455 (fmtresult::adjust_for_width_or_precision): New function.
27456 (fmtresult:type_max_digits): New function.
27457 (bytes_remaining): Handle likely and unlikely counters.
27458 (min_bytes_remaining): Remove.
27459 (format_percent): Simplify.
27460 (format_integer, format_floating): Set likely and unlikely counters.
27461 (get_string_length, format_character, format_string): Same.
27462 (format_plain, should_warn_p): New function.
27463 (maybe_warn): Call should_warn_p. Update diagnostic messages
27464 and handle those for all directives, including plain strings.
27465 (format_directive): Handle likely and unlikely counters.
27466 Remove unnecessary quoting from diagnostics. Add an informational
27468 (add_bytes): Remove.
27469 (pass_sprintf_length::compute_format_length): Simplify.
27470 (try_substitute_return_value): Handle likely and unlikely counters.
27472 2017-01-26 Carl Love <cel@us.ibm.com>
27474 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
27475 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
27477 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
27480 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
27481 endianess for subregs into account.
27482 * lra-constraints.c (lra_constraints): Do risky transformations
27483 always on the first iteration.
27484 * lra-lives.c (check_pseudos_live_through_calls): Add arg
27485 last_call_used_reg_set.
27486 (process_bb_lives): Define and use last_call_used_reg_set.
27487 * lra.c (lra): Always continue after lra_constraints on the first
27490 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
27492 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
27494 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
27496 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27498 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
27499 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
27500 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
27501 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
27502 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
27503 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
27504 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
27505 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
27506 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
27508 2017-01-26 Marek Polacek <polacek@redhat.com>
27511 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
27512 for the third operand.
27514 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27516 PR middle-end/79236
27517 * omp-low.c (struct omp_context): Add simt_stmt field.
27518 (scan_omp_for): Return omp_context *.
27519 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
27520 context to the _simt_ SIMD stmt.
27521 (lower_omp_for): For combined SIMD with sibling _simt_
27522 SIMD, make sure to use the same decls in _looptemp_
27523 clauses as in the sibling.
27525 2017-01-26 David Sherwood <david.sherwood@arm.com>
27527 PR middle-end/79212
27528 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
27531 2017-01-26 Jakub Jelinek <jakub@redhat.com>
27534 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
27535 emit fld b; fld a; if possible.
27537 * brig-builtins.def: Update copyright years.
27538 * config/arm/arm_acle_builtins.def: Update copyright years.
27540 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
27543 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
27544 constraint instead of o for the stxsd instruction.
27546 2017-01-25 Carl Love <cel@us.ibm.com>
27548 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
27549 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
27551 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
27553 * doc/invoke.texi (C++ Dialect Options): Fix typo.
27555 2017-01-25 Richard Biener <rguenther@suse.de>
27557 PR tree-optimization/69264
27558 * target.def (vector_alignment_reachable): Improve documentation.
27559 * doc/tm.texi: Regenerate.
27560 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
27562 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
27563 earlier changes with respect to TYPE_USER_ALIGN.
27564 (vector_alignment_reachable_p): Likewise. Improve dumping.
27566 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27569 * config/arm/arm.md (xordi3): Force constant operand into a register
27572 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27574 * doc/invoke.texi (-fstore-merging): Correct default optimization
27575 levels at which it is enabled.
27576 (-O): Move -fstore-merging from list to...
27579 2017-01-25 Richard Biener <rguenther@suse.de>
27582 * omp-expand.c: Include debug.h.
27583 (expand_omp_taskreg): Make sure to generate early debug before
27584 outlining anything from a function.
27585 (expand_omp_target): Likewise.
27586 (grid_expand_target_grid_body): Likewise.
27588 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
27591 * asan.c (get_translation_unit_decl): New function.
27592 (asan_add_global): Extract modules file name from globals
27593 TRANSLATION_UNIT_DECL name.
27595 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
27598 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
27599 for long calls with APCS frame and VFP.
27601 2017-01-24 David Malcolm <dmalcolm@redhat.com>
27603 * cfg.c (original_copy_tables_initialized_p): New function.
27604 * cfg.h (original_copy_tables_initialized_p): New decl.
27605 * cfgrtl.c (relink_block_chain): Guard the call to
27606 free_original_copy_tables with a call to
27607 original_copy_tables_initialized_p.
27608 * cgraph.h (symtab_node::native_rtl_p): New decl.
27609 * cgraphunit.c (symtab_node::native_rtl_p): New function.
27610 (symtab_node::needed_p): Don't assert for early assembly output
27611 for __RTL functions.
27612 (cgraph_node::finalize_function): Set "force_output" for __RTL
27614 (cgraph_node::analyze): Bail out early for __RTL functions.
27615 (analyze_functions): Update assertion to support __RTL functions.
27616 (cgraph_node::expand): Bail out early for __RTL functions.
27617 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
27619 * function.h (struct function): Update comment for field
27621 * gimple-expr.c: Include "tree-pass.h".
27622 (gimple_has_body_p): Return false for __RTL functions.
27623 * Makefile.in (OBJS): Add run-rtl-passes.o.
27624 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
27626 (gcc::pass_manager::get_clean_slate): New accessor.
27627 * passes.c: Include "insn-addr.h".
27628 (should_skip_pass_p): Add logging. Update logic for running
27629 "expand" to be compatible with both __GIMPLE and __RTL. Guard
27630 property-provider override so it is only done for gimple passes.
27632 (skip_pass): New function.
27633 (execute_one_pass): Call skip_pass when skipping passes.
27634 * read-md.c (md_reader::read_char): Support filtering
27635 the input to a subset of line numbers.
27636 (md_reader::md_reader): Initialize fields
27637 m_first_line and m_last_line.
27638 (md_reader::read_file_fragment): New function.
27639 * read-md.h (md_reader::read_file_fragment): New decl.
27640 (md_reader::m_first_line): New field.
27641 (md_reader::m_last_line): New field.
27642 * read-rtl-function.c (function_reader::create_function): Only
27643 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
27644 curr_properties. Set DECL_INITIAL to a dummy block.
27645 (read_rtl_function_body_from_file_range): New function.
27646 * read-rtl-function.h (read_rtl_function_body_from_file_range):
27648 * run-rtl-passes.c: New file.
27649 * run-rtl-passes.h: New file.
27651 2017-01-24 Jeff Law <law@redhat.com>
27653 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
27656 2017-01-24 Bin Cheng <bin.cheng@arm.com>
27658 PR tree-optimization/79159
27659 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
27660 (record_nonwrapping_iv): Improve boundary using above function if no
27661 value range information.
27663 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
27664 Martin Jambor <mjambor@suse.cz>
27666 * brig-builtins.def: New file.
27667 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
27668 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
27669 (DEF_HSAIL_SAT_BUILTIN): Likewise.
27670 (DEF_HSAIL_INTR_BUILTIN): Likewise.
27671 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
27672 * builtin-types.def (BT_INT8): New.
27673 (BT_INT16): Likewise.
27674 (BT_UINT8): Likewise.
27675 (BT_UINT16): Likewise.
27676 (BT_FN_ULONG): Likewise.
27677 (BT_FN_UINT_INT): Likewise.
27678 (BT_FN_UINT_ULONG): Likewise.
27679 (BT_FN_UINT_LONG): Likewise.
27680 (BT_FN_UINT_PTR): Likewise.
27681 (BT_FN_ULONG_PTR): Likewise.
27682 (BT_FN_INT8_FLOAT): Likewise.
27683 (BT_FN_INT16_FLOAT): Likewise.
27684 (BT_FN_UINT32_FLOAT): Likewise.
27685 (BT_FN_UINT16_FLOAT): Likewise.
27686 (BT_FN_UINT8_FLOAT): Likewise.
27687 (BT_FN_UINT64_FLOAT): Likewise.
27688 (BT_FN_UINT16_UINT32): Likewise.
27689 (BT_FN_UINT32_UINT16): Likewise.
27690 (BT_FN_UINT16_UINT16_UINT16): Likewise.
27691 (BT_FN_INT_PTR_INT): Likewise.
27692 (BT_FN_UINT_PTR_UINT): Likewise.
27693 (BT_FN_LONG_PTR_LONG): Likewise.
27694 (BT_FN_ULONG_PTR_ULONG): Likewise.
27695 (BT_FN_VOID_UINT64_UINT64): Likewise.
27696 (BT_FN_UINT8_UINT8_UINT8): Likewise.
27697 (BT_FN_INT8_INT8_INT8): Likewise.
27698 (BT_FN_INT16_INT16_INT16): Likewise.
27699 (BT_FN_INT_INT_INT): Likewise.
27700 (BT_FN_UINT_FLOAT_UINT): Likewise.
27701 (BT_FN_FLOAT_UINT_UINT): Likewise.
27702 (BT_FN_ULONG_UINT_UINT): Likewise.
27703 (BT_FN_ULONG_UINT_PTR): Likewise.
27704 (BT_FN_ULONG_ULONG_ULONG): Likewise.
27705 (BT_FN_UINT_UINT_UINT): Likewise.
27706 (BT_FN_VOID_UINT_PTR): Likewise.
27707 (BT_FN_UINT_UINT_PTR: Likewise.
27708 (BT_FN_UINT32_UINT64_PTR): Likewise.
27709 (BT_FN_INT_INT_UINT_UINT): Likewise.
27710 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
27711 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
27712 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
27713 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
27714 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
27715 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
27716 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
27717 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
27718 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
27719 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
27720 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
27721 * doc/frontends.texi: List BRIG FE.
27722 * doc/install.texi (Testing): Add BRIG tesring requirements.
27723 * doc/invoke.texi (Overall Options): Mention BRIG.
27724 * doc/standards.texi (Standards): Doucment BRIG HSA version.
27726 2017-01-24 Richard Biener <rguenther@suse.de>
27728 PR translation/79208
27729 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
27731 2017-01-24 Martin Jambor <mjambor@suse.cz>
27734 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
27735 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
27736 and known_contexts.
27738 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
27740 PR middle-end/79123
27741 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
27742 casts from signed to unsigned really don't have a range.
27744 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
27746 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
27747 GMP_RNDx for compatiblity.
27749 2017-01-24 Martin Liska <mliska@suse.cz>
27752 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
27753 that would prevent us to call alloca with -1 as argument.
27755 2017-01-24 Jakub Jelinek <jakub@redhat.com>
27757 * dwarf2out.c (output_compilation_unit_header, output_file_names):
27758 Avoid -Wformat-security warning.
27760 2017-01-23 Andrew Pinski <apinski@cavium.com>
27762 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
27765 2017-01-23 Martin Sebor <msebor@redhat.com>
27767 PR middle-end/78703
27768 * gimple-ssa-sprintf.c (warn_level): New global.
27769 (format_integer): Use it here and throughout the rest of the file.
27770 Use the same switch to compute sign as base.
27771 (maybe_warn): New function.
