1 2017-11-09 Jakub Jelinek <jakub@redhat.com>
3 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
5 (store_immediate_info::store_immediate_info): Add bitnotp argument,
7 (imm_store_chain_info::coalesce_immediate_stores): Break group
8 if bit_not_p is different.
9 (count_multiple_uses, split_group,
10 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
11 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
12 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
13 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
14 to store_immediate_info ctor.
16 2017-11-09 Jim Wilson <jimw@sifive.com>
18 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
19 (scan_prog_file): Likewise.
21 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
23 * bb-reorder.c (max_entry_frequency): Remove.
24 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
25 connect_traces, push_to_next_round_p): Remove prototypes.
26 (find_traces_1_round): Use counts only.
27 (push_to_next_round_p): Likewise.
28 (find_traces): Likewise.
29 (rotate_loop): Likewise.
30 (find_traces_1_round): Likewise.
31 (connect_traces): Likewise.
32 (edge_order): Likewise.
34 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
36 * config/arm/arm.c (output_return_instruction): Add comments to
37 indicate requirement for cmse_nonsecure_entry return to account
38 for the size of clearing instruction output here.
39 (thumb_exit): Likewise.
40 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
41 return in hardfloat mode.
43 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
45 * config/rs6000/rs6000.c (machine_function): Add a bool,
46 "toc_is_wrapped_separately".
47 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
48 if it wasn't explicitly set or unset, we are optimizing for speed, and
49 doing separate shrink-wrapping.
50 (rs6000_get_separate_components): Enable the TOC component if
51 saving the TOC register in the prologue.
52 (rs6000_components_for_bb): Handle the TOC component.
53 (rs6000_emit_prologue_components): Store the TOC register where needed.
54 (rs6000_set_handled_components): Mark TOC as handled, if handled.
55 (rs6000_emit_prologue): Don't save the TOC if that is already done.
57 2017-11-09 Martin Jambor <mjambor@suse.cz>
59 * ipa-param-manipulation.c: New file.
60 * ipa-param-manipulation.h: Likewise.
61 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
62 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
63 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
64 (ipa_parm_adjustment): Likewise.
65 (ipa_parm_adjustment_vec): Likewise.
66 (ipa_get_vector_of_formal_parms): Moved declaration to
67 ipa-param-manipulation.h.
68 (ipa_get_vector_of_formal_parm_types): Likewise.
69 (ipa_modify_formal_parameters): Likewise.
70 (ipa_modify_call_arguments): Likewise.
71 (ipa_combine_adjustments): Likewise.
72 (ipa_dump_param_adjustments): Likewise.
73 (ipa_modify_expr): Likewise.
74 (ipa_get_adjustment_candidate): Likewise.
75 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
76 ipa-param-manipulation.c.
77 (ipa_get_vector_of_formal_parm_types): Likewise.
78 (ipa_modify_formal_parameters): Likewise.
79 (ipa_modify_call_arguments): Likewise.
80 (ipa_modify_expr): Likewise.
81 (get_ssa_base_param): Likewise.
82 (ipa_get_adjustment_candidate): Likewise.
83 (index_in_adjustments_multiple_times_p): Likewise.
84 (ipa_combine_adjustments): Likewise.
85 (ipa_dump_param_adjustments): Likewise.
86 * tree-sra.c: Also include ipa-param-manipulation.h
87 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
90 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
91 Alan Hayward <alan.hayward@arm.com>
92 David Sherwood <david.sherwood@arm.com>
94 * doc/sourcebuild.texi (vect_masked_store): Document.
96 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
97 Alan Hayward <alan.hayward@arm.com>
98 David Sherwood <david.sherwood@arm.com>
100 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
102 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
103 Alan Hayward <alan.hayward@arm.com>
104 David Sherwood <david.sherwood@arm.com>
106 * doc/sourcebuild.texi (vect_variable_length): Document.
108 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
109 Alan Hayward <alan.hayward@arm.com>
110 David Sherwood <david.sherwood@arm.com>
112 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
114 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
115 Alan Hayward <alan.hayward@arm.com>
116 David Sherwood <david.sherwood@arm.com>
118 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
120 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
121 Alan Hayward <alan.hayward@arm.com>
122 David Sherwood <david.sherwood@arm.com>
124 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
125 previously undocumented selectors.
126 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
128 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
130 * doc/rtl.texi (const_vector): Say that elements can be
132 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
133 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
134 (gen_vec_duplicate): Use it instead of CONSTANT_P.
135 * optabs.c (expand_vector_broadcast): Likewise.
137 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
138 Alan Hayward <alan.hayward@arm.com>
139 David Sherwood <david.sherwood@arm.com>
141 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
142 scaled index even if the unscaled address was invalid.
143 Don't increase the complexity of using a scale in that case.
145 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
146 Alan Hayward <alan.hayward@arm.com>
147 David Sherwood <david.sherwood@arm.com>
149 * doc/rtl.texi: Rewrite the subreg rules so that they partition
150 the inner register into REGMODE_NATURAL_SIZE bytes rather than
151 UNITS_PER_WORD bytes.
152 * emit-rtl.c (validate_subreg): Divide subregs into blocks
153 based on REGMODE_NATURAL_SIZE of the inner mode.
154 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
155 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
156 * expmed.c (lowpart_bit_field_p): Divide the value up into
157 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
158 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
159 whether something is likely to occupy more than one register.
161 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
164 * ipa-inline-transform.c (update_noncloned_frequencies): Use
165 profile_count::adjust_for_ipa_scaling.
166 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
167 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
169 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
171 2017-11-09 Jakub Jelinek <jakub@redhat.com>
173 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
174 (split_group): Add total_orig and total_new arguments, estimate the
175 number of statements related to the store group without store merging
176 and with store merging.
177 (imm_store_chain_info::output_merged_store): Adjust split_group
178 callers, punt if estimated number of statements with store merging
179 is not smaller than estimated number of statements without it.
181 (handled_load): Remove has_single_use checks.
182 (pass_store_merging::process_store): Likewise.
184 2017-11-09 Richard Biener <rguenther@suse.de>
186 PR tree-optimization/82902
187 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
189 2017-11-09 Martin Liska <mliska@suse.cz>
192 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
195 2017-11-09 Martin Liska <mliska@suse.cz>
197 PR tree-optimization/82669
198 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
200 2017-11-09 Martin Liska <mliska@suse.cz>
202 PR gcov-profile/48463
203 * coverage.c (coverage_begin_function): Output also end locus
204 of a function and information whether the function is
206 * gcov-dump.c (tag_function): Parse and print the information.
207 * gcov.c (INCLUDE_MAP): Add include.
208 (INCLUDE_SET): Likewise.
209 (struct line_info): Move earlier in the source file because
210 of vector<line_info> in function_info structure.
211 (line_info::line_info): Likewise.
212 (line_info::has_block): Likewise.
213 (struct source_info): Add new member index.
214 (source_info::get_functions_at_location): New function.
215 (function_info::group_line_p): New function.
216 (output_intermediate_line): New function.
217 (output_intermediate_file): Use the mentioned function.
218 (struct function_start): New.
219 (struct function_start_pair_hash): Likewise.
220 (process_file): Add code that identifies group functions.
221 Assign lines either to global or function scope.
222 (generate_results): Skip artificial functions.
223 (find_source): Assign index for each source file.
224 (read_graph_file): Read new flag artificial and end_line.
225 (add_line_counts): Assign it either to global of function scope.
226 (accumulate_line_counts): Isolate core of the function to
227 accumulate_line_info and call it for both function and global
229 (accumulate_line_info): New function.
230 (output_line_beginning): Fix GNU coding style.
231 (print_source_line): New function.
232 (output_line_details): Likewise.
233 (output_function_details): Likewise.
234 (output_lines): Iterate both source (global) scope and function
236 (struct function_line_start_cmp): New class.
237 * doc/gcov.texi: Reflect changes in documentation.
239 2017-11-09 Jakub Jelinek <jakub@redhat.com>
242 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
243 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
244 and similarly for not instead of neg.
246 2017-11-08 Andi Kleen <ak@linux.intel.com>
248 * config/i386/i386.opt: Add -mforce-indirect-call.
249 * config/i386/predicates.md: Check for flag_force_indirect_call.
250 * doc/invoke.texi: Document -mforce-indirect-call
252 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
254 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
256 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
257 config/riscv/riscv.c (predict.h): New include.
258 (riscv_slow_unaligned_access_p): No longer static.
259 (riscv_block_move_straight): Add require.
260 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
262 2017-11-08 Jakub Jelinek <jakub@redhat.com>
265 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
266 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
267 nonimmediate_operand predicate for operand 1 instead of
270 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
272 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
274 * config/aarch64/constraints.md (Uml): New constraint.
275 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
278 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
280 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
281 of two vec_duplicates into a vec_concat.
283 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
285 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
286 Simplify vec_merge of vec_duplicate and vec_concat.
287 * config/aarch64/constraints.md (Utq): New constraint.
288 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
291 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
293 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
294 Simplify vec_merge of vec_duplicate and const_vector.
295 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
297 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
298 mode iterator. Update predicate on operand 1 to
299 handle non-const_vec constants. Delete constraints.
300 (*aarch64_combinez_be<mode>): Likewise for operand 2.
302 2017-11-08 Jakub Jelinek <jakub@redhat.com>
304 PR tree-optimization/78821
305 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
307 (store_operand_info::store_operand_info): Initialize it to false.
308 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
309 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
310 store in the group, and if chain_info is non-NULL, to ignore altogether
312 (compatible_load_p): Fail if bit_not_p does not match.
313 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
314 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
315 (pass_store_merging::process_store): Adjust
316 terminate_all_aliasing_chains calls to pass NULL in all current spots,
317 call terminate_all_aliasing_chains newly when adding a store into
318 a chain with non-NULL chain_info.
320 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
322 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
324 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
326 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
328 (aarch64_layout_frame): Initialise emit_frame_chain.
329 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
330 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
332 2017-11-08 Martin Liska <mliska@suse.cz>
334 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
335 of gimple_call_internal_p.
337 2017-11-07 Tom de Vries <tom@codesourcery.com>
339 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
341 2017-11-07 Tom de Vries <tom@codesourcery.com>
343 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
346 2017-11-08 Martin Liska <mliska@suse.cz>
349 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
351 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
353 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
355 2017-11-07 Jakub Jelinek <jakub@redhat.com>
358 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
359 (*cmp<mode>_ccz_1): New insn with $k alternative.
362 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
363 RTX_COMM_COMPARE as commutative as well.
364 (ix86_binary_operator_ok): Formatting fix.
365 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
366 *<code><mode>3<mask_name><round_saeonly_name>,
367 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
368 *<s>mul<mode>3_highpart<mask_name>,
369 *vec_widen_umult_even_v16si<mask_name>,
370 *vec_widen_umult_even_v8si<mask_name>,
371 *vec_widen_umult_even_v4si<mask_name>,
372 *vec_widen_smult_even_v16si<mask_name>,
373 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
374 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
375 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
376 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
377 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
378 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
379 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
380 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
381 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
382 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
383 ix86_binary_operator_ok. Formatting fixes.
384 (*<plusminus_insn><mode>3<mask_name><round_name>,
385 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
388 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
390 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
391 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
392 eq and ne if TARGET_ISEL.
393 (cmp): New code_iterator.
394 (UNS, UNSU_, UNSIK): New code_attrs.
395 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
396 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
398 ("eq<mode>3"): ... this.
399 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
401 ("ne<mode>3"): ... this.
403 2017-11-07 Julia Koval <julia.koval@intel.com>
406 * common/config/i386/i386-common.c
407 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
408 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
409 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
410 * config/i386/i386.opt: Ditto.
411 * config/i386/i386.c (ix86_target_string): Ditto.
412 (ix86_option_override_internal): Ditto.
413 (ix86_init_mpx_builtins): Move MPX to args2.
414 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
415 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
416 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
417 __builtin_ia32_vgf2p8affineinvqb_v32qi,
418 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
419 __builtin_ia32_vgf2p8affineinvqb_v16qi,
420 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
422 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
425 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
426 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
427 (zero-extendsidi peephole2): Remove peephole.
429 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
432 * stor-layout.c: Include attribs.h.
433 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
434 explicit lookup of "aligned" attribute.
436 2017-11-07 Andrew Waterman <andrew@sifive.com>
438 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
439 (riscv_expand_block_move): Likewise.
440 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
442 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
443 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
444 * config/riscv/riscv.c (riscv_block_move_straight): New function.
445 (riscv_adjust_block_mem): Likewise.
446 (riscv_block_move_loop): Likewise.
447 (riscv_expand_block_move): Likewise.
448 gcc/config/riscv/riscv.md (movmemsi): New pattern.
450 2017-11-07 Michael Clark <michaeljclark@mac.com>
452 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
453 (MUSL_DYNAMIC_LINKER): Likewise.
455 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
457 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
458 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
461 2017-11-07 Richard Biener <rguenther@suse.de>
463 * match.pd: Fix build.
465 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
466 Jackson Woodruff <jackson.woodruff@arm.com>
468 PR tree-optimization/71026
469 * match.pd: Canonicalize negate in division.
471 2017-11-07 Sudakshina Das <sudi.das@arm.com>
474 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
476 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
478 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
479 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
481 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
483 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
484 non-scalar integral types.
485 * match.pd (negate_expr_p): Handle MINUS_EXPR.
486 (-(A-B), -(~A)): New transformations.
488 2017-11-07 Tom de Vries <tom@codesourcery.com>
490 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
491 semicolon after "do {} while (0)".
492 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
493 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
494 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
495 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
496 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
498 2017-11-07 Tom de Vries <tom@codesourcery.com>
500 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
501 after "do {} while (0)".
502 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
503 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
504 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
505 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
506 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
508 2017-11-07 Tom de Vries <tom@codesourcery.com>
511 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
513 (arm_rtx_costs_internal): Add missing semicolon after
514 HANDLE_NARROW_SHIFT_ARITH call.
516 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
518 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
519 disable isel if it was not set explicitly.
521 2017-11-06 James Bowman <james.bowman@ftdichip.com>
523 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
525 (add_type_attribute) likewise.
527 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
529 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
530 of struct ix86_frame.
531 (ix86_initial_elimination_offset): Likewise.
532 (ix86_expand_split_stack_prologue): Likewise.
534 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
536 * tree-vrp.h (enum value_range_type): Update stale comment.
538 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
539 Alan Hayward <alan.hayward@arm.com>
540 David Sherwood <david.sherwood@arm.com>
542 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
543 (aarch64_expand_vec_perm_const): Take the number of units too.
544 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
545 (aarch64_expand_vec_perm_const): Likewise.
546 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
547 (vec_perm<mode>): Update accordingly.
549 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
550 Alan Hayward <alan.hayward@arm.com>
551 David Sherwood <david.sherwood@arm.com>
553 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
554 Take the number of units too.
555 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
556 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
557 but check for a vector mode before rather than after the call.
558 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
559 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
560 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
561 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
562 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
563 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
564 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
565 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
566 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
567 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
568 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
569 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
570 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
571 (aarch64_sqdmull2_n<mode>): Update accordingly.
573 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
574 Alan Hayward <alan.hayward@arm.com>
575 David Sherwood <david.sherwood@arm.com>
577 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
578 the number of units too.
579 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
580 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
581 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
582 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
583 (vec_store_lanesxi<mode>): Update accordingly.
585 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
586 Alan Hayward <alan.hayward@arm.com>
587 David Sherwood <david.sherwood@arm.com>
589 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
590 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
591 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
592 of units rather than the mode.
593 * config/aarch64/iterators.md (nunits): New mode attribute.
594 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
595 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
596 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
597 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
598 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
599 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
600 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
601 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
602 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
603 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
604 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
605 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
606 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
607 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
608 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
609 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
610 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
611 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
612 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
613 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
614 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
615 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
616 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
617 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
618 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
619 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
620 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
621 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
622 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
623 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
624 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
625 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
626 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
627 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
628 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
629 (aarch64_simd_vec_setv2di): Likewise.
631 2017-11-06 Carl Love <cel@us.ibm.com>
633 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
635 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
636 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
638 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
640 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
641 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
642 Add power 8 macro expansions.
643 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
644 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
645 power 8 instructions. (VSX_XXBR): Add iterator.
647 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
649 * config/arm/arm.md (predicable_short_it): Change default to "no",
650 improve documentation, remove uses that are identical to the default.
651 (enabled_for_depr_it): Rename to enabled_for_short_it.
652 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
653 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
654 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
655 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
656 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
658 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
661 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
662 float128 helper macros, which are no longer used after deleting
663 the old 'q' built-in functions, and moving the round to odd
664 built-in functions to being special built-in functions.
665 (BU_FLOAT128_2): Likewise.
666 (BU_FLOAT128_1_HW): Likewise.
667 (BU_FLOAT128_2_HW): Likewise.
668 (BU_FLOAT128_3_HW): Likewise.
669 (FABSQ): Delete old 'q' built-in functions.
670 (COPYSIGNQ): Likewise.
671 (SQRTF128_ODD): Move round to odd built-in functions to be
672 special built-in functions, so that we can handle
673 -mabi=ieeelongdouble.
674 (TRUNCF128_ODD): Likewise.
675 (ADDF128_ODD): Likewise.
676 (SUBF128_ODD): Likewise.
677 (MULF128_ODD): Likewise.
678 (DIVF128_ODD): Likewise.
679 (FMAF128_ODD): Likewise.
680 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
681 built-in names to 'f128'.
682 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
683 old 'q' built-in functions, as the machine independent code for
684 'f128' built-in functions handles this.
685 (rs6000_expand_builtin): Add expansion for float128 round to odd
686 functions, keying off on -mabi=ieeelongdouble of whether to use
687 the KFmode or TFmode variant.
688 (rs6000_init_builtins): Initialize the _Float128 round to odd
690 * doc/extend.texi (PowerPC Built-in Functions): Document the old
691 _Float128 'q' built-in functions are now mapped into the new
692 'f128' built-in functions.
694 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
696 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
697 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
699 2017-11-06 David Malcolm <dmalcolm@redhat.com>
702 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
703 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
704 * toplev.c: Include "ipa-fnsummary.h".
705 (toplev::finalize): Call ipa_fnsummary_c_finalize.
707 2017-11-06 Jakub Jelinek <jakub@redhat.com>
709 PR tree-optimization/82838
710 * gimple-ssa-store-merging.c
711 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
712 on a separate gimple_seq which is then appended to seq.
714 2017-11-06 Jeff Law <law@redhat.com>
717 * config/i386/i386.c (PROBE_INTERVAL): Remove.
718 (get_probe_interval): New functions.
719 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
720 (ix86_adjust_stack_and_probe): Likewise.
721 (output_adjust_stack_and_probe): Likewise.
722 (ix86_emit_probe_stack_range): Likewise.
723 (ix86_expand_prologue): Likewise.
725 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
727 PR tree-optimization/82816
728 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
729 if the modes of the two types are the same.
730 (convert_plusminus_to_widen): Likewise.
732 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
734 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
736 (usadv16qi): New define_expand.
737 (usadv8hi): New define_expand.
739 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
742 * ipa-inline-transform.c (update_noncloned_frequencies): Always
744 (inline_transform): Likewise.
745 * predict.c (counts_to_freqs): Remove useless conditional.
746 * profile-count.h (profile_count::apply_scale): Move sanity check.
747 * tree-inline.c (copy_bb): Always scale.
748 (copy_cfg_body): Likewise.
750 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
753 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
755 (*cmp_ite0): Add enabled_for_depr_it attribute.
756 (*cmp_ite1): Likewise.
758 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
760 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
763 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
765 * tree-vrp.c (vrp_int_const_binop): Return true on success and
766 return the value by pointer.
767 (extract_range_from_multiplicative_op_1): Update accordingly.
768 Return as soon as an operation fails.
770 2017-11-05 Tom de Vries <tom@codesourcery.com>
773 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
774 (DEF_SANITIZER_BUILTIN): ... here.
775 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
776 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
778 2017-11-05 Tom de Vries <tom@codesourcery.com>
781 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
783 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
784 ASM_OUTPUT_BEFORE_CASE_LABEL call.
785 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
787 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
788 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
789 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
791 2017-11-05 Tom de Vries <tom@codesourcery.com>
794 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
797 2017-11-04 Michael Clark <michaeljclark@mac.com>
799 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
800 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
802 (*addsi3_extended): Likewise.
803 (*addsi3_extended2): Likewise.
804 (<optab>si3): Likewise.
805 (<optab>di3): Likewise.
806 (<optab><mode>3): Likewise.
807 (<*optabe>si3_internal): Likewise.
808 (zero_extendqi<SUPERQI:mode>2): Likewise.
809 (*add<mode>hi3): Likewise.
810 (*xor<mode>hi3): Likewise.
811 (<optab>di3): Likewise.
812 (*<optab>si3_extend): Likewise.
813 (*sge<u>_<X:mode><GPR:mode>): Likewise.
814 (*slt<u>_<X:mode><GPR:mode>): Likewise.
815 (*sle<u>_<X:mode><GPR:mode>): Likewise.
817 2017-11-04 Andrew Waterman <andrew@sifive.com>
819 * config/riscv/riscv.c (riscv_option_override): Conditionally set
820 TARGET_STRICT_ALIGN based upon -mtune argument.
822 2017-11-04 Andrew Waterman <andrew@sifive.com>
824 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
826 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
828 config/i386/i386.c (choose_basereg): Use optional scratch
829 register and add assertion.
830 (x86_emit_outlined_ms2sysv_save): Use scratch register when
831 needed, and don't allocate stack.
832 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
833 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
834 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
836 2017-11-03 Jeff Law <law@redhat.com>
838 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
839 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
840 to probe at the start of a noreturn function.
842 2017-11-03 Jakub Jelinek <jakub@redhat.com>
844 PR tree-optimization/78821
845 * gimple-ssa-store-merging.c: Update the file comment.
846 (MAX_STORE_ALIAS_CHECKS): Define.
847 (struct store_operand_info): New type.
848 (store_operand_info::store_operand_info): New constructor.
849 (struct store_immediate_info): Add rhs_code and ops data members.
850 (store_immediate_info::store_immediate_info): Add rhscode, op0r
851 and op1r arguments to the ctor, initialize corresponding data members.
852 (struct merged_store_group): Add load_align_base and load_align
854 (merged_store_group::merged_store_group): Initialize them.
855 (merged_store_group::do_merge): Update them.
856 (merged_store_group::apply_stores): Pick the constant for
857 encode_tree_to_bitpos from one of the two operands, or skip
858 encode_tree_to_bitpos if neither operand is a constant.
859 (class pass_store_merging): Add process_store method decl. Remove
860 bool argument from terminate_all_aliasing_chains method decl.
861 (pass_store_merging::terminate_all_aliasing_chains): Remove
862 var_offset_p argument and corresponding handling.
863 (stmts_may_clobber_ref_p): New function.
864 (compatible_load_p): New function.
865 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
866 if there is overlap and rhs_code is not INTEGER_CST. For
867 non-overlapping stores terminate group if rhs is not mergeable.
868 (get_alias_type_for_stmts): Change first argument from
869 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
870 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
871 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
873 (get_location_for_stmts): Change first argument from
874 auto_vec<gimple *> & to vec<gimple *> &.
875 (struct split_store): Remove orig_stmts data member, add orig_stores.
876 (split_store::split_store): Create orig_stores rather than orig_stmts.
877 (find_constituent_stmts): Renamed to ...
878 (find_constituent_stores): ... this. Change second argument from
879 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
880 to info structures rather than the statements.
881 (split_group): Rename ALLOW_UNALIGNED argument to
882 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
883 it. Adjust find_constituent_stores caller.
884 (imm_store_chain_info::output_merged_store): Handle rhs_code other
885 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
886 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
887 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
888 (mem_valid_for_store_merging): New function.
889 (handled_load): New function.
890 (pass_store_merging::process_store): New method.
891 (pass_store_merging::execute): Use process_store method. Adjust
892 terminate_all_aliasing_chains caller.
894 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
896 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
897 Return true for more constants, symbols and label references.
898 (aarch64_valid_floating_const): Remove unused function.
900 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
903 * config/aarch64/aarch64.c (aarch64_layout_frame):
904 Undo forcing of LR at bottom of frame.
906 2017-11-03 Jeff Law <law@redhat.com>
909 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
910 for int_registers_saved.
912 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
913 extracted from tree-ssa-dom.c.
914 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
915 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
916 (record_equivalences_from_incoming_edge): Add additional argument
917 to single_pred_edge_ignoring_loop_edges call.
918 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
919 (uncprop_dom_walker::before_dom_children): Add additional argument
920 to single_pred_edge_ignoring_loop_edges call.
921 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
922 single_pred_edge_ignoring_loop_edges rather than open coding.
923 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
925 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
927 * match.pd (-(-A)): Rewrite.
929 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
931 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
932 (rs6000_emit_int_cmove): New declaration.
933 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
934 (rs6000_emit_sISEL): Delete.
935 (rs6000_emit_int_cmove): Make non-static.
936 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
937 instead of rs6000_emit_sISEL.
939 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
941 * asan.c (create_cond_insert_point): Maintain profile.
942 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
944 * basic-block.h (struct basic_block_def): Remove frequency.
945 (EDGE_FREQUENCY): Use to_frequency
946 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
948 (find_traces): Update to use to_frequency.
949 (find_traces_1_round): Likewise; use only IPA counts.
950 (bb_to_key): Likewise.
951 (connect_traces): Use IPA counts only.
952 (copy_bb_p): Update to use to_frequency.
953 (fix_up_crossing_landing_pad): Likewise.
954 (sanitize_hot_paths): Likewise.
955 * bt-load.c (basic_block_freq): Likewise.
956 * cfg.c (init_flow): Set count_max to uninitialized.
957 (check_bb_profile): Remove frequencies; check counts.
958 (dump_bb_info): Do not dump frequencies.
959 (update_bb_profile_for_threading): Update counts only.
960 (scale_bbs_frequencies_int): Likewise.
961 (MAX_SAFE_MULTIPLIER): Remove.
962 (scale_bbs_frequencies_gcov_type): Update counts only.
963 (scale_bbs_frequencies_profile_count): Update counts only.
964 (scale_bbs_frequencies): Update counts only.
965 * cfg.h (struct control_flow_graph): Add count-max.
966 (update_bb_profile_for_threading): Update prototype.
967 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
968 (find_many_sub_basic_blocks): Likewise.
969 * cfgcleanup.c (try_forward_edges): Likewise.
970 (try_crossjump_to_edge): Likewise.
971 * cfgexpand.c (expand_gimple_cond): Likewise.
972 (expand_gimple_tailcall): Likewise.
973 (construct_init_block): Likewise.
974 (construct_exit_block): Likewise.
975 * cfghooks.c (verify_flow_info): Check consistency of counts.
976 (dump_bb_for_graph): Do not dump frequencies.
977 (split_block_1): Do not update frequencies.
978 (split_edge): Do not update frequencies.
979 (make_forwarder_block): Do not update frequencies.
980 (duplicate_block): Do not update frequencies.
981 (account_profile_record): Do not update frequencies.
982 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
983 for global heuristics.
984 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
985 (expected_loop_iterations_unbounded): Use counts only.
986 * cfgloopmanip.c (scale_loop_profile): Simplify.
987 (create_empty_loop_on_edge): Simplify
989 (duplicate_loop_to_header_edge): Simplify
990 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
991 (update_br_prob_note): Take care of removing note when profile
993 (relink_block_chain): Do not dump frequency.
994 (rtl_account_profile_record): Use to_frequency.
995 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
996 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
997 (cgraph_update_edges_for_call_stmt_node): Likewise.
998 (cgraph_edge::verify_count_and_frequency): Update.
999 (cgraph_node::verify_node): Temporarily disable frequency verification.
1000 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1001 to_cgraph_frequency.
1002 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1003 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1005 (cgraph_node::expand_thunk): Update profile.
1006 * except.c (dw2_build_landing_pads): Do not update frequency.
1007 * final.c (compute_alignments): Use to_frequency.
1008 (dump_basic_block_info): Do not dump frequency.
1009 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1010 (dump_gimple_bb_header): Do not dump frequency.
1011 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1013 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1014 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1015 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1016 (init_before_recovery): Do not update frequency.
1017 (sched_create_recovery_edges): Do not update frequency.
1018 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1019 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1020 (ipa_cp_c_finalize): Set max_count to uninitialized.
1021 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1022 (param_change_prob): Use counts.
1023 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1025 * ipa-split.c (consider_split): Use to_frequency.
1026 (split_function): Use to_frequency.
1027 * ira-build.c (loop_compare_func): Likewise.
1028 (mark_loops_for_removal): Likewise.
1029 (mark_all_loops_for_removal): Likewise.
1030 * loop-doloop.c (doloop_modify): Do not update frequency.
1031 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1033 * lto-streamer-in.c (input_function): Update count_max.
1034 * omp-expand.c (expand_omp_taskreg): Update count_max.
1035 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1036 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1037 (maybe_hot_count_p): Use ipa counts only.
1038 (maybe_hot_bb_p): Simplify.
1039 (maybe_hot_edge_p): Simplify.
1040 (probably_never_executed): Do not take frequency argument.
1041 (probably_never_executed_bb_p): Do not pass frequency.
1042 (probably_never_executed_edge_p): Likewise.
1043 (combine_predictions_for_bb): Check that profile is nonzero.
1044 (propagate_freq): Do not set frequency.
1045 (drop_profile): Simplify.
1046 (counts_to_freqs): Simplify.
1047 (expensive_function_p): Use to_frequency.
1048 (propagate_unlikely_bbs_forward): Simplify.
1049 (determine_unlikely_bbs): Simplify.
1050 (estimate_bb_frequencies): Add hack to silence graphite issues.
1051 (compute_function_frequency): Use ipa counts.
1052 (pass_profile::execute): Update.
1053 (rebuild_frequencies): Use counts only.
1054 (force_edge_cold): Use counts only.
1055 * profile-count.c (profile_count::dump): Dump new count types.
1056 (profile_count::differs_from_p): Check compatiblity.
1057 (profile_count::to_frequency): New function.
1058 (profile_count::to_cgraph_frequency): New function.
1059 * profile-count.h (struct function): Declare.
1060 (enum profile_quality): Add profile_guessed_local and
1061 profile_guessed_global0.
1062 (class profile_proability): Decrease number of bits to 29;
1063 update from_reg_br_prob_note and to_reg_br_prob_note.
1064 (class profile_count: Update comment; decrease number of bits
1065 to 61. Check compatibility.
1066 (profile_count::compatible_p): New private member function.
1067 (profile_count::ipa_p): New member function.
1068 (profile_count::operator<): Handle global zero correctly.
1069 (profile_count::operator>): Handle global zero correctly.
1070 (profile_count::operator<=): Handle global zero correctly.
1071 (profile_count::operator>=): Handle global zero correctly.
1072 (profile_count::nonzero_p): New member function.
1073 (profile_count::force_nonzero): New member function.
1074 (profile_count::max): New member function.
1075 (profile_count::apply_scale): Handle IPA scalling.
1076 (profile_count::guessed_local): New member function.
1077 (profile_count::global0): New member function.
1078 (profile_count::ipa): New member function.
1079 (profile_count::to_frequency): Declare.
1080 (profile_count::to_cgraph_frequency): Declare.
1081 * profile.c (OVERLAP_BASE): Delete.
1082 (compute_frequency_overlap): Delete.
1083 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1084 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1085 * sched-ebb.c (rank): Use counts only.
1086 * shrink-wrap.c (handle_simple_exit): Use counts only.
1087 (try_shrink_wrapping): Use counts only.
1088 (place_prologue_for_one_component): Use counts only.
1089 * tracer.c (find_best_predecessor): Use to_frequency.
1090 (find_trace): Use to_frequency.
1091 (tail_duplicate): Use to_frequency.
1092 * trans-mem.c (expand_transaction): Do not update frequency.
1093 * tree-call-cdce.c: Do not update frequency.
1094 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1095 (gimple_merge_blocks): Likewise.
1096 (gimple_split_edge): Likewise.
1097 (gimple_duplicate_sese_region): Likewise.
1098 (gimple_duplicate_sese_tail): Likewise.
1099 (move_sese_region_to_fn): Likewise.
1100 (gimple_account_profile_record): Likewise.
1101 (insert_cond_bb): Likewise.
1102 * tree-complex.c (expand_complex_div_wide): Likewise.
1103 * tree-eh.c (lower_resx): Update profile.
1104 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1106 (initialize_cfun): Do not initialize frequencies
1107 (freqs_to_counts): Delete.
1108 (copy_cfg_body): Ignore count parameter.
1109 (copy_body): Update.
1110 (expand_call_inline): Update count_max.
1111 (optimize_inline_calls): Update count_max.
1112 (tree_function_versioning): Update count_max.
1113 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1114 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1116 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1117 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1118 (try_peel_loop): Likewise.
1119 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1121 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1122 (tree_transform_and_unroll_loop): Do not use frequencies
1123 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1124 Use reliable prediction only.
1125 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1126 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1127 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1128 probability scaling.
1129 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1130 not update frequency
1131 (any_remaining_duplicated_blocks): Likewise.
1132 (update_profile): Likewise.
1133 (estimated_freqs_path): Delete.
1134 (freqs_to_counts_path): Delete.
1135 (clear_counts_path): Delete.
1136 (ssa_fix_duplicate_block_edges): Likewise.
1137 (duplicate_thread_path): Likewise.
1138 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1139 * tree-tailcall.c (decrease_profile): Do not update frequency.
1140 (eliminate_tail_call): Likewise.
1141 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1142 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1143 (optimize_mask_stores): Likewise.
1144 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1145 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1146 (ubsan_expand_ptr_ifn): Update profile.
1147 * value-prof.c (gimple_ic): Simplify.
1148 * value-prof.h (gimple_ic): Update prototype.
1149 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1150 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1152 (want_inline_self_recursive_call_p): Likewise.
1153 (resolve_noninline_speculation): Only cummulate defined counts.
1154 (inline_small_functions): Use nonzero_p.
1155 (ipa_inline): Do not access freed node.
1157 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1159 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1160 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1162 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1164 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1165 non-legitimate address.
1167 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1169 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1170 (*lt0_<mode>di, *lt0_<mode>si): New.
1172 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1174 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1175 TARGET_PAIRED_FLOAT.
1177 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1178 Jim Wilson <jim.wilson@linaro.org>
1180 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1181 * config/aarch64/aarch64-tune.md: Regenerated.
1182 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1183 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1185 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1187 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1188 (arc_expand_prologue): Restore blink for millicode.
1189 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1191 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1194 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1195 gen_vec_duplicate after forcing the scalar into a register.
1197 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1199 * combine (try_combine): Print the insns input to try_combine to the
1202 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1205 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1206 Remove second argument from aarch64_process_target_attr call.
1207 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1209 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1211 (aarch64_handle_attr_arch): Remove second argument.
1212 (aarch64_handle_attr_cpu): Ditto.
1213 (aarch64_handle_attr_tune): Ditto.
1214 (aarch64_handle_attr_isa_flags): Ditto.
1215 (aarch64_process_one_target_attr): Ditto.
1216 (aarch64_process_target_attr): Ditto.
1217 (aarch64_option_valid_attribute_p): Remove second argument.
1218 on aarch64_process_target_attr call.
1220 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1222 * diagnostic.c: Include "selftest-diagnostic.h".
1223 (selftest::assert_location_text): New function.
1224 (selftest::test_diagnostic_get_location_text): New function.
1225 (selftest::diagnostic_c_tests): Call it.
1227 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1229 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1230 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1231 (class selftest::test_diagnostic_context): Move to...
1232 * selftest-diagnostic.c: New file.
1233 * selftest-diagnostic.h: New file.
1235 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1237 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1238 offset range for FT32B.
1239 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1240 * config/ft32/ft32.md: Add TARGET_NOPM.
1241 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1242 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1244 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1246 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1248 2017-11-02 Jeff Law <law@redhat.com>
1250 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1251 virtual keyword on FINAL OVERRIDE members.
1253 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1254 virtuals together. Add virtual destructor.
1255 (substitute_and_fold_engine): Similarly.
1257 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1259 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1261 2017-11-02 Richard Biener <rguenther@suse.de>
1263 PR tree-optimization/82795
1264 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1266 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1268 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1270 (gcc_GAS_CHECK_FEATURE): Remove.
1271 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1272 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1273 * configure: Regenerate.
1275 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1277 * config/arc/arc.c (hwloop_optimize): Account for empty
1280 2017-11-02 Richard Biener <rguenther@suse.de>
1283 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1284 Truncate ARRAY_REF index and element size.
1286 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1288 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1290 2017-11-01 Jeff Law <law@redhat.com>
1292 * tree-ssa-ccp.c (ccp_folder): New class derived from
1293 substitute_and_fold_engine.
1294 (ccp_folder::get_value): New member function.
1295 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1296 (ccp_fold_stmt): Remove prototype.
1297 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1298 * tree-ssa-copy.c (copy_folder): New class derived from
1299 substitute_and_fold_engine.
1300 (copy_folder::get_value): Renamed from get_value.
1301 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1302 * tree-vrp.c (vrp_folder): New class derived from
1303 substitute_and_fold_engine.
1304 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1305 (vrp_folder::get_value): New member function.
1306 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1307 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1308 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1309 provide a class interface to folder/substitute routines.
1310 (ssa_prop_fold_stmt_fn): Remove typedef.
1311 (ssa_prop_get_value_fn): Likewise.
1312 (subsitute_and_fold): Remove prototype.
1313 (replace_uses_in): Likewise.
1314 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1315 Renamed from replace_uses_in. Call the virtual member function
1316 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1317 (substitute_and_fold_dom_walker): Remove initialization of
1318 data member entries for calbacks. Add substitute_and_fold_engine
1319 member and initialize it.
1320 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1321 member functions for get_value, replace_phi_args_in c
1322 replace_uses_in, and fold_stmt calls.
1323 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1324 substitute_and_fold. Remove assert. Update ctor call.
1326 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1327 (ssa_prop_visit_phi_fn): Likewise.
1328 (class ssa_propagation_engine): New class to provide an interface
1330 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1332 (ssa_prop_visit_phi): Likewise.
1333 (ssa_propagation_engine::simulate_stmt): Moved into class.
1334 Call visit_phi/visit_stmt from the class rather than via
1335 file scoped static variables.
1336 (ssa_propagation_engine::simulate_block): Moved into class.
1337 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1338 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1339 set file scoped statics for the visit_stmt/visit_phi callbacks.
1340 * tree-complex.c (complex_propagate): New class derived from
1341 ssa_propagation_engine.
1342 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1343 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1344 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1346 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1347 ssa_propagation_engine.
1348 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1349 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1350 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1351 * tree-ssa-copy.c (copy_prop): New class derived from
1352 ssa_propagation_engine.
1353 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1354 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1355 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1356 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1357 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1358 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1359 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1361 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1363 PR rtl-optimization/82778
1364 PR rtl-optimization/82597
1365 * compare-elim.c (struct comparison): Add in_a_setter field.
1366 (find_comparison_dom_walker::before_dom_children): Remove killed
1367 bitmap and df_simulate_find_defs call, instead walk the defs.
1368 Compute last_setter and initialize in_a_setter. Merge definitions
1369 with first initialization for a few variables.
1370 (try_validate_parallel): Use insn_invalid_p instead of
1371 recog_memoized. Return insn rather than just the pattern.
1372 (try_merge_compare): Fix up comment. Don't uselessly test if
1373 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1375 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1378 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1379 Alan Hayward <alan.hayward@arm.com>
1380 David Sherwood <david.sherwood@arm.com>
1382 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1383 aarch64_hard_regno_nregs to get the number of registers
1386 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1387 Alan Hayward <alan.hayward@arm.com>
1388 David Sherwood <david.sherwood@arm.com>
1390 * config/aarch64/constraints.md (Upl): Rename to...
1392 * config/aarch64/aarch64.md
1393 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1396 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1397 Alan Hayward <alan.hayward@arm.com>
1398 David Sherwood <david.sherwood@arm.com>
1400 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1401 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1404 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1405 Alan Hayward <alan.hayward@arm.com>
1406 David Sherwood <david.sherwood@arm.com>
1408 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1409 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1410 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1412 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1413 of a single element.
1414 * config/aarch64/iterators.md: Add a comment above the permute
1415 unspecs to say that they are generated directly by
1416 aarch64_expand_vec_perm_const.
1417 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1419 2017-11-01 Nathan Sidwell <nathan@acm.org>
1421 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1423 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1425 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1426 and medany code models, and describe what they do.
1428 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1430 Revert accidental duplicate:
1432 * combine.c (can_change_dest_mode): Reject changes in
1433 REGMODE_NATURAL_SIZE.
1435 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1437 PR rtl-optimization/64682
1438 PR rtl-optimization/69567
1439 PR rtl-optimization/69737
1440 PR rtl-optimization/82683
1441 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1442 register mentioned in the note, drop the note, unless it came from I3,
1443 in which case it should go to I3 again.
1445 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1447 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1448 and return false if not.
1449 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1451 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1453 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1454 range is known to be empty.
1456 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1457 Alan Hayward <alan.hayward@arm.com>
1458 David Sherwood <david.sherwood@arm.com>
1460 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1461 and CONST_VECTOR_NUNITS instead of computing the number of units from
1462 the byte sizes of the vector and element.
1463 (simplify_binary_operation_1): Likewise.
1464 (simplify_const_binary_operation): Likewise.
1465 (simplify_ternary_operation): Likewise.
1467 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1468 Alan Hayward <alan.hayward@arm.com>
1469 David Sherwood <david.sherwood@arm.com>
1471 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1472 (int_mem_offset): ...this new function.
1473 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1474 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1475 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1478 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1479 Alan Hayward <alan.hayward@arm.com>
1480 David Sherwood <david.sherwood@arm.com>
1482 * lower-subreg.c (interesting_mode_p): New function.
1483 (compute_costs, find_decomposable_subregs, decompose_register)
1484 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1485 (resolve_clobber, dump_choices): Use it.
1487 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1488 Alan Hayward <alan.hayward@arm.com>
1489 David Sherwood <david.sherwood@arm.com>
1491 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1493 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1494 Alan Hayward <alan.hayward@arm.com>
1495 David Sherwood <david.sherwood@arm.com>
1497 * alias.c (find_base_value, find_base_term): Only process integer
1498 truncations. Check the precision rather than the size.
1500 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1501 Alan Hayward <alan.hayward@arm.com>
1502 David Sherwood <david.sherwood@arm.com>
1504 * machmode.h (is_narrower_int_mode): New function
1505 * optabs.c (expand_float, expand_fix): Use it.
1506 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1508 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1509 Alan Hayward <alan.hayward@arm.com>
1510 David Sherwood <david.sherwood@arm.com>
1512 * rtl.h (narrower_subreg_mode): New function.
1513 * ira-color.c (update_costs_from_allocno): Use it.
1515 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1516 Alan Hayward <alan.hayward@arm.com>
1517 David Sherwood <david.sherwood@arm.com>
1519 * optabs-query.h (convert_optab_p): New function, split out from...
1520 (convert_optab_handler): ...here.
1521 (widening_optab_handler): Delete.
1522 (find_widening_optab_handler): Remove permit_non_widening parameter.
1523 (find_widening_optab_handler_and_mode): Likewise. Provide an
1524 override that operates on mode class wrappers.
1525 * optabs-query.c (widening_optab_handler): Delete.
1526 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1527 parameter. Assert that the two modes are the same class and that
1528 the "from" mode is narrower than the "to" mode. Use
1529 convert_optab_handler instead of widening_optab_handler.
1530 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1531 instead of widening_optab_handler.
1532 * expr.c (expand_expr_real_2): Update calls to
1533 find_widening_optab_handler.
1534 * optabs.c (expand_widen_pattern_expr): Likewise.
1535 (expand_binop_directly): Take the insn_code as a parameter.
1536 (expand_binop): Only call find_widening_optab_handler for
1537 conversion optabs; use optab_handler otherwise. Update calls
1538 to find_widening_optab_handler and expand_binop_directly.
1539 Use convert_optab_handler instead of widening_optab_handler.
1540 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1541 find_widening_optab_handler and use scalar_mode rather than
1543 (convert_plusminus_to_widen): Likewise.
1545 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1546 Alan Hayward <alan.hayward@arm.com>
1547 David Sherwood <david.sherwood@arm.com>
1549 * machmode.h (fixed_size_mode): New class.
1550 * rtl.h (get_pool_mode): Return fixed_size_mode.
1551 * gengtype.c (main): Add fixed_size_mode.
1552 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1553 (get_raw_arg_mode): Likewise.
1554 * doc/tm.texi: Regenerate.
1555 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1556 * targhooks.c (default_get_reg_raw_mode): Likewise.
1557 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1558 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1559 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1560 (msp430_get_raw_result_mode): Likewise.
1561 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1562 * dbxout.c (dbxout_parms): Require fixed-size modes.
1563 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1564 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1565 * omp-low.c (lower_oacc_reductions): Likewise.
1566 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1567 (simplify_subreg): Update accordingly.
1568 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1569 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1570 that aren't fixed-size.
1571 (get_pool_mode): Return a fixed_size_mode.
1572 (output_constant_pool_2): Take a fixed_size_mode.
1574 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1575 Alan Hayward <alan.hayward@arm.com>
1576 David Sherwood <david.sherwood@arm.com>
1578 * doc/rtl.texi (vec_series): Document.
1579 (const): Say that the operand can be a vec_series.
1580 * rtl.def (VEC_SERIES): New rtx code.
1581 * rtl.h (const_vec_series_p_1): Declare.
1582 (const_vec_series_p): New function.
1583 * emit-rtl.h (gen_const_vec_series): Declare.
1584 (gen_vec_series): Likewise.
1585 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1586 (gen_vec_series): Likewise.
1587 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1588 * simplify-rtx.c (simplify_unary_operation): Handle negations
1590 (simplify_binary_operation_series): New function.
1591 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1592 (test_vector_ops_series): New function.
1593 (test_vector_ops): Call it.
1594 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1595 gen_const_vec_series.
1596 (altivec_lvsr): Likewise.
1597 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1599 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1600 Alan Hayward <alan.hayward@arm.com>
1601 David Sherwood <david.sherwood@arm.com>
1603 * doc/rtl.texi (const): Update description of address constants.
1604 Say that vector constants are allowed too.
1605 * common.md (E, F): Use CONSTANT_P instead of checking for
1607 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1608 checking for CONST_VECTOR.
1609 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1611 * expr.c (expand_expr_real_2): Check for vector modes instead
1612 of checking for CONST_VECTOR.
1613 * rtl.h (const_vec_p): New function.
1614 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1615 (unwrap_const_vec_duplicate): Handle them here too.
1617 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1618 David Malcolm <dmalcolm@redhat.com>
1619 Alan Hayward <alan.hayward@arm.com>
1620 David Sherwood <david.sherwood@arm.com>
1622 * rtl.h (vec_duplicate_p): New function.
1623 * selftest-rtl.c (assert_rtx_eq_at): New function.
1624 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1625 (assert_rtx_eq_at): Declare.
1626 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1627 * selftest-run-tests.c (selftest::run_tests): Call it.
1628 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1629 (simplify_unary_operation_1): Recursively handle vector duplicates.
1630 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1632 (simplify_subreg): Handle subregs of vector duplicates.
1633 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1634 (selftest::simplify_rtx_c_tests): New functions.
1636 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1637 Alan Hayward <alan.hayward@arm.com>
1638 David Sherwood <david.sherwood@arm.com>
1640 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1641 (gen_vec_duplicate): Likewise.
1642 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1644 (gen_const_vector): ...here.
1645 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1646 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1647 whose elements are all equal.
1648 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1649 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1650 (simplify_relational_operation): Likewise.
1651 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1653 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1654 (aarch64_expand_vector_init): Likewise.
1655 * config/arm/arm.c (neon_vdup_constant): Likewise.
1656 (neon_expand_vector_init): Likewise.
1657 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1658 (arm_block_set_unaligned_vect): Likewise.
1659 (arm_block_set_aligned_vect): Likewise.
1660 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1661 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1662 (expand_vec_perm_even_odd_pack): Likewise.
1663 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1664 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1665 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1666 gen_const_vec_duplicate.
1667 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1668 * config/mips/mips.c (mips_gen_const_int_vector): Use
1669 gen_const_vec_duplicate.
1670 (mips_expand_vector_init): Use CONST0_RTX.
1671 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1672 (define_split): Use gen_const_vec_duplicate.
1673 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1674 (define_split): Use gen_const_vec_duplicate.
1675 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1676 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1677 * config/spu/spu.c (spu_const): Likewise.
1679 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1680 Alan Hayward <alan.hayward@arm.com>
1681 David Sherwood <david.sherwood@arm.com>
1683 * combine.c (can_change_dest_mode): Reject changes in
1684 REGMODE_NATURAL_SIZE.
1686 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1688 * configure.ac (--enable-libssp): New.
1689 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1690 trying to determine target-specific default. Adjust indentation.
1691 * configure: Regenerated.
1692 * doc/install.texi (Configuration): Expand --disable-libssp
1695 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1697 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1700 2017-10-31 Martin Jambor <mjambor@suse.cz>
1703 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1705 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1707 * auto-profile.c (autofdo_source_profile::read): Use
1708 UNKNOWN_LOCATION rather than 0.
1709 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1710 (warning_at): ...this overload.
1711 (warning_at_rich_loc_n): Rename to...
1712 (warning_n): ...this overload.
1713 (error_at_rich_loc): Rename to...
1714 (error_at): ...this overload.
1715 (pedwarn_at_rich_loc): Rename to...
1716 (pedwarn): ...this overload.
1717 (permerror_at_rich_loc): Rename to...
1718 (permerror): ...this overload.
1719 (inform_at_rich_loc): Rename to...
1720 (inform): ...this overload.
1721 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1722 (diagnostic_n_impl_richloc): Rename to...
1723 (diagnostic_n_impl): ...this rich_location *-based decl.
1724 (inform_at_rich_loc): Rename to...
1725 (inform): ...this, and add an assertion.
1726 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1727 (warning_at_rich_loc): Rename to...
1728 (warning_at): ...this, and add an assertion.
1729 (warning_at_rich_loc_n): Rename to...
1730 (warning_n): ...this, and add an assertion.
1731 (warning_n): Update location_t-based implementation for removal of
1732 location_t-based diagnostic_n_impl.
1733 (pedwarn_at_rich_loc): Rename to...
1734 (pedwarn): ...this, and add an assertion.
1735 (permerror_at_rich_loc): Rename to...
1736 (permerror): ...this, and add an assertion.
1737 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1738 (error_at_rich_loc): Rename to...
1739 (error_at): ...this, and add an assertion.
1740 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1741 (driver::do_spec_on_infiles): Likewise.
1742 * substring-locations.c (format_warning_va): Update for renaming
1743 of inform_at_rich_loc.
1745 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1747 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1748 _Float<N>X built-in functions so that the variant without the
1749 "__builtin_" prefix is only enabled for the GNU C and Objective C
1750 languages when they are in non-strict ANSI/ISO mode.
1751 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1752 * target.def (floatn_builtin_p): Add a target hook to control
1753 whether _Float<N> and _Float<N>X built-in functions without the
1754 "__builtin_" prefix are enabled, and return true for C and
1755 Objective C in the default hook. Include langhooks.h in
1757 * targhooks.h (default_floatn_builtin_p): Likewise.
1758 * targhooks.c (default_floatn_builtin_p): Likewise.
1759 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1760 floatn_builtin_p target hook.
1761 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1763 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1764 Eric Botcazou <ebotcazou@adacore.com>
1766 PR rtl-optimization/81803
1767 * lra-constraints.c (curr_insn_transform): Also reload the whole
1768 register for a strict subreg no wider than a word if this is for
1769 a WORD_REGISTER_OPERATIONS target.
1771 2017-10-31 Jason Merrill <jason@redhat.com>
1773 * gdbinit.in: Skip over inlines from timevar.h.
1775 2017-10-31 Martin Liska <mliska@suse.cz>
1777 * doc/gcov.texi: Document new option.
1778 * gcov.c (print_usage): Likewise print it.
1779 (process_args): Support the argument.
1780 (format_count): New function.
1781 (format_gcov): Use the function.
1783 2017-10-31 Martin Liska <mliska@suse.cz>
1785 * gcov.c (struct name_map): do not use typedef.
1786 Define operator== and operator<.
1787 (name_search): Remove.
1788 (name_sort): Remove.
1789 (main): Do not allocate names.
1790 (process_file): Add vertical space.
1791 (generate_results): Use std::find.
1792 (release_structures): Do not release memory.
1793 (find_source): Use std::find.
1795 2017-10-31 Martin Liska <mliska@suse.cz>
1797 * gcov.c (struct line_info): Remove it's typedef.
1798 (line_info::line_info): Add proper ctor.
1799 (line_info::has_block): Do not use a typedef.
1800 (struct source_info): Do not use typedef.
1801 (circuit): Likewise.
1802 (get_cycles_count): Likewise.
1803 (output_intermediate_file): Iterate via vector iterator.
1804 (add_line_counts): Use std::vector methods.
1805 (accumulate_line_counts): Likewise.
1806 (output_lines): Likewise.
1808 2017-10-31 Martin Liska <mliska@suse.cz>
1810 * gcov.c (struct source_info): Remove typedef.
1811 (source_info::source_info): Add proper ctor.
1812 (accumulate_line_counts): Use struct, not it's typedef.
1813 (output_gcov_file): Likewise.
1814 (output_lines): Likewise.
1815 (main): Do not allocate an array.
1816 (output_intermediate_file): Use size of vector container.
1817 (process_file): Resize the vector.
1818 (generate_results): Do not preallocate, use newly added vector
1820 (release_structures): Do not release sources.
1821 (find_source): Use vector methods.
1822 (add_line_counts): Do not use typedef.
1824 2017-10-31 Martin Liska <mliska@suse.cz>
1826 * doc/gcov.texi: Document that.
1827 * gcov.c (add_line_counts): Mark lines with a non-executed
1829 (output_line_beginning): Handle such lines.
1830 (output_lines): Pass new argument.
1831 (output_intermediate_file): Print it in intermediate format.
1833 2017-10-31 Martin Liska <mliska@suse.cz>
1835 * color-macros.h: New file.
1836 * diagnostic-color.c: Factor out color related to macros to
1838 * doc/gcov.texi: Document -k option.
1839 * gcov.c (INCLUDE_STRING): Include string.h.
1840 (print_usage): Add -k option.
1841 (process_args): Parse it.
1842 (pad_count_string): New function.
1843 (output_line_beginning): Likewise.
1844 (DEFAULT_LINE_START): New macro.
1845 (output_lines): Support color output.
1847 2017-10-31 Martin Liska <mliska@suse.cz>
1849 PR gcov-profile/82633
1850 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1851 their interaction with GCOV infrastructure.
1852 * configure.ac: Add -fkeep-{inline,static}-functions to
1854 * configure: Regenerate.
1856 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1859 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1861 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1864 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1865 into a register if it does not fit into an immediate offset field.
1867 2017-10-31 Olivier Hainque <hainque@adacore.com>
1869 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1871 2017-10-31 Julia Koval <julia.koval@intel.com>
1873 * config.gcc: Add gfniintrin.h.
1874 * config/i386/gfniintrin.h: New.
1875 * config/i386/i386-builtin-types.def
1876 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1877 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1878 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1879 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1880 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1881 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1882 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1883 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1884 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1885 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1886 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1887 * config/i386/immintrin.h: Include gfniintrin.h.
1888 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1890 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1892 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1894 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1896 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1897 (arm_ashldi3_1bit): Remove pattern.
1898 (ashrdi3): Remove shift by 1 expansion.
1899 (arm_ashrdi3_1bit): Remove pattern.
1900 (lshrdi3): Remove shift by 1 expansion.
1901 (arm_lshrdi3_1bit): Remove pattern.
1902 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1903 cost of ashldi3 by 1.
1904 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1905 (<shift>di3_neon): Likewise.
1907 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1909 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1910 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1911 and (*aarch64_simd_mov<VQ:mode>).
1912 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1913 pattern alternative.
1914 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1915 attributes to match pattern alternative.
1917 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1919 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1920 * config/rs6000/emmintrin.h: New file.
1921 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1923 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1925 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1926 * (movdi_vfp_cortexa8): Remove pattern.
1928 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1930 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1932 (Specific, alpha*-dec-osf5.1): Remove.
1933 (Specific, mips-sgi-irix5): Remove.
1934 (Specific, mips-sgi-irix6): Remove.
1936 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1939 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1940 arguments to clear_bit_region_be.
1942 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1944 * gimplify.c: Include memmodel.h.
1946 2017-10-30 Martin Jambor <mjambor@suse.cz>
1948 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1949 condition whether loop should be executed at all.
1951 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1953 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1954 gimple folding of vec_madd() intrinsics.
1955 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1956 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1957 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1959 2017-10-30 Richard Biener <rguenther@suse.de>
1961 PR tree-optimization/82762
1963 2017-10-23 Richard Biener <rguenther@suse.de>
1965 PR tree-optimization/82129
1967 2017-08-01 Richard Biener <rguenther@suse.de>
1969 PR tree-optimization/81181
1970 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1971 (compute_antic): ... end of iteration here.
1973 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1975 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1976 -std=iso9899:2017 and -std=gnu17.
1977 * doc/standards.texi (C Language): Document C17 support.
1978 * doc/cpp.texi (Overview): Mention -std=c17.
1979 (Standard Predefined Macros): Document C11 and C17 values of
1980 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1981 * doc/extend.texi (Inline): Do not list individual options for
1982 standards newer than C99.
1983 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1985 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1988 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1990 * asan.c (asan_finish_file): Align asan globals array by shadow
1993 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1996 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1997 (struct store_immediate_info): Add bitregion_start and bitregion_end
1999 (store_immediate_info::store_immediate_info): Add brs and bre
2000 arguments and initialize bitregion_{start,end} from those.
2001 (struct merged_store_group): Add bitregion_start, bitregion_end,
2002 align_base and mask fields. Drop unnecessary struct keyword from
2003 struct store_immediate_info. Add do_merge method.