27772 (format_directive): Factor out warnings into maybe_warn.
27773 Add debugging output. Use warn_level.
27774 (add_bytes): Use warn_level.
27775 (pass_sprintf_length::compute_format_length): Add debugging output.
27776 (try_substitute_return_value): Same.
27777 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
27779 PR middle-end/78703
27780 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
27781 (struct fmtresult, format_integer, format_floating): Adjust.
27782 (fmtresult::fmtresult): Set max correctly in two argument ctor.
27783 (get_string_length, format_string,format_directive): Same.
27784 (pass_sprintf_length::compute_format_length): Same.
27785 (try_substitute_return_value): Simplify slightly.
27787 PR middle-end/78703
27788 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
27789 (fmtresult::operator+=): Outlined.
27790 (struct fmtresult): Add ctors.
27791 (struct conversion_spec): Rename...
27792 (struct directive): ...to this. Add and remove data members.
27793 (directive::set_width, directive::set_precision): New functions.
27794 (format_percent): Use fmtresult ctor.
27795 (get_width_and_precision): Remove.
27796 (format_integer): Make naming changes. Avoid computing width and
27798 (format_floating): Same. Adjust indentation.
27799 (format_character, format_none): New functions.
27800 (format_string): Moved character handling to format_character.
27801 (format_directive): Remove arguments, change return type.
27802 (parse_directive): New function.
27803 (pass_sprintf_length::compute_format_length): Move directive
27804 parsing to parse_directive.
27806 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27808 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
27809 (assign_assembler_name_if_needed): ... this.
27810 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
27811 (assign_assembler_name_if_needed): ... this.
27812 (free_lang_data_in_cgraph): Adjust callers.
27813 * cgraphunit.c (cgraph_node::analyze): Likewise.
27814 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
27817 2017-01-23 Richard Biener <rguenther@suse.de>
27819 PR tree-optimization/79088
27820 PR tree-optimization/79188
27821 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
27822 resetting loop bounds after last path deletion. Reset loop
27823 bounds of the target loop, make code match the comments.
27824 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
27825 Make sure loops need no fixups.
27827 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27829 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
27830 exponent support with double type for first argument.
27831 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
27832 type returned by __builtin_vec_extract_sig,
27833 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
27834 functions from "vector int" to "vector unsigned int" or from
27835 "vector long long int" to "vector unsigned long long int".
27836 Changed type returned by __builtin_vec_extract_exp,
27837 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
27838 functions from "vector int" to "vector unsigned int" or from
27839 "vector long long int" to "vector unsigned long long int".
27840 Changed return type of __builtin_vec_test_data_class,
27841 __builtin_vec_test_data_class_sp, and
27842 __builtin_vec_test_data_class_dp from "vector int" to
27843 "vector bool int" or from "vector long long int" to "vector bool
27844 long long int" and changed second argument type from "unsigned
27845 int" to "int". Added new overloaded function forms "vector float
27846 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
27847 "vector float __builtin_vec_insert_exp_sp (vector float, vector
27848 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
27849 double, vector unsigned long long int)" and "vector double
27850 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
27851 long int)". Changed return type of
27852 __builtin_scalar_test_data_class and
27853 __builtin_scalar_test_data_class_sp and
27854 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
27855 int" and changed second argument from "unsigned int" to "int".
27856 Changed type returned by __builtin_scalar_test_neg,
27857 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
27858 from "int" to "bool int". Added new overloaded function form
27859 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
27860 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
27861 exponent double-precision with floating point first argument.
27862 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
27863 documentation of scalar_test_data_class, scalar_test_neg,
27864 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
27865 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
27866 vec_test_data_class built-in functions to reflect refinements in
27867 their type signatures.
27869 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
27871 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
27873 (aarch64_elf_asm_destructor): Likewise.
27875 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
27877 PR rtl-optimization/78634
27878 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
27879 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
27880 * ifcvt.c (noce_try_cmove): Add missing cost check.
27882 PR rtl-optimization/71724
27883 * combine.c (if_then_else_cond): Look for situations where it is
27884 beneficial to undo the work of one of the recursive calls.
27886 2017-01-23 Bin Cheng <bin.cheng@arm.com>
27888 PR tree-optimization/70754
27889 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
27890 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
27891 combined stmt before it if not NULL.
27892 (combine_chains): Process refs reversely and compute dominance point
27895 2017-01-23 Martin Liska <mliska@suse.cz>
27897 PR tree-optimization/79196
27898 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
27899 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
27901 (strlen_optimize_stmt): Call the renamed function.
27903 2017-01-23 Michael Matz <matz@suse.de>
27905 PR tree-optimization/78384
27906 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
27908 2017-01-23 Richard Biener <rguenther@suse.de>
27910 PR tree-optimization/79186
27911 * tree-vrp.c (register_new_assert_for): Make sure we've seen
27912 both incoming edges before moving an assert.
27914 2017-01-23 Martin Jambor <mjambor@suse.cz>
27916 * ipa-prop.c (load_from_param_1): Removed.
27917 (load_from_unmodified_param): Bits from load_from_param_1 put back
27919 (load_from_param): Removed.
27920 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
27921 with stmt. Reverted back to use of load_from_unmodified_param.
27923 2017-01-23 Martin Jambor <mjambor@suse.cz>
27926 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
27927 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
27928 field a pointer to garbage collected vector, mark lattices and
27929 ipcp_orig_node with GTY((skip)).
27930 (ipa_get_param_count): Adjust to descriptors being a pointer.
27931 (ipa_get_param): Likewise.
27932 (ipa_get_type): Likewise.
27933 (ipa_get_param_move_cost): Likewise.
27934 (ipa_set_param_used): Likewise.
27935 (ipa_get_controlled_uses): Likewise.
27936 (ipa_set_controlled_uses): Likewise.
27937 (ipa_is_param_used): Likewise.
27938 (ipa_node_params_t): Move into garbage collector. New methods insert
27940 (ipa_node_params_sum): Annotate wth GTY(()).
27941 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
27943 (ipa_load_from_parm_agg): Adjust declaration.
27944 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
27945 * ipa-profile.c (ipa_profile): Likewise.
27946 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
27947 (ipa_populate_param_decls): Make descriptors parameter garbage
27949 (ipa_dump_param): Adjust to descriptors being a pointer.
27950 (ipa_alloc_node_params): Likewise.
27951 (ipa_initialize_node_params): Likewise.
27952 (load_from_param_1): Make descriptors parameter garbage collected.
27953 (load_from_unmodified_param): Likewise.
27954 (load_from_param): Likewise.
27955 (ipa_load_from_parm_agg): Likewise.
27956 (ipa_node_params::~ipa_node_params): Removed.
27957 (ipa_free_all_node_params): Remove call to delete operator.
27958 (ipa_node_params_t::insert): New.
27959 (ipa_node_params_t::remove): Likewise.
27960 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
27961 copy known_csts and known_contexts vectors.
27962 (ipa_read_node_info): Adjust to descriptors being a pointer.
27963 (ipcp_modif_dom_walker): Make m_descriptors field garbage
27965 (ipcp_transform_function): Make descriptors variable garbage
27968 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
27970 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
27971 * config/i386/avx512dqintrin.h: Ditto.
27972 * config/i386/avx512fintrin.h: Ditto.
27973 * config/i386/i386.c: Handle new builtins.
27974 * config/i386/i386-builtin.def: Add new builtins.
27975 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
27976 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
27978 2017-01-23 Jakub Jelinek <jakub@redhat.com>
27979 Martin Liska <mliska@suse.cz>
27981 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
27982 * asan.c (asan_expand_poison_ifn): Support stores and use
27983 appropriate ASAN report function.
27984 * internal-fn.c (expand_ASAN_POISON_USE): New function.
27985 * internal-fn.def (ASAN_POISON_USE): Declare.
27986 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
27987 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
27988 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
27989 ASAN_POISON calls w/o LHS.
27990 * tree-ssa.c (execute_update_addresses_taken): Create clobber
27991 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
27992 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
27993 * gimplify.c (asan_poison_variables): Add attribute
27994 use_after_scope_memory to variables that really needs to live
27996 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
27997 having the attribute.
27999 2017-01-23 Martin Liska <mliska@suse.cz>
28001 * asan.c (create_asan_shadow_var): New function.
28002 (asan_expand_poison_ifn): Likewise.
28003 * asan.h (asan_expand_poison_ifn): New declaration.
28004 * internal-fn.c (expand_ASAN_POISON): Likewise.
28005 * internal-fn.def (ASAN_POISON): New builtin.
28006 * sanopt.c (pass_sanopt::execute): Expand
28007 asan_expand_poison_ifn.
28008 * tree-inline.c (copy_decl_for_dup_finish): Make function
28010 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
28011 * tree-ssa.c (is_asan_mark_p): New function.
28012 (execute_update_addresses_taken): Rewrite local variables
28013 (identified just by use-after-scope as addressable) into SSA.
28015 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28017 * doc/install.texi (Specific): opensource.apple.com uses https
28018 now. Remove trailing slash.
28020 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
28022 * README.Portability: Remove note on an Irix compatibility issue.
28024 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
28026 * gcov.c (INCLUDE_ALGORITHM): Define.
28027 (INCLUDE_VECTOR): Define.
28028 No longer include <vector> and <algorithm> directly.
28030 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
28032 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
28034 * doc/invoke.texi (Code Gen Options): Ditto.
28036 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
28039 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
28041 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
28043 rtl-optimization/79125
28044 * cprop.c (local_cprop_pass): Handle cases where we make an
28045 unconditional trap.
28047 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
28051 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
28052 read from, for big endian.
28054 2017-01-20 Jiong Wang <jiong.wang@arm.com>
28056 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
28057 register pauth builtins for LP64 only.
28059 2017-01-20 Marek Polacek <polacek@redhat.com>
28062 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
28065 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28067 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
28069 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
28070 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
28071 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
28073 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28076 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
28077 in big-endian mode when they are not a single duplicated value.
28079 2017-01-20 Richard Biener <rguenther@suse.de>
28081 * BASE-VER: Bump to 7.0.1.
28083 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
28085 * omp-low.c (omplow_simd_context): New struct. Use it...
28086 (lower_rec_simd_input_clauses): ...here and...
28087 (lower_rec_input_clauses): ...here to hold common data. Adjust all
28088 references to idx, lane, max_vf, is_simt.
28090 2017-01-20 Graham Markall <graham.markall@embecosm.com>
28092 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
28095 2017-01-20 Martin Jambor <mjambor@suse.cz>
28097 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
28098 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
28100 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
28101 (GTFILES): Rename hsa.c to hsa-common.c.
28102 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
28103 * hsa-dump.c: Likewise.
28104 * hsa-gen.c: Likewise.
28105 * hsa-regalloc.c: Likewise.
28106 * ipa-hsa.c: Likewise.
28107 * omp-expand.c: Likewise.
28108 * omp-low.c: Likewise.