2004 (clear_bit_region_be): Use memset instead of loop storing zeros.
2005 (merged_store_group::do_merge): New method.
2006 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2007 stores as long as the surrounding bitregions have no gaps.
2008 (merged_store_group::merge_overlapping): Use do_merge.
2009 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2010 is byte aligned, rather than requiring that start and width are
2011 byte aligned. Drop unnecessary struct keyword from
2012 struct store_immediate_info. Allocate and populate also mask array.
2013 Make start of the arrays relative to bitregion_start rather than
2014 start and size them according to bitregion_{end,start} difference.
2015 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2016 struct store_immediate_info.
2017 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2018 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2019 struct keyword from struct store_immediate_info.
2020 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2021 between stores as long as the surrounding bitregions have no gaps.
2023 (struct split_store): Add orig non-static data member.
2024 (split_store::split_store): Initialize orig to false.
2025 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2026 if there is exactly a single original stmt. Change stmts argument
2027 to pointer from reference, if NULL, don't push anything to it. Add
2028 first argument, use it to optimize skipping over orig stmts that
2029 are known to be before bitpos already. Simplify.
2030 (split_group): Return unsigned int count how many stores are or
2031 would be needed rather than a bool. Add allow_unaligned argument.
2032 Change split_stores argument from reference to pointer, if NULL,
2033 only do a dry run computing how many stores would be produced.
2034 Rewritten algorithm to use both alignment and misalign if
2035 !allow_unaligned and handle bitfield stores with gaps.
2036 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2037 from bitregion_start instead of start. Compute allow_unaligned
2038 here, if true, do 2 split_group dry runs to compute which one
2039 produces fewer stores and prefer aligned if equal. Punt if
2040 new count is bigger or equal than original before emitting any
2041 statements, rather than during that. Remove no longer needed
2042 new_ssa_names tracking. Replace num_stmts with
2043 split_stores.length (). Use 32-bit stack allocated entries
2044 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2045 if possible. Handle bitfields with gaps.
2046 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2047 Compute bitregion_{start,end} for the stores and construct
2048 store_immediate_info with that. Formatting fixes.
2050 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2053 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2054 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2056 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2058 * gimplify.c: Include tm_p.h.
2060 * common.opt (gcoff): Re-add as ignored option.
2061 (gcoff1, gcoff2, gcoff3): Likewise.
2063 * Makefile.in (OBJS): Delete sdbout.o.
2064 (GTFILES): Delete $(srcdir)/sdbout.c.
2065 * debug.h: Delete sdb_debug_hooks.
2066 * final.c: Delete sdbout.h include.
2067 (final_scan_insn): Delete SDB_DEBUG check.
2068 (rest_of_clean_state): Likewise.
2069 * output.h: Delete sdb_begin_function_line.
2072 * toplev.c: Delete sdbout.h include.
2073 (process_options): Delete SDB_DEBUG check.
2074 * tree-core.h (tree_type_common): Delete pointer field of
2076 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2077 TYPE_SYMTAB_POINTER.
2078 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2079 (TYPE_SYMTAB_IS_POINTER): Delete.
2080 (TYPE_SYMTAB_IS_DIE): Renumber.
2081 * xcoffout.c: Refer to former sdbout.c file.
2082 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2084 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2085 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2086 (Debugging Options): Delete -gcoff.
2087 (-gstabs): Delete SDB reference.
2089 (-gcoff@var{level}): Delete.
2090 * doc/passes.texi (Debugging information output): Delete SDB and
2091 sdbout.c references.
2092 * doc/tm.texi: Regenerate.
2093 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2094 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2096 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2097 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2098 (SDB_DEBUGGING_INFO): Delete.
2099 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2100 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2101 * target.def (output_source_filename): Delete COFF reference.
2103 * common.opt (gcoff): Delete.
2104 (gxcoff+): Update Negative chain.
2105 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2107 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2108 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2109 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2111 (expand_function_start): Change sdb reference to past tense.
2112 (expand_function_end): Change sdb reference to past tense.
2113 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2114 * opts.c (debug_type_names): Delete coff entry.
2115 (common_handle_option): Delete OPT_gcoff case.
2116 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2118 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2119 * config/cris/cris.h: Delete SDB reference in comment.
2120 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2121 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2122 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2123 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2125 (ix86_expand_prologue): Likewise.
2126 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2127 * config/ia64/ia64.h: Likewise.
2128 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2129 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2131 * config/mmix/mmix.h: Likewise.
2132 * config/nds32/nds32.c: Likewise.
2133 * config/stormy/storym16.h: Likewise.
2134 * config/visium/visium.h: Likewise.
2135 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2137 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2139 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2140 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2141 FRAME_POINTER_REGNUM point at high end of local var area.
2143 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2145 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2146 Move comment around. Do not reset best_edge for a copiable
2147 destination if the copy would cause a partition change.
2148 (better_edge_p): Remove redundant check.
2150 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2152 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2154 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2156 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2157 for math functions that have _Float<N> and _Float<N>X variants.
2158 (mathfn_built_in_2): Add support for math functions that have
2159 _Float<N> and _Float<N>X variants.
2160 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2161 (expand_builtin_mathfn_ternary): Add support for fma with
2162 _Float<N> and _Float<N>X variants.
2163 (expand_builtin): Likewise.
2164 (fold_builtin_3): Likewise.
2165 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2166 create math function _Float<N> and _Float<N>X variants as external
2168 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2169 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2170 the __builtin_ prefix and if not strict ansi, without the prefix.
2171 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2172 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2173 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2174 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2175 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2176 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2177 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2178 function signatures for fma _Float<N> and _Float<N>X variants.
2179 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2180 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2181 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2182 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2183 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2184 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2185 * gencfn-macros.c (print_case_cfn): Add support for math functions
2186 that have _Float<N> and _Float<N>X variants.
2187 (print_define_operator_list): Likewise.
2188 (fltfn_suffixes): Likewise.
2190 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2191 for math functions that have _Float<N> and _Float<N>X variants.
2192 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2193 and _Float<N>X variants.
2194 (COPYSIGN): Likewise.
2197 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2198 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2200 (integer_valued_read_call_p): Likewise.
2201 * fold-const-call.c (fold_const_call_ss): Likewise.
2202 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2203 _Float<N> and _Float<N>X variants.
2204 (fold_const_call_ssss): Add support for fma _Float<N> and
2205 _Float<N>X variants.
2206 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2207 support for copysign and fma _Float<N> and _Float<N>X variants.
2208 (backprop::process_builtin_call_use): Likewise.
2209 * tree-call-cdce.c (can_test_argument_range); Add support for
2210 sqrt _Float<N> and _Float<N>X variants.
2211 (edom_only_function): Likewise.
2212 (get_no_error_domain): Likewise.
2213 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2214 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2215 copysign _Float<N> and _Float<N>X variants.
2216 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2217 handled by machine independent code.
2218 (FMAF128): Likewise.
2219 * doc/cpp.texi (Common Predefined Macros): Document defining
2220 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2221 fma _Float<N> and _Float<N>X variants.
2223 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2226 * config/i386/i386-modes.def (CCFPU): Remove definition.
2227 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2228 (ix86_cc_modes_compatible): Ditto.
2229 (ix86_expand_carry_flag_compare): Ditto.
2230 (ix86_expand_int_movcc): Ditto.
2231 (ix86_expand_int_addcc): Ditto.
2232 (ix86_reverse_condition): Ditto.
2233 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2234 Return true/false for unordered/ordered fp comparisons.
2235 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2236 (ix86_prepare_fp_compare_args): Update for rename.
2237 (ix86_expand_fp_compare): Update for rename. Generate unordered
2238 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2239 (ix86_expand_sse_compare_and_jump): Ditto.
2240 * config/i386/predicates.md (fcmov_comparison_operator):
2241 Remove CCFPU mode handling.
2242 (ix86_comparison_operator): Ditto.
2243 (ix86_carry_flag_operator): Ditto.
2244 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2245 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2246 (*cmpu<mode>_cc_i387): Ditto.
2247 (FPCMP): Remove mode iterator.
2248 (unord): Remove mode attribute.
2249 (unord_subst): New define_subst transformation
2250 (unord): New define_subst attribute.
2252 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2253 (*cmpi<unord>xf_i387): Ditto.
2254 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2255 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2256 using unord_subst transformation.
2257 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2258 (round_saeonly): Also handle CCFP mode.
2259 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2260 Remove UNSPEC_SAHF unspec handling.
2262 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2264 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2266 2017-10-27 Jeff Law <law@redhat.com>
2268 * gimple-ssa-sprintf.c: Include domwalk.h.
2269 (class sprintf_dom_walker): New class, derived from dom_walker.
2270 (sprintf_dom_walker::before_dom_children): New function.
2271 (struct call_info): Moved into sprintf_dom_walker class
2272 (compute_formath_length, handle_gimple_call): Likewise.
2273 (sprintf_length::execute): Call the dominator walker rather
2274 than walking the statements.
2276 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2277 gimple statement locations.
2278 (check_array_bounds): Corresponding changes. Get the statement's
2279 location directly from wi->stmt.
2281 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2284 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2286 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2288 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2289 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2291 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2294 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2295 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2296 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2297 maybe_get_pool_constant.
2298 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2301 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2303 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2304 2.26 caveat. Update gas and gld versions.
2305 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2308 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2310 * cgraph.h (set_malloc_flag): Declare.
2311 * cgraph.c (set_malloc_flag_1): New function.
2312 (set_malloc_flag): Likewise.
2313 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2314 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2316 (read_ipa_call_summary): Add support for reading is_return_callee.
2317 (write_ipa_call_summary): Stream is_return_callee.
2318 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2319 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2320 ipa-prop.h, ipa-fnsummary.h.
2321 (pure_const_names): Change to static.
2322 (malloc_state_e): Define.
2323 (malloc_state_names): Define.
2324 (funct_state_d): Add field malloc_state.
2325 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2326 (check_retval_uses): New function.
2327 (malloc_candidate_p): Likewise.
2328 (analyze_function): Add support for malloc attribute.
2329 (pure_const_write_summary): Stream malloc_state.
2330 (pure_const_read_summary): Add support for reading malloc_state.
2331 (dump_malloc_lattice): New function.
2332 (propagate_malloc): New function.
2333 (warn_function_malloc): New function.
2334 (ipa_pure_const::execute): Call propagate_malloc and
2335 ipa_free_fn_summary.
2336 (pass_local_pure_const::execute): Add support for malloc attribute.
2337 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2338 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2340 2017-10-27 Martin Liska <mliska@suse.cz>
2342 PR gcov-profile/82457
2343 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2344 for fork-like functions to be properly instrumented.
2346 2017-10-27 Richard Biener <rguenther@suse.de>
2349 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2350 info when we redirected EH.
2352 2017-10-26 Michael Collison <michael.collison@arm.com>
2354 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2356 (<optab>_trunchf<GPI:mode>2: New pattern.
2357 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2358 * config/aarch64/iterators.md (wv): New mode attribute.
2359 (vf, VF): New mode attributes.
2360 (vgp, VGP): New mode attributes.
2361 (s): Update attribute with SImode and DImode prefixes.
2363 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2365 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2366 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2367 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2368 (nios2_option_overide): Initialize it. Don't allow R0-relative
2369 addressing with PIC.
2370 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2371 (nios2_symbolic_constant_p): Likewise.
2372 (nios2_legitimate_address_p): Likewise.
2373 (nios2_r0rel_section_name_p): New.
2374 (nios2_symbol_ref_in_r0rel_data_p): New.
2375 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2376 (r0rel_constant_p): New.
2377 (nios2_print_operand_address): Handle r0rel_constant_p.
2378 (nios2_cdx_narrow_form_p): Likewise.
2379 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2380 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2381 (Nios II Options): Document -mr0rel-sec.
2383 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2385 * config/nios2/nios2.c: Include xregex.h.
2386 (nios2_gprel_sec_regex): New.
2387 (nios2_option_overide): Initialize it. Don't allow GP-relative
2388 addressing with PIC.
2389 (nios2_small_section_name_p): Check for regex match.
2390 * config/nios2/nios2.opt (mgprel-sec=): New option.
2391 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2392 (Nios II Options): Document -mgprel-sec.
2394 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2396 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2398 2017-10-26 Tom de Vries <tom@codesourcery.com>
2400 PR tree-optimization/82707
2401 * gimple.c (gimple_copy): Fix unsharing of
2402 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2404 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2406 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2407 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2408 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2409 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2410 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2411 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2412 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2413 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2414 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2415 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2416 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2417 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2418 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2419 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2420 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2421 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2422 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2424 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2426 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2428 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2429 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2430 warning to rs6000.c. Remove the Undocumented flag, since it has
2432 (-mabi=ibmlongdouble): Likewise.
2433 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2434 already set, set the default format for long double.
2435 (rs6000_debug_reg_global): Print whether long double is IBM or
2437 (rs6000_option_override_internal): Rework setting long double
2438 format. Only warn if the user is changing the long double default
2439 and they did not use -Wno-psabi.
2440 * doc/invoke.texi (PowerPC options): Update the documentation for
2441 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2443 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2444 Alan Hayward <alan.hayward@arm.com>
2445 David Sherwood <david.sherwood@arm.com>
2447 * rtl.h (wider_subreg_mode): New function.
2448 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2449 rather than an unsigned int *.
2450 * ira-color.c (regno_max_ref_width): Replace with...
2451 (regno_max_ref_mode): ...this new variable.
2452 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2453 Use wider_subreg_mode.
2454 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2455 rather than an unsigned int *.
2456 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2457 (process_alt_operands): Likewise.
2458 (invariant_p): Likewise.
2459 * lra-spills.c (assign_mem_slot): Likewise.
2460 (add_pseudo_to_slot): Likewise.
2461 * lra.c (collect_non_operand_hard_regs): Likewise.
2462 (add_regs_to_insn_regno_info): Likewise.
2463 * reload1.c (regno_max_ref_width): Replace with...
2464 (regno_max_ref_mode): ...this new variable.
2465 (reload): Update accordingly. Update call to
2466 ira_sort_regnos_for_alter_reg.
2467 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2468 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2469 (scan_paradoxical_subregs): Likewise.
2471 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2473 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2474 (aarch64_frame): Add emit_frame_chain boolean.
2475 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2476 Move eh_return case to aarch64_layout_frame.
2477 (aarch64_layout_frame): Initialize emit_frame_chain.
2478 (aarch64_expand_prologue): Use emit_frame_chain.
2480 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2482 * config/aarch64/aarch64.c (aarch64_layout_frame):
2483 Ensure LR is always stored at the bottom of the callee-saves.
2484 Remove rarely used frame layout which saves callee-saves at top of
2485 frame, so the store of LR can be used as a valid probe in all cases.
2487 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2489 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2490 Improve unaligned TImode/TFmode base/offset split.
2492 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2493 Alan Hayward <alan.hayward@arm.com>
2494 David Sherwood <david.sherwood@arm.com>
2496 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2497 * combine.c (find_single_use_1): Likewise.
2498 (expand_field_assignment): Likewise.
2499 (move_deaths): Likewise.
2500 * lra-constraints.c (simplify_operand_subreg): Likewise.
2501 (curr_insn_transform): Likewise.
2502 * lra.c (collect_non_operand_hard_regs): Likewise.
2503 (add_regs_to_insn_regno_info): Likewise.
2504 * rtlanal.c (reg_referenced_p): Likewise.
2505 (covers_regno_no_parallel_p): Likewise.
2507 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2509 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2510 Don't print any bits outside the precision of the value.
2511 * wide-int.cc (test_printing): Add some new tests.
2513 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2515 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2516 supports -xbrace_comment option.
2517 * configure: Regenerate.
2518 * config.in: Regenerate.
2519 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2520 (ASM_CPU_SPEC): Use it.
2522 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2524 * target.def (static_rtx_alignment): New hook.
2525 * targhooks.h (default_static_rtx_alignment): Declare.
2526 * targhooks.c (default_static_rtx_alignment): New function.
2527 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2528 * doc/tm.texi: Regenerate.
2529 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2530 instead of targetm.constant_alignment. Remove call to
2532 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2533 (cris_preferred_mininum_alignment): New function, split out from...
2534 (cris_constant_alignment): ...here.
2535 (cris_static_rtx_alignment): New function.
2536 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2538 (ix86_constant_alignment): ...here.
2539 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2540 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2541 (mmix_static_rtx_alignment): New function.
2542 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2543 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2545 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2548 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2549 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2551 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2553 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2554 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2555 mark as a sign-extending load.
2556 (local_pic_load_u): Define.
2558 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2561 * fold-const.c (operand_equal_for_comparison_p): Also return true
2562 if ARG0 is a simple variant of ARG1 with narrower precision.
2563 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2565 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2567 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2569 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2570 scatter_static, scatter_per_elt.
2571 * x86-tune-costs.h: Add new cost entries.
2573 2017-10-25 Richard Biener <rguenther@suse.de>
2575 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2576 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2577 class pass_fre): Move to ...
2578 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2579 class pass_fre): ... here and adjust for statistics.
2581 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2584 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2585 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2586 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2587 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2590 2017-10-25 Richard Biener <rguenther@suse.de>
2592 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2593 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2594 eliminate_push_avail, eliminate_insert): Move inside...
2595 (class eliminate_dom_walker): ... this class in preparation
2597 (fini_eliminate): Remove by merging with ...
2598 (eliminate): ... this function. Adjust for class changes.
2599 (pass_pre::execute): Remove fini_eliminate call.
2600 (pass_fre::execute): Likewise.
2602 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2605 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2606 (VPERMI2, VPERMI2I): New mode iterators.
2607 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2608 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2610 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2611 mode iterator. Remove 3 old define_insn patterns.
2612 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2613 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2614 VPERMI2 mode iterator, remove the other two expanders.
2615 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2616 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2617 instructions, remove the other two patterns.
2618 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2619 mode iterator, remove the other two patterns.
2620 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2621 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2622 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2623 and adjust argument order accordingly.
2624 (ix86_expand_vec_perm): Adjust caller.
2625 (expand_vec_perm_1): Likewise.
2626 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2627 (expand_vec_perm_vpermt2_vpshub2): ... this.
2628 (ix86_expand_vec_perm_const_1): Adjust caller.
2629 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2632 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2633 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2634 (vec_shl_<mode>): Remove unused expander.
2635 (avx512bw_<shift_insn><mode>3): New define_insn.
2636 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2637 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2639 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2642 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2645 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2647 PR rtl-optimization/82396
2648 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2649 (autopref_multipass_init): Simplify initialization.
2650 (autopref_rank_data): Simplify sort order.
2651 * gcc/sched-int.h (autopref_multipass_data_): Remove
2652 multi_mem_insn_p, min_offset and max_offset.
2654 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2657 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2658 Check special value of flag_omit_frame_pointer.
2659 (aarch64_can_eliminate): Likewise.
2660 (aarch64_override_options_after_change_1): Simplify handling of
2661 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2663 2017-10-24 Richard Biener <rguenther@suse.de>
2665 PR tree-optimization/82697
2666 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2667 zero for conditional load and unconditional store.
2669 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2671 * doc/install.texi: Document bootstrap-cet.
2673 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2676 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2677 ENDBR instruction at function entrance if function is only
2680 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2683 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2684 patterns to better describe from which operation the CF is computed.
2685 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2686 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2687 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2688 is 0, use _0 suffixed expanders instead of emitting a comparison
2691 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2693 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2694 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2696 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2699 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2700 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2701 * loop-iv.c (iv_get_reaching_def): Likewise.
2702 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2703 variable is promoted and the partition contains undefined values.
2705 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2707 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2709 (nios2_address_cost): Define.
2710 (nios2_legitimize_address): Recognize (exp + constant) directly.
2711 (TARGET_ADDRESS_COST): Define.
2713 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2715 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2716 (nios2_symbolic_memory_operand_p): Declare.
2717 (nios2_split_large_constant): Declare.
2718 (nios2_split_symbolic_memory_operand): Declare.
2719 * config/nios2/nios2.c: Adjust includes.
2720 (nios2_symbolic_constant_allowed): New.
2721 (nios2_symbolic_constant_p): New.
2722 (nios2_plus_symbolic_constant_p): New.
2723 (nios2_valid_addr_expr_p): Recognize addresses involving
2725 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2726 (nios2_symbolic_memory_operand_p): New.
2727 (nios2_large_constant_p): New.
2728 (nios2_split_large_constant): New.
2729 (nios2_split_plus_large_constant): New.
2730 (nios2_split_symbolic_memory_operand): New.
2731 (nios2_legitimize_address): Code refactoring. Handle addresses
2732 involving symbolic constants.
2733 (nios2_emit_move_sequence): Likewise.
2734 (nios2_print_operand): Improve error output.
2735 (nios2_print_operand_address): Handle LO_SUM.
2736 (nios2_cdx_narrow_form_p): Likewise.
2737 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2738 operands involving symbolic constants.
2739 (movhi_internal, movsi_internal): Likewise.
2740 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2741 (extendhisi2, extendqi<mode>2): Likewise.
2743 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2745 * tree-pass.h (PROP_rtl_split_insns): Define.
2746 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2748 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2750 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2752 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2755 * target.def (const_not_ok_for_debug_p): Default to
2756 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2757 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2758 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2759 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2760 which targetm.const_not_ok_for_debug_p returned true.
2761 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2763 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2765 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2766 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2767 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2769 (ix86_const_not_ok_for_debug_p): New function.
2770 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2771 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2773 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2776 * system.h: Conditionally include "unique-ptr.h" if
2777 INCLUDE_UNIQUE_PTR is defined.
2778 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2779 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2781 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2783 * config/rl78/rl78.md: New define_expand "subdi3".
2785 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2788 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2789 DF_REF_INSN if DF_REF_INSN_INFO is false.
2791 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2793 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2794 xmm_move instead of sse_move.
2795 (sse_store_index): New function.
2796 (ix86_register_move_cost): Be more sensible about mismatch stall;
2797 model AVX moves correctly; make difference between sse->integer and
2799 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2800 moves; make difference between SSE and AVX.
2801 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2802 and zmm_move. Increase size of sse load and store tables;
2803 add unaligned load and store tables; add ssemmx_to_integer.
2804 * x86-tune-costs.h: Update all entries according to real
2805 move latencies from Agner Fog's manual and chip documentation.
2807 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2810 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2811 * config/i386/constraints.md (Wf): New constraint.
2812 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2813 (cmp<dwi>_doubleword): Removed.
2814 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2815 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2816 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2817 expand with cmp<dwi>_doubleword. For LTU and GEU use
2818 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2820 * common.opt (gcolumn-info): Enable by default.
2821 * doc/invoke.texi (gcolumn-info): Document new default.
2823 2017-10-23 Richard Biener <rguenther@suse.de>
2825 PR tree-optimization/82672
2826 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2827 Fold the stmt if we propagated into it.
2829 2017-10-23 Richard Biener <rguenther@suse.de>
2831 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2832 (bitmap_remove_expr_from_set): ... this. All callers call this
2833 for non-constant values.
2834 (bitmap_set_subtract): Rename to...
2835 (bitmap_set_subtract_expressions): ... this. Adjust and
2837 (bitmap_set_contains_value): Remove superfluous check.
2838 (bitmap_set_replace_value): Inline into single caller ...
2839 (bitmap_value_replace_in_set): ... here and simplify.
2840 (dependent_clean): Merge into ...
2841 (clean): ... this using an overload. Adjust.
2842 (prune_clobbered_mems): Adjust.
2843 (compute_antic_aux): Likewise.
2844 (compute_partial_antic_aux): Likewise.
2846 2017-10-23 Richard Biener <rguenther@suse.de>
2848 PR tree-optimization/82129
2850 2017-08-01 Richard Biener <rguenther@suse.de>
2852 PR tree-optimization/81181
2853 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2854 (compute_antic): ... end of iteration here.
2856 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2858 * target.def (starting_frame_offset): New hook.
2859 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2860 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2861 * doc/tm.texi.in: Regenerate.
2862 * hooks.h (hook_hwi_void_0): Declare.
2863 * hooks.c (hook_hwi_void_0): New function.
2864 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2865 STARTING_FRAME_OFFSET.
2866 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2867 * reload1.c (reload): Likewise.
2868 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2869 instead of STARTING_FRAME_OFFSET.
2870 * function.c (try_fit_stack_local): Likewise.
2871 (assign_stack_local_1): Likewise
2872 (instantiate_virtual_regs): Likewise.
2873 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2874 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2875 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2876 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2877 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2878 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2879 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2880 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2881 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2882 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2883 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2884 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2885 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2886 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2887 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2888 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2889 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2890 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2891 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2892 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2893 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2894 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2895 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2896 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2897 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2898 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2899 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2900 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2901 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2902 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2903 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2904 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2905 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2906 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2907 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2908 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2909 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2910 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2911 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2912 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2913 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2914 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2915 return a HOST_WIDE_INT.
2916 (avr_builtin_setjmp_frame_value): Use it instead of
2917 STARTING_FRAME_OFFSET.
2918 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2919 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2920 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2922 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2923 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2924 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2925 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2926 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2927 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2928 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2929 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2930 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2931 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2932 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2933 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2935 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2936 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2937 * config/mips/mips.c (mips_compute_frame_info): Refer to
2938 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2939 (mips_starting_frame_offset): New function.
2940 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2941 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2942 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2943 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2944 and return a HOST_WIDE_INT.
2945 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2946 (mmix_initial_elimination_offset): Refer to
2947 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2948 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2949 * config/pa/pa.c (pa_starting_frame_offset): New function.
2950 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2951 (pa_expand_prologue): Likewise.
2952 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2953 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2954 !FRAME_GROWS_DOWNWARD handling to...
2955 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2956 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2957 !FRAME_GROWS_DOWNWARD handling to...
2958 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2959 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2960 !FRAME_GROWS_DOWNWARD handling to...
2961 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2962 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2964 (rs6000_starting_frame_offset): New function.
2965 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2966 !FRAME_GROWS_DOWNWARD handling to...
2967 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2968 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2969 !FRAME_GROWS_DOWNWARD handling to...
2970 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2971 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2972 !FRAME_GROWS_DOWNWARD handling to...
2973 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2974 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2975 (rs6000_starting_frame_offset): New function.
2976 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2977 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2978 * config/vax/vax.c (vax_starting_frame_offset): New function.
2979 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2980 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2981 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2982 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2983 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2984 * system.h (STARTING_FRAME_OFFSET): Poison.
2986 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2988 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2989 SCALAR_TYPE_MODE instead of TYPE_MODE.
2991 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2992 Alan Hayward <alan.hayward@arm.com>
2993 David Sherwood <david.sherwood@arm.com>
2995 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2997 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2998 Alan Hayward <alan.hayward@arm.com>
2999 David Sherwood <david.sherwood@arm.com>
3001 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3003 2017-10-23 Richard Biener <rguenther@suse.de>
3005 PR tree-optimization/82129
3006 * tree-ssa-pre.c (bitmap_set_and): Remove.
3007 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3008 canonicalizing expressions in the set to those with lowest
3009 ID rather than taking that from the first edge.
3011 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3013 * combine.c (rtx_equal_for_field_assignment_p): Use
3014 byte_lowpart_offset.
3016 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3017 Alan Hayward <alan.hayward@arm.com>
3018 David Sherwood <david.sherwood@arm.com>
3020 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3021 to a SUBREG_PROMOTED_VAR.
3023 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3024 Alan Hayward <alan.hayward@arm.com>
3025 David Sherwood <david.sherwood@arm.com>
3027 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3028 (expand_debug_source_expr): Likewise.
3029 * combine.c (combine_simplify_rtx): Likewise.
3030 * cse.c (fold_rtx): Likewise.
3031 * optabs.c (expand_float): Likewise.
3032 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3033 (simplify_binary_operation_1): Likewise.
3035 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3036 Alan Hayward <alan.hayward@arm.com>
3037 David Sherwood <david.sherwood@arm.com>
3039 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3040 (record_promoted_value): Likewise.
3041 * expr.c (expand_expr_real_2): Likewise.
3042 * ree.c (update_reg_equal_equiv_notes): Likewise.
3043 (combine_set_extension): Likewise.
3044 * rtlanal.c (low_bitmask_len): Likewise.
3045 * simplify-rtx.c (neg_const_int): Likewise.
3046 (simplify_binary_operation_1): Likewise.
3048 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3049 Alan Hayward <alan.hayward@arm.com>
3050 David Sherwood <david.sherwood@arm.com>
3052 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3053 * regcprop.c (maybe_mode_change): Likewise.
3054 * reload1.c (alter_reg): Likewise.
3056 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3058 * inchash.h (inchash::hash::add_wide_int): New function.
3059 * lto-streamer-out.c (hash_tree): Use it.
3061 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3063 * inchash.h (inchash::hash::add_wide_int): Rename to...
3064 (inchash::hash::add_hwi): ...this.
3065 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3066 (polymorphic_call_target_hasher::hash): Likewise.
3067 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3068 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3069 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3070 * lto-streamer-out.c (hash_tree): Likewise.
3071 * optc-save-gen.awk: Likewise.