28109 * toplev.c: Likewise.
28111 2017-01-20 Marek Polacek <polacek@redhat.com>
28114 * doc/invoke.texi: Document -Wduplicated-branches.
28115 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
28116 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
28117 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
28118 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
28119 return 0 only when not OEP_LEXICOGRAPHIC.
28120 (fold_build_cleanup_point_expr): Use the expression
28121 location when building CLEANUP_POINT_EXPR.
28122 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
28123 * tree.c (add_expr): Handle error_mark_node.
28125 2017-01-20 Martin Liska <mliska@suse.cz>
28128 * tree-profile.c (init_ic_make_global_vars): Do not call
28130 (gimple_init_gcov_profiler): Likewise.
28132 2017-01-20 Martin Liska <mliska@suse.cz>
28135 * cgraph.h (maybe_create_reference): Remove argument and
28137 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
28139 * ipa-cp.c (create_specialized_node): Likewise.
28140 * symtab.c (symtab_node::maybe_create_reference): Handle
28141 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
28143 2017-01-20 Martin Liska <mliska@suse.cz>
28145 * read-rtl-function.c (function_reader::create_function): Use
28146 build_decl instread of build_decl_stat.
28148 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
28150 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
28151 * config/i386/avx512dqintrin.h: Ditto.
28152 * config/i386/avx512fintrin.h: Ditto.
28153 * config/i386/i386-builtin-types.def: Add new types.
28154 * config/i386/i386.c: Handle new types.
28155 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
28156 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
28157 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
28158 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
28159 (__builtin_ia32_kshiftridi): New.
28160 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
28162 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
28166 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
28167 define to rs6000_init_stack_protect_guard.
28168 (rs6000_init_stack_protect_guard): New function.
28170 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28171 Yunqiang Su <yunqiang.su@imgtec.com>
28173 * config.gcc (supported_defaults): Add madd4.
28174 (with_madd4): Add validation.
28175 (all_defaults): Add madd4.
28176 * config/mips/mips.opt (mmadd4): New option.
28177 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28179 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28181 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
28182 (ISA_HAS_FUSED_MADD4): Likewise.
28183 * doc/invoke.texi (-mmadd4): Document the new option.
28184 * doc/install.texi (--with-madd4): Document the new option.
28186 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28188 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
28189 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
28190 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
28191 (aarch64_init_pauth_hint_builtins): New.
28192 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
28193 (aarch64_expand_builtin): Expand new builtins.
28195 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28197 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
28198 * combine-stack-adj.c (no_unhandled_cfa): Handle
28199 REG_CFA_TOGGLE_RA_MANGLE.
28200 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
28201 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
28202 info for return address signing.
28203 (aarch64_expand_epilogue): Likewise.
28205 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28207 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
28208 * config/aarch64/aarch64-protos.h
28209 (aarch64_return_address_signing_enabled): New declaration.
28210 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
28212 (aarch64_expand_prologue): Sign return address before it's pushed onto
28214 (aarch64_expand_epilogue): Authenticate return address fetched from
28216 (aarch64_override_options): Sanity check for ILP32 and ISA level.
28217 (aarch64_attributes): New function attributes for "sign-return-address".
28218 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
28219 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
28220 ("*do_return"): Generate combined instructions according to key index.
28221 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
28222 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
28224 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
28225 * config/aarch64/aarch64.opt (msign-return-address=): New.
28226 * doc/extend.texi (AArch64 Function Attributes): Documents
28227 "sign-return-address=".
28228 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
28230 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
28232 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
28233 overall option summary.
28235 2017-01-19 Jiong Wang <jiong.wang@arm.com>
28237 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
28238 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
28239 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
28240 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
28242 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
28244 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
28245 -mpower9-minmax by default for -mcpu=power9.
28246 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
28247 128-bit floating point.
28249 2017-01-20 Alan Modra <amodra@gmail.com>
28251 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
28252 optimizing for size.
28254 2017-01-20 Alan Modra <amodra@gmail.com>
28257 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
28258 for strcmp and strncmp from corresponding builtin decl.
28260 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28262 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
28263 instead of i386/rtems-64.h.
28264 * config/i386/rtems-64.h: Remove.
28266 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
28270 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
28272 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
28274 2017-01-19 Tamar Christina <tamar.christina@arm.com>
28276 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
28277 Change int to HOST_WIDE_INT.
28278 * config/aarch64/aarch64-protos.h
28279 (aarch64_simd_gen_const_vector_dup): Likewise.
28280 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
28282 2017-01-19 David Malcolm <dmalcolm@redhat.com>
28284 * langhooks-def.h (lhd_type_for_size): New decl.
28285 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
28286 * langhooks.c (lhd_type_for_size): New function, taken from
28289 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
28291 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
28292 define_bypass for CR latency.
28293 (power9-cracked-alu): Update bypass latency and remove power9-branch.
28294 (power9-alu2): Add define_bypass for CR latency.
28296 (power9-mul): Update insn latency.
28297 (power9-mul-compare): Update insn latency, bypass latency and remove
28300 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28302 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
28304 * config/aarch64/aarch64.md
28305 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
28306 aarch64_nopcrelative_literal_loads.
28307 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
28309 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
28311 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
28312 TARGET_LOONGSON_3A.
28313 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
28315 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
28318 * config.gcc (supported_defaults): Add lxc1-sxc1.
28319 (with_lxc1_sxc1): Add validation.
28320 (all_defaults): Add lxc1-sxc1.
28321 * config/mips/mips.opt (mlxc1-sxc1): New option.
28322 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
28324 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
28325 __mips_no_lxc1_sxc1.
28326 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
28327 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
28328 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
28330 2017-01-19 Richard Biener <rguenther@suse.de>
28332 PR tree-optimization/72488
28333 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
28334 sure to restore SSA info.
28335 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
28337 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
28339 PR rtl-optimization/79121
28340 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
28341 of the inner type when shifting an extended value.
28343 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28346 * symtab.c (symtab_node::equal_address_to): Fix comparing of
28347 interposable aliases.
28349 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
28352 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
28353 Use the evmergelohi instruction.
28354 (mov_si<mode>_e500_subreg4_2_le): Likewise.
28355 (mov_sitf_e500_subreg8_2_be): Likewise.
28356 (mov_sitf_e500_subreg12_2_le): Likewise.
28357 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
28358 (mov_si<mode>_e500_subreg4_2_be): Likewise.
28359 (mov_sitf_e500_subreg8_2_le): Likewise.
28360 (mov_sitf_e500_subreg12_2_be): Likewise.
28362 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28364 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
28365 attribute from vecsimple to vecperm.
28366 (altivec_vbpermq2): Likewise.
28368 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28371 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
28373 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28374 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
28375 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
28376 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
28377 case where N arg is SIZE_MAX.
28378 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
28379 (cmpstrsi): Add pattern.
28381 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
28383 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
28384 __builtin_vec_revb builtins.
28385 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
28386 built-in functions to support generation of the ISA 3.0 XXBR<x>
28387 vector byte reverse instructions.
28388 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
28389 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
28390 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
28391 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
28392 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
28393 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
28394 (P9V_BUILTIN_VEC_REVB): Likewise.
28395 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
28396 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
28397 (p9_xxbrq_v16qi): Likewise.
28398 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
28399 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
28400 (p9_xxbrh_v8hi): Likewise.
28401 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
28402 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
28403 vec_revb built-in functions.
28405 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
28407 PR rtl-optimization/78952
28408 * config/i386/i386.md (any_extract): New code iterator.
28409 (*insvqi_2): Use any_extract for source operand.
28410 (*insvqi_3): Use any_shiftrt for source operand.
28412 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
28414 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
28416 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
28418 2017-01-18 Matthias Klose <doko@ubuntu.com>
28420 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
28422 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28424 * config/rs6000/altivec.h (vec_bperm): Change #define.
28425 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
28426 (altivec_vbpermq2): New define_insn.
28427 (altivec_vbpermd): Likewise.
28428 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
28429 function interface.
28430 (VBPERMD): Likewise.
28431 (VBPERM): New polymorphic function interface.
28432 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
28433 Add entries for P9V_BUILTIN_VEC_VBPERM.
28434 * doc/extend.texi: Add interfaces for vec_bperm.
28436 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28438 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
28439 first letter of error messages.
28440 (s390_resolve_overloaded_builtin): Likewise.
28441 * config/s390/s390.c (s390_expand_builtin): Likewise.
28442 (s390_invalid_arg_for_unprototyped_fn): Likewise.
28443 (s390_valid_target_attribute_inner_p): Likewise.
28444 * config/s390/s390.md ("tabort"): Likewise.
28446 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
28448 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
28449 (ISA_AVOID_DIV_HILO): New macro.
28450 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
28451 (ISA_HAS_DDIV): Likewise.
28453 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28455 * doc/invoke.texi (fabi-version): Correct number of occurrences.
28457 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28459 * doc/invoke.texi (fabi-version): Spelling fix.
28461 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28464 * doc/invoke.texi (fabi-version): Mention mangling fix for
28467 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
28470 * doc/invoke.texi (fabi-version): Document discriminator mangling.
28472 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
28475 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
28476 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
28478 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
28480 (stack_protect_test): Ditto.
28481 * config/rs6000/rs6000.opt (mstack-protector-guard=,
28482 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
28484 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
28485 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
28486 -mstack-protector-guard-offset=.
28487 (RS/6000 and PowerPC Options): Ditto.
28489 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28491 * config/i386/i386.h (MASK_CLASS_P): New define.
28492 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
28493 there are no registers from different register sets also when
28494 mask registers are used. Update function comment.
28495 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
28496 to (*k/*r) and (*k/*km) alternatives.
28498 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
28500 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
28501 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
28502 (EH_RETURN_HANDLER_RTX): New define.
28503 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
28504 Force frame pointer in EH return functions.
28505 (aarch64_expand_epilogue): Add barrier for eh_return.
28506 (aarch64_final_eh_return_addr): Remove.
28507 (aarch64_eh_return_handler_rtx): New function.
28508 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
28510 (aarch64_eh_return_handler_rtx): New prototype.
28512 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28514 * config/rs6000/altivec.h (vec_rlmi): New #define.
28515 (vec_vrlnm): Likewise.
28516 (vec_rlnm): Likewise.
28517 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
28518 (UNSPEC_VRLNM): Likewise.
28519 (VIlong): New mode iterator.
28520 (altivec_vrl<VI_char>mi): New define_insn.
28521 (altivec_vrl<VI_char>nm): Likewise.
28522 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
28524 (VRLDNM): Likewise.
28525 (RLNM): New polymorphic function entry.
28526 (VRLWMI): New monomorphic function entry.
28527 (VRLDMI): Likewise.
28528 (RLMI): New polymorphic function entry.
28529 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
28530 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
28531 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
28534 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28537 * dwarf2out.c (field_byte_offset): Restore the
28538 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
28539 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
28540 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
28543 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28546 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
28548 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28551 * dwarf2out.c (add_data_member_location_attribute): For constant
28552 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
28553 instead of DW_AT_data_member_location, DW_AT_bit_offset and
28554 DW_AT_byte_size attributes.
28556 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
28558 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
28559 after forcing to constant memory when the code model is medium.
28561 2017-01-17 Julia Koval <julia.koval@intel.com>
28564 * config/i386/avx512fintrin.h
28565 (_mm512_i32gather_ps): Change __addr type to void const*.
28566 (_mm512_mask_i32gather_ps): Ditto.
28567 (_mm512_i32gather_pd): Ditto.
28568 (_mm512_mask_i32gather_pd): Ditto.
28569 (_mm512_i64gather_ps): Ditto.
28570 (_mm512_mask_i64gather_ps): Ditto.
28571 (_mm512_i64gather_pd): Ditto.
28572 (_mm512_mask_i64gather_pd): Ditto.
28573 (_mm512_i32gather_epi32): Ditto.
28574 (_mm512_mask_i32gather_epi32): Ditto.
28575 (_mm512_i32gather_epi64): Ditto.
28576 (_mm512_mask_i32gather_epi64): Ditto.
28577 (_mm512_i64gather_epi32): Ditto.
28578 (_mm512_mask_i64gather_epi32): Ditto.
28579 (_mm512_i64gather_epi64): Ditto.
28580 (_mm512_mask_i64gather_epi64): Ditto.
28581 (_mm512_i32scatter_ps): Change __addr type to void*.
28582 (_mm512_mask_i32scatter_ps): Ditto.
28583 (_mm512_i32scatter_pd): Ditto.
28584 (_mm512_mask_i32scatter_pd): Ditto.
28585 (_mm512_i64scatter_ps): Ditto.
28586 (_mm512_mask_i64scatter_ps): Ditto.
28587 (_mm512_i64scatter_pd): Ditto.
28588 (_mm512_mask_i64scatter_pd): Ditto.
28589 (_mm512_i32scatter_epi32): Ditto.
28590 (_mm512_mask_i32scatter_epi32): Ditto.
28591 (_mm512_i32scatter_epi64): Ditto.
28592 (_mm512_mask_i32scatter_epi64): Ditto.
28593 (_mm512_i64scatter_epi32): Ditto.
28594 (_mm512_mask_i64scatter_epi32): Ditto.
28595 (_mm512_i64scatter_epi64): Ditto.
28596 (_mm512_mask_i64scatter_epi64): Ditto.
28597 * config/i386/avx512pfintrin.h
28598 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
28599 (_mm512_mask_prefetch_i32gather_ps): Ditto.
28600 (_mm512_mask_prefetch_i64gather_pd): Ditto.
28601 (_mm512_mask_prefetch_i64gather_ps): Ditto.
28602 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
28603 (_mm512_prefetch_i32scatter_ps): Ditto.
28604 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
28605 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
28606 (_mm512_prefetch_i64scatter_pd): Ditto.
28607 (_mm512_prefetch_i64scatter_ps): Ditto.
28608 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
28609 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
28610 * config/i386/avx512vlintrin.h
28611 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
28612 (_mm_mmask_i32gather_ps): Ditto.
28613 (_mm256_mmask_i32gather_pd): Ditto.
28614 (_mm_mmask_i32gather_pd): Ditto.
28615 (_mm256_mmask_i64gather_ps): Ditto.
28616 (_mm_mmask_i64gather_ps): Ditto.
28617 (_mm256_mmask_i64gather_pd): Ditto.
28618 (_mm_mmask_i64gather_pd): Ditto.
28619 (_mm256_mmask_i32gather_epi32): Ditto.
28620 (_mm_mmask_i32gather_epi32): Ditto.
28621 (_mm256_mmask_i32gather_epi64): Ditto.
28622 (_mm_mmask_i32gather_epi64): Ditto.
28623 (_mm256_mmask_i64gather_epi32): Ditto.
28624 (_mm_mmask_i64gather_epi32): Ditto.
28625 (_mm256_mmask_i64gather_epi64): Ditto.
28626 (_mm_mmask_i64gather_epi64): Ditto.
28627 (_mm256_i32scatter_ps): Change __addr type to void*.
28628 (_mm256_mask_i32scatter_ps): Ditto.
28629 (_mm_i32scatter_ps): Ditto.
28630 (_mm_mask_i32scatter_ps): Ditto.
28631 (_mm256_i32scatter_pd): Ditto.
28632 (_mm256_mask_i32scatter_pd): Ditto.
28633 (_mm_i32scatter_pd): Ditto.
28634 (_mm_mask_i32scatter_pd): Ditto.
28635 (_mm256_i64scatter_ps): Ditto.
28636 (_mm256_mask_i64scatter_ps): Ditto.
28637 (_mm_i64scatter_ps): Ditto.
28638 (_mm_mask_i64scatter_ps): Ditto.
28639 (_mm256_i64scatter_pd): Ditto.
28640 (_mm256_mask_i64scatter_pd): Ditto.
28641 (_mm_i64scatter_pd): Ditto.
28642 (_mm_mask_i64scatter_pd): Ditto.
28643 (_mm256_i32scatter_epi32): Ditto.
28644 (_mm256_mask_i32scatter_epi32): Ditto.
28645 (_mm_i32scatter_epi32): Ditto.
28646 (_mm_mask_i32scatter_epi32): Ditto.
28647 (_mm256_i32scatter_epi64): Ditto.
28648 (_mm256_mask_i32scatter_epi64): Ditto.
28649 (_mm_i32scatter_epi64): Ditto.
28650 (_mm_mask_i32scatter_epi64): Ditto.
28651 (_mm256_i64scatter_epi32): Ditto.
28652 (_mm256_mask_i64scatter_epi32): Ditto.
28653 (_mm_i64scatter_epi32): Ditto.
28654 (_mm_mask_i64scatter_epi32): Ditto.
28655 (_mm256_i64scatter_epi64): Ditto.
28656 (_mm256_mask_i64scatter_epi64): Ditto.
28657 (_mm_i64scatter_epi64): Ditto.
28658 (_mm_mask_i64scatter_epi64): Ditto.
28659 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
28660 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
28661 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
28662 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
28663 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
28664 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
28665 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
28666 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
28667 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
28668 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
28669 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
28670 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
28671 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
28672 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
28673 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
28674 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
28675 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
28676 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
28677 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
28678 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
28679 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
28680 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
28681 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
28682 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
28683 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
28684 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
28685 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
28686 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
28687 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
28688 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
28689 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
28690 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
28691 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
28692 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
28693 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
28694 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
28695 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
28696 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
28697 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
28698 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
28699 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
28700 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
28701 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
28702 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
28703 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
28704 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
28705 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
28706 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
28707 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
28708 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
28709 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
28710 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
28711 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
28712 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
28713 definitions accordingly.
28715 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
28716 Kuan-Lin Chen <kuanlinchentw@gmail.com>
28719 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
28722 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
28725 * ira-conflicts.c (ira_build_conflicts): Update total conflict
28726 hard regs for inner regno.
28728 2017-01-17 Martin Liska <mliska@suse.cz>
28731 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
28732 assumption and add comment.
28734 2017-01-17 Nathan Sidwell <nathan@acm.org>
28736 * ipa-visibility.c (localize_node): New function, broken out of ...
28737 (function_and_variable_visibility): ... here. Call it.
28739 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
28741 PR middle-end/77445
28742 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
28743 correctly set frequency of oudgoing edge.
28744 (duplicate_thread_path): Fix profile updating.
28746 2017-01-17 Jakub Jelinek <jakub@redhat.com>
28749 * configure.ac: Add GCC_BASE_VER.
28750 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
28751 version from BASE-VER file.
28752 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
28753 (gcc.o): Depend on $(BASEVER).
28754 * common.opt (dumpfullversion): New option.
28755 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
28756 * doc/invoke.texi: Document -dumpfullversion.
28757 * doc/install.texi: Document --with-gcc-major-version-only.
28758 * configure: Regenerated.
28760 2017-01-17 Richard Biener <rguenther@suse.de>
28762 PR tree-optimization/71433
28763 * tree-vrp.c (register_new_assert_for): Merge same asserts
28764 on all incoming edges.
28765 (process_assert_insertions_for): Handle insertions at the
28768 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
28770 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
28771 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
28773 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
28776 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
28779 2017-01-17 Alan Modra <amodra@gmail.com>
28782 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
28783 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
28784 symbolic stack limit when pic.
28786 2017-01-16 Martin Sebor <msebor@redhat.com>
28788 PR tree-optimization/78608
28789 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
28791 2017-01-16 Jeff Law <law@redhat.com>
28794 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
28795 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
28796 for several include directories that may be relative to sysroot.
28797 * config/i386/x-mingw32 (gplus_includedir): Define.
28798 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
28799 (native_system_includedir): Likewise.
28800 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
28801 override if TARGET_SYSTEM_ROOT is defined.
28802 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
28804 PR tree-optimization/79090
28805 PR tree-optimization/33562
28806 PR tree-optimization/61912
28807 PR tree-optimization/77485
28808 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
28809 and computed trims into the dump file.
28811 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
28813 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
28815 2017-01-16 Jakub Jelinek <jakub@redhat.com>
28818 * gimplify.c (gimplify_init_constructor): If want_value and
28819 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
28823 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
28824 sequence. Formatting fixes.
28825 (doloop_optimize): Formatting fixes.
28828 * gcc.c (debug_level_greater_than_spec_func): New function.
28829 (static_spec_functions): Add debug-level-gt spec function.
28830 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
28832 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28833 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
28834 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
28835 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
28836 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
28837 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
28839 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
28841 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
28842 QImode fixups to general and mask registers only.
28844 2017-01-16 Carl Love <cel@us.ibm.com>
28846 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
28847 for built-in functions
28848 vector signed char vec_nabs (vector signed char)
28849 vector signed short vec_nabs (vector signed short)
28850 vector signed int vec_nabs (vector signed int)
28851 vector signed long long vec_nabs (vector signed long long)
28852 vector float vec_nabs (vector float)
28853 vector double vec_nabs (vector double)
28854 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
28856 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
28857 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
28858 * doc/extend.texi: Update the documentation file for the new built-in
28861 2017-01-16 Martin Sebor <msebor@redhat.com>
28863 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
28866 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28868 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
28869 UNSPEC_VSX__XXSPLTD to require special splat handling.
28871 2017-01-16 David Malcolm <dmalcolm@redhat.com>
28874 * system.h: Poison strndup.
28876 2017-01-16 Alan Modra <amodra@gmail.com>
28879 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
28882 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
28884 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
28886 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
28888 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
28889 call recog here. Assert that INSN_CODE (insn) is non-negative.
28891 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28894 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
28896 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
28897 in the currently scheduled RTL fragment.