3072 * tree.c (add_expr): Likewise.
3074 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3077 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3078 for ordered inequality comparisons even with TARGET_IEEE_FP.
3080 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3083 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3084 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3085 Expand with cmp<dwi>_doubleword.
3087 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3089 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3090 List CET intrinsics.
3091 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3092 specific to -fcf-protection option.
3094 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3096 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3097 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3098 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3099 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3100 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3101 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3102 (extra_objs): Add cet.o for Linux/x86 targets.
3103 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3104 * config/i386/cet.c: New file.
3105 * config/i386/cetintrin.h: Likewise.
3106 * config/i386/t-cet: Likewise.
3107 * config/i386/cpuid.h (bit_SHSTK): New.
3108 (bit_IBT): Likewise.
3109 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3110 pass IBT and SHSTK bits.
3111 * config/i386/i386-builtin-types.def
3112 (VOID_FTYPE_UNSIGNED_PVOID): New.
3113 (VOID_FTYPE_UINT64_PVOID): Likewise.
3114 * config/i386/i386-builtin.def: Add CET intrinsics.
3115 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3116 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3117 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3118 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3120 * config/i386/i386.c (rest_of_insert_endbranch): New.
3121 (pass_data_insert_endbranch): Likewise.
3122 (pass_insert_endbranch): Likewise.
3123 (make_pass_insert_endbranch): Likewise.
3124 (ix86_notrack_prefixed_insn_p): Likewise.
3125 (ix86_target_string): Add -mibt, -mshstk flags.
3126 (ix86_option_override_internal): Add flag_cf_protection
3128 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3129 (ix86_print_operand): Add 'notrack' prefix output.
3130 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3131 (ix86_expand_builtin): Expand CET intrinsics.
3132 (x86_output_mi_thunk): Add 'endbranch' instruction.
3133 * config/i386/i386.h (TARGET_IBT): New.
3134 (TARGET_IBT_P): Likewise.
3135 (TARGET_SHSTK): Likewise.
3136 (TARGET_SHSTK_P): Likewise.
3137 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3138 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3139 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3140 (builtin_setjmp_setup): New pattern.
3141 (builtin_longjmp): Likewise.
3142 (rdssp<mode>): Likewise.
3143 (incssp<mode>): Likewise.
3144 (saveprevssp): Likewise.
3145 (rstorssp): Likewise.
3146 (wrss<mode>): Likewise.
3147 (wruss<mode>): Likewise.
3148 (setssbsy): Likewise.
3149 (clrssbsy): Likewise.
3150 (nop_endbr): Likewise.
3151 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3153 * config/i386/immintrin.h: Include <cetintrin.h>.
3154 * config/i386/linux-common.h
3155 (file_end_indicate_exec_stack_and_cet): New prototype.
3156 (TARGET_ASM_FILE_END): New.
3158 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3160 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3161 latencies instead of having separate table; make difference between
3162 integer and float costs.
3163 * i386.h (processor_costs): Remove scalar_stmt_cost,
3164 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3165 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3167 * x86-tune-costs.h: Remove entries which has been removed in
3168 procesor_costs from all tables; make cond_taken_branch_cost
3169 and cond_not_taken_branch_cost COST_N_INSNS based.
3171 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3173 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3175 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3177 * config/i386/i386.md (isa): Remove fma_avx512f.
3178 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3179 <avx512>_fmadd_<mode>_mask3<round_name>,
3180 <avx512>_fmsub_<mode>_mask<round_name>,
3181 <avx512>_fmsub_<mode>_mask3<round_name>,
3182 <avx512>_fnmadd_<mode>_mask<round_name>,
3183 <avx512>_fnmadd_<mode>_mask3<round_name>,
3184 <avx512>_fnmsub_<mode>_mask<round_name>,
3185 <avx512>_fnmsub_<mode>_mask3<round_name>,
3186 <avx512>_fmaddsub_<mode>_mask<round_name>,
3187 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3188 <avx512>_fmsubadd_<mode>_mask<round_name>,
3189 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3190 (*vec_widen_umult_even_v16si<mask_name>,
3191 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3192 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3194 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3196 * extend.texi: Add 'nocf_check' documentation.
3197 * gimple.texi: Add second parameter to
3198 gimple_build_call_from_tree.
3199 * invoke.texi: Add -fcf-protection documentation.
3200 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3202 2017-10-20 Richard Biener <rguenther@suse.de>
3204 PR tree-optimization/82473
3205 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3206 the largest input type.
3208 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3210 * c-attribs.c (handle_nocf_check_attribute): New function.
3211 (c_common_attribute_table): Add 'nocf_check' handling.
3212 * gimple-parser.c: Add second argument NULL to
3213 gimple_build_call_from_tree.
3214 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3215 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3217 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3218 * common.opt: Add fcf-protection flag.
3219 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3220 * flag-types.h: Add enum cf_protection_level.
3221 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3222 Add 'nocf_check' attribute propagation to gimple call.
3223 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3224 (gimple_build_call_from_tree): Update prototype.
3225 (gimple_call_nocf_check_p): New function.
3226 (gimple_call_set_nocf_check): Likewise.
3227 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3228 * ipa-icf.c: Add nocf_check attribute in statement hash.
3229 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3230 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3231 * toplev.c (process_options): Add flag_cf_protection handling.
3233 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3235 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3237 2017-10-20 Richard Biener <rguenther@suse.de>
3239 PR tree-optimization/82603
3240 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3241 remove false predicated stores.
3243 2017-10-20 Richard Biener <rguenther@suse.de>
3245 * graphite-isl-ast-to-gimple.c
3246 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3247 Remove return value and simplify, dump copied stmt after lhs
3249 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3250 Reduce dump verbosity.
3251 (gsi_insert_earliest): Likewise.
3252 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3253 * graphite.c (print_global_statistics): Adjust dumping.
3254 (print_graphite_scop_statistics): Likewise.
3255 (print_graphite_statistics): Do not dump loops here.
3256 (graphite_transform_loops): But here.
3258 2017-10-20 Nicolas Roche <roche@adacore.com>
3260 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3261 * configure: Regenerate.
3263 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3266 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3267 functions when optimizing replace GIMPLE_RETURN stmts with
3268 calls to __builtin_unreachable ().
3271 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3272 for -fsanitize=thread link of executables.
3273 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3274 link of executables.
3277 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3279 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3280 define_insns for logical vector shifts to use VI248_AVX512BW
3281 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3282 condition, useless isa and prefix attributes. Change the first
3283 2 of these define_insns to ...
3284 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3285 define_insn for avx512vl.
3286 (<shift_insn><mode>3): ... and this, new define_insn without
3287 masking for non-avx512vl.
3290 * config/i386/sse.md (*andnot<mode>3,
3291 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3292 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3293 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3294 not applied use empty suffix even for TARGET_AVX512VL.
3295 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3296 is applied, supply evex,evex or evex,evex,evex instead of just
3299 2017-10-20 Julia Koval <julia.koval@intel.com>
3301 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3302 (OPTION_MASK_ISA_GFNI_UNSET): New.
3303 (ix86_handle_option): Handle OPT_mgfni.
3304 * config/i386/cpuid.h (bit_GFNI): New.
3305 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3306 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3307 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3308 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3309 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3310 * config/i386/i386.opt: Add mgfni.
3312 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3314 * config/msp430/msp430.c (msp430_option_override): Disable
3315 -fdelete-null-pointer-checks.
3316 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3318 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3320 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3321 of x87 and SSE instructions.
3323 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3325 * asan.c (create_cond_insert_point): Do not update edge count.
3326 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3327 (afdo_propagate_circuit): Likewise.
3328 (afdo_calculate_branch_prob): Likewise.
3329 (afdo_annotate_cfg): Likewise.
3330 * basic-block.h (struct edge_def): Remove count.
3331 (edge_def::count): New accessor.
3332 * bb-reorder.c (rotate_loop): Update.
3333 (find_traces_1_round): Update.
3334 (connect_traces): Update.
3335 (sanitize_hot_paths): Update.
3336 * cfg.c (unchecked_make_edge): Update.
3337 (make_single_succ_edge): Update.
3338 (check_bb_profile): Update.
3339 (dump_edge_info): Update.
3340 (update_bb_profile_for_threading): Update.
3341 (scale_bbs_frequencies_int): Update.
3342 (scale_bbs_frequencies_gcov_type): Update.
3343 (scale_bbs_frequencies_profile_count): Update.
3344 (scale_bbs_frequencies): Update.
3345 * cfganal.c (connect_infinite_loops_to_exit): Update.
3346 * cfgbuild.c (compute_outgoing_frequencies): Update.
3347 (find_many_sub_basic_blocks): Update.
3348 * cfgcleanup.c (try_forward_edges): Update.
3349 (try_crossjump_to_edge): Update
3350 * cfgexpand.c (expand_gimple_cond): Update
3351 (expand_gimple_tailcall): Update
3352 (construct_exit_block): Update
3353 * cfghooks.c (verify_flow_info): Update
3354 (redirect_edge_succ_nodup): Update
3355 (split_edge): Update
3356 (make_forwarder_block): Update
3357 (duplicate_block): Update
3358 (account_profile_record): Update
3359 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3360 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3361 * cfgloopmanip.c (scale_loop_profile): Update.
3363 (lv_adjust_loop_entry_edge): Update.
3364 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3365 (force_nonfallthru_and_redirect): Update.
3366 (purge_dead_edges): Update.
3367 (rtl_flow_call_edges_add): Update.
3368 * cgraphunit.c (init_lowered_empty_function): Update.
3369 (cgraph_node::expand_thunk): Update.
3370 * gimple-pretty-print.c (dump_probability): Update.
3371 (dump_edge_probability): Update.
3372 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3373 * haifa-sched.c (sched_create_recovery_edges): Update.
3374 * hsa-gen.c (convert_switch_statements): Update.
3375 * ifcvt.c (dead_or_predicable): Update.
3376 * ipa-inline-transform.c (inline_transform): Update.
3377 * ipa-split.c (split_function): Update.
3378 * ipa-utils.c (ipa_merge_profiles): Update.
3379 * loop-doloop.c (add_test): Update.
3380 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3381 * lto-streamer-in.c (input_cfg): Update.
3382 (input_function): Update.
3383 * lto-streamer-out.c (output_cfg): Update.
3384 * modulo-sched.c (sms_schedule): Update.
3385 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3386 * predict.c (maybe_hot_edge_p): Update.
3387 (unlikely_executed_edge_p): Update.
3388 (probably_never_executed_edge_p): Update.
3389 (dump_prediction): Update.
3390 (drop_profile): Update.
3391 (propagate_unlikely_bbs_forward): Update.
3392 (determine_unlikely_bbs): Update.
3393 (force_edge_cold): Update.
3394 * profile.c (compute_branch_probabilities): Update.
3395 * reg-stack.c (better_edge): Update.
3396 * shrink-wrap.c (handle_simple_exit): Update.
3397 * tracer.c (better_p): Update.
3398 * trans-mem.c (expand_transaction): Update.
3399 (split_bb_make_tm_edge): Update.
3400 * tree-call-cdce.c: Update.
3401 * tree-cfg.c (gimple_find_sub_bbs): Update.
3402 (gimple_split_edge): Update.
3403 (gimple_duplicate_sese_region): Update.
3404 (gimple_duplicate_sese_tail): Update.
3405 (gimple_flow_call_edges_add): Update.
3406 (insert_cond_bb): Update.
3407 (execute_fixup_cfg): Update.
3408 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3409 * tree-complex.c (expand_complex_div_wide): Update.
3410 * tree-eh.c (lower_resx): Update.
3411 (unsplit_eh): Update.
3412 (cleanup_empty_eh_move_lp): Update.
3413 * tree-inline.c (copy_edges_for_bb): Update.
3414 (freqs_to_counts): Update.
3415 (copy_cfg_body): Update.
3416 * tree-ssa-dce.c (remove_dead_stmt): Update.
3417 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3418 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3419 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3420 (unloop_loops): Update.
3421 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3422 * tree-ssa-loop-split.c (connect_loops): Update.
3423 (split_loop): Update.
3424 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3425 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3426 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3427 * tree-ssa-reassoc.c (branch_fixup): Update.
3428 * tree-ssa-tail-merge.c (replace_block_by): Update.
3429 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3430 (compute_path_counts): Update.
3431 (update_profile): Update.
3432 (recompute_probabilities): Update.
3433 (update_joiner_offpath_counts): Update.
3434 (estimated_freqs_path): Update.
3435 (freqs_to_counts_path): Update.
3436 (clear_counts_path): Update.
3437 (ssa_fix_duplicate_block_edges): Update.
3438 (duplicate_thread_path): Update.
3439 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3440 (case_bit_test_cmp): Update.
3441 (collect_switch_conv_info): Update.
3442 (gen_inbound_check): Update.
3443 (do_jump_if_equal): Update.
3444 (emit_cmp_and_jump_insns): Update.
3445 * tree-tailcall.c (decrease_profile): Update.
3446 (eliminate_tail_call): Update.
3447 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3448 (vect_do_peeling): Update.
3449 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3450 * ubsan.c (ubsan_expand_null_ifn): Update.
3451 (ubsan_expand_ptr_ifn): Update.
3452 * value-prof.c (gimple_divmod_fixed_value): Update.
3453 (gimple_mod_pow2): Update.
3454 (gimple_mod_subtract): Update.
3455 (gimple_ic): Update.
3456 (gimple_stringop_fixed_value): Update.
3458 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3461 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3463 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3465 PR rtl-optimization/82395
3466 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3467 based on non_spilled_static_chain_regno_p.
3469 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3471 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3472 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3473 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3475 2017-10-19 Martin Sebor <msebor@redhat.com>
3477 PR tree-optimization/82596
3478 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3480 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3482 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3483 (get_mem_refs_of_builtin_call): Likewise.
3484 * builtins.c (expand_builtin_apply): Adjust call to
3485 allocate_dynamic_stack_space.
3486 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3487 the third argument to allocate_dynamic_stack_space, otherwise -1.
3488 (expand_builtin): Deal with all alloca variants.
3489 (is_inexpensive_builtin): Likewise.
3490 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3491 * calls.c (special_function_p): Deal with all alloca variants.
3492 (initialize_argument_information): Adjust call to
3493 allocate_dynamic_stack_space.
3494 (expand_call): Likewise.
3495 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3496 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3497 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3498 use it for the stack usage computation.
3499 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3500 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3501 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3502 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3503 (in_loop_p): Remove first argument and useless check.
3504 (pass_walloca::execute): Remove useless test and adjust call to above.
3505 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3506 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3507 (gimplify_call_expr): Deal with all alloca variants.
3508 * hsa-gen.c (gen_hsa_alloca): Likewise.
3509 (gen_hsa_insns_for_call): Likewise.
3510 * ipa-pure-const.c (special_builtin_state): Likewise.
3511 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3512 * tree-object-size.c (alloc_object_size): Likewise.
3513 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3514 (call_may_clobber_ref_p_1): Likewise.
3515 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3516 (ccp_fold_stmt): Likewise.
3517 (optimize_stack_restore): Likewise.
3518 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3519 (mark_all_reaching_defs_necessary_1): Likewise.
3520 (propagate_necessity): Likewise.
3521 (eliminate_unnecessary_stmts): Likewise.
3522 * tree.c (build_common_builtin_nodes): Build
3523 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3524 (build_alloca_call_expr): New function.
3525 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3526 (CASE_BUILT_IN_ALLOCA): Likewise.
3527 (build_alloca_call_expr): Declare.
3528 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3530 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3533 * dwarf2out.c (new_die_raw): New static inline function.
3534 (new_die): Use it to create the DIE.
3535 (add_AT_external_die_ref): Likewise.
3536 (clone_die): Likewise.
3537 (clone_as_declaration): Likewise.
3538 (dwarf2out_vms_debug_main_pointer): Likewise.
3539 (base_type_die): Likewise. Remove early return for corner cases.
3540 Do not call add_pubtype on the DIE here.
3541 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3542 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3543 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3544 native order exists for base types, attach the DIE manually and call
3545 add_pubtype on it. Do not equate a reverse order DIE to the type.
3547 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3549 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3550 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3552 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3554 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3556 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3557 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3558 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3559 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3560 * opts.c (sanitizer_opts): Add builtin.
3561 * ubsan.c (instrument_builtin): New function.
3562 (pass_ubsan::execute): Call it.
3563 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3564 * doc/invoke.texi: Document -fsanitize=builtin.
3566 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3567 builtins, store max (log2 (align), 0) into uchar field instead of
3568 align into uptr field.
3569 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3570 store uchar 0 field instead of uptr 0 field.
3571 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3572 instead of passing one address of struct with 2 locations pass
3573 two addresses of structs with 1 location each.
3574 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3575 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3576 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3577 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3578 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3579 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3580 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3581 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3583 2017-10-19 Martin Liska <mliska@suse.cz>
3586 * file-find.c (remove_prefix): Remove.
3587 * file-find.h (remove_prefix): Likewise.
3588 * gcc-ar.c: Remove smartness of lookup.
3590 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3592 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3593 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3594 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3596 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3599 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3600 (setcc + and to xor + setcc): New peephole2.
3602 2017-10-19 Tom de Vries <tom@codesourcery.com>
3604 * doc/sourcebuild.texi (Test Directives, Variants of
3605 dg-require-support): Add dg-require-stack-size.
3607 2017-10-19 Martin Liska <mliska@suse.cz>
3610 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3611 that have a large alignment.
3612 (gimplify_target_expr): Likewise.
3614 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3616 PR rtl-optimization/82602
3617 * ira.c (rtx_moveable_p): Return false for volatile asm.
3619 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3622 * config/i386/i386-modes.def (CCGZ): New CC mode.
3623 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3624 * config/i386/predicates.md (ix86_comparison_operator):
3626 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3627 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3628 with double-word subtraction.
3629 (put_condition_code): Handle CCGZmode.
3631 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3633 * wide-int.cc (debug (const wide_int &)): New.
3634 (debug (const wide_int *)): New.
3635 (debug (const widest_int &)): New.
3636 (debug (const widest_int *)): New.
3638 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3641 * lra-constraints.c (curr_insn_transform): Use non-input operand
3642 instead of output one for matched reload.
3644 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3646 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3647 (tree-ssa-loop-ivopts.h): New header file.
3648 (struct builtin_info): New fields.
3649 (classify_builtin_1): Compute and record base and offset parts for
3650 memset builtin partition by calling strip_offset.
3651 (offset_cmp, fuse_memset_builtins): New functions.
3652 (finalize_partitions): Fuse adjacent memset partitions by calling
3654 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3655 Expose the interface.
3656 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3658 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3660 PR tree-optimization/82574
3661 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3662 that data reference must be executed exactly once per iteration
3663 against the outermost loop in nest.
3664 (classify_partition): Update call to above function.
3666 2017-10-18 Richard Biener <rguenther@suse.de>
3668 PR tree-optimization/82591
3669 * graphite.c (graphite_transform_loops): Move code gen message
3671 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3672 Here. Handle scop_to_isl_ast failing.
3673 (scop_to_isl_ast): Limit the number of ISL operations.
3675 2017-10-18 Richard Biener <rguenther@suse.de>
3677 * graphite-isl-ast-to-gimple.c
3678 (translate_isl_ast_to_gimple::set_rename): Simplify.
3679 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3680 (graphite_copy_stmts_from_block): ... here.
3681 (copy_bb_and_scalar_dependences): Simplify.
3682 (add_parameters_to_ivs_params): Canonicalize.
3683 (generate_entry_out_of_ssa_copies): Simplify.
3684 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3685 by passing in ISL dimension.
3686 (parameter_index_in_region_1): Rename to ...
3687 (parameter_index_in_region): ... this.
3688 (extract_affine): Adjust assert, pass down parameter index.
3689 (add_param_constraints): Use range-info when available.
3690 (build_scop_context): Adjust.
3691 * sese.c (new_sese_info): Adjust.
3692 (free_sese_info): Likewise.
3693 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3694 Remove unused typedefs.
3695 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3697 2017-10-18 Martin Liska <mliska@suse.cz>
3699 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3701 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3703 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3704 (s390_sched_init): Do not reset s390_sched_state if we entered the
3705 current basic block via a fallthru edge and all others are unlikely.
3707 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3709 * config/s390/s390.c (NUM_SIDES): New variable.
3710 (LONGRUNNING_THRESHOLD): New variable.
3711 (LATENCY_FACTOR): New variable.
3712 (s390_sched_score): Decrease score for long-running instructions on
3714 (s390_sched_variable_issue): Perform bookkeeping for long-running
3717 2017-10-18 Richard Biener <rguenther@suse.de>
3719 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3720 Simplify with removal of the parameter rename map.
3721 (set_rename): Likewise.
3722 (should_copy_to_new_region): Likewise.
3723 (graphite_copy_stmts_from_block): Likewise.
3724 (copy_bb_and_scalar_dependences): Remove initialization of
3725 unused copied_bb_map.
3727 (copy_internal_parameters): Likewise.
3728 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3729 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3730 Use INTEGRAL_TYPE_P.
3731 (parameter_index_in_region_1): Rename to ...
3732 (assign_parameter_index_in_region): ... this. Assert we have
3733 a parameter we handle.
3734 (scan_tree_for_params): Adjust.
3735 * sese.h (parameter_rename_map_t): Remove.
3736 (struct sese_info_t): Remove unused parameter_rename_map and
3737 copied_bb_map members.
3738 * sese.c (new_sese_info): Adjust.
3739 (free_sese_info): Likewise.
3741 2017-10-18 Martin Liska <mliska@suse.cz>
3744 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3745 on an abnormal edge.
3747 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3749 * doc/invoke.texi (ffunction-sections and fdata-sections):
3752 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3754 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3755 the use statement can throw internally.
3757 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3759 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3760 any RTX present on the RHS of a SET.
3761 * compare-elim.c (try_eliminate_compare): Restore comment.
3763 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3765 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3766 may be also called on tcc_type nodes.
3767 * langhooks.c (lhd_tree_size): Likewise.
3769 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3771 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3772 format_warning_at_substring.
3773 (maybe_warn): Convert source_range * param to a location_t. Pass
3774 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3775 (format_directive): Remove code to extract source_ranges and
3776 source_range * in favor of just a location_t.
3777 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3779 * substring-locations.c (format_warning_va): Convert
3780 source_range * param to a location_t.
3781 (format_warning_at_substring): Likewise.
3782 * substring-locations.h (format_warning_va): Likewise.
3783 (format_warning_at_substring): Likewise.
3785 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3787 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3789 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3790 and scatter/gather ops.
3792 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3793 vec_gather_load and vec_scatter_store.
3794 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3795 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3796 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3797 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3798 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3799 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3801 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3803 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3804 Detect FTST instruction and handle its register pops. Only pop
3805 second operand if can_pop_second_op is true.
3806 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3807 set can_pop_second_op to false in the compare_for_stack_reg call.
3809 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3810 output_fp_compare for stack register operands.
3811 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3812 instructions here. Do not emit stack register pops here. Assert
3813 that FCOMPP pops next to top stack register. Rewrite function.
3815 2017-10-17 Nathan Sidwell <nathan@acm.org>
3818 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3819 use DECL_ASSEMBLER_NAME_RAW.
3822 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3825 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3826 Wilco Dijkstra <wilco.dijkstra@arm.com>
3828 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3829 converstion to Pmode from the buf_addr.
3831 2017-10-17 Richard Biener <rguenther@suse.de>
3833 * graphite-dependences.c (scop_get_reads_and_writes): Change
3834 output parameters to references.
3836 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3838 PR 71026/tree-optimization
3839 * fold-const.c (distribute_real_division): Removed.
3840 (fold_binary_loc): Remove calls to distribute_real_divison.
3842 2017-10-17 Richard Biener <rguenther@suse.de>
3844 * graphite-scop-detection.c
3845 (scop_detection::stmt_has_simple_data_refs_p): Always use
3846 the full nest as region.
3847 (try_generate_gimple_bb): Likewise.
3848 * sese.c (scalar_evolution_in_region): Simplify now that
3849 SCEV can handle instantiation in regions.
3850 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3851 in the non-loop part of a function if requested.
3853 2017-10-17 Richard Biener <rguenther@suse.de>
3855 PR tree-optimization/82563
3856 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3858 (graphite_regenerate_ast_isl): Call it.
3859 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3861 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3863 PR tree-optimization/82549
3864 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3865 Formatting fixes. Instead of calling make_bit_field_ref with negative
3868 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3870 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3871 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3872 _mm_maskz_reduce_ss): New.
3873 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3874 __builtin_ia32_reducess_mask): Ditto..
3875 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3876 * config/i386/sse.md (reduces<mode>): Renamed to ...
3877 (reduces<mode><mask_scalar_name>): ... this.
3878 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3880 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3881 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3883 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3885 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3886 * selftest-run-tests.c (selftest::run_tests): Call
3887 selftest::unique_ptr_tests_cc_tests.
3888 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3889 * unique-ptr-tests.cc: New file.
3891 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3894 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3896 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3897 (make_hard_regno_born, make_hard_regno_dead): Update
3898 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3900 2017-10-16 Jeff Law <law@redhat.com>
3902 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3904 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3906 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3908 2017-10-16 Olivier Hainque <hainque@adacore.com>
3910 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3911 with_cpu if we were configured for an e500v2 target cpu name.
3913 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3915 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3916 * doc/invoke.texi: Document +nodsp as a valid extension for
3919 2017-10-16 Martin Liska <mliska@suse.cz>
3921 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3922 (test_set_range): Likewise.
3923 (test_range_functions): Rename to ...
3924 (test_bit_in_range): ... this.
3925 (sbitmap_c_tests): Add new test.
3927 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3929 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3931 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3932 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3934 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3936 * config/aarch64/aarch64-builtins.c
3937 (aarch64_types_quadopu_lane_qualifiers): New.
3938 (TYPES_QUADOPU_LANE): New.
3939 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3940 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3941 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3942 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3943 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3944 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3945 (Vdottype, DOTPROD): New.
3946 (sur): Add SDOT and UDOT.
3948 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3950 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3951 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3952 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3954 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3955 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3956 Enable TARGET_DOTPROD.
3957 (cortex-a75.cortex-a55): Likewise.
3958 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3960 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3962 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3963 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3965 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3967 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3968 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3969 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3970 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3971 * config/arm/types.md (neon_dot, neon_dot_q): New.
3972 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3974 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3976 * config/arm/arm.h (TARGET_DOTPROD): New.
3977 * config/arm/arm.c (arm_arch_dotprod): New.
3978 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3979 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3980 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3981 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3982 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3983 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3984 * doc/invoke.texi (armv8.2-a): Document dotprod
3986 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3988 * i386.c (ix86_vec_cost): New function.
3989 (ix86_rtx_costs): Handle vector operations better.
3990 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3991 * x86-tune-costs.h: Add new costs to all tables.
3993 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3995 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3997 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3998 divsd, sqrtss and sqrtsd
3999 * x86-tune-costs.h: Add new entries to all costs.
4000 (znver1_cost): Fix to match real instruction latencies.
4002 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4003 Michael Collison <michael.collison@arm.com>
4005 * compare-elim.c: Include emit-rtl.h.
4006 (can_merge_compare_into_arith): New function.
4007 (try_validate_parallel): Likewise.
4008 (try_merge_compare): Likewise.
4009 (try_eliminate_compare): Call the above when no previous clobber
4011 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4014 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4018 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4019 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4023 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4024 to be any operand_equal_p operands. For & (B - 1) require
4025 B to be power of 2. Recognize
4026 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4028 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4031 * optabs.c (expand_memory_blockage): Fix call of
4032 targetm.have_memory_blockage.
4034 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4037 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4038 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4039 objects to extra_objs instead of overwriting it.
4041 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4043 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4044 Use any_fp_register_operand as operand[3] predicate. Simplify
4045 equality test for operands[2] and operands[4] memory location.
4046 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4047 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4048 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4049 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4050 any_fp_register_operand as operand[1] predicate. Simplify
4051 equality test for operands[0] and operands[3] memory location.
4052 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4053 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4054 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4056 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4058 * target-insns.def: Add memory_blockage.
4059 * optabs.c (expand_memory_blockage): New function.
4060 (expand_asm_memory_barrier): Rename ...
4061 (expand_asm_memory_blockage): ... to this.
4062 (expand_mem_thread_fence): Call expand_memory_blockage
4063 instead of expand_asm_memory_barrier.
4064 (expand_mem_singnal_fence): Ditto.
4065 (expand_atomic_load): Ditto.
4066 (expand_atomic_store): Ditto.
4067 * doc/md.texi (Standard Pattern Names For Generation):
4068 Document memory_blockage instruction pattern.
4070 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4072 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4073 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4074 * config/rl78/rl78.md: New define_expand "adddi3".
4076 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4078 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4081 2017-10-13 Jeff Law <law@redhat.com>
4083 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4085 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4088 * internal-fn.c (expand_mul_overflow): If both operands have
4089 the same highpart of -1 or 0 and the topmost bit of lowpart
4090 is different, overflow is if res <= 0 rather than res < 0.
4092 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4094 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4095 TARGET_P9_VECTOR code for unaligned_load case.