28899 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
28901 PR rtl-optimization/78751
28902 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
28905 2017-01-14 Jeff Law <law@redhat.com>
28907 PR tree-optimization/79090
28908 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
28909 variable length stores.
28910 (compute_trims): Delete dead assignment to *trim_tail.
28911 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
28914 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
28916 PR rtl-optimization/78626
28917 PR rtl-optimization/78727
28918 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
28919 of a block, and split such blocks after everything else is finished.
28921 2017-01-14 Alan Modra <amodra@gmail.com>
28924 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
28925 target legitimate_combined_insn.
28926 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
28927 (rs6000_legitimate_combined_insn): New function.
28928 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
28930 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
28931 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
28932 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
28934 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
28936 * doc/frontends.texi (G++ and GCC): Remove references to Java.
28938 2017-01-13 Jeff Law <law@redhat.com>
28940 PR tree-optimization/33562
28941 PR tree-optimization/61912
28942 PR tree-optimization/77485
28943 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
28945 (delete_dead_assignment): Likewise.
28946 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
28947 statement to delete_dead_call and delete_dead_assignment.
28949 2017-01-13 David Malcolm <dmalcolm@redhat.com>
28952 * substring-locations.c (format_warning_va): Strengthen case 1 so
28953 that both endpoints of the substring must be within the format
28954 range for just the substring to be printed.
28956 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
28958 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
28959 * config/i386/i386.c (ix86_target_string): Add missing options
28960 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
28961 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
28962 flags_other and ix86_target_other to flags2_other. Display unknown
28964 (ix86_valid_target_attribute_inner_p): Add missing options and
28965 reorder options by implied ISAs, as in ix86_target_string.
28967 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28969 * hash-table.h (hash_table::too_empty_p): New function.
28970 (hash_table::expand): Use it.
28971 (hash_table::traverse): Likewise.
28972 (hash_table::empty_slot): Use sizeof (value_type) instead of
28973 sizeof (PTR) to convert bytes to elements. Shrink the table
28974 if the current size is excessive for the current number of
28977 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
28979 * ira-costs.c (record_reg_classes): Break from the inner loop
28980 early once alt_fail is known to be true. Update outer loop
28981 handling accordingly.
28983 2017-01-13 Jeff Law <law@redhat.com>
28985 * tree-ssa-dse.c (decrement_count): New function.
28986 (increment_start_addr, maybe_trim_memstar_call): Likewise.
28987 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
28988 when we know the partially dead statement is a mem* function.
28990 PR tree-optimization/61912
28991 PR tree-optimization/77485
28992 * tree-ssa-dse.c: Include expr.h.
28993 (maybe_trim_constructor_store): New function.
28994 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
28996 PR tree-optimization/33562
28997 PR tree-optimization/61912
28998 PR tree-optimization/77485
28999 * doc/invoke.texi: Document new dse-max-object-size param.
29000 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
29001 * tree-ssa-dse.c: Include params.h.
29002 (dse_store_status): New enum.
29003 (initialize_ao_ref_for_dse): New, partially extracted from
29005 (valid_ao_ref_for_dse, normalize_ref): New.
29006 (setup_live_bytes_from_ref, compute_trims): Likewise.
29007 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
29008 (maybe_trim_partially_dead_store): Likewise.
29009 (maybe_trim_complex_store): Likewise.
29010 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
29011 Track what bytes live from the original store. Return tri-state
29012 for dead, partially dead or live.
29013 (dse_dom_walker): Add constructor, destructor and new private members.
29014 (delete_dead_call, delete_dead_assignment): New extracted from
29016 (dse_optimize_stmt): Make a member of dse_dom_walker.
29017 Use initialize_ao_ref_for_dse.
29019 PR tree-optimization/33562
29020 PR tree-optimization/61912
29021 PR tree-optimization/77485
29022 * sbitmap.h (bitmap_count_bits): Prototype.
29023 (bitmap_clear_range, bitmap_set_range): Likewise.
29024 * sbitmap.c (bitmap_clear_range): New function.
29025 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
29027 2017-01-13 Martin Liska <mliska@suse.cz>
29030 * function.c (set_cfun): Add new argument force.
29031 * function.h (set_cfun): Likewise.
29032 * ipa-inline-transform.c (inline_call): Use the function when
29033 strict alising from is dropped for function we inline to.
29035 2017-01-13 Richard Biener <rguenther@suse.de>
29037 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
29038 for dumping GIMPLE INTEGER_CSTs.
29040 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
29042 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
29043 to 201112L since C++17.
29045 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
29048 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
29049 if -fsanitize=kernel-address is present.
29051 2017-01-13 Richard Biener <rguenther@suse.de>
29053 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
29054 as _Literal ( type ) number in case usual suffixes do not
29055 preserve all information.
29057 2017-01-13 Richard Biener <rguenther@suse.de>
29059 PR tree-optimization/77283
29060 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
29061 and ssa-iterators.h.
29062 (is_feasible_trace): Implement a cost model based on joiner
29065 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
29068 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
29069 char or short to __float128/_Float128 directly.
29071 2017-01-12 Martin Sebor <msebor@redhat.com>
29073 to -Wformat-overflow.
29074 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
29075 (min_bytes_remaining): Same.
29076 (get_string_length): Same.
29077 (format_string): Same.
29078 (format_directive): Same.
29080 (pass_sprintf_length::handle_gimple_call): Same.
29082 2017-01-12 Jakub Jelinek <jakub@redhat.com>
29084 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
29085 info.nowrite calls with no lhs that can't throw. Return bool
29086 whether gsi_remove has been called or not.
29087 (pass_sprintf_length::handle_gimple_call): Return bool whether
29088 try_substitute_return_value called gsi_remove. Formatting fix.
29089 (pass_sprintf_length::execute): Don't use gsi_remove if
29090 handle_gimple_call returned true.
29093 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
29094 be removed due to side-effects, don't remove following barrier nor
29095 turn the successor edge into fallthru edge.
29097 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29100 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
29101 element-reversing loads and stores as not swappable.
29103 2017-01-12 Nathan Sidwell <nathan@acm.org>
29104 Nicolai Stange <nicstange@gmail.com>
29106 * combine.c (try_combine): Don't ignore result of overlap checking
29107 loop. Combine overlap & asm check into single loop.
29109 2017-01-12 Richard Biener <rguenther@suse.de>
29111 * tree-pretty-print.c (dump_generic_node): Provide -gimple
29112 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
29114 2017-01-12 Richard Biener <rguenther@suse.de>
29116 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
29117 and TS_TARGET_OPTION directly derive from TS_BASE.
29118 * tree-core.h (tree_optimization_option): Derive from tree_base.
29119 (tree_target_option): Likewise.
29121 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
29123 * config/i386/i386.c (memory_address_length): Increase len
29124 only when rip_relative_addr_p returns false.
29126 2017-01-11 Julia Koval <julia.koval@intel.com>
29128 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
29129 (OPTION_MASK_ISA_SGX_SET): New.
29130 (ix86_handle_option): Handle OPT_msgx.
29131 * config.gcc: Added sgxintrin.h.
29132 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
29133 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
29134 * config/i386/i386.c (ix86_target_string): Add -msgx.
29136 (ix86_option_override_internal): Handle new options.
29137 (ix86_valid_target_attribute_inner_p): Add sgx.
29138 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
29139 * config/i386/i386.opt: Add msgx.
29140 * config/i386/sgxintrin.h: New file.
29141 * config/i386/x86intrin.h: Add sgxintrin.h.
29143 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29146 * fold-const.c (maybe_nonzero_address): Return 1 for function
29148 (tree_single_nonzero_warnv_p): Don't handle function local objects
29152 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
29155 2017-01-11 David Malcolm <dmalcolm@redhat.com>
29158 * opts.c: Include "spellcheck.h"
29159 (struct string_fragment): New struct.
29160 (struct edit_distance_traits<const string_fragment &>): New
29162 (get_closest_sanitizer_option): New function.
29163 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
29165 2017-01-11 Jakub Jelinek <jakub@redhat.com>
29167 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
29169 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
29170 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
29171 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
29172 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
29173 for initial die_offset if dwarf_split_debug_info.
29174 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
29175 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
29177 (output_skeleton_debug_sections): Formatting fix. Use
29178 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
29179 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
29181 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
29183 * config/arm/cortex-a53.md: Add bypasses for
29184 cortex_a53_r2f_cvt.
29185 (cortex_a53_r2f): Only use for transfers.
29186 (cortex_a53_f2r): Likewise.
29187 (cortex_a53_r2f_cvt): Add reservation for conversions.
29188 (cortex_a53_f2r_cvt): Likewise.
29190 2017-01-11 Tamar Christina <tamar.christina@arm.com>
29192 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
29193 to all inlined functions, change static to extern.
29195 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
29198 * config/arm/arm.c (legitimize_pic_address): Handle reference to
29200 (arm_assemble_integer): Likewise.
29202 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
29204 * config.gcc: Use new awk script to check CPU, FPU and architecture
29205 parameters for --with-... options.
29206 * config/arm/parsecpu.awk: New file
29207 * config/arm/arm-cpus.in: New file.
29208 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
29210 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
29212 * config/arm/t-arm: Update dependency rules.
29213 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
29214 of processing .def files.
29215 * config/arm/genopt.sh: Deleted.
29216 * config/arm/gentune.sh: Deleted.
29217 * config/arm/arm-cores.def: Deleted.
29218 * config/arm/arm-arches.def: Deleted.
29219 * config/arm/arm-fpus.def: Deleted.
29220 * config/arm/arm-tune.md: Regenerated.
29221 * config/arm/arm-tables.opt: Regenerated.
29222 * config/arm/arm-cpu.h: New generated file.
29223 * config/arm/arm-cpu-data.h: New generated file.
29224 * config/arm/arm-cpu-cdata.h: New generated file.
29226 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
29229 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
29231 (input_varpool_node): Unpack dynamically_initialized bit.
29233 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
29235 PR rtl-optimization/79032
29236 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
29237 the alignment of the adjusted memory reference against that of MODE,
29238 instead of the alignment of the original memory reference.
29240 2017-01-11 Martin Jambor <mjambor@suse.cz>
29242 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
29244 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
29245 decorated functions.
29247 2017-01-11 Richard Biener <rguenther@suse.de>
29249 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
29250 set range/nonnull info for PHI results. Do not set it on
29251 stmts marked for removal.
29253 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
29255 * expr.c (store_field): In the bitfield case, fetch the return value
29256 from the registers before applying a single big-endian adjustment.
29257 Always do a final load for a BLKmode value not larger than a word.
29259 2017-01-10 David Malcolm <dmalcolm@redhat.com>
29262 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29263 that we correctly handle column numbers greater than
29264 LINE_MAP_MAX_COLUMN_NUMBER.
29266 2017-01-10 Martin Sebor <msebor@redhat.com>
29268 PR middle-end/78245
29269 * gimple-ssa-sprintf.c (get_destination_size): Call
29270 {init,fini}object_sizes.