4097 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4099 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4101 2017-10-13 Nathan Sidwell <nathan@acm.org>
4103 * tree-core.h (tree_contains_struct): Make bool.
4104 * tree.c (tree_contains_struct): Likewise.
4105 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4106 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4107 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4108 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4109 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4111 2017-10-13 Richard Biener <rguenther@suse.de>
4113 * graphite-isl-ast-to-gimple.c
4114 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4115 parameters and dominance check.
4116 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4117 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4118 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4119 Do not update SSA form here or do intermediate IL verification.
4120 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4121 (graphite_initialize): Remove check on the number of loops in
4122 the function and inline into graphite_transform_loops.
4123 (graphite_finalize): Inline into graphite_transform_loops.
4124 (graphite_transform_loops): Perform SSA update and IL verification
4126 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4128 2017-10-13 Richard Biener <rguenther@suse.de>
4130 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4131 graphite_expression_type_precision): Avoid global constructor
4133 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4134 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4135 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4136 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4137 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4138 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4140 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4143 * config/i386/i386.h (ix86_red_zone_size): New.
4144 * config/i386/i386.md (push peephole2s): Replace
4145 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4147 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4148 Alan Hayward <alan.hayward@arm.com>
4149 David Sherwood <david.sherwood@arm.com>
4151 * combine.c (can_change_dest_mode): Reject changes in
4152 REGMODE_NATURAL_SIZE.
4154 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4155 Alan Hayward <alan.hayward@arm.com>
4156 David Sherwood <david.sherwood@arm.com>
4158 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4159 (expand_debug_source_expr): Likewise.
4160 * combine.c (combine_simplify_rtx): Likewise.
4161 * cse.c (fold_rtx): Likewise.
4162 * fwprop.c (canonicalize_address): Likewise.
4163 * targhooks.c (default_shift_truncation_mask): Likewise.
4165 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4166 Alan Hayward <alan.hayward@arm.com>
4167 David Sherwood <david.sherwood@arm.com>
4169 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4170 (widened_mode): Likewise.
4171 (expand_unop): Likewise.
4172 * ree.c (transform_ifelse): Likewise.
4173 (merge_def_and_ext): Likewise.
4174 (combine_reaching_defs): Likewise.
4175 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4177 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4178 Alan Hayward <alan.hayward@arm.com>
4179 David Sherwood <david.sherwood@arm.com>
4181 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4182 * combine.c (gen_lowpart_for_combine): Likewise.
4183 * dwarf2out.c (rtl_for_decl_location): Likewise.
4184 * final.c (alter_subreg): Likewise.
4185 * rtlhooks.c (gen_lowpart_general): Likewise.
4186 (gen_lowpart_if_possible): Likewise.
4188 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4189 Alan Hayward <alan.hayward@arm.com>
4190 David Sherwood <david.sherwood@arm.com>
4192 * calls.c (expand_call): Use subreg_lowpart_offset.
4193 * cse.c (cse_insn): Likewise.
4194 * regcprop.c (copy_value): Likewise.
4195 (copyprop_hardreg_forward_1): Likewise.
4197 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4200 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4201 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4202 =Q constraints to +Q and into insn condition add check
4203 that operands[0] and operands[1] are equal.
4204 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4205 =Q constraints to +Q and into insn condition add check
4206 that operands[0] is equal to either operands[1] or operands[2].
4209 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4210 instead of handling MINUS_EXPR twice (once for each argument),
4211 canonicalize operand order and handle just once, use rtype where
4212 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4215 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4216 any values of __C while still being pattern recognizable as a simple
4219 2017-10-13 Richard Biener <rguenther@suse.de>
4221 PR tree-optimization/82451
4223 2017-10-02 Richard Biener <rguenther@suse.de>
4225 PR tree-optimization/82355
4226 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4227 a mapping for the enclosing loop but avoid generating one for
4229 (copy_bb_and_scalar_dependences): Remove premature codegen
4230 error on PHIs in blocks duplicated into multiple places.
4231 * graphite-scop-detection.c
4232 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4233 in the region use it as loop and nest to analyze the DR in.
4234 (try_generate_gimple_bb): Likewise.
4235 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4236 (add_loop_constraints): For blocks in a loop not in the region
4237 create a dimension with a single iteration.
4238 * sese.h (gbb_loop_at_index): Remove assert.
4240 * cfgloop.c (loop_preheader_edge): For the loop tree root
4241 return the single successor of the entry block.
4242 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4243 Reset the SCEV hashtable and niters.
4244 * graphite-scop-detection.c
4245 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4246 assert that we only have POLYNOMIAL_CHREC that vary in loops
4247 contained in the region.
4248 (scop_detection::graphite_can_represent_expr): Adjust.
4249 (scop_detection::stmt_has_simple_data_refs_p): For loops
4250 not in the region set loop to NULL. The nest is now the
4251 entry edge to the region.
4252 (try_generate_gimple_bb): Likewise.
4253 * sese.c (scalar_evolution_in_region): Adjust for
4254 instantiate_scev change.
4255 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4256 Make nest parameter the edge into the region.
4257 (create_data_ref): Likewise.
4258 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4259 entry edge into a region and adjust instantiate_scev calls.
4260 (create_data_ref): Likewise.
4261 (graphite_find_data_references_in_stmt): Likewise.
4262 (find_data_references_in_stmt): Pass the loop preheader edge
4263 from the nest argument.
4264 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4265 parameter the edge into the region.
4266 (instantiate_parameters): Use the loop preheader edge as entry.
4267 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4269 (get_instantiated_value_entry): Make instantiate_below parameter
4270 the edge into the region.
4271 (instantiate_scev_name): Likewise. Adjust dominance checks,
4272 when we cannot use loop-based instantiation instantiate by
4273 walking use-def chains.
4274 (instantiate_scev_poly): Adjust.
4275 (instantiate_scev_binary): Likewise.
4276 (instantiate_scev_convert): Likewise.
4277 (instantiate_scev_not): Likewise.
4278 (instantiate_array_ref): Remove.
4279 (instantiate_scev_3): Likewise.
4280 (instantiate_scev_2): Likewise.
4281 (instantiate_scev_1): Likewise.
4282 (instantiate_scev_r): Do not blindly handle N-operand trees.
4283 Do not instantiate array-refs. Handle all constants and invariants.
4284 (instantiate_scev): Make instantiate_below parameter
4285 the edge into the region.
4286 (resolve_mixers): Use the loop preheader edge for the region
4287 parameter to instantiate_scev_r.
4288 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4290 2017-10-13 Richard Biener <rguenther@suse.de>
4292 PR tree-optimization/82525
4293 * graphite-isl-ast-to-gimple.c
4294 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4296 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4297 Fail code generation when we cannot represent the isl integer.
4298 (binary_op_to_tree): Elide modulo operations that are no-ops
4299 in the type we code generate. Remove now superfluous code
4302 2017-10-13 Richard Biener <rguenther@suse.de>
4304 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4305 (scop_detection::harmful_loop_in_region): Remove premature
4306 IV type restriction.
4307 (scop_detection::graphite_can_represent_scev): We can handle
4308 pointer IVs just fine.
4310 2017-10-13 Alan Modra <amodra@gmail.com>
4312 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4313 "Clobbers and Scratch Registers". Add paragraph on
4314 alternative to clobbers for scratch registers and OpenBLAS
4317 2017-10-13 Alan Modra <amodra@gmail.com>
4319 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4320 example of a memory input for a string of known length. Move
4321 commentary out of table. Add a number of new examples
4322 covering array memory inputs.
4324 2017-10-12 Martin Liska <mliska@suse.cz>
4326 PR tree-optimization/82493
4327 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4328 (test_range_functions): New function.
4329 (sbitmap_c_tests): Likewise.
4330 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4331 * selftest.h (sbitmap_c_tests): New function.
4333 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4335 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4337 * config/rs6000/amo.h: Fix spacing issue.
4339 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4342 * config/i386/i386.md (*ashl<mode>3_mask_1,
4343 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4344 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4347 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4349 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4350 (profile_probability): Set max_probability
4351 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4353 * profile-count.c (profile_probability::differs_from_p): Do not
4354 rely on max_probaiblity == 10000
4356 2017-10-12 Jeff Law <law@redhat.com>
4358 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4361 2017-10-12 Martin Sebor <msebor@redhat.com>
4365 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4366 (handle_alias_pairs): Call it.
4367 * common.opt (-Wattribute-alias): New option.
4368 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4369 * doc/invoke.texi (-Wattribute-alias): Document.
4371 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4374 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4376 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4378 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4379 (make_hard_regno_born, make_hard_regno_dead): Update
4380 bb_killed_pseudos and bb_gen_pseudos.
4382 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4384 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4386 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4388 * config/alpha/alpha.c (alpha_split_conditional_move):
4389 Use std::swap instead of manually swapping.
4390 (alpha_stdarg_optimize_hook): Ditto.
4391 (alpha_canonicalize_comparison): Ditto.
4393 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4395 * tree-loop-distribution.c (struct builtin_info): New struct.
4396 (struct partition): Refactor fields into struct builtin_info.
4397 (partition_free): Free struct builtin_info.
4398 (build_size_arg_loc, build_addr_arg_loc): Delete.
4399 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4400 information from struct builtin_info.
4401 (find_single_drs): New function refactored from classify_partition.
4402 Also moved builtin validity checks to this function.
4403 (compute_access_range, alloc_builtin): New functions.
4404 (classify_builtin_st, classify_builtin_ldst): New functions.
4405 (classify_partition): Refactor code into functions find_single_drs,
4406 classify_builtin_st and classify_builtin_ldst.
4407 (distribute_loop): Don't do runtime alias check when distributing
4409 (find_seed_stmts_for_distribution): New function.
4410 (pass_loop_distribution::execute): Refactor code finding seed
4411 stmts into above function. Support distribution for the innermost
4412 two-level loop nest. Adjust dump information.
4414 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4416 * tree-loop-distribution.c: Adjust the general comment.
4417 (NUM_PARTITION_THRESHOLD): New macro.
4418 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4419 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4420 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4421 in call to build_partition_graph.
4422 (finalize_partitions): New parameter. Make loop distribution more
4423 conservative by fusing more partitions.
4424 (distribute_loop): Don't do runtime alias check in case of loop nest
4426 (find_seed_stmts_for_distribution): New function.
4427 (prepare_perfect_loop_nest): New function.
4428 (pass_loop_distribution::execute): Refactor code finding seed stmts
4429 and loop nest into above functions. Support loop nest distribution.
4430 Adjust dump information accordingly.
4432 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4434 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4435 and set PTYPE_SEQUENTIAL for merged partition.
4437 2017-10-12 Richard Biener <rguenther@suse.de>
4439 PR tree-optimization/69728
4441 2017-09-19 Richard Biener <rguenther@suse.de>
4443 PR tree-optimization/69728
4444 * graphite-sese-to-poly.c (schedule_error): New global.
4445 (add_loop_schedule): Handle empty domain by failing the
4447 (build_original_schedule): Handle schedule_error.
4449 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4450 domain by returning an unchanged schedule.
4452 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4454 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4455 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4457 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4459 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4462 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4465 * expr.c (store_field): Don't optimize away bitsize == 0 store
4466 from CALL_EXPR with addressable return type.
4468 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4470 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4471 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4472 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4473 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4474 TARGET_ISEL instead of TARGET_ISEL<sel>.
4476 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4478 * config/rs6000/rs6000.c
4479 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4481 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4483 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4485 (reg_or_cint_operand): Fix comment.
4486 (reg_or_zero_operand): New predicate.
4487 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4488 * config/rs6000/rs6000.c (output_isel): Delete.
4489 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4490 instead of reg_or_cint_operand. Output instruction directly (not via
4492 (isel_unsigned_<mode>): Ditto.
4493 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4494 gpc_reg_operand. Add an instruction alternative for this. Output
4495 instruction directly.
4496 (*isel_reversed_unsigned_<mode>): Ditto.
4498 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4500 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4501 (TARGET_CANONICALIZE_COMPARISON): Define.
4503 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4506 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4507 Check whether the dest is REG before adding REG_EQUIV note.
4509 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4512 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4514 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4515 (make_hard_regno_born, make_hard_regno_dead): Update
4516 bb_killed_pseudos and bb_gen_pseudos.
4518 2017-10-11 Nathan Sidwell <nathan@acm.org>
4520 * incpath.h (enum incpath_kind): Name enum, prefix values.
4521 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4522 * incpath.c (heads, tails): Use INC_MAX.
4523 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4524 (merge_include_chains, split_quote_chain,
4525 register_include_chains): Update incpath_kind names.
4526 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4527 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4529 (add_framework_path, darwin_register_objc_includes): Likewise.
4530 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4532 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4534 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4535 Do not use float_operator operator predicate.
4536 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4537 * config/i386/predicates.md (float_operator): Remove predicate.
4539 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4541 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4542 (*jccxf_i387): Ditto.
4543 (*jcc<mode>_i387): Ditto.
4544 (*jccu<mode>_i387): Ditto.
4545 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4546 (*jcc_*_i387 splitters): Remove.
4547 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4548 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4549 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4552 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4554 * profile-count.h (slow_safe_scale_64bit): New function.
4555 (safe_scale_64bit): New inline.
4556 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4557 * profile-count.c: Include wide-int.h
4558 (slow_safe_scale_64bit): New.
4560 2017-10-11 Nathan Sidwell <nathan@acm.org>
4562 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4563 HAS_DECL_ASSEMBLER_NAME_P.
4564 * gimple-expr.c (gimple_decl_printable_name: Check
4565 HAS_DECL_ASSEMBLER_NAME_P too.
4566 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4567 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4568 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4569 * passes.c (rest_of_decl_compilation): Check
4570 HAS_DECL_ASSEMBLER_NAME_P too.
4571 * recog.c (verify_changes): Likewise.
4572 * tree-pretty-print.c (dump_decl_name): Likewise.
4573 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4575 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4576 (SET_DECL_ASSEMBLER_NAME): Use it.
4577 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4578 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4579 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4581 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4583 * config.gcc (i386, x86_64): Add extra objects.
4584 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4585 (ix86_min_insn_size): Declare.
4586 (ix86_issue_rate): Declare.
4587 (ix86_adjust_cost): Declare.
4588 (ia32_multipass_dfa_lookahead): Declare.
4589 (ix86_macro_fusion_p): Declare.
4590 (ix86_macro_fusion_pair_p): Declare.
4591 (ix86_bd_has_dispatch): Declare.
4592 (ix86_bd_do_dispatch): Declare.
4593 (ix86_core2i7_init_hooks): Declare.
4594 (ix86_atom_sched_reorder): Declare.
4595 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4596 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4597 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4598 (rip_relative_addr_p): Rename to ...
4599 (ix86_rip_relative_addr_p): ... this one; export.
4600 (memory_address_length): Update.
4601 (ix86_issue_rate): Move to x86-tune-sched.c.
4602 (ix86_flags_dependent): Move to x86-tune-sched.c.
4603 (ix86_agi_dependent): Move to x86-tune-sched.c.
4604 (exact_dependency_1): Move to x86-tune-sched.c.
4605 (exact_store_load_dependency): Move to x86-tune-sched.c.
4606 (ix86_adjust_cost): Move to x86-tune-sched.c.
4607 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4608 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4609 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4610 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4611 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4612 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4613 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4614 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4615 (min_insn_size): Rename to ...
4616 (ix86_min_insn_size): ... this one; export.
4617 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4618 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4619 (core2i7_first_cycle_multipass_backtrack): Move to
4620 x86-tune-sched-core.c.
4621 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4622 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4623 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4624 (ix86_avoid_jump_mispredicts): Update.
4625 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4626 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4627 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4628 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4629 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4630 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4631 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4632 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4633 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4634 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4635 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4636 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4637 (BIG): Move to ix86-tune-sched-bd.c.
4638 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4639 (enum insn_path): Move to ix86-tune-sched-bd.c.
4640 (get_mem_group): Move to ix86-tune-sched-bd.c.
4641 (is_cmp): Move to ix86-tune-sched-bd.c.
4642 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4643 (is_branch): Move to ix86-tune-sched-bd.c.
4644 (is_prefetch): Move to ix86-tune-sched-bd.c.
4645 (init_window): Move to ix86-tune-sched-bd.c.
4646 (allocate_window): Move to ix86-tune-sched-bd.c.
4647 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4648 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4649 (process_end_window): Move to ix86-tune-sched-bd.c.
4650 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4651 (find_constant): Move to ix86-tune-sched-bd.c.
4652 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4653 (has_immediate): Move to ix86-tune-sched-bd.c.
4654 (get_insn_path): Move to ix86-tune-sched-bd.c.
4655 (get_insn_group): Move to ix86-tune-sched-bd.c.
4656 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4657 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4658 (add_insn_window): Move to ix86-tune-sched-bd.c.
4659 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4660 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4661 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4662 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4663 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4664 (do_dispatch): Move to ix86-tune-sched-bd.c.
4665 (has_dispatch): Move to ix86-tune-sched-bd.c.
4666 * config/i386/t-i386: Add new object files.
4667 * config/i386/x86-tune-costs.h: New file.
4668 * config/i386/x86-tune-sched-atom.c: New file.
4669 * config/i386/x86-tune-sched-bd.c: New file.
4670 * config/i386/x86-tune-sched-core.c: New file.
4671 * config/i386/x86-tune-sched.c: New file.
4673 2017-10-11 Liu Hao <lh_mouse@126.com>
4675 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4676 the generic version below instead.
4677 (should_colorize): Recognize Windows consoles as terminals
4679 * pretty-print.c [__MINGW32__] (write_all): New function.
4680 [__MINGW32__] (find_esc_head): Likewise.
4681 [__MINGW32__] (find_esc_terminator): Likewise.
4682 [__MINGW32__] (eat_esc_sequence): Likewise.
4683 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4685 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4688 2017-10-11 Richard Biener <rguenther@suse.de>
4690 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4691 Properly call analyze_scalar_evolution with the loop of the stmt.
4693 2017-10-11 Richard Biener <rguenther@suse.de>
4695 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4696 * tree-core.h (tree_base): Add chrec_var union member.
4697 * tree.h (CHREC_VAR): Remove.
4698 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4699 * tree-chrec.h (build_polynomial_chrec): Adjust.
4700 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4701 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4703 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4705 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4706 * match.pd: ... here.
4707 ((T) X == (T) Y): Relax condition.
4709 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4711 PR tree-optimization/82472
4712 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4714 (break_alias_scc_partitions): Update postorder number.
4716 2017-10-11 Martin Liska <mliska@suse.cz>
4719 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4721 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4723 2017-10-11 Martin Liska <mliska@suse.cz>
4725 * print-rtl.c (print_insn): Move declaration of idbuf
4726 to same scope as name.
4728 2017-10-11 Martin Liska <mliska@suse.cz>
4733 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4736 2017-10-11 Martin Liska <mliska@suse.cz>
4739 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4742 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4744 * config.gcc (default_gnu_indirect_function): Default to yes for
4745 arm*-*-linux* with glibc.
4747 2017-10-11 Richard Biener <rguenther@suse.de>
4749 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4750 default-defs and types we do not want to analyze.
4751 (interpret_loop_phi): Replace unreachable code with an assert.
4752 (compute_scalar_evolution_in_loop): Remove and inline ...
4753 (analyze_scalar_evolution_1): ... here, replacing condition with
4754 what makes the intent clearer. Remove handling of cases
4755 get_scalar_evolution now handles.
4757 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4759 PR rtl-optimization/81434
4760 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4761 comment for main loop. In sched_group_found if, also add checks for
4762 pass and min_cost_group.
4764 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4766 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4767 (rs6000_insn_cost): New function.
4768 * config/rs6000/rs6000.md (cost): New attribute.
4770 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4771 H.J. Lu <hongjiu.lu@intel.com>
4775 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4776 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4777 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4778 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4779 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4780 OPTION_MASK_ISA_MMX and some other bit set require both
4781 mmx and the ISAs without the mmx bit.
4782 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4783 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4784 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4785 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4786 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4787 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4788 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4789 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4790 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4791 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4792 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4793 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4794 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4795 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4796 Add OPTION_MASK_ISA_MMX.
4798 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4800 * config.gcc (armv7*-*-freebsd*): New target.
4801 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4803 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4805 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4806 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4809 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4811 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4812 a template parameter.
4813 (WIDE_INT_REF_FOR): Update accordingly.
4814 * tree.h (wi::int_traits <const_tree>): Delete.
4815 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4816 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4817 (wi::tree_to_wide_ref): New typedef.
4818 (wi::to_wide): New function.
4819 * calls.c (get_size_range): Use wi::to_wide when operating on
4821 * cgraph.c (cgraph_node::create_thunk): Likewise.
4822 * config/i386/i386.c (ix86_data_alignment): Likewise.
4823 (ix86_local_alignment): Likewise.
4824 * dbxout.c (stabstr_O): Likewise.
4825 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4826 * expr.c (const_vector_from_tree): Likewise.
4827 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4828 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4829 (fold_negate_expr_1, int_const_binop_1, const_binop)
4830 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4831 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4832 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4833 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4834 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4835 (fold_not_const, round_up_loc): Likewise.
4836 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4837 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4838 (alloca_call_type): Likewise.
4839 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4840 * godump.c (go_output_typedef): Likewise.
4841 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4842 * internal-fn.c (get_min_precision): Likewise.
4843 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4844 * ipa-polymorphic-call.c
4845 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4846 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4847 (ipa_modify_call_arguments): Likewise.
4848 * match.pd: Likewise.
4849 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4850 * print-tree.c (print_node_brief, print_node): Likewise.
4851 * stmt.c (expand_case): Likewise.
4852 * stor-layout.c (layout_type): Likewise.
4853 * tree-affine.c (tree_to_aff_combination): Likewise.
4854 * tree-cfg.c (group_case_labels_stmt): Likewise.
4855 * tree-data-ref.c (dr_analyze_indices): Likewise.
4856 (prune_runtime_alias_test_list): Likewise.
4857 * tree-dump.c (dequeue_and_dump): Likewise.
4858 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4859 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4860 * tree-pretty-print.c (dump_generic_node): Likewise.
4861 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4862 (simple_iv_with_niters): Likewise.
4863 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4864 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4865 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4866 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4867 (refine_value_range_using_guard, number_of_iterations_ne_max)
4868 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4869 (get_cst_init_from_scev, record_nonwrapping_iv)
4870 (scev_var_range_cant_overflow): Likewise.
4871 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4872 * tree-ssa-pre.c (compute_avail): Likewise.
4873 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4874 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4875 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4876 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4877 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4878 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4879 (dump_case_nodes, try_switch_expansion): Likewise.
4880 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4881 (vect_do_peeling): Likewise.
4882 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4883 * tree-vect-stmts.c (vectorizable_load): Likewise.
4884 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4885 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4886 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4887 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4888 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4889 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4890 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4891 (evrp_dom_walker::before_dom_children): Likewise.
4892 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4893 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4894 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4895 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4896 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4897 (get_range_pos_neg): Likewise.
4898 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4899 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4900 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4901 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4902 * config/avr/avr.c (avr_fold_builtin): Likewise.
4903 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4904 * config/msp430/msp430.c (msp430_attr): Likewise.
4905 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4906 * config/powerpcspe/powerpcspe-c.c
4907 (altivec_resolve_overloaded_builtin): Likewise.
4908 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4909 (rs6000_expand_ternop_builtin): Likewise.
4910 * config/rs6000/rs6000-c.c
4911 (altivec_resolve_overloaded_builtin): Likewise.
4912 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4913 (rs6000_expand_ternop_builtin): Likewise.
4914 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4916 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4918 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4919 when copying loop nest with only one inner loop.
4921 2017-10-10 Richard Biener <rguenther@suse.de>
4923 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4924 blocks if SCEV is active.
4925 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4927 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4928 (scev_initialize): Assert we are not yet initialized.
4930 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4932 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4933 inner loop's exit stmt by making it always exit the loop, otherwise
4934 we would generate an infinite empty loop.
4936 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4938 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4939 renaming variables in new preheader if it's deleted.
4941 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4943 * tree-loop-distribution.c (struct partition): Remove unused field
4944 loops of the structure.
4945 (partition_alloc, partition_free): Ditto.
4946 (build_rdg_partition_for_vertex): Ditto.
4948 2017-10-09 Jeff Law <law@redhat.com>
4950 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4951 return type to match prototype and documentation.
4953 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4955 * config/rs6000/rs6000.c (processor_costs): Move to ...
4956 * config/rs6000/rs6000.h: ... here.
4957 (rs6000_cost): Declare.
4959 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4961 * except.c (setjmp_fn): New global variable.
4962 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4963 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4964 if DONT_USE_BUILTIN_SETJMP is defined.
4966 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4968 * target.def (insn_cost): New hook.
4969 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4970 * doc/tm.texi: Regenerate.
4971 * rtlanal.c (insn_cost): Use the new hook.
4973 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4975 * combine.c (combine_validate_cost): Compute the new insn_cost,
4976 not just pattern_cost.
4977 (try_combine): Adjust comment.
4979 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4981 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4983 * combine.c (uid_insn_cost): Adjust comment.
4984 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4986 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4987 * dse.c (find_shift_sequence): Ditto.
4988 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4989 (bb_valid_for_noce_process_p): Use pattern_cost.
4990 * rtl.h (insn_rtx_cost): Delete.
4991 (pattern_cost): New prototype.
4992 (insn_cost): New prototype.
4993 * rtlanal.c (insn_rtx_cost): Rename to...
4994 (pattern_cost): ... this.
4997 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4999 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5000 (*jcc<mode>_0_r_i387): Ditto.
5001 (*jccxf_r_i387): Ditto.
5002 (*jcc<mode>_r_i387): Ditto.
5003 (*jccu<mode>_r_i387): Ditto.
5004 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5005 (*jcc): Rename from *jcc_1.
5007 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5009 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5010 deferred rescans after the lvx/stvx recombination pre-pass.
5012 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5014 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5015 memory operation instruction support.
5016 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5017 (rs6000-ibm-aix[789]*): Likewise.
5018 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5019 Document new functions.
5021 2017-10-09 Richard Biener <rguenther@suse.de>
5023 PR tree-optimization/82397
5024 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5025 equality only for semantically equal trees.
5027 2017-10-09 Richard Biener <rguenther@suse.de>
5029 PR tree-optimization/82449
5030 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5031 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5032 allow constant addresses.
5033 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5036 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5038 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5041 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5044 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5047 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5050 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5052 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5055 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5056 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5058 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5060 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5061 (WI_BINARY_PREDICATE_RESULT): Likewise.
5062 (wi::binary_traits::operator_result): New type.
5063 (wi::binary_traits::predicate_result): Likewise.
5064 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5065 (generic_wide_int::operator==, generic_wide_int::operator!=)
5066 (generic_wide_int::operator&, generic_wide_int::and_not)
5067 (generic_wide_int::operator|, generic_wide_int::or_not)
5068 (generic_wide_int::operator^, generic_wide_int::operator+
5069 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5071 (operator~, unary operator-, operator==, operator!=, operator&)
5072 (operator|, operator^, operator+, binary operator-, operator*): New
5074 * expr.c (get_inner_reference): Use wi::bit_and_not.
5075 * fold-const.c (fold_binary_loc): Likewise.
5076 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5077 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5078 (bit_value_binop): Likewise.
5079 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5080 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5081 (extract_range_from_binary_expr_1): Likewise.
5082 (masked_increment): Likewise.
5083 (simplify_bit_ops_using_ranges): Likewise.
5085 2017-10-09 Martin Jambor <mjambor@suse.cz>
5088 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5089 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5090 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5092 (hsa_fixup_mov_insn_type): New function.
5093 (hsa_op_with_type::get_in_type): Use it.
5094 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5095 immediates in an assert.
5096 (hsa_op_with_type::extend_int_to_32bit): New method.
5097 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5098 types. Convert to dest type if necessary.
5099 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5100 (reg_for_gimple_ssa): Pass false as min32int to
5101 hsa_type_for_scalar_tree_type.
5102 (gen_hsa_addr): Fixup type when creating addresable temporary.
5103 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5104 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5105 necessary. Call hsa_fixup_mov_insn_type.
5106 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5107 extend operands and convert to dest type if necessary.
5108 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5109 to dest type if necessary.
5110 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5111 if conversion nt necessary and size matches.
5112 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5113 to dest type if necessary.
5114 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5115 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5117 (gen_hsa_clrsb): Likewise.
5118 (gen_hsa_ffs): Likewise.
5119 (gen_hsa_divmod): Extend operands and convert to dest type if
5121 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5123 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5125 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5126 Remove empty default arguments. Use a brace block as output
5128 (conditional return): Ditto.
5130 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5131 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5133 (group_ending_nop): Ditto.
5134 (doloop_end): Ditto.
5135 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5136 (splitters for those): Ditto.
5138 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5140 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5141 a conditional jump (and the compare for it) so that pc_rtx is the
5143 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5144 for the deleted and renamed ctr<mode>_internal[234] patterns.
5145 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5146 Delete second conditional return pattern.
5147 (ctr<mode>_internal2): Delete this second bdnz pattern.
5148 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5149 (ctr<mode>_internal4): Delete this second bdz pattern.