29271 * tree-object-size.c (addr_object_size): Adjust.
29272 (pass_through_call): Adjust.
29273 (pass_object_sizes::execute): Adjust.
29274 * tree-object-size.h (fini_object_sizes): Declare.
29276 2017-01-10 Martin Sebor <msebor@redhat.com>
29278 PR tree-optimization/78775
29279 * builtins.c (get_size_range): Move...
29280 * calls.c: ...to here.
29281 (alloc_max_size): Accept zero argument.
29282 (operand_signed_p): Remove.
29283 (maybe_warn_alloc_args_overflow): Call get_size_range.
29284 * calls.h (get_size_range): Declare.
29286 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
29288 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
29289 from TI's devices.csv file as of September 2016.
29290 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
29292 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
29294 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
29295 * doc/invoke.texi: Likewise.
29296 * doc/md.texi: Likewise.
29297 * doc/objc.texi: Likewise.
29299 2017-01-10 Joshua Conner <joshconner@google.com>
29301 * config/arm/fuchsia-elf.h: New file.
29302 * config/fuchsia.h: New file.
29303 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
29304 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
29306 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
29308 2016-01-10 Richard Biener <rguenther@suse.de>
29310 PR tree-optimization/79034
29311 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
29312 Propagate out degenerate PHIs in the joiner.
29314 2017-01-10 Martin Liska <mliska@suse.cz>
29316 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
29317 (sort_congruence_classes_by_decl_uid): Likewise.
29318 (sort_congruence_class_groups_by_decl_uid): Likewise.
29319 (sem_item_optimizer::merge_classes): Sort class, groups in these
29320 classes and members in the groups by DECL_UID of declarations.
29321 This would make merge operations stable.
29323 2017-01-10 Martin Liska <mliska@suse.cz>
29325 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
29326 usage of m_classes_vec.
29327 (sem_item_optimizer::~sem_item_optimizer): Likewise.
29328 (sem_item_optimizer::get_group_by_hash): Likewise.
29329 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
29330 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
29331 (sem_item_optimizer::verify_classes): Likewise.
29332 (sem_item_optimizer::process_cong_reduction): Likewise.
29333 (sem_item_optimizer::dump_cong_classes): Likewise.
29334 (sem_item_optimizer::merge_classes): Likewise.
29335 * ipa-icf.h (congruence_class_hash): Rename from
29336 congruence_class_group_hash. Remove declaration of m_classes_vec.
29338 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
29340 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
29341 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
29342 * config.gcc: Add avx512vpopcntdqintrin.h.
29343 * config/i386/avx512vpopcntdqintrin.h: New.
29344 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
29345 * config/i386/i386-builtin-types.def: Add new types.
29346 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
29347 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
29348 __builtin_ia32_vpopcountq_v8di_mask): New.
29349 * config/i386/i386-c.c (ix86_target_macros_internal): Define
29350 __AVX512VPOPCNTDQ__.
29351 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
29352 (PTA_AVX512VPOPCNTDQ): Define.
29353 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
29354 TARGET_AVX512VPOPCNTDQ_P): Define.
29355 * config/i386/i386.opt: Add mavx512vpopcntdq.
29356 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
29357 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
29359 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29361 PR middle-end/77484
29362 * predict.def (PRED_CALL): Set to 67.
29364 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
29366 * expr.c (store_field): In the bitfield case, if the value comes from
29367 a function call and is of an aggregate type returned in registers, do
29368 not modify the field mode; extract the value in all cases if the mode
29369 is BLKmode and the size is not larger than a word.
29371 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
29374 * config/i386/cpuid.h: Fix undefined behavior.
29376 2017-01-04 Jeff Law <law@redhat.com>
29378 PR tree-optimization/79007
29379 PR tree-optimization/67955
29380 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
29381 conservative for pt.null when flag_non_call_exceptions is on.
29383 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29385 PR translation/79019
29386 PR translation/79020
29387 * params.def (PARAM_INLINE_MIN_SPEEDUP,
29388 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
29389 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
29391 * config/avr/avr.opt (maccumulate-args): Likewise.
29392 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
29393 * common.opt (freport-bug): Likewise.
29394 * cif-code.def (CIF_FINAL_ERROR): Likewise.
29395 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
29396 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
29397 translatable string.
29398 * config/i386/i386.c (function_value_32): Likewise.
29399 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
29400 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
29402 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
29403 * common/config/msp430/msp430-common.c (msp430_handle_option):
29405 * symtab.c (symtab_node::verify_base): Likewise.
29406 * opts.c (set_debug_level): Likewise.
29407 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
29408 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
29409 missing whitespace to translatable strings.
29410 * config/avr/avr.md (bswapsi2): Fix typo in comment.
29411 * config/sh/superh.h: Likewise.
29412 * config/i386/xopintrin.h: Likewise.
29413 * config/i386/znver1.md: Likewise.
29414 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
29415 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
29416 * double-int.h (struct double_int): Likewise.
29417 * double-int.c (div_and_round_double): Likewise.
29418 * wide-int.cc: Likewise.
29419 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
29420 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
29421 * cfgcleanup.c (crossjumps_occured): Renamed to ...
29422 (crossjumps_occurred): ... this.
29423 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
29426 PR tree-optimization/78899
29427 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
29428 returning bool return struct loop *, NULL for failure and the new
29430 (versionable_outer_loop_p): Don't version outer loop if it has
29431 dont_vectorized bit set.
29432 (tree_if_conversion): When versioning outer loop, ensure
29433 tree_if_conversion is performed also on the inner loop of the
29434 non-vectorizable outer loop copy.
29435 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
29436 LOOP_VECTORIZED in inner loop of the scalar outer loop and
29437 prevent vectorization of it.
29438 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
29439 the outer loop vectorization of the non-scalar version is attempted
29440 before vectorization of the inner loop in scalar version. If
29441 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
29442 vectorization of its inner loop.
29443 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
29444 has 2 inner loops, rename also on edges from bb whose single pred
29445 is outer_loop->header. Fix typo in function comment.
29447 2017-01-09 Martin Sebor <msebor@redhat.com>
29450 * asan.c (asan_emit_stack_protection): Increase local buffer size
29451 to avoid snprintf truncation warning.
29453 2017-01-09 Andrew Pinski <apinski@cavium.com>
29455 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
29456 to reference thunderx2t99 for the tuning structure
29457 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
29459 (thunderx2t99_extra_costs): This.
29460 * config/aarch64/aarch64-tune.md: Regenerate.
29461 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
29462 (vulcan_addrcost_table): This.
29463 (vulcan_regmove_cost): Rename to ...
29464 (thunderx2t99_regmove_cost): This.
29465 (vulcan_vector_cost): Rename to ...
29466 (thunderx2t99_vector_cost): this.
29467 (vulcan_branch_cost): Rename to ...
29468 (thunderx2t99_branch_cost): This.
29469 (vulcan_tunings): Rename to ...
29470 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
29471 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
29473 2017-01-09 Martin Jambor <mjambor@suse.cz>
29477 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
29478 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
29479 (propagate_vr_accross_jump_function): Use the above function for all
29480 value range computations for pass-through jump functions and type
29481 converasion from explicit value range values.
29482 (ipcp_propagate_stage): Do not attempt to deduce types of formal
29483 parameters from TYPE_ARG_TYPES.
29484 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
29485 (ipa_write_node_info): Stream type of the actual argument.
29486 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
29488 2017-01-09 Martin Liska <mliska@suse.cz>
29491 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
29492 (lookup_compiler): Do not show error message with have_E.
29494 2017-01-09 Jakub Jelinek <jakub@redhat.com>
29496 PR tree-optimization/78938
29497 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
29498 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
29499 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
29500 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
29503 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29505 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
29508 2017-01-09 Richard Biener <rguenther@suse.de>
29510 PR tree-optimization/78997
29511 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
29512 name condition properly.
29514 2017-01-09 Richard Biener <rguenther@suse.de>
29517 * dwarf2out.c (is_cxx): New overload with context.
29518 (is_naming_typedef_decl): Use it.
29520 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29522 * invoke.texi (Option Summary): Correct spacing in option lists
29523 and add line breaks to fix over-long lines.
29525 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29527 PR middle-end/17660
29529 * extend.texi (Common Variable Attributes): Add xref to GCC
29530 Internals manual to explain mode attribute keywords.
29532 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
29535 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
29536 and Preprocessor Options.
29537 (Options for Linking): Document -pthread here....
29538 (RS/6000 and PowerPC Options): ...not here.
29539 (Solaris 2 Options): ...or here.
29540 * doc/cppopts.texi: Document -pthread.
29542 2017-01-08 Martin Sebor <msebor@redhat.com>
29544 PR middle-end/77708
29545 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
29546 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
29547 New member functions.
29548 (format_directive): Used them.
29550 (pass_sprintf_length::handle_gimple_call): Same.
29551 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
29552 to avoid truncation for any argument.
29553 (extract_affine_mul): Same.
29554 * tree.c (get_file_function_name): Same.
29556 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29558 PR middle-end/77484
29559 * predict.def (PRED_INDIR_CALL): Set to 86.
29561 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29563 PR preprocessor/54124
29564 * doc/cppopts.texi: Reformat -d subtable to list the full name
29565 of the options. Add cross-reference to the docs for the general
29566 compiler -d options.
29567 * doc/invoke.texi (Developer Options): Add cross-reference to the
29568 preprocessor-specific -d option documentation.
29570 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29572 PR preprocessor/13498
29573 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
29574 redudant material, and reflect new command-line options.
29575 (System Headers): Likewise.
29577 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
29579 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
29580 -isystem, and -idirafter. Copy-edit.
29581 * doc/cppopts.texi: Copy-edit. Remove contradiction about
29582 default for -ftrack-macro-expansion. Delete obsolete and
29583 badly-formatted implementation details about -fdebug-cpp output.
29584 * doc/cppwarnopts.texi: Copy-edit.
29586 2017-01-07 David Malcolm <dmalcolm@redhat.com>
29589 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
29590 that the transition from a max line width >= 1<<10 to narrower
29591 lines works correctly.
29593 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
29595 * doc/options.texi (PerFunction): New.
29596 * opt-functions.awk (switch_flags): Map both Optimization and
29597 PerFunction to CL_OPTIMIZATION.
29598 * opth-gen.awk: Test for PerFunction flag along with
29600 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
29601 it only when the latter is present. Skip those that don't in
29602 the hash function generator.
29603 * common.opt (fvar-tracking): Mark as PerFunction instead of
29605 (fvar-tracking-assignments): Likewise.
29606 (fvar-tracking-assignments-toggle): Likewise.
29607 (fvar-tracking-uninit): Likewise.