5151 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5153 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5154 (always_initialized_rtx_for_ssa_name_p): New predicate.
5155 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5156 (finish_out_of_ssa): Free new field of SA.
5157 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5158 * tree-ssa-coalesce.c: Include tree-ssa.h.
5159 (get_parm_default_def_partitions): Remove extern keyword.
5160 (get_undefined_value_partitions): New function.
5161 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5162 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5164 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5166 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5168 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5170 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5172 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5173 for targets that preffer 128bit.
5175 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5177 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5179 2017-10-08 Olivier Hainque <hainque@adacore.com>
5181 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5182 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5184 (thumb_set_return_address): Likewise.
5186 2017-10-08 Olivier Hainque <hainque@adacore.com>
5188 * common/config/arm/arm-common.c (arm_except_unwind_info):
5189 Handle DWARF2_UNWIND_INFO.
5191 2017-10-07 Michael Collison <michael.collison@arm.com>
5193 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5196 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5198 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5199 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5200 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5201 defined, force the creation of a new block for a dispatch label.
5203 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5205 * invoke.texi (Wsuggest-attribute=cold): Document.
5206 * common.opt (Wsuggest-attribute=cold): New
5207 * ipa-pure-const.c (warn_function_cold): New function.
5208 * predict.c (compute_function_frequency): Use it.
5209 * predict.h (warn_function_cold): Declare.
5211 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5213 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5216 2017-10-06 Martin Liska <mliska@suse.cz>
5218 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5219 keyword for member functions.
5220 (struct sanopt_tree_couple): New struct.
5221 (struct sanopt_tree_couple_hash): New function.
5222 (struct sanopt_ctx): Add new hash_map.
5223 (has_dominating_ubsan_ptr_check): New function.
5224 (record_ubsan_ptr_check_stmt): Likewise.
5225 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5226 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5227 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5229 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5232 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5233 aarch64_simd_valid_immediate on CONST_VECTORs.
5234 (aarch64_reg_or_bic_imm): Likewise.
5236 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5238 PR rtl-optimization/82396
5239 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5241 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5243 * graphite-dependences.c (scop_get_reads): Move code to...
5244 (scop_get_must_writes): Move code to...
5245 (scop_get_may_writes): Move code to...
5246 (scop_get_reads_and_writes): ... here.
5247 (scop_get_dependences): Call scop_get_reads_and_writes.
5249 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5251 PR tree-optimization/82434
5252 * fold-const.h (can_native_encode_type_p,
5253 can_native_encode_string_p): Remove.
5254 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5255 don't encode anything, just return what would be otherwise returned.
5256 (native_encode_fixed, native_encode_complex, native_encode_vector):
5258 (native_encode_string): Likewise. Inline by hand
5259 can_native_encode_string_p.
5260 (can_native_encode_type_p): Remove.
5261 (can_native_encode_string_p): Remove.
5262 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5263 STRING_CSTs using can_native_encode_string_p, test all
5264 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5265 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5266 argument from native_encode_expr.
5267 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5268 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5271 2017-10-06 Richard Biener <rguenther@suse.de>
5273 PR tree-optimization/82397
5274 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5275 operand_equal_p but rely on data_ref_compare_tree for detecting
5277 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5278 to match up with dr_group_sort_cmp.
5280 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5283 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5285 * config/s390/s390-builtin-types.def: Regenerate.
5287 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5290 * config/s390/s390-builtin-types.def: Regenerate.
5291 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5292 Change flag from B_VXE to B_VX.
5293 (s390_vec_min_dbl): Remove B_VXE flag.
5295 2017-10-06 Richard Biener <rguenther@suse.de>
5297 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5298 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5299 translate_isl_ast_to_gimple::is_valid_rename,
5300 translate_isl_ast_to_gimple::get_rename,
5301 translate_isl_ast_to_gimple::get_def_bb_for_const,
5302 translate_isl_ast_to_gimple::get_new_name,
5303 translate_isl_ast_to_gimple::collect_all_ssa_names,
5304 translate_isl_ast_to_gimple::copy_loop_phi_args,
5305 translate_isl_ast_to_gimple::collect_all_ssa_names,
5306 translate_isl_ast_to_gimple::copy_loop_phi_args,
5307 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5308 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5309 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5310 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5311 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5312 translate_isl_ast_to_gimple::copy_cond_phi_args,
5313 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5314 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5315 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5316 translate_isl_ast_to_gimple::rename_uses,
5317 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5318 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5319 (set_rename_for_each_def): Likewise.
5320 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5321 here. Handle rewriting SCEV analyzable uses here.
5322 (copy_bb_and_scalar_dependences): Generate code for PHI
5324 (graphite_regenerate_ast_isl): Adjust.
5325 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5326 (add_write, add_read): New functions.
5327 (build_cross_bb_scalars_def): Use it and simplify.
5328 (build_cross_bb_scalars_use): Likewise.
5329 (graphite_find_cross_bb_scalar_vars): Inline into...
5330 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5331 simulating out-of-SSA. Compute liveout and add dependencies.
5332 (build_scops): Force an empty entry block.
5333 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5335 (sese_build_liveouts): Declare.
5336 (sese_trivially_empty_bb_p): Likewise.
5337 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5338 compute liveout and debug_liveout.
5339 (sese_bad_liveouts_use): Remove.
5340 (sese_reset_debug_liveouts_bb): Likewise.
5341 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5342 (sese_build_liveouts): Build liveout and debug_liveout and store
5344 (new_sese_info): Adjust.
5345 (free_sese_info): Likewise.
5346 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5347 do not build liveout here.
5348 (move_sese_in_condition): Adjust region entry.
5349 (scev_analyzable_p): Match up with chrec_apply requirements.
5350 (sese_trivially_empty_bb_p): New.
5351 * tree-into-ssa.c (get_reaching_def): Properly support generating
5352 default-defs for incremental rewrite of anonymous names.
5354 2017-10-06 Richard Biener <rguenther@suse.de>
5356 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5357 precision do not perform modulo reduction.
5359 2017-10-06 Richard Biener <rguenther@suse.de>
5361 PR tree-optimization/82436
5362 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5363 conservatively choose the vectorization factor when checking
5364 whether we can perform the required load permutation.
5365 (vect_transform_slp_perm_load): Assert when we may not fail.
5367 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5369 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5370 message for incompatible -msdata=* and -mcall-* options.
5372 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5374 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5375 rate for post-reload scheduling.
5377 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5379 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5381 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5383 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5384 to improve monte carlo in scimark.
5386 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5388 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5389 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5390 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5391 pentium4_cost, nocona_cost): Set reassociation width to 1.
5392 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5393 width to 2 for fp operations and 1 otherwise.
5394 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5396 (atom_cost): Set reassociation width to 2.
5397 (slm_cost, generic_cost): Set fp reassociation width
5398 to 2 and 1 otherwise.
5399 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5400 (core_cost): Set fp reassociation width to 4 and vector to 2.
5401 (ix86_reassociation_width): Rewrite using cost table; special case
5402 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5403 and TARGET_AVX128_OPTIMAL.
5404 * config/i386/i386.h (processor_costs): Add
5405 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5406 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5407 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5408 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5409 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5410 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5412 2017-10-05 Nathan Sidwell <nathan@acm.org>
5414 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5416 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5418 * config/arm/arm.c (arm_test_fpu_data): New.
5419 (arm_run_selftests): Call arm_test_fpu_data.
5421 2017-10-04 Nathan Sidwell <nathan@acm.org>
5423 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5425 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5427 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5429 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5430 check type for aarch64_simd_valid_immediate.
5431 (aarch64_output_simd_mov_immediate): Update prototype.
5432 (aarch64_simd_valid_immediate): Update prototype.
5433 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5434 support for ORR-immediate.
5435 (and<mode>3): modified pattern to add support for BIC-immediate.
5436 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5437 now checks for valid immediate for BIC and ORR based on new enum
5439 (aarch64_output_simd_mov_immediate): Function now used to output
5440 BIC/ORR imm as well based on new enum argument.
5441 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5443 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5444 (aarch64_reg_or_bic_imm): Likewise.
5446 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5448 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5449 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5451 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5455 PR rtl-optimization/82396
5456 * haifa-sched.c (autopref_multipass_init): Simplify
5458 (autopref_rank_data): Simplify sort order.
5459 * sched-int.h (autopref_multipass_data_): Remove
5460 multi_mem_insn_p, min_offset and max_offset.
5462 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5464 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5466 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5468 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5471 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5473 * doc/sourcebuild.texi: Document vect_long_mult.
5475 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5477 PR tree-optimization/82413
5478 * fold-const.c (build_range_check): Use widest_int when comparing
5479 the maximum ETYPE value with HIGH.
5481 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5483 PR rtl-optimization/82396
5484 * haifa-sched.c (autopref_multipass_init): Simplify
5486 (autopref_rank_data): Simplify sort order.
5487 * sched-int.h (autopref_multipass_data_): Remove
5488 multi_mem_insn_p, min_offset and max_offset.
5490 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5492 PR tree-optimization/82381
5493 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5494 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5497 PR tree-optimization/82374
5498 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5499 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5500 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5501 current_function_decl to the new decl.
5503 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5505 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5506 helper macro for IEEE float128 hardware built-in functions.
5507 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5509 (TRUNCF128_ODD): Likewise.
5510 (ADDF128_ODD): Likewise.
5511 (SUBF128_ODD): Likewise.
5512 (MULF128_ODD): Likewise.
5513 (DIVF128_ODD): Likewise.
5514 (FMAF128_ODD): Likewise.
5515 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5516 UNSPEC_TRUNC_ROUND_TO_ODD.
5517 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5518 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5519 floating point round to odd instructions.
5520 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5521 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5522 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5523 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5524 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5525 (trunc<mode>sf2_hw): Change the truncate with round to odd
5526 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5527 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5528 to odd hardware instructions.
5529 (sub<mode>3_odd): Likewise.
5530 (mul<mode>3_odd): Likewise.
5531 (div<mode>3_odd): Likewise.
5532 (sqrt<mode>2_odd): Likewise.
5533 (fma<mode>4_odd): Likewise.
5534 (fms<mode>4_odd): Likewise.
5535 (nfma<mode>4_odd): Likewise.
5536 (nfms<mode>4_odd): Likewise.
5537 (trunc<mode>df2_odd): Change the truncate with round to odd
5538 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5540 * doc/extend.texi (PowerPC built-in functions): Update documentation
5541 for existing IEEE float128-bit built-in functions. Add built-in
5542 functions that generate the IEEE 128-bit floating point round to
5545 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5547 PR rtl-optimization/77729
5548 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5549 to (X&(C1&~C2))|C2 transformations.
5551 2017-10-03 Martin Jambor <mjambor@suse.cz>
5553 PR tree-optimization/82363
5554 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5555 mismatch, mark lacc written regardless of racc.
5557 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5559 PR tree-optimization/82381
5560 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5561 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5562 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5563 fallthrough into reassoc_stmt_dominates_stmt_p.
5566 * combine.c (combine_instructions): Don't combine in unreachable
5569 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5572 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5573 function to not use the have_cpu variable. Do not set cpu_index,
5574 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5576 (rs6000_valid_attribute_p): Remove duplicate initializations of
5577 old_optimize and func_optimize.
5578 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5579 (rs6000_activate_target_options): Make global.
5580 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5583 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5585 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5586 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5587 Set *poffset to 0 and *psize and *pmax_size to -1 if
5588 *poffset + *psize overflows in HOST_WIDE_INT.
5590 PR tree-optimization/82387
5591 PR tree-optimization/82388
5592 PR tree-optimization/82389
5593 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5594 instead of live_bytes non-NULL.
5596 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5599 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5602 2017-10-02 Richard Biener <rguenther@suse.de>
5604 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5605 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5606 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5608 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5610 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5611 requested precision matches the type's.
5612 * calls.c (alloc_max_size): Calculate the new candidate size as
5613 a widest_int and use wi::to_widest when comparing it with the
5614 current candidate size.
5615 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5616 zero rather than integer_zero_node.
5617 * match.pd: Check for a no-op conversion before using wi::add
5618 rather than after. Use tree_to_uhwi when summing small shift
5619 counts into an unsigned int.
5621 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5622 Alan Hayward <alan.hayward@arm.com>
5623 David Sherwood <david.sherwood@arm.com>
5626 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5627 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5628 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5629 POINTER_AND_FP_REGS.
5631 2017-10-02 Richard Biener <rguenther@suse.de>
5633 PR tree-optimization/82355
5634 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5635 a mapping for the enclosing loop but avoid generating one for
5637 (copy_bb_and_scalar_dependences): Remove premature codegen
5638 error on PHIs in blocks duplicated into multiple places.
5639 * graphite-scop-detection.c
5640 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5641 in the region use it as loop and nest to analyze the DR in.
5642 (try_generate_gimple_bb): Likewise.
5643 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5644 (add_loop_constraints): For blocks in a loop not in the region
5645 create a dimension with a single iteration.
5646 * sese.h (gbb_loop_at_index): Remove assert.
5648 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5650 * omp-expand.c (adjust_context_scope): New function.
5651 (expand_parallel_call): Call adjust_context_scope.
5653 2017-10-01 Jeff Law <law@redhat.com>
5655 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5656 dom_opt_dom_walker class with direct access to private members.
5657 Add comments. Call test_for_singularity.
5658 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5659 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5660 m_dummy_cond anymore.
5661 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5663 (pass_dominator:execute): Build the dummy_cond here and pass it
5664 to the dom_opt_dom_walker ctor.
5665 (test_for_singularity): New function.
5667 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5668 Maya Rashish <coypu@sdf.org>
5670 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5671 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5672 (alpha*-*-netbsd*) Use nbsd_tm_file.
5673 (arm*-*-netbsdelf*) Likewise.
5674 (i[34567]86-*-netbsdelf*) Likewise.
5675 (x86_64-*-netbsd*) Likewise.
5676 (mips*-*-netbsd*) Likewise.
5677 (powerpc-*-netbsd*) Likewise.
5678 (sh*-*-netbsd*) Likewise.
5679 (sparc-*-netbsdelf*) Likewise.
5680 (sparc64-*-netbsd*) Likewise.
5681 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5683 (vax-*-netbsdelf*) Likewise.
5684 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5685 (UINT_FAST8_TYPE) Likewise.
5686 (INT_FAST16_TYPE) Check CHAR_FAST16.
5687 (UINT_FAST16_TYPE) Likewise.
5689 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5692 * config/i386/i386.md
5693 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5694 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5695 *divmodsi4_zext_2): New define_insn_and_split.
5696 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5697 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5698 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5699 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5700 New define_insn_and_split.
5701 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5702 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5703 operands[1] having DImode when mode is SImode.
5705 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5706 always SImode for DIV and MOD in REG_EQUAL notes.
5708 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5711 * match.pd: Fix handling of NaNs in pattern.
5713 2017-09-29 Jeff Law <law@redhat.com>
5715 * sbitmap.c (bitmap_bit_in_range_p): New function.
5716 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5717 * tree-ssa-dse.c (live_bytes_read): New function.
5718 (dse_classify_store): Ignore reads of dead bytes.
5720 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5721 typos and whitespace errors.
5722 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5723 * config/s390/s390.c (s390_emit_prologue): Likewise.
5725 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5728 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5729 with a symbol for LRA.
5731 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5733 PR rtl-optimization/82338
5734 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5736 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5738 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5739 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5740 (qsort_chk): Declare.
5741 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5742 (qsort_chk): New function.
5744 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5746 PR tree-optimization/82337
5747 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5748 phi definition if the PHI result appears in an abnormal PHI.
5749 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5750 basis appears in an abnormal PHI.
5752 2017-09-29 Richard Biener <rguenther@suse.de>
5754 * graphite-isl-ast-to-gimple.c
5755 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5756 (binary_op_to_tree): Use it.
5757 (get_rename_from_scev): Likewise.
5758 (copy_loop_phi_nodes): Likewise.
5759 (copy_bb_and_scalar_dependences): Likewise.
5760 (translate_pending_phi_nodes): Likewise.
5762 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5765 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5766 for movabsq $(i32 << shift), r64.
5768 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5770 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5771 index when encoding %esp as %rsp to avoid 0x67 prefix.
5773 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5775 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5776 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5778 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5780 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5781 Extensions with more than 16 double VFP registers.
5782 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5783 to_clear_mask and all code related to it. Replace the remaining
5784 entry by a sbitmap and adapt code accordingly.
5786 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5788 * brig-builtins.def: Change pure attributes to const.
5790 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5792 * config.gcc (default_gnu_indirect_function): Default to yes for
5793 sparc*-*-linux* with glibc.
5795 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5797 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5798 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5799 when creating .init_array and .fini_array sections with priority
5802 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5805 * config/aarch64/aarch64.c
5806 (aarch64_builtin_support_vector_misalignment): Always return false
5807 when misalignment is unknown.
5809 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5811 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5812 this function to return false if the definition used by the swap
5813 instruction is artificial, or if the memory address from which the
5814 constant value is loaded is not represented by a base address held
5815 in a register or if the base address register is a frame or stack
5816 pointer. Additionally, return false if the base address of the
5817 loaded constant is a SYMBOL_REF but is not considered to be a
5819 (replace_swapped_load_constant): New function.
5820 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5821 loaded constant vector with a load of a swapped constant vector.
5823 2017-09-27 Carl Love <cel@us.ibm.com>
5825 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5826 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5827 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5830 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5832 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5833 first, always call autopref_rank_data otherwise.
5835 2017-09-27 Richard Biener <rguenther@suse.de>
5837 * graphite-scop-detection.c (find_scop_parameters): Move
5838 loop bound handling ...
5839 (gather_bbs::before_dom_children): ... here, avoiding the need
5840 to build scop_info->loop_nest.
5841 (record_loop_in_sese): Remove.
5842 * sese.h (sese_info_t::loop_nest): Remove.
5843 * sese.c (new_sese_info): Do not allocate loop_nest.
5844 (free_sese_info): Do not free loop_nest.
5846 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5849 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5850 lhs from calls if the lhs has addressable type.
5852 2017-09-27 Richard Biener <rguenther@suse.de>
5854 * graphite.h (scop::max_alias_set): New member.
5855 * graphite-scop-detection.c: Remove references to non-existing
5856 --param in comments.
5857 (build_alias_sets): Record the maximum alias set used for drs.
5858 (build_scops): Support zero as unlimited for
5859 --param graphite-max-arrays-per-scop.
5860 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5862 (build_poly_sr_1): ... here. Compute alias set based on the
5863 maximum alias set used for drs rather than
5864 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5866 2017-09-27 Richard Biener <rguenther@suse.de>
5868 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5869 --param loop-block-tile-size=0 to disable tiling.
5871 2017-09-27 Richard Biener <rguenther@suse.de>
5873 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5874 (graphite-max-nb-scop-params): Document special value zero.
5875 * domwalk.h (dom_walker::STOP): New symbolical constant.
5876 (dom_walker::dom_walker): Add optional parameter for bb to
5878 (dom_walker::~dom_walker): Declare.
5879 (dom_walker::before_dom_children): Document STOP return value.
5880 (dom_walker::m_user_bb_to_rpo): New member.
5881 (dom_walker::m_bb_to_rpo): Likewise.
5882 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5883 mapping here if not provided by the user.
5884 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5885 provided by the user.
5886 (dom_walker::STOP): Define.
5887 (dom_walker::walk): Do not compute bb to RPO mapping here.
5888 Support STOP return value from before_dom_children to stop
5890 * graphite-optimize-isl.c (optimize_isl): If the schedule
5891 is the same still generate code if -fgraphite-identity
5892 or -floop-parallelize-all are given.
5893 * graphite-scop-detection.c: Include cfganal.h.
5894 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5896 (gather_bbs::before_dom_children): Return STOP for BBs
5898 (build_scops): Compute bb to RPO mapping and pass it to
5899 the domwalk. Treat --param graphite-max-nb-scop-params=0
5900 as not limiting the number of params.
5901 * graphite.c (graphite_initialize): Remove limit on the
5902 number of basic-blocks in a function.
5903 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5904 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5905 default value of 10.
5907 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5909 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5910 Adjust code to eliminate needing to do the shift right 32-bits
5911 operation after XSCVDPSPN.
5913 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5915 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5916 ((X / Y) != 0 -> X >= Y): Likewise.
5918 2017-09-26 Carl Love <cel@us.ibm.com>
5920 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5921 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5922 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5923 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5924 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5925 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5926 definitions and overloading.
5927 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5928 statement for P9V_BUILTIN_XST_LEN_R.
5929 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5930 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5931 define_expand and define_insn for the instructions and builtins.
5932 * doc/extend.texi: Update the built-in documentation file for the new
5934 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5935 define_insn for the instructions
5937 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5940 * gcc/config/netbsd-protos.h: New file.
5941 * gcc/config/netbsd.c: New file.
5942 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5943 * gcc/config/t-netbsd: New file.
5944 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5945 (tmake_file) Add t-netbsd.
5946 (extra_objs) Add netbsd.o.
5948 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5952 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5954 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5956 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5957 sign extension from a vector register to a GPR by doing a 32-bit
5958 direct move and then an EXTSW.
5959 (extendsi<mode>2 splitter): Likewise.
5960 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5961 right or vector extract after doing XSCVDPSPN. Use
5962 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5964 (movdi_from_sf_zero_ext): Likewise.
5965 (reload_gpr_from_vsxsf): Likewise.
5966 (p8_mfvsrd_4_disf): Delete, no longer used.
5967 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5968 then needing to move the SFmode to a GPR to use the XSCVDPSP
5969 instruction instead of FRSP and XSCVDPSPN.
5970 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5971 it is adjacent to the other XSCVSPDP insns.
5972 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5973 SFmode to be in traditional Altivec registers.
5974 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5975 (vsx_xscvspdpn): Likewise.
5976 (vsx_xscvspdpn_scalar): Likewise.
5978 2017-09-26 Martin Jambor <mjambor@suse.cz>
5980 * tree-sra.c (compare_access_positions): Put integral types first,
5981 stabilize sorting of integral types, remove conditions putting
5982 non-full-precision integers last.
5983 (sort_and_splice_var_accesses): Disable scalarization if a
5984 non-integert would be represented by a non-full-precision integer.
5986 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5988 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5989 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5990 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5991 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5992 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5993 conditionals inside the function instead of around it. Call
5994 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5995 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5997 2017-09-26 Richard Biener <rguenther@suse.de>
5999 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6001 (scop_detection::build_scop_breadth): ... this. Removed.
6002 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6003 (scop_detection::harmful_stmt_in_bb): Likewise.
6004 (scop_detection::graphite_can_represent_stmt): Likewise.
6005 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6006 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6007 (scop_detection::can_represent_loop_1): ... this. Removed.
6008 (scop_detection::harmful_loop_in_region): Simplify after inlining
6009 the above and remove more quadraticness.
6010 (build_scops): Adjust.
6011 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6014 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6017 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6018 REGNO (base) == SP_REG if base is a REG.
6021 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6022 if it is different SSA_NAME.
6023 (optimize_range_tests_cmp_bitwise): New function.
6024 (optimize_range_tests): Call it.
6026 2017-09-26 Richard Biener <rguenther@suse.de>
6028 PR tree-optimization/82321
6029 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6030 for the def being inside the loop.
6032 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6034 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6036 * config/s390/s390-builtins.def: Fix constraint on op4.
6038 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6040 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6041 independent expanders.
6042 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6043 ("vec_ordered", "vec_unordered"): New expanders.
6045 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6047 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6050 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6052 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6053 vec_unpacks_lo_v16qi.
6054 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6056 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6058 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6059 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6060 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6062 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6064 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6066 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6068 ("*vec_slb<mode>"): New insn pattern.
6069 ("vec_shr_<mode>"): New expander.
6070 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6071 and force the shift count operand to V16QImode.
6072 ("vec_srb<mode>"): Set shift count mode to V16QI.
6074 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6076 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6077 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6078 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6080 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6083 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6085 2017-09-26 Richard Biener <rguenther@suse.de>
6087 PR tree-optimization/82320
6088 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6091 2017-09-25 Jeff Law <law@redhat.com>
6093 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6094 prototype for new argument.
6095 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6096 mostly extracted from rs6000_emit_allocate_stack.
6097 (rs6000_emit_probe_stack_range_stack_clash): New function.
6098 (rs6000_emit_allocate_stack): Call
6099 rs6000_emit_probe_stack_range_stack_clash as needed.
6100 (rs6000_emit_probe_stack_range): Add additional argument
6101 to call to gen_probe_stack_range{si,di}.
6102 (output_probe_stack_range): New.
6103 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6104 (output_probe_stack_range_stack_clash): New.
6105 (rs6000_emit_prologue): Emit notes into dump file as requested.
6106 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6107 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6108 Add additional operand and pass it to output_probe_stack_range.
6110 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6112 PR tree-optimization/82163
6113 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6114 (checking_verify_loop_closed_ssa): New parameter.
6115 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6116 (check_loop_closed_ssa_stmt): Delete.
6117 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6118 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6119 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6122 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6124 * brig-builtins.def: Treat HSAIL barrier builtins as
6125 setjmp/longjump style functions.
6127 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6129 * target.def (constant_alignment): New hook.
6130 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6131 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6132 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6133 * doc/tm.texi: Regenerate.
6134 * targhooks.h (default_constant_alignment): Declare.
6135 (constant_alignment_word_strings): Likewise.
6136 * targhooks.c (default_constant_alignment): New function.
6137 (constant_alignment_word_strings): Likewise.
6138 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6139 instead of CONSTANT_ALIGNMENT.
6140 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6141 (force_const_mem): Likewise.
6142 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6143 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6144 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6145 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6146 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6148 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6149 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6150 constant_alignment_word_strings.
6151 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6152 (CONSTANT_ALIGNMENT): Likewise.
6153 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6154 (arm_constant_alignment): New function.
6155 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6156 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6157 constant_alignment_word_strings.
6158 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6159 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6160 constant_alignment_word_strings.
6161 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6162 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6163 (cris_constant_alignment): New function.
6164 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6165 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6166 (epiphany_constant_alignment): New function.
6167 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6168 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6169 constant_alignment_word_strings.
6170 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6171 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6172 constant_alignment_word_strings.
6173 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6174 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6175 constant_alignment_word_strings.
6176 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6177 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6178 * config/i386/i386.c (ix86_constant_alignment): Make static.
6179 Use the same interface as the target hook.
6180 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6181 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6182 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6183 constant_alignment_word_strings.
6184 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6185 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6186 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6187 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6188 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6189 constant_alignment_word_strings.
6190 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6191 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6192 constant_alignment_word_strings.
6193 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6194 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6195 constant_alignment_word_strings.
6196 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6197 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6199 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6200 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6201 * config/mips/mips.c (mips_constant_alignment): New function.
6202 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6203 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6204 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6205 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6206 (mmix_constant_alignment): Make static. Use the same interface
6208 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6209 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6210 constant_alignment_word_strings.
6211 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6212 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6213 constant_alignment_word_strings.
6214 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6215 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6216 constant_alignment_word_strings.
6217 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6218 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6219 (rs6000_constant_alignment): New function.
6220 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6221 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6222 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6223 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6224 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6225 (rs6000_constant_alignment): New function.
6226 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6227 * config/s390/s390.c (s390_constant_alignment): New function.
6228 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6229 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6230 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6231 constant_alignment_word_strings.
6232 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6233 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6234 (sparc_constant_alignment): New function.
6235 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6236 * config/spu/spu.c (spu_constant_alignment): New function.
6237 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6238 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6239 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6240 constant_alignment_word_strings.
6241 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6242 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6243 constant_alignment_word_strings.
6244 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6245 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6246 constant_alignment_word_strings.
6247 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6248 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6249 (visium_constant_alignment): New function.
6250 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6251 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6252 (xtensa_constant_alignment): New function.
6253 * system.h (CONSTANT_ALIGNMENT): Poison.
6255 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6257 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6258 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6259 (rs6000_builtin_valid_without_lhs): New helper function.
6260 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6261 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6263 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6265 * target.h (vec_perm_indices): Use unsigned short rather than
6267 (auto_vec_perm_indices): Likewise.
6268 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6269 Use unsigned int rather than unsigned char.
6270 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6272 2017-09-25 Richard Biener <rguenther@suse.de>
6274 * cfgloop.h (sort_sibling_loops): Declare.
6275 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6276 (sort_sibling_loops): New function sorting the sibling loop list
6278 * graphite.c (graphite_transform_loops): Sort sibling loops.
6280 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6282 * target.def (vec_perm_const_ok): Change sel parameter to
6284 * optabs-query.c (can_vec_perm_p): Update accordingly.
6285 * doc/tm.texi: Regenerate.
6286 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6287 auto_vec_perm_indices and remove separate nelt field.
6288 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6289 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6290 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6291 (aarch64_expand_vec_perm_const): Update accordingly.
6292 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6293 to vec_perm_indices.
6294 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6295 auto_vec_perm_indices and remove separate nelt field.
6296 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6297 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6298 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6300 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6301 to vec_perm_indices.
6302 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6303 sel to vec_perm_indices.
6304 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6305 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6306 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6308 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6311 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6314 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6315 on the FUNCTION_DECL function context if it has a DIE that is a
6318 2017-09-25 Richard Biener <rguenther@suse.de>
6320 PR tree-optimization/82285
6321 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6324 2017-09-25 Tom de Vries <tom@codesourcery.com>
6328 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6331 2017-09-25 Richard Biener <rguenther@suse.de>
6333 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6334 ISL errors other than isl_error_quota happen. Dump if the
6335 schedule is the same.
6336 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6337 errors instead of aborting inside ISL.
6339 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6342 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6343 of libgcc_eh for m64.
6344 * config/i386/darwin64.h: Likewise.
6346 2017-09-25 Richard Biener <rguenther@suse.de>
6349 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6350 attribute for incomplete types nor twice for complete ones.
6352 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6355 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6356 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6359 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6362 * config/i386/i386.opt (mprefer-avx256): Use
6363 ix86_target_flags variable.
6364 * config/i386/i386.c (ix86_target_string): Move
6365 -mprefer-avx256 to flag2_opts.
6367 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6370 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6371 and x != -1 | y != -1 into (x & y) != -1.
6373 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6375 * config.gcc: Add new case statement to set
6376 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6377 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6378 s390x-*-linux* case statements. Added aarch64 to the list of
6379 supported architectures.
6381 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6383 PR tree-optimization/82289
6384 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6385 STMT_VINFO_RELEVANT_P.
6387 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6388 Alan Hayward <alan.hayward@arm.com>
6389 David Sherwood <david.sherwood@arm.com>
6391 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6392 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6393 (extract_range_from_binary_expr_1): Don't call
6394 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6396 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6397 Alan Hayward <alan.hayward@arm.com>
6398 David Sherwood <david.sherwood@arm.com>
6400 * target.def (preferred_vector_alignment): New hook.
6401 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6403 * doc/tm.texi: Regenerate.
6404 * targhooks.h (default_preferred_vector_alignment): Declare.
6405 * targhooks.c (default_preferred_vector_alignment): New function.
6406 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6408 (DR_TARGET_ALIGNMENT): New macro.
6409 (aligned_access_p): Update commentary.
6410 (vect_known_alignment_in_bytes): New function.
6411 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6413 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6414 Calculate the misalignment based on the target alignment rather than
6416 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6417 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6418 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6419 the target alignment, rather than masking the element misalignment
6420 with the number of elements in a vector. Also use the target
6421 alignment when calculating the maximum number of peels.
6422 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6423 instead of TYPE_ALIGN_UNIT.
6424 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6425 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6426 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6427 (vect_create_data_ref_ptr): Likewise.
6428 (vect_setup_realignment): Realign by ANDing with
6429 -DR_TARGET_MISALIGNMENT.
6430 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6431 the number of peels based on DR_TARGET_ALIGNMENT.
6432 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6433 with the guaranteed alignment boundary when deciding whether
6435 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6436 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6437 (ensure_base_align): Remove stmt_info parameter. Get the
6438 target base alignment from DR_TARGET_ALIGNMENT.
6439 (vectorizable_store): Update call accordingly. Interpret
6440 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6442 (vectorizable_load): Likewise.
6444 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6445 Alan Hayward <alan.hayward@arm.com>
6446 David Sherwood <david.sherwood@arm.com>
6448 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6449 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6450 (vect_enhance_data_refs_alignment): Likewise.
6452 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6454 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6455 error. Only quit immediately if parsing is complete.
6456 (BEGIN): Initialize fatal_err and parse_done.
6457 (begin fpu, end fpu): Check number of arguments.
6458 (begin arch, end arch): Likewise.
6459 (begin cpu, end cpu): Likewise.
6460 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6461 (optalias): Likewise.
6463 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6465 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6466 * config/arm/arm-isa.h: Delete. Move definitions to ...
6467 * arm-cpus.in: ... here. Use new feature and fgroup values.
6468 * config/arm/arm.c (arm_option_override): Use lower case for feature
6470 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6471 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6472 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6474 (print_isa_bits_for): New function.
6475 (gen_isa): New function.
6476 (gen_comm_data): Use print_isa_bits_for.
6477 (define feature): New keyword.
6478 (define fgroup): New keyword.
6479 * config/arm/t-arm (TM_H): Remove.
6480 (GTM_H): Add arm-isa.h.
6481 (arm-isa.h): Add rule to generate file.
6482 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6483 case for feature bit names.
6485 2017-09-22 Richard Biener <rguenther@suse.de>
6487 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6489 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6490 print of no dependency loops ...
6491 * graphite.c (graphite_transform_loops): ... here.
6492 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6494 (same_close_phi_node, remove_duplicate_close_phi,
6495 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6496 (canonicalize_loop_closed_ssa): ... here and simplify.
6497 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6498 (optimize_isl): Use dump_printf_loc to tell when we stopped
6499 optimizing because of an ISL timeout.
6501 2017-09-22 Richard Biener <rguenther@suse.de>
6503 PR tree-optimization/82291
6504 * tree-if-conv.c (predicate_mem_writes): Make sure to
6505 remove writes in blocks predicated with false.
6507 2017-09-22 Richard Biener <rguenther@suse.de>
6509 * sese.c: Include cfganal.h.
6510 (if_region_set_false_region): Remove.
6511 (create_if_region_on_edge): Likewise.
6512 (move_sese_in_condition): Re-implement without destroying
6515 2017-09-22 Richard Biener <rguenther@suse.de>
6517 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6518 Verify both BBs contain loop PHI nodes before dispatching to
6520 (graphite_regenerate_ast_isl): Do not recompute dominators,
6521 do not verify three times. Restructure for clarity.
6522 * graphite-scop-detection.c (same_close_phi_node,
6523 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6524 defined_in_loop_p, canonicalize_loop_closed_ssa,
6525 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6526 checking and SSA rewrite, move to ...
6527 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6528 (graphite_initialize): Do not pass in ctx, do not reset the
6529 SCEV cache, compute only dominators.
6530 (graphite_transform_loops): Allocate ISL ctx after
6531 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6532 Maintain post-dominators only around build_scops.
6533 * sese.c (if_region_set_false_region): Make static. Free
6534 and recompute dominators.
6535 (move_sese_in_condition): Assert we don't get called with
6536 post-dominators computed.
6537 * sese.h (if_region_set_false_region): Remove.
6539 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6541 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6542 mode attribute for TARGET_AVX512VL.
6544 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6546 * config/i386/i386.opt (mprefer-avx256): New option.
6547 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6549 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6550 for TARGET_PREFER_AVX256.
6551 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6553 2017-09-21 Jeff Law <law@redhat.com>
6555 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6556 Fix dump output if the only stack space is for pushed registers.
6558 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6560 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6563 2017-09-21 Martin Sebor <msebor@redhat.com>
6566 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6567 code (in C++) or code that triggers warnings.
6569 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6571 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6573 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6575 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6576 * sched-rgn.c: Ditto.
6577 * sel-sched-ir.c: Ditto.
6579 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6581 * toplev.h (set_random_seed): Adjust return type.
6582 * toplev.c (init_local_tick): Move eager initialization of random_seed
6583 to get_random_seed. Adjust comment.
6584 (init_random_seed): Inline to get_random_seed, delete.
6585 (get_random_seed): Initialize random_seed lazily.
6586 (set_random_seed): Do not return previous value.
6587 (print_switch_value): Do not call get_random_seed.
6589 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6591 * cgraph.c (delete_function_version): New, broken out from...
6592 (cgraph_node::delete_function_version): ...here. Rename to
6593 cgraph_node::delete_function_version_by_decl. Update all uses.
6594 (cgraph_node::remove): Call delete_function_version.
6596 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6599 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6600 VAR_DECLs to which addressable non-volatile parameters are mapped
6601 and for id->retvar after the return value assignment. Clear
6602 id->retval and id->retbnd after inlining.
6604 2017-09-21 Richard Biener <rguenther@suse.de>
6606 PR tree-optimization/82276
6607 PR tree-optimization/82244
6608 * tree-vrp.c (build_assert_expr_for): Set
6609 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6611 (remove_range_assertions): Revert earlier change.
6613 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6616 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6618 2017-09-21 Richard Biener <rguenther@suse.de>
6620 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6621 Restore valid IL after code generation errors.
6622 * graphite.c (graphite_transform_loops): Diagnose code
6623 generation issues as MSG_MISSED_OPTIMIZATION and continue
6624 with processing SCOPs.
6626 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6627 Alan Hayward <alan.hayward@arm.com>
6628 David Sherwood <david.sherwood@arm.com>
6630 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6631 rather than choosing between plus_constant and gen_rtx_<CODE>.
6632 * expr.c (emit_push_insn): Likewise.
6633 (expand_expr_real_2): Likewise.
6635 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6636 Alan Hayward <alan.hayward@arm.com>
6637 David Sherwood <david.sherwood@arm.com>
6639 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6641 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6642 Alan Hayward <alan.hayward@arm.com>
6643 David Sherwood <david.sherwood@arm.com>
6645 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6646 calling tree_to_uhwi.
6648 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6649 Alan Hayward <alan.hayward@arm.com>
6650 David Sherwood <david.sherwood@arm.com>
6652 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6653 INTEGER_CST rather than a negative test for ADDR_EXPR.
6655 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6656 Alan Hayward <alan.hayward@arm.com>
6657 David Sherwood <david.sherwood@arm.com>
6659 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6660 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6662 2017-09-21 Richard Biener <rguenther@suse.de>
6664 PR tree-optimization/71351
6665 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6666 graphite_create_new_loop_guard): Remove, fold remaining parts
6668 (translate_isl_ast_node_for): ... here and simplify.
6670 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6673 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6674 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6675 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6676 alternative always use QI mode, for -Os imov (=R,R) alternative
6677 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6680 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6681 Jeff Law <law@redhat.com>
6683 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6684 (allocate_stack_space): New function, partially extracted from
6686 (s390_emit_prologue): Track offset to most recent stack probe.
6687 Code to allocate space moved into allocate_stack_space.
6688 Dump actions when no stack is allocated.
6689 (s390_prologue_plus_offset): New function.
6690 (s390_emit_stack_probe): Likewise.
6692 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6694 * common.opt (Wa, Wl, Wp, g, gz=): Add
6696 (gno-column-info): Remove.
6697 (gcolumn-info): Drop RejectNegative.
6699 (gno-record-gcc-switches): Remove.
6700 (grecord-gcc-switches): Drop RejectNegative.
6701 (gno-split-dwarf): Remove.
6702 (gsplit-dwarf): Drop RejectNegative.
6703 (gno-strict-dwarf): Remove.
6704 (gstrict-dwarf): Drop RejectNegative.
6705 * config/darwin.opt (gfull, gused): Add RejectNegative.
6706 * dwarf2out.c (gen_producer_string): Drop
6707 gno-record-gcc-switches handler.
6708 * optc-gen.awk: Add g to prefixes with negative forms.
6709 * opts-common.c (remapping_prefix_p): New.
6710 (find_opt): Check it.
6711 (generate_canonical_option): Test g prefix.
6712 (option_map): Add -gno- mapping.
6713 (add_misspelling_candidates): Check remapping_prefix_p.
6715 2017-09-20 Jeff Law <law@redhat.com>
6717 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6718 thinko in stack clash protection support.
6720 * explow.c (compute_stack_clash_protection_loop_data): Use
6721 CONST_INT_P instead of explicit test. Verify object is a
6722 CONST_INT_P before looking at INTVAL.
6723 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6724 instead of explicit test.
6726 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6729 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6730 address instead of to r1 and r11.
6732 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6734 * config.gcc: Support "knm".
6735 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6736 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6738 * config/i386/i386.c (m_KNM): Define.
6739 (processor_target_table): Add "knm".
6741 (ix86_option_override_internal): Add "knm".
6742 (ix86_issue_rate): Add PROCESSOR_KNM.
6743 (ix86_adjust_cost): Ditto.
6744 (ia32_multipass_dfa_lookahead): Ditto.
6745 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6746 (fold_builtin_cpu): Add M_INTEL_KNM.
6747 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6748 (processor_type): Add PROCESSOR_KNM.
6749 * config/i386/x86-tune.def: Add m_KNM.
6750 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6752 2017-09-20 Richard Biener <rguenther@suse.de>
6754 PR tree-optimization/80213
6755 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6756 are allowed in empty BBs as well.
6757 (canonicalize_loop_closed_ssa): Also look for other complex
6759 (scop_detection::get_sese): Include the loop-closed PHI block
6761 (scop_detection::merge_sese): Remove code adding extra blocks.
6762 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6763 (build_scops): Assert the final returned scop is invalid.
6765 2017-09-20 Richard Biener <rguenther@suse.de>
6767 PR tree-optimization/82264
6768 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6770 (vn_phi_lookup): Likewise.
6771 (vn_phi_insert): Likewise.
6773 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6775 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6776 that fits into uhwi or shwi, add DW_AT_const_value regardless
6777 of early_dwarf without going through RTL, using add_AT_unsigned
6780 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6781 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6782 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6783 (DEBUG_MACRO_SECTION): Likewise.
6784 (DEBUG_LTO_MACRO_SECTION): Likewise.
6785 (DEBUG_STR_DWO_SECTION): Likewise.
6786 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6787 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6788 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6789 (DEBUG_LTO_LINE_STR_SECTION): Define.
6790 (init_sections_and_labels): Initialize debug_line_str_section
6791 variable. Initialize debug_loc_section for -gdwarf-5 to
6792 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6794 2017-09-20 Richard Biener <rguenther@suse.de>
6796 * graphite-sese-to-poly.c (extract_affine): Properly handle
6797 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6799 2017-09-20 Richard Biener <rguenther@suse.de>
6801 PR tree-optimization/81373
6802 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6803 Force SESE live-out defs to be handled even if they are
6806 2017-09-19 Jeff Law <law@redhat.com>
6808 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6809 nothing for stack adjustments with REG_STACK_CHECK.
6810 * sched-deps.c (parse_add_or_inc): Reject insns with
6811 REG_STACK_CHECK from dependency breaking.
6812 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6813 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6814 * reg-notes.def (STACK_CHECK): New note.
6816 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6817 (ix86_expand_prologue): Dump stack clash info as needed.
6818 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6820 * function.c (dump_stack_clash_frame_info): New function.
6821 * function.h (dump_stack_clash_frame_info): Prototype.
6822 (enum stack_clash_probes): New enum.
6824 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6825 flag_stack_clash_protection.
6826 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6827 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6828 (arm_frame_pointer_required): Likewise.
6829 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6830 (ia64_expand_prologue): Likewise.
6831 * config/mips/mips.c (mips_expand_prologue): Likewise.
6832 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6833 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6834 (sparc_flat_expand_prologue): Likewise.
6835 * config/spu/spu.c (spu_expand_prologue): Likewise.
6837 * explow.c: Include "params.h".
6838 (anti_adjust_stack_and_probe_stack_clash): New function.
6839 (get_stack_check_protect): Likewise.
6840 (compute_stack_clash_protection_loop_data): Likewise.
6841 (emit_stack_clash_protection_loop_start): Likewise.
6842 (emit_stack_clash_protection_loop_end): Likewise.
6843 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6844 Use anti_adjust_stack_and_probe_stack_clash.
6845 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6846 (emit_stack_clash_protection_loop_start): Likewise.
6847 (emit_stack_clash_protection_loop_end): Likewise.
6848 * rtl.h (get_stack_check_protect): Prototype.
6849 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6850 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6851 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6853 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6855 * doc/tm.texi: Rebuilt.
6856 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6857 get_stack_check_protect.
6858 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6859 * config/arm/arm.c (arm_expand_prologue): Likewise.
6860 (arm_frame_pointer_required): Likewise.
6861 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6862 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6863 * config/mips/mips.c (mips_expand_prologue): Likewise.
6864 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6865 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6866 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6867 (sparc_flat_expand_prologue): Likewise.
6869 * common.opt (-fstack-clash-protection): New option.
6870 * flag-types.h (enum stack_check_type): Note difference between
6871 -fstack-check= and -fstack-clash-protection.
6872 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6873 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6874 * toplev.c (process_options): Issue warnings/errors for cases
6875 not handled with -fstack-clash-protection.
6876 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6877 (-fstack-check): Note additional problem with -fstack-check=generic.
6878 Note that -fstack-check is primarily for Ada and refer users
6879 to -fstack-clash-protection for stack-clash-protection.
6880 Document new params for stack clash protection.
6882 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6884 * config/i386/i386.c (ix86_split_long_move): Do not handle
6885 address used for LEA in a special way.
6887 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6889 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6891 2017-09-19 Martin Sebor <msebor@redhat.com>
6894 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6895 of incompatible types.
6897 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6899 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6900 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6901 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6902 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6904 2017-09-19 Richard Biener <rguenther@suse.de>
6906 PR tree-optimization/82244
6907 * tree-vrp.c (remove_range_assertions): Do not propagate
6908 a constant to abnormals but replace the assert with a copy.
6910 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6912 PR rtl-optimization/57878
6913 PR rtl-optimization/68988
6914 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6915 avoidance test involving non_reload_pseudos. Move frequency test
6916 below the general fragmentation avoidance test.
6918 2017-09-19 Richard Biener <rguenther@suse.de>
6920 PR tree-optimization/69728
6921 * graphite-sese-to-poly.c (schedule_error): New global.
6922 (add_loop_schedule): Handle empty domain by failing the
6924 (build_original_schedule): Handle schedule_error.
6926 2017-09-19 Richard Biener <rguenther@suse.de>
6928 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6929 Do not iterate to sibling loops but only to siblings of inner
6932 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6935 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6936 that the registers are different.
6938 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6940 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6941 to processor_model and "amdfam17h" to arch_names_table.
6942 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6944 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6947 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6949 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6950 Alan Hayward <alan.hayward@arm.com>
6951 David Sherwood <david.sherwood@arm.com>
6953 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6955 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6957 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6958 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6959 vect_schedule_slp_instance.
6960 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6961 Update call to vect_slp_analyze_node_operations. Simplify return
6963 (vect_slp_analyze_bb_1): Update call accordingly.
6964 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6965 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6966 (vect_schedule_slp): Update call accordingly.
6968 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6969 Alan Hayward <alan.hayward@arm.com>
6970 David Sherwood <david.sherwood@arm.com>
6972 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6973 with types that aren't in fact scalar.
6975 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6977 * tree-vect-slp.c (vect_record_max_nunits): New function,
6979 (vect_build_slp_tree_1): ...here.
6980 (vect_build_slp_tree_2): Call it for phis too.
6982 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6984 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6985 to vect_get_vec_def_for_operand when getting the mask operand.
6987 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6988 Alan Hayward <alan.hayward@arm.com>
6989 David Sherwood <david.sherwood@arm.com>
6991 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
6994 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6995 Alan Hayward <alan.hayward@arm.com>
6996 David Sherwood <david.sherwood@arm.com>
6998 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
6999 calculation for vector booleans.
7001 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7002 Alan Hayward <alan.hayward@arm.com>
7003 David Sherwood <david.sherwood@arm.com>
7005 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7007 (vect_transform_stmt): ...here.
7008 (vect_analyze_stmt): Use it instead of calling
7009 vectorizable_live_operation directly.
7011 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7013 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7014 non-SIMT targets in acc vector loops.
7016 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7018 * configure.ac: Add arc and check if assembler supports gdwarf2.
7019 * configure: Regenerate.
7021 2017-09-18 Richard Biener <rguenther@suse.de>
7023 PR tree-optimization/82220
7024 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7025 epilogue niters from the min_profitable_iters compute.
7027 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7030 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7031 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7032 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7034 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7037 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7038 switching to a new text section.
7040 2017-09-18 Richard Biener <rguenther@suse.de>
7042 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7044 (build_alias_set): Reject aliases with no access function.
7046 2017-09-18 Richard Biener <rguenther@suse.de>
7048 PR tree-optimization/79622
7049 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7051 (build_cross_bb_scalars_use): Likewise.
7053 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7055 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7057 2017-09-18 Alan Modra <amodra@gmail.com>
7060 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7061 stack_pointer_rtx for count 0. Update comments. Break up
7062 large rtl expression.
7064 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7066 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7067 Increase to 20 bytes.
7068 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7069 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7070 or avx version of the stub.
7072 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7075 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7076 compute the minimum stack alignment. Also update preferred stack
7077 boundary for leaf functions.
7079 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7081 PR tree-optimization/82228
7082 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7085 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7087 * common/config/nds32/nds32-common.c
7088 (nds32_option_optimization_table): Refine formatting.
7089 (nds32_option_optimization_table): Use -fsched-pressure and
7090 -fomit-frame-pointer for specific optimization level.
7092 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7094 * config/nds32/nds32.c: Refine formatting and comments.
7095 * config/nds32/nds32.h: Likewise.
7096 * config/nds32/nds32.md: Likewise.
7097 * config/nds32/nds32-cost.c: Likewise.
7098 * config/nds32/nds32-isr.c: Likewise.
7099 * config/nds32/nds32-md-auxiliary.c: Likewise.
7100 * config/nds32/nds32-multiple.md: Likewise.
7101 * config/nds32/nds32-predicates.c: Likewise.
7103 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7104 Jakub Jelinek <jakub@redhat.com>
7106 Add support for -std=c++2a.
7107 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7109 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7111 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7114 * doc/extend.texi (Common Function Attributes): Add
7115 references to ARM, AArch64, and S/390 specific attributes.
7116 (Function Specific Option Pragmas): Add AArch64 and S/390
7117 to list of back ends that support the target pragma.
7119 2017-09-15 Nathan Sidwell <nathan@acm.org>
7121 * doc/standards.texi: Fix C++17 description. Update URLs for
7124 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7126 * common.opt (Wcast-align=strict): New warning option.
7127 * doc/invoke.texi: Document -Wcast-align=strict.
7129 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7131 * cgraph.h (cgraph_thunk_info): Add comments.
7132 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7133 assert for VIRTUAL_* arguments stricter.
7135 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7137 PR tree-optimization/71026
7138 * match.pd: Move RDIV patterns from fold-const.c
7139 * fold-const.c (distribute_real_division): Removed.
7140 (fold_binary_loc): Remove calls to distribute_real_divison.
7142 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7144 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7145 c++1z and gnu++1z as deprecated. Change other references to
7146 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7147 Change -Wc++1z-compat to -Wc++17-compat.
7148 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7149 * dwarf2out.c (highest_c_language): Handle C++17.
7150 (gen_compile_unit_die): Likewise.
7152 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7154 PR rtl-optimization/82192
7155 * combine.c (make_extraction): Don't look through non-paradoxical
7156 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7159 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7160 Alan Hayward <alan.hayward@arm.com>
7161 David Sherwood <david.sherwood@arm.com>
7163 * target.def (function_arg_offset): New hook.
7164 * targhooks.h (default_function_arg_offset): Declare.
7165 * targhooks.c (default_function_arg_offset): New function.
7166 * function.c (locate_and_pad_parm): Use
7167 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7168 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7169 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7170 * doc/tm.texi: Regenerate.
7171 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7172 * config/spu/spu.c (spu_function_arg_offset): New function.
7173 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7174 * system.h (FUNCTION_ARG_OFFSET): Poison.
7176 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7177 Alan Hayard <alan.hayward@arm.com>
7178 David Sherwood <david.sherwood@arm.com>
7180 * target.def (truly_noop_truncation): New hook.
7181 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7182 than TRULY_NOOP_TRUNCATION.
7183 * hooks.h (hook_bool_uint_uint_true): Declare.
7184 * hooks.c (hook_bool_uint_uint_true): New function.
7185 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7186 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7187 * doc/tm.texi: Regenerate.
7188 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7189 rather than TRULY_NOOP_TRUNCATION in comments.
7190 (simplify_comparison): Likewise.
7191 (record_truncated_value): Likewise.
7192 * expmed.c (extract_bit_field_1): Likewise.
7193 (extract_split_bit_field): Likewise.
7194 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7195 instead of TRULY_NOOP_TRUNCATION.
7196 * function.c (assign_parm_setup_block): Likewise.
7197 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7198 * rtlhooks.c: Include target.h.
7199 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7200 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7201 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7202 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7203 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7204 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7205 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7206 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7207 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7208 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7209 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7210 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7211 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7212 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7213 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7214 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7215 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7216 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7217 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7218 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7219 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7220 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7221 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7222 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7223 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7224 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7225 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7226 rather than TRULY_NOOP_TRUNCATION in comments.
7227 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7228 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7229 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7230 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7231 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7232 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7233 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7234 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7235 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7236 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7237 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7238 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7239 rather than TRULY_NOOP_TRUNCATION in comments.
7240 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7241 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7242 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7243 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7244 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7245 TRULY_NOOP_TRUNCATION condition.
7246 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7247 (TRULY_NOOP_TRUNCATION): Delete.
7248 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7249 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7250 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7251 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7252 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7253 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7254 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7255 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7256 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7257 rather than TRULY_NOOP_TRUNCATION in comments.
7258 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7259 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7260 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7261 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7262 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7263 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7265 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7268 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7269 (*cmp_ior): Likewise.
7270 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7271 (*ior_scc_scc_cmp): Likewise.
7272 (*and_scc_scc): Likewise.
7273 (*and_scc_scc_cmp): Likewise.
7275 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7276 Alan Hayard <alan.hayward@arm.com>
7277 David Sherwood <david.sherwood@arm.com>
7279 * target.def (can_change_mode_class): New hook.
7280 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7281 (hard_regno_nregs): Likewise.
7282 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7283 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7284 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7285 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7286 (LOAD_EXTEND_OP): Update accordingly.
7287 * doc/tm.texi: Regenerate.
7288 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7289 CANNOT_CHANGE_MODE_CLASS.
7290 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7291 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7292 * combine.c (simplify_set): Update accordingly.
7293 * emit-rtl.c (validate_subreg): Likewise.
7294 * recog.c (general_operand): Likewise.
7295 * regcprop.c (mode_change_ok): Likewise.
7296 * reload1.c (choose_reload_regs): Likewise.
7297 (inherit_piecemeal_p): Likewise.
7298 * rtlanal.c (simplify_subreg_regno): Likewise.
7299 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7300 instead of CANNOT_CHANGE_MODE_CLASS.
7301 (reload_cse_simplify_operands): Likewise.
7302 * reload.c (push_reload): Use targetm.can_change_mode_class
7303 instead of CANNOT_CHANGE_MODE_CLASS.
7304 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7305 REG_CANNOT_CHANGE_MODE_P.
7306 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7307 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7308 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7309 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7310 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7311 (arm_can_change_mode_class): New function.
7312 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7313 than CANNOT_CHANGE_MODE_CLASS in comments.
7314 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7315 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7316 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7317 (ix86_can_change_mode_class): ...this new function, inverting the
7318 sense of the return value.
7319 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7320 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7321 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7322 (ia64_can_change_mode_class): New function.
7323 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7324 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7325 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7326 (m32c_can_change_mode_class): ...this new function, inverting the
7327 sense of the return value.
7328 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7329 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7330 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7331 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7332 (mips_can_change_mode_class): ...this new function, inverting the
7333 sense of the return value.
7334 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7335 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7336 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7337 (msp430_can_change_mode_class): New function.
7338 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7339 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7340 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7341 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7342 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7343 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7344 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7345 (pa_cannot_change_mode_class): Replace with...
7346 (pa_can_change_mode_class): ...this new function, inverting the
7347 sense of the return value.
7348 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7349 than CANNOT_CHANGE_MODE_CLASS in comments.
7350 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7351 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7352 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7353 (pdp11_cannot_change_mode_class): Replace with...
7354 (pdp11_can_change_mode_class): ...this new function, inverting the
7355 sense of the return value.
7356 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7357 * config/powerpcspe/powerpcspe-protos.h
7358 (rs6000_cannot_change_mode_class_ptr): Delete.
7359 * config/powerpcspe/powerpcspe.c
7360 (rs6000_cannot_change_mode_class_ptr): Delete.
7361 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7362 (rs6000_option_override_internal): Assign to
7363 targetm.can_change_mode_class instead of
7364 rs6000_cannot_change_mode_class_ptr.
7365 (rs6000_cannot_change_mode_class): Replace with...
7366 (rs6000_can_change_mode_class): ...this new function, inverting the
7367 sense of the return value.
7368 (rs6000_debug_cannot_change_mode_class): Replace with...
7369 (rs6000_debug_can_change_mode_class): ...this new function.
7370 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7371 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7372 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7373 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7374 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7376 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7377 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7378 (rs6000_option_override_internal): Assign to
7379 targetm.can_change_mode_class instead of
7380 rs6000_cannot_change_mode_class_ptr.
7381 (rs6000_cannot_change_mode_class): Replace with...
7382 (rs6000_can_change_mode_class): ...this new function, inverting the
7383 sense of the return value.
7384 (rs6000_debug_cannot_change_mode_class): Replace with...
7385 (rs6000_debug_can_change_mode_class): ...this new function.
7386 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7387 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7388 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7389 (s390_can_change_mode_class): ...this new function, inverting the
7390 sense of the return value.
7391 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7392 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7393 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7394 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7395 (sh_cannot_change_mode_class): Replace with...
7396 (sh_can_change_mode_class): ...this new function, inverting the
7397 sense of the return value.