29609 2017-01-07 Jakub Jelinek <jakub@redhat.com>
29611 PR translation/79018
29612 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
29615 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
29618 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
29619 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
29620 TARGET_LONG_JUMP_TABLE_OFFSETS.
29621 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
29622 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
29623 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
29624 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
29625 * config/m68k/m68k.md (tablejump expander): Likewise.
29626 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
29627 TARGET_LONG_JUMP_TABLE_OFFSETS.
29628 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
29629 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
29631 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
29632 David Holsgrove <david.holsgrove@xilinx.com>
29634 * common/config/microblaze/microblaze-common.c
29635 (TARGET_EXCEPT_UNWIND_INFO): Remove.
29636 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
29638 * config/microblaze/microblaze.c (microblaze_must_save_register)
29639 (microblaze_expand_epilogue, microblaze_return_addr): Handle
29641 (microblaze_eh_return): New function.
29642 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
29643 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
29644 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
29645 * config/microblaze/microblaze.md (eh_return): New pattern.
29647 2017-01-06 Jakub Jelinek <jakub@redhat.com>
29649 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
29650 GCC_DIAGNOSTIC_STRINGIFY): Define.
29652 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
29654 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29656 * config/arm/arm.md (<mcrr>): New.
29658 * config/arm/arm.c (arm_arch5te): New.
29659 (arm_option_override): Set arm_arch5te.
29660 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
29662 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
29663 (arm_mcrr_qualifiers): ... this. New.
29664 (MRRC_QUALIFIERS): Define to...
29665 (arm_mrrc_qualifiers): ... this. New.
29666 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
29668 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
29669 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
29670 (MRRCI, mrrc, MRRC): New.
29671 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
29672 VUNSPEC_MRRC2): New.
29674 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29676 * config/arm/arm.md (<mcr>): New.
29678 * config/arm/arm.c (arm_coproc_builtin_available): Add
29679 support for mcr, mrc, mcr2 and mrc2.
29680 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
29681 (arm_mcr_qualifiers): ... this. New.
29682 (MRC_QUALIFIERS): Define to ...
29683 (arm_mrc_qualifiers): ... this. New.
29684 (MCR_QUALIFIERS): Define to ...
29685 (arm_mcr_qualifiers): ... this. New.
29686 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
29688 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
29689 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
29690 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
29691 VUNSPEC_MRC2): New.
29693 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29695 * config/arm/arm.md (*ldc): New.
29699 * config/arm/arm.c (arm_coproc_builtin_available): Add
29700 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
29701 (arm_coproc_ldc_stc_legitimate_address): New.
29702 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
29703 'qualifier_const_pointer'.
29704 (LDC_QUALIFIERS): Define to...
29705 (arm_ldc_qualifiers): ... this. New.
29706 (STC_QUALIFIERS): Define to...
29707 (arm_stc_qualifiers): ... this. New.
29708 * config/arm/arm-protos.h
29709 (arm_coproc_ldc_stc_legitimate_address): New.
29710 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
29711 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
29712 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
29713 stc2, stcl, stc2l): New.
29714 * config/arm/constraints.md (Uz): New.
29715 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
29716 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
29717 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
29718 VUNSPEC_STC2L): New.
29720 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29722 * config/arm/arm.md (<cdp>): New.
29723 * config/arm/arm.c (neon_const_bounds): Rename this ...
29724 (arm_const_bounds): ... this.
29725 (arm_coproc_builtin_available): New.
29726 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
29727 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
29728 (CDP_QUALIFIERS): Define to...
29729 (arm_cdp_qualifiers): ... this. New.
29731 (arm_expand_builtin_args): Add case for 6 arguments.
29732 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
29733 (arm_const_bounds): ... this.
29734 (arm_coproc_builtin_available): New.
29735 * config/arm/arm_acle.h (__arm_cdp): New.
29737 * config/arm/arm_acle_builtins.def (cdp): New.
29739 * config/arm/iterators.md (CDPI,CDP,cdp): New.
29740 * config/arm/neon.md: Rename all 'neon_const_bounds' to
29741 'arm_const_bounds'.
29742 * config/arm/types.md (coproc): New.
29743 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
29744 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
29745 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
29746 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
29748 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29750 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
29751 (UBINOP_QUALIFIERS): New.
29753 (acle_builtin_data): New. Change comment.
29754 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
29755 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
29756 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
29757 arm_acle_builtins.def.
29758 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
29759 (arm_init_acle_builtins): New.
29760 (CRC32_BUILTIN): Remove.
29761 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
29762 crc32cb, crc32ch and crc32cw.
29763 (arm_init_crc32_builtins): Remove.
29764 (arm_init_builtins): Use arm_init_acle_builtins rather
29765 than arm_init_crc32_builtins.
29766 (arm_expand_acle_builtin): New.
29767 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
29768 * config/arm/arm_acle_builtins.def: New.
29770 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
29772 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
29773 (arm_builtin_datum): ... this.
29774 (arm_init_neon_builtin): Rename to ...
29775 (arm_init_builtin): ... this. Add a new parameters PREFIX
29776 and USE_SIG_IN_NAME.
29777 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
29778 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
29779 'arm_builtin_datum'.
29780 (arm_init_vfp_builtins): Likewise.
29781 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
29782 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
29783 (arm_expand_neon_args): Rename to ...
29784 (arm_expand_builtin_args): ... this. Rename builtin_arg
29785 enum values and differentiate between ARG_BUILTIN_MEMORY
29786 and ARG_BUILTIN_NEON_MEMORY.
29787 (arm_expand_neon_builtin_1): Rename to ...
29788 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
29789 values, arm_expand_builtin_args and add bool parameter NEON.
29790 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
29791 (arm_expand_vfp_builtin): Likewise.
29792 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
29794 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
29796 PR middle-end/77484
29797 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
29798 * predict.c (tree_estimate_probability_bb): Reverse direction of
29799 polymorphic call predictor.
29801 2017-01-06 David Malcolm <dmalcolm@redhat.com>
29803 * passes.c (execute_one_pass): Split out pass-skipping logic into...
29804 (determine_pass_name_match): ...this new function and...
29805 (should_skip_pass_p): ...this new function.
29807 2017-01-06 Nathan Sidwell <nathan@acm.org>
29809 * ipa-visibility.c (function_and_variable_visibility): Reformat
29810 comments and long lines. Remove extrneous if.
29811 * symtab.c (symtab_node::make_decl_local): Fix code format.
29812 (symtab_node::set_section_for_node): Fix comment typo.
29814 2017-01-06 Martin Liska <mliska@suse.cz>
29817 * lra-constraints.c: Rename invariant to lra_invariant.
29818 * predict.c (set_even_probabilities): Initialize e to NULL.
29820 2017-01-05 Martin Sebor <msebor@redhat.com>
29822 PR tree-optimization/78910
29823 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
29824 (format_integer): Correct off-by-one error in the handling
29825 of precision with negative numbers in signed conversions..
29827 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
29829 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
29831 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29833 PR tree-optimization/71016
29834 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
29835 factor_out_conditional_conversion. Formatting fix.
29836 (factor_out_conditional_conversion): Add cond_stmt argument.
29837 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
29838 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
29841 2017-01-05 David Malcolm <dmalcolm@redhat.com>
29843 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
29844 read-rtl-function.o, and selftest-rtl.o.
29845 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
29846 (selftest::aarch64_test_loading_full_dump): New function.
29847 (selftest::aarch64_run_selftests): New function.
29848 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
29849 selftest::aarch64_run_selftests.
29850 * config/i386/i386.c
29851 (selftest::ix86_test_loading_dump_fragment_1): New function.
29852 (selftest::ix86_test_loading_call_insn): New function.
29853 (selftest::ix86_test_loading_full_dump): New function.
29854 (selftest::ix86_test_loading_unspec): New function.
29855 (selftest::ix86_run_selftests): Call the new functions.
29856 * emit-rtl.c (maybe_set_max_label_num): New function.
29857 * emit-rtl.h (maybe_set_max_label_num): New decl.
29858 * function.c (instantiate_decls): Guard call to
29859 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
29860 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
29862 * gensupport.c (gen_reader::gen_reader): Pass "false"
29863 for new "compact" param of rtx_reader.
29864 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
29865 rather than an empty string for NULL strings.
29866 * read-md.c: Potentially include config.h rather than bconfig.h.
29867 Wrap include of errors.h with #ifdef GENERATOR_FILE.
29868 (have_error): New global, copied from errors.c.
29869 (md_reader::read_name): Rename to...
29870 (md_reader::read_name_1): ...this, adding "out_loc" param,
29871 and converting "missing name or number" to returning false, rather
29873 (md_reader::read_name): Reimplement in terms of read_name_1.
29874 (md_reader::read_name_or_nil): New function.
29875 (md_reader::read_string): Handle "(nil)" by returning NULL.
29876 (md_reader::md_reader): Add new param "compact".
29877 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
29878 (md_reader::read_file): New method.
29879 * read-md.h (md_reader::md_reader): Add new param "compact".
29880 (md_reader::read_file): New method.
29881 (md_reader::is_compact): New accessor.
29882 (md_reader::read_name): Convert return type from void to file_location.
29883 (md_reader::read_name_or_nil): New decl.
29884 (md_reader::read_name_1): New decl.
29885 (md_reader::m_compact): New field.
29886 (noop_reader::noop_reader): Pass "false" for new "compact" param
29888 (rtx_reader::rtx_reader): Add new "compact" param.
29889 (rtx_reader::read_rtx_operand): Make virtual and convert return
29890 type from void to rtx.
29891 (rtx_reader::read_until): New decl.
29892 (rtx_reader::handle_any_trailing_information): New virtual function.
29893 (rtx_reader::postprocess): New virtual function.
29894 (rtx_reader::finalize_string): New virtual function.
29895 (rtx_reader::m_in_call_function_usage): New field.
29896 (rtx_reader::m_reuse_rtx_by_id): New field.
29897 * read-rtl-function.c: New file.
29898 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
29899 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
29900 (selftest::verify_three_block_rtl_cfg): New decl.
29901 * read-rtl-function.h: New file.
29902 * read-rtl.c: Potentially include config.h rather than bconfig.h.
29903 For host, include function.h, memmodel.h, and emit-rtl.h.
29904 (one_time_initialization): New function.
29905 (struct compact_insn_name): New struct.
29906 (compact_insn_names): New array.
29907 (find_code): Handle insn codes in compact dumps.
29908 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
29909 (bind_subst_iter_and_attr): Likewise.
29910 (add_condition_to_string): Likewise.
29911 (add_condition_to_rtx): Likewise.
29912 (apply_attribute_uses): Likewise.
29913 (add_current_iterators): Likewise.
29914 (apply_iterators): Likewise.
29915 (initialize_iterators): Guard usage of apply_subst_iterator with
29916 #ifdef GENERATOR_FILE.
29917 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
29918 (md_reader::read_mapping): Likewise.
29919 (add_define_attr_for_define_subst): Likewise.
29920 (add_define_subst_attr): Likewise.
29921 (read_subst_mapping): Likewise.
29922 (check_code_iterator): Likewise.
29923 (rtx_reader::read_rtx): Likewise. Move one-time initialization
29925 (one_time_initialization): New function.
29926 (rtx_reader::read_until): New method.
29927 (read_flags): New function.
29928 (parse_reg_note_name): New function.
29929 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
29930 Handle reuse_rtx ids.
29931 Wrap iterator lookup within #ifdef GENERATOR_FILE.
29932 Add parsing support for RTL dumps, mirroring the special-cases in
29933 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
29934 values, and calling handle_any_trailing_information.
29935 (rtx_reader::read_rtx_operand): Convert return type from void
29936 to rtx, returning return_rtx. Handle case 'e'. Call
29937 finalize_string on XSTR and XTMPL fields.
29938 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
29939 "(nil)" values were omitted. Call the postprocess vfunc on the
29941 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
29942 class ctor. Initialize m_in_call_function_usage. Call
29943 one_time_initialization.
29944 * rtl-tests.c (selftest::test_uncond_jump): Call
29945 set_new_first_and_last_insn.
29946 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
29947 * selftest-rtl.c: New file.
29948 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
29949 (selftest::get_insn_by_uid): New decl.
29950 * selftest-run-tests.c (selftest::run_tests): Call
29951 read_rtl_function_c_tests.
29952 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
29953 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
29956 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
29958 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
29959 operands in a special way. Assert that pos+len <= mode precision.
29961 2017-01-05 Jakub Jelinek <jakub@redhat.com>
29963 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
29964 3 argument Alias with unlimited for the negative form.
29965 (fno-vect-cost-model): Removed.
29967 2017-01-05 Martin Liska <mliska@suse.cz>
29969 * hsa-gen.c (gen_hsa_divmod): New function.
29970 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
29972 2017-01-05 Martin Liska <mliska@suse.cz>
29975 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
29978 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29980 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
29981 small constant length operands.
29983 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29985 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
29986 between loop iterations.
29988 2017-01-05 Martin Liska <mliska@suse.cz>
29991 * gimplify.c (gimplify_decl_expr): Compare to
29992 asan_poisoned_variables instread of checking flags.
29993 (gimplify_target_expr): Likewise.
29994 (gimplify_expr): Likewise.
29995 (gimplify_function_tree): Conditionally initialize
29996 asan_poisoned_variables.
29998 2017-01-04 Jeff Law <law@redhat.com>
30000 PR tree-optimizatin/78812
30001 * rtl.h (contains_mem_rtx_p): Prototype.
30002 * ifcvt.c (containts_mem_rtx_p): Move from here to...
30003 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
30004 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
30005 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
30006 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
30008 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30010 * input.c (assert_char_at_range): Default-initialize actual_range.
30012 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30014 * df-scan.c (df_ref_create_structure): Make regno unsigned,
30015 to match the caller.
30017 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30019 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
30020 insns after final jump in test to emit dummy move.
30022 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30024 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
30025 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
30027 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
30029 * multiple_target.c (create_dispatcher_calls): Init e_next.
30030 * tree-ssa-loop-split.c (split_loop): Init border.
30031 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
30034 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
30039 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
30040 (altivec_register_operand): Do not return true if the operand
30041 contains a SUBREG mixing SImode and SFmode.
30042 (vsx_register_operand): Likewise.
30043 (vsx_reg_sfsubreg_ok): New predicate.
30044 (vfloat_operand): Do not return true if the operand contains a
30045 SUBREG mixing SImode and SFmode.
30046 (vint_operand): Likewise.
30047 (vlogical_operand): Likewise.
30048 (gpc_reg_operand): Likewise.
30049 (int_reg_operand): Likewise.
30050 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
30051 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
30052 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
30054 (rs6000_emit_move_si_sf_subreg): New helper function.
30055 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
30056 fixup SUBREGs involving SImode and SFmode.
30057 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
30058 numbers for the new peephole2 optimization.
30059 (peephole2 for SFmode unions): New peephole2 to optimize cases in
30060 the GLIBC math library that do AND/IOR/XOR operations on single
30061 precision floating point.
30062 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
30063 target macros to say whether we need to avoid SUBREGs mixing
30065 (TARGET_ALLOW_SF_SUBREG): Likewise.
30066 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
30067 (UNSPEC_SI_FROM_SF): Likewise.
30068 (iorxor): Change spacing.
30069 (and_ior_xor): New iterator for AND, IOR, and XOR.
30070 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
30071 (movdi_from_sf_zero_ext): Likewise.
30072 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
30073 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
30074 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
30075 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
30076 (fms<mode>4): Likewise.
30077 (fnma<mode>4): Likewise.
30078 (fnms<mode>4): Likewise.
30079 (nfma<mode>4): Likewise.
30080 (nfms<mode>4): Likewise.
30082 2017-01-04 Marek Polacek <polacek@redhat.com>
30085 * doc/invoke.texi: Document -Wpointer-compare.
30087 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30089 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
30092 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
30093 descriptions for -gdwarf-5 and emit them as uleb128 instead of
30096 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
30099 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
30100 documentation of the powerpc_popcntb_ok attribute.
30101 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
30102 code to issue warning messages if a requested CPU configuration is
30103 not supported by the binary (assembler and loader) toolchain.
30104 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
30105 made to define a built-in function that has been disabled.
30106 (paired_init_builtins): Add assertion to prevent ICE if attempt is
30107 made to define a built-in function that has been disabled.
30108 (altivec_init_builtins): Add comment explaining why definition
30109 of the DST built-in functions is not preceded by an assertion
30110 check. Add assertions to prevent ICE if attempts are made to
30111 define an altivec predicate or an abs* built-in function that has
30113 (htm_init_builtins): Add comment explaining why definition of the
30114 htm built-in functions is not preceded by an assertion check.
30116 2017-01-04 Jeff Law <law@redhat.com>
30118 PR tree-optimizatin/67955
30119 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
30120 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
30121 the points-to solution does not include pt_null. Use DECL_PT_UID
30124 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
30126 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
30127 Use gen_int_mode instead of gen_lopwart for const_int operands.
30129 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30131 PR tree-optimization/71563
30132 * match.pd: Simplify X << Y into X if Y is known to be 0 or
30133 out of range value - has low bits known to be zero.
30135 2017-01-04 Alan Modra <amodra@gmail.com>
30137 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
30138 * configure: Regenerate.
30139 * config.in: Regenerate.
30141 2017-01-04 Jakub Jelinek <jakub@redhat.com>
30144 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
30145 a substring of the message, but strcmp with the whole message. Ifdef
30146 ENABLE_NLS, translate the message first using dgettext.
30148 2017-01-03 Jeff Law <law@redhat.com>
30150 PR tree-optimizatin/78856
30151 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
30152 (mark_threaded_blocks): Remove code to truncate thread paths that
30153 cross multiple loop headers. Instead invalidate the cached loop
30154 iteration information and handle case of a thread path walking
30155 into an irreducible region.
30157 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
30160 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
30161 assertions. Add support for doing the signbit if the IEEE 128-bit
30162 floating point value is in a GPR.
30163 * config/rs6000/rs6000.md (Fsignbit): Delete.
30164 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
30165 Update the length attribute if the value is in a GPR.
30166 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
30167 the sign or zero extension instruction, since the value is always 0/1.
30168 (signbit<mode>2_dm2): Delete using <Fsignbit>.
30171 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
30172 extracting SImode to a GPR register so that we can generate a
30173 store, limit the vector to be in a traditional Altivec register
30174 for the vextuwrx instruction.
30176 2017-01-03 Ian Lance Taylor <iant@google.com>
30178 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
30180 2017-01-03 Martin Sebor <msebor@redhat.com>
30182 PR tree-optimization/78696
30183 * gimple-ssa-sprintf.c (format_floating): Correct handling of
30184 precision. Use MPFR for %f for greater fidelity. Correct handling
30186 (pass_sprintf_length::compute_format_length): Set width and precision
30187 specified by asrerisk to void_node for vararg functions.
30188 (try_substitute_return_value): Adjust dump output.
30190 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
30192 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
30194 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
30196 * doc/invoke.texi (SPARC options): Document -mlra as the default.
30197 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
30198 -mlra/-mno-lra was passed to the compiler.
30200 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
30202 PR rtl-optimization/65618
30203 * emit-rtl.c (try_split): Move initialization of "before" and
30204 "after" to just before the call to emit_insn_after_setloc.
30206 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
30208 * doc/md.texi (Standard Names): Remove reference to Java frontend.
30210 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
30212 * dwarf2out.c (gen_enumeration_type_die): When
30213 -gno-strict-dwarf, add a DW_AT_encoding attribute.
30215 2017-01-03 Jakub Jelinek <jakub@redhat.com>
30217 PR tree-optimization/78965
30218 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
30219 Change first argument from const call_info & to call_info &. For %n
30220 set info.nowrite to false.
30222 PR middle-end/78901
30223 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
30224 possibly throwing calls.
30226 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
30227 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
30228 and fns handling, rather than in a separate case SSA_NAME.
30230 2017-01-02 Jeff Law <law@redhat.com>
30232 * config/darwin-driver.c (darwin_driver_init): Const-correctness
30233 fixes for first_period and second_period variables.
30235 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
30238 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
30239 (*insvqi_1): New insn pattern.
30240 (*insvqi_1_mem_rex64): Ditto.
30241 (*insvqi_2): Ditto.
30242 (*insvqi_3): Rename from *insvqi.
30244 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
30246 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
30248 * doc/cfg.texi (Edges): Remove reference to Java.
30249 (Maintaining the CFG): Ditto.
30251 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30253 PR middle-end/77674
30254 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
30255 transparent aliases.
30257 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
30259 PR middle-end/77484
30260 * predict.def (PRED_CALL): Update hitrate.
30261 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
30262 * predict.c (tree_estimate_probability_bb): Split CALL predictor
30263 into direct/indirect/polymorphic variants.
30265 2017-01-01 Jakub Jelinek <jakub@redhat.com>
30267 Update copyright years.
30269 * gcc.c (process_command): Update copyright notice dates.
30270 * gcov-dump.c (print_version): Ditto.
30271 * gcov.c (print_version): Ditto.
30272 * gcov-tool.c (print_version): Ditto.
30273 * gengtype.c (create_file): Ditto.
30274 * doc/cpp.texi: Bump @copying's copyright year.
30275 * doc/cppinternals.texi: Ditto.
30276 * doc/gcc.texi: Ditto.
30277 * doc/gccint.texi: Ditto.
30278 * doc/gcov.texi: Ditto.
30279 * doc/install.texi: Ditto.
30280 * doc/invoke.texi: Ditto.
30282 Copyright (C) 2017 Free Software Foundation, Inc.
30284 Copying and distribution of this file, with or without modification,
30285 are permitted in any medium without royalty provided the copyright
30286 notice and this notice are preserved.