7398 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7399 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7400 (sparc_can_change_mode_class): New function.
7401 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7402 * config/spu/spu.c (spu_can_change_mode_class): New function.
7403 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7404 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7405 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7406 (visium_can_change_mode_class): New function.
7407 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7409 2017-09-15 Richard Biener <rguenther@suse.de>
7411 PR tree-optimization/82217
7412 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7413 but not undefined case.
7415 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7418 * postreload.c (reload_cse_simplify_operands): Skip
7419 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7421 2017-09-15 Richard Biener <rguenther@suse.de>
7423 PR tree-optimization/68823
7424 * graphite-scop-detection.c (build_alias_set): If we have a
7425 possible dependence check whether we can handle them by just
7426 looking at the DRs DR_ACCESS_FNs.
7427 (build_scops): If build_alias_set fails, fail the SCOP.
7429 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7431 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7432 to support float128 built-in functions that require the ISA 3.0
7434 (BU_FLOAT128_3_HW): Likewise.
7435 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7437 (FMAF128): Likewise.
7439 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7440 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7441 floating point instructions.
7442 (rs6000_invalid_builtin): Likewise.
7443 (rs6000_builtin_mask_names): Likewise.
7444 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7445 (RS6000_BTM_FLOAT128_HW): Likewise.
7446 (RS6000_BTM_COMMON): Likewise.
7447 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7449 * doc/extend.texi (RS/6000 built-in functions): Document the
7450 IEEE 128-bit floating point square root and fused multiply-add
7453 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7455 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7456 reg (r2) isn't in the set of registers defined in the prologue.
7458 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7459 Alan Hayward <alan.hayward@arm.com>
7460 David Sherwood <david.sherwood@arm.com>
7462 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7463 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7464 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7465 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7466 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7468 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7469 max_vectorization_factor.
7470 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7472 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7473 Alan Hayward <alan.hayward@arm.com>
7474 David Sherwood <david.sherwood@arm.com>
7476 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7477 (vect_worthwhile_without_simd_p): Declare.
7478 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7479 (vectorizable_reduction): Use it.
7480 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7481 (vectorizable_operation): Likewise.
7483 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7484 Alan Hayward <alan.hayward@arm.com>
7485 David Sherwood <david.sherwood@arm.com>
7487 * tree-vectorizer.h (vect_get_num_copies): New function.
7488 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7489 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7490 (vectorizable_induction): Likewise.
7491 (vectorizable_live_operation): Likewise.
7492 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7493 (vectorizable_bswap): Likewise.
7494 (vectorizable_call): Likewise.
7495 (vectorizable_conversion): Likewise.
7496 (vectorizable_assignment): Likewise.
7497 (vectorizable_shift): Likewise.
7498 (vectorizable_operation): Likewise.
7499 (vectorizable_store): Likewise.
7500 (vectorizable_load): Likewise.
7501 (vectorizable_condition): Likewise.
7502 (vectorizable_comparison): Likewise.
7503 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7505 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7506 Alan Hayward <alan.hayward@arm.com>
7507 David Sherwood <david.sherwood@arm.com>
7509 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7510 of vect_init_vector.
7512 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7513 Alan Hayward <alan.hayward@arm.com>
7514 David Sherwood <david.sherwood@arm.com>
7516 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7517 an inline wrapper that provides a location.
7518 (gimple_build_vector): Likewise.
7519 * gimple-fold.c (gimple_build_vector_from_val): New function.
7520 (gimple_build_vector): Likewise.
7521 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7522 functions to build the initial value. Always return a gimple value.
7523 (get_initial_defs_for_reduction): Likewise. Only compute
7525 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7526 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7527 (vectorizable_induction): Use gimple_build_vector rather than
7530 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7531 Alan Hayward <alan.hayward@arm.com>
7532 David Sherwood <david.sherwood@arm.com>
7534 * target.h (vec_perm_indices): New typedef.
7535 (auto_vec_perm_indices): Likewise.
7536 * optabs-query.h: Include target.h
7537 (can_vec_perm_p): Take a vec_perm_indices *.
7538 * optabs-query.c (can_vec_perm_p): Likewise.
7539 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7540 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7541 * tree-vect-generic.c (lower_vec_perm): Likewise.
7542 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7543 (vect_grouped_load_supported): Likewise.
7544 (vect_shift_permute_load_chain): Likewise.
7545 (vect_permute_store_chain): Use auto_vec_perm_indices.
7546 (vect_permute_load_chain): Likewise.
7547 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7548 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7549 Update uses of can_vec_perm_p.
7550 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7551 mode with a number of elements. Take a vec_perm_indices *.
7552 (vect_create_epilog_for_reduction): Update accordingly.
7553 Use auto_vec_perm_indices.
7554 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7555 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7556 (vect_transform_slp_perm_load): Likewise.
7557 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7558 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7559 (vect_gen_perm_mask_checked): Likewise.
7560 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7561 (vect_gen_perm_mask_checked): Likewise.
7562 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7563 (vectorizable_store): Likewise.
7564 (vectorizable_load): Likewise.
7565 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7566 (vectorizable_bswap): Likewise.
7568 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7569 Alan Hayward <alan.hayward@arm.com>
7570 David Sherwood <david.sherwood@arm.com>
7572 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7573 * tree.c (build_vector): Likewise.
7574 (build_vector_from_ctor): Update accordingly.
7575 (build_vector_from_val): Likewise.
7576 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7577 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7578 * tree-vect-generic.c (add_rshift): Likewise.
7579 (expand_vector_divmod): Likewise.
7580 (optimize_vector_constructor): Likewise.
7581 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7582 (vect_transform_slp_perm_load): Likewise.
7583 (vect_schedule_slp_instance): Likewise.
7584 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7585 (vectorizable_call): Likewise.
7586 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7587 * expmed.c (make_tree): Likewise.
7588 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7589 a vector passed to build_vector.
7590 (fold_convert_const): Likewise.
7591 (exact_inverse): Likewise.
7592 (fold_ternary_loc): Likewise.
7593 (fold_relational_const): Likewise.
7594 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7595 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7596 (const_unop): Likewise. Store the reduction accumulator in a
7597 variable rather than an array.
7598 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7599 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7600 the new vector, rather than constructing it after the input arrays.
7601 (native_interpret_vector): Use auto_vec<tree> when building
7602 a vector passed to build_vector. Add elements in order.
7603 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7604 auto_vec<tree> when building a vector passed to build_vector.
7605 (vect_create_epilog_for_reduction): Likewise.
7606 (vectorizable_induction): Likewise.
7607 (get_initial_def_for_reduction): Likewise. Fix indentation of
7609 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7611 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7612 passed to build_vector.
7614 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7615 Alan Hayward <alan.hayward@arm.com>
7616 David Sherwood <david.sherwood@arm.com>
7618 * tree-core.h (tree_base::u): Add an "nelts" field.
7619 (tree_vector): Use VECTOR_CST_NELTS as the length.
7620 * tree.c (tree_size): Likewise.
7621 (make_vector): Initialize VECTOR_CST_NELTS.
7622 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7623 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7624 TYPE_VECTOR_SUBPARTS.
7625 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7626 as the number of units, setting it from VECTOR_CST_NELTS.
7627 (const_vector_from_tree): Likewise.
7628 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7629 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7630 (fold_negate_expr_1): Likewise.
7631 (fold_convert_const): Likewise.
7632 (const_binop): Likewise. Differentiate the number of output and
7634 (const_unop): Likewise.
7635 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7636 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7637 in cases that did the opposite.
7639 2017-09-14 Richard Biener <rguenther@suse.de>
7641 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7644 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7646 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7648 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7651 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7652 if and where to split a bb, except for splitting before debug insn
7653 sequences followed by non-label real insn. Delete debug insns
7654 in between basic blocks.
7656 * combine.c (make_compound_operation_int): Formatting fixes.
7658 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7659 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7660 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7661 * config/sol2.h (LINK_EH_SPEC): Likewise.
7662 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7663 * config/s390/linux.h (LINK_SPEC): Likewise.
7664 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7665 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7666 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7667 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7668 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7669 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7670 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7671 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7673 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7675 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7677 (ENDFILE_LINUX_SPEC): Likewise.
7678 (LINK_EH_SPEC): Likewise.
7679 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7680 (LINK_OS_LINUX_SPEC32): Likewise.
7681 (LINK_OS_LINUX_SPEC64): Likewise.
7682 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7683 (LINK_OS_LINUX_SPEC): Likewise.
7685 2017-09-13 Martin Liska <mliska@suse.cz>
7688 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7689 CASE_HIGH is NULL_TREE.
7691 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7692 Alan Hayward <alan.hayward@arm.com>
7693 David Sherwood <david.sherwood@arm.com>
7695 * target.def (secondary_memory_needed): New hook.
7696 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7697 instead of SECONDARY_MEMORY_NEEDED.
7698 (secondary_memory_needed_mode): Likewise.
7699 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7700 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7701 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7702 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7703 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7704 * doc/tm.texi: Regenerate.
7705 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7706 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7707 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7708 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7709 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7710 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7711 mode argument first and change the reg_class arguments to reg_class_t.
7712 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7713 Make static. Update the call to inline_secondary_memory_needed.
7714 (ix86_register_move_cost): Update the call to
7715 inline_secondary_memory_needed.
7716 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7717 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7719 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7720 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7722 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7723 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7724 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7725 and match hook interface. Add comment from mips.h.
7726 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7727 * config/mmix/mmix.md (truncdfsf2): Refer to
7728 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7730 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7731 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7732 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7733 (pa_secondary_memory_needed): New function.
7734 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7735 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7736 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7737 (pdp11_secondary_memory_needed): Make static and match hook interface.
7738 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7739 * config/powerpcspe/powerpcspe-protos.h
7740 (rs6000_secondary_memory_needed_ptr): Delete.
7741 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7743 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7744 (rs6000_option_override_internal): Assign to
7745 targetm.secondary_memory_needed rather than
7746 rs6000_secondary_memory_needed_ptr.
7747 (rs6000_secondary_memory_needed): Match hook interface.
7748 (rs6000_debug_secondary_memory_needed): Likewise.
7749 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7750 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7751 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7752 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7753 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7754 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7756 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7757 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7758 (rs6000_option_override_internal): Assign to
7759 targetm.secondary_memory_needed rather than
7760 rs6000_secondary_memory_needed_ptr.
7761 (rs6000_secondary_memory_needed): Match hook interface.
7762 (rs6000_debug_secondary_memory_needed): Likewise.
7763 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7764 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7765 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7766 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7767 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7768 (sparc_secondary_memory_needed): New function.
7769 * lra-constraints.c (check_and_process_move): Refer to
7770 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7772 (curr_insn_transform): Likewise.
7773 (process_alt_operands): Use targetm.secondary_memory_needed
7774 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7775 (check_secondary_memory_needed_p): Likewise.
7776 (choose_split_class): Likewise.
7777 * reload.c: Unconditionally include code that was previously
7778 conditional on SECONDARY_MEMORY_NEEDED.
7779 (push_secondary_reload): Use targetm.secondary_memory_needed
7780 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7781 (push_reload): Likewise.
7782 * reload1.c: Unconditionally include code that was previously
7783 conditional on SECONDARY_MEMORY_NEEDED.
7784 (choose_reload_regs): Use targetm.secondary_memory_needed
7785 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7786 (gen_reload): Likewise.
7787 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7789 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7790 Alan Hayward <alan.hayward@arm.com>
7791 David Sherwood <david.sherwood@arm.com>
7793 * target.def (secondary_memory_needed_mode): New hook:
7794 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7795 * targhooks.h (default_secondary_memory_needed_mode): New function.
7796 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7797 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7798 * doc/tm.texi: Regenerate.
7799 * lra-constraints.c (check_and_process_move): Use
7800 targetm.secondary_memory_needed_mode instead of
7801 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7802 (curr_insn_transform): Likewise.
7803 * reload.c (get_secondary_mem): Likewise.
7804 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7805 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7807 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7808 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7809 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7810 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7811 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7813 * config/powerpcspe/powerpcspe-protos.h
7814 (rs6000_secondary_memory_needed_mode): Delete.
7815 * config/powerpcspe/powerpcspe.c
7816 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7817 (rs6000_secondary_memory_needed_mode): Make static.
7818 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7819 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7821 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7823 (rs6000_secondary_memory_needed_mode): Make static.
7824 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7825 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7826 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7827 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7828 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7830 (sparc_secondary_memory_needed_mode): New function.
7831 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7833 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7835 * config/aarch64/constraints.md (Umq): New constraint.
7836 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7838 (mov<mode>): Update condition.
7840 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7842 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7843 when bitposition is the same.
7845 2017-09-13 Richard Biener <rguenther@suse.de>
7847 * dwarf2out.c (output_die_symbol): Remove.
7848 (output_die): Do not output a DIEs symbol.
7850 2017-09-13 Richard Biener <rguenther@suse.de>
7853 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7854 default-def to avoid breaking iterator update with the weird
7855 interaction with cgraph_update_edges_for_call_stmt_node.
7857 2017-09-13 Richard Biener <rguenther@suse.de>
7859 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7860 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7861 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7862 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7864 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7866 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7867 Disable pc relative literal load irrespective of
7868 TARGET_FIX_ERR_A53_84341 for default.
7870 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7872 * config/sparc/sparc.c (output_return): Output the source location of
7873 the insn in the delay slot, if any.
7874 (output_sibcall): Likewise.
7876 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7879 * common.opt (-static-pie): New alias.
7880 (shared): Negate static-pie.
7881 (-no-pie): Update help text.
7883 (static-pie): New option.
7884 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7885 -static-pie support.
7886 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7887 (LINK_EH_SPEC): Likewise.
7888 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7889 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7890 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7891 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7892 (init_gcc_specs): Likewise.
7893 (init_spec): Likewise.
7894 (display_help): Update help message for -pie.
7895 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7898 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7900 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7901 (movdi_aarch64): Likewise.
7902 (movti_aarch64): Likewise.
7904 2017-09-12 Simon Wright <simon@pushface.org>
7907 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7908 calculation of the minor version, always output as 0.
7910 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7913 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7914 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7915 on it early, rather than manual conversion late. For
7916 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7917 instead of performing manual conversion.
7919 2017-09-12 Carl Love <cel@us.ibm.com>
7921 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7922 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7924 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7925 VMULOSW): Add definitions.
7926 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7927 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7928 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7929 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7930 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7932 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7934 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7936 (movti_aarch64): Likewise.
7937 (movdf_aarch64): Likewise.
7938 (movtf_aarch64): Likewise.
7939 (load_pairdi): Likewise.
7940 (store_pairdi): Likewise.
7941 (load_pairdf): Likewise.
7942 (store_pairdf): Likewise.
7943 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7944 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7945 (ldr_got_small_<mode>): Likewise.
7946 (ldr_got_small_28k_<mode>): Likewise.
7947 (ldr_got_tiny): Likewise.
7948 * config/aarch64/iterators.md (ldst_sz): New.
7949 (ldpstp_sz): Likewise.
7950 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7952 (thunderx_load): Split load_8 to load_16.
7953 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7955 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7956 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7957 (xgene1_store_pair): Split store_8 to store_16.
7958 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7959 (falkor_st_0_st_sd): Split store_8 to store_16.
7961 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7963 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7964 and store1/2/3/4 to store_4/8/12/16.
7965 * config/aarch64/aarch64.md: Update for rename.
7966 * config/arm/arm.md: Likewise.: Likewise.
7967 * config/arm/arm.c: Likewise.
7968 * config/arm/thumb1.md: Likewise.
7969 * config/arm/thumb2.md: Likewise.
7970 * config/arm/vfp.md: Likewise.
7971 * config/arm/arm-generic.md: Likewise.
7972 * config/arm/arm1020e.md: Likewise.
7973 * config/arm/arm1026ejs.md: Likewise.
7974 * config/arm/arm1136jfs.md: Likewise.
7975 * config/arm/arm926ejs.md: Likewise.
7976 * config/arm/cortex-a15.md: Likewise.
7977 * config/arm/cortex-a17.md: Likewise.
7978 * config/arm/cortex-a5.md: Likewise.
7979 * config/arm/cortex-a53.md: Likewise.
7980 * config/arm/cortex-a57.md: Likewise.
7981 * config/arm/cortex-a7.md: Likewise.
7982 * config/arm/cortex-a8.md: Likewise.
7983 * config/arm/cortex-a9.md: Likewise.
7984 * config/arm/cortex-m4.md: Likewise.
7985 * config/arm/cortex-m7.md: Likewise.
7986 * config/arm/cortex-r4.md: Likewise.
7987 * config/arm/exynos-m1.md: Likewise.
7988 * config/arm/fa526.md: Likewise.
7989 * config/arm/fa606te.md: Likewise.
7990 * config/arm/fa626te.md: Likewise.
7991 * config/arm/fa726te.md: Likewise.
7992 * config/arm/fmp626.md: Likewise.
7993 * config/arm/iwmmxt.md: Likewise.
7994 * config/arm/ldmstm.md: Likewise.
7995 * config/arm/marvell-pj4.md: Likewise.
7996 * config/arm/xgene1.md: Likewise.
7997 * config/aarch64/thunderx.md: Likewise.
7998 * config/aarch64/thunderx2t99.md: Likewise.
7999 * config/aarch64/falkor.md: Likewise.
8001 2017-09-12 Martin Liska <mliska@suse.cz>
8003 * attribs.c (private_lookup_attribute): New function.
8004 * attribs.h (private_lookup_attribute): Declared here.
8005 (lookup_attribute): Called from this place.
8007 2017-09-12 Richard Biener <rguenther@suse.de>
8009 PR tree-optimization/82157
8010 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8011 stmts with side-effects.
8013 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8014 Alan Hayward <alan.hayward@arm.com>
8015 David Sherwood <david.sherwood@arm.com>
8017 * target.def (hard_regno_nregs): New hook.
8018 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8019 * targhooks.h (default_hard_regno_nregs): Declare.
8020 * targhooks.c (default_hard_regno_nregs): New function.
8021 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8022 (TARGET_HARD_REGNO_NREGS): ...this hook.
8023 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8024 (CLASS_MAX_NREGS): Likewise.
8025 * doc/tm.texi: Regenerate.
8026 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8027 instead of HARD_REGNO_NREGS.
8028 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8029 HARD_REGNO_NREGS in the comment.
8030 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8031 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8032 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8033 Return an unsigned int.
8034 (TARGET_HARD_REGNO_NREGS): Redefine.
8035 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8036 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8037 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8038 (arc_hard_regno_nregs): New function.
8039 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8040 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8041 (arm_hard_regno_nregs): New function.
8042 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8043 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8044 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8045 (TARGET_HARD_REGNO_NREGS): Redefine.
8046 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8047 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8048 (HARD_REGNO_NREGS): Delete.
8049 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8050 (cr16_hard_regno_nregs): New function.
8051 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8052 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8053 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8054 (cris_hard_regno_nregs): New function.
8055 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8056 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8057 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8058 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8059 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8060 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8061 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8062 (frv_hard_regno_nregs): Make static. Take and return an
8064 (frv_class_max_nregs): Remove outdated copy of documentation.
8065 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8066 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8067 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8068 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8069 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8070 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8071 (TARGET_HARD_REGNO_NREGS): Redefine.
8072 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8073 (CLASS_MAX_NREGS): Update comment.
8074 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8075 (ia64_hard_regno_nregs): New function.
8076 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8077 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8078 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8079 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8080 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8082 (m32c_hard_regno_nregs): Likewise. Make static.
8083 (TARGET_HARD_REGNO_NREGS): Redefine.
8084 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8085 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8086 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8087 (m68k_hard_regno_nregs): New function.
8088 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8089 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8090 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8091 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8092 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8093 Take and return an unsigned int.
8094 (TARGET_HARD_REGNO_NREGS): Redefine.
8095 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8096 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8097 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8098 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8099 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8100 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8101 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8102 (msp430_hard_regno_nregs): Make static. Take and return an
8104 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8105 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8106 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8107 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8108 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8109 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8110 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8111 (TARGET_HARD_REGNO_NREGS): Redefine.
8112 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8113 (PA_HARD_REGNO_NREGS): ...this.
8114 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8115 (PA_HARD_REGNO_NREGS): ...this.
8116 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8117 (pa_hard_regno_nregs): New function.
8118 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8119 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8120 (pdp11_hard_regno_nregs): New function.
8121 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8122 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8123 (rs6000_hard_regno_nregs_hook): New function.
8124 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8125 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8126 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8127 Take and return an unsigned int. Move earlier in file.
8128 (TARGET_HARD_REGNO_NREGS): Redefine.
8129 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8130 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8131 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8132 (rl78_hard_regno_nregs): Make static. Take and return an
8134 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8135 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8136 (rs6000_hard_regno_nregs_hook): New function.
8137 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8138 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8139 (TARGET_HARD_REGNO_NREGS): Redefine.
8140 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8141 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8142 instead of HARD_REGNO_NREGS.
8143 (s390_hard_regno_nregs): New function.
8144 (s390_hard_regno_mode_ok): Add comment from s390.h.
8145 (TARGET_HARD_REGNO_NREGS): Redefine.
8146 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8147 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8148 (sh_hard_regno_nregs): New function.
8149 (sh_pass_in_reg_p): Use it.
8150 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8151 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8152 (sparc_hard_regno_nregs): New function.
8153 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8154 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8155 (spu_function_arg_advance): Use it, supplying a valid register number.
8156 (TARGET_HARD_REGNO_NREGS): Redefine.
8157 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8158 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8159 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8160 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8161 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8162 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8163 (CLASS_MAX_NREGS): Remove copy of old documentation.
8164 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8165 (visium_hard_regno_nregs): New function.
8166 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8167 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8168 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8169 xtensa_hard_regno_nregs): New function.
8170 * system.h (HARD_REGNO_NREGS): Poison.
8172 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8174 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8175 hard_regno_nregs instead of HARD_REGNO_NREGS.
8176 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8177 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8178 (c6x_expand_epilogue): Likewise.
8179 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8180 (frv_read_iacc_argument): Likewise.
8181 * config/sh/sh.c: Include regs.h.
8182 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8183 (regs_used): Likewise.
8184 (output_stack_adjust): Likewise.
8185 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8186 * expmed.c: Include regs.h.
8187 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8188 * ree.c: Include regs.h.
8189 (combine_reaching_defs): Use hard_regno_nregs instead of
8191 (add_removable_extension): Likewise.
8193 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8195 * regs.h (hard_regno_nregs): Turn into a function.
8196 (end_hard_regno): Update accordingly.
8197 * caller-save.c (setup_save_areas): Likewise.
8198 (save_call_clobbered_regs): Likewise.
8199 (replace_reg_with_saved_mem): Likewise.
8200 (insert_restore): Likewise.
8201 (insert_save): Likewise.
8202 * combine.c (can_change_dest_mode): Likewise.
8203 (move_deaths): Likewise.
8204 (distribute_notes): Likewise.
8205 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8206 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8207 (rs6000_split_multireg_move): Likewise.
8208 (rs6000_register_move_cost): Likewise.
8209 (rs6000_memory_move_cost): Likewise.
8210 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8211 (rs6000_split_multireg_move): Likewise.
8212 (rs6000_register_move_cost): Likewise.
8213 (rs6000_memory_move_cost): Likewise.
8214 * cselib.c (cselib_reset_table): Likewise.
8215 (cselib_lookup_1): Likewise.
8216 * emit-rtl.c (set_mode_and_regno): Likewise.
8217 * function.c (aggregate_value_p): Likewise.
8218 * ira-color.c (setup_profitable_hard_regs): Likewise.
8219 (check_hard_reg_p): Likewise.
8220 (calculate_saved_nregs): Likewise.
8221 (assign_hard_reg): Likewise.
8222 (improve_allocation): Likewise.
8223 (calculate_spill_cost): Likewise.
8224 * ira-emit.c (modify_move_list): Likewise.
8225 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8226 (ira_hard_reg_in_set_p): Likewise.
8227 * ira.c (setup_reg_mode_hard_regset): Likewise.
8228 (clarify_prohibited_class_mode_regs): Likewise.
8229 (check_allocation): Likewise.
8230 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8231 (lra_setup_reg_renumber): Likewise.
8232 (setup_try_hard_regno_pseudos): Likewise.
8233 (spill_for): Likewise.
8234 (assign_hard_regno): Likewise.
8235 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8236 * lra-constraints.c (in_class_p): Likewise.
8237 (lra_constraint_offset): Likewise.
8238 (simplify_operand_subreg): Likewise.
8239 (lra_constraints): Likewise.
8240 (split_reg): Likewise.
8241 (split_if_necessary): Likewise.
8242 (invariant_p): Likewise.
8243 (inherit_in_ebb): Likewise.
8244 * lra-lives.c (process_bb_lives): Likewise.
8245 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8246 (get_hard_regs): Likewise.
8247 (do_remat): Likewise.
8248 * lra-spills.c (assign_spill_hard_regs): Likewise.
8249 * mode-switching.c (create_pre_exit): Likewise.
8250 * postreload.c (reload_combine_recognize_pattern): Likewise.
8251 * recog.c (peep2_find_free_register): Likewise.
8252 * regcprop.c (kill_value_regno): Likewise.
8253 (set_value_regno): Likewise.
8254 (copy_value): Likewise.
8255 (maybe_mode_change): Likewise.
8256 (find_oldest_value_reg): Likewise.
8257 (copyprop_hardreg_forward_1): Likewise.
8258 * regrename.c (check_new_reg_p): Likewise.
8259 (regrename_do_replace): Likewise.
8260 * reload.c (push_reload): Likewise.
8261 (combine_reloads): Likewise.
8262 (find_dummy_reload): Likewise.
8263 (operands_match_p): Likewise.
8264 (find_reloads): Likewise.
8265 (find_equiv_reg): Likewise.
8266 (reload_adjust_reg_for_mode): Likewise.
8267 * reload1.c (count_pseudo): Likewise.
8268 (count_spilled_pseudo): Likewise.
8269 (find_reg): Likewise.
8270 (clear_reload_reg_in_use): Likewise.
8271 (free_for_value_p): Likewise.
8272 (allocate_reload_reg): Likewise.
8273 (choose_reload_regs): Likewise.
8274 (reload_adjust_reg_for_temp): Likewise.
8275 (emit_reload_insns): Likewise.
8276 (delete_output_reload): Likewise.
8277 * rtlanal.c (subreg_get_info): Likewise.
8278 * sched-deps.c (sched_analyze_reg): Likewise.
8279 * sel-sched.c (init_regs_for_mode): Likewise.
8280 (mark_unavailable_hard_regs): Likewise.
8281 (choose_best_reg_1): Likewise.
8282 (verify_target_availability): Likewise.
8283 * valtrack.c (dead_debug_insert_temp): Likewise.
8284 * var-tracking.c (track_loc_p): Likewise.
8285 (emit_note_insn_var_location): Likewise.
8286 * varasm.c (make_decl_rtl): Likewise.
8287 * reginfo.c (choose_hard_reg_mode): Likewise.
8288 (init_reg_modes_target): Refer directly to
8289 this_target_regs->x_hard_regno_nregs.
8291 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8293 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8294 instead of hard_regno_nregs.
8296 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8298 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8299 end_hard_regno instead of hard_regno_nregs.
8300 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8301 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8302 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8303 * ira-color.c (improve_allocation): Likewise.
8304 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8305 * lra-lives.c (mark_regno_live): Likewise.
8306 (mark_regno_dead): Likewise.
8307 * lra-remat.c (operand_to_remat): Likewise.
8308 * lra.c (collect_non_operand_hard_regs): Likewise.
8309 * postreload.c (reload_combine_note_store): Likewise.
8310 (move2add_valid_value_p): Likewise.
8311 * reload.c (regno_clobbered_p): Likewise.
8313 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8315 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8317 * config/v850/v850.c (v850_reorg): Likewise.
8318 * reload.c (refers_to_regno_for_reload_p): Likewise.
8319 (find_equiv_reg): Likewise.
8320 * reload1.c (reload_reg_reaches_end_p): Likewise.
8322 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8324 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8326 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8327 * config/arm/arm.c (output_move_neon): Likewise.
8328 (arm_attr_length_move_neon): Likewise.
8329 (neon_split_vcombine): Likewise.
8330 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8331 (c6x_mark_reg_written): Likewise.
8332 (c6x_dwarf_register_span): Likewise.
8333 * config/i386/i386.c (ix86_save_reg): Likewise.
8334 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8335 (rws_access_reg): Likewise.
8336 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8337 * mode-switching.c (create_pre_exit): Likewise.
8338 * ree.c (combine_reaching_defs): Likewise.
8339 (add_removable_extension): Likewise.
8340 * regcprop.c (find_oldest_value_reg): Likewise.
8341 (copyprop_hardreg_forward_1): Likewise.
8342 * reload.c (reload_inner_reg_of_subreg): Likewise.
8343 (push_reload): Likewise.
8344 (combine_reloads): Likewise.
8345 (find_dummy_reload): Likewise.
8346 (reload_adjust_reg_for_mode): Likewise.
8347 * reload1.c (find_reload_regs): Likewise.
8348 (forget_old_reloads_1): Likewise.
8349 (reload_reg_free_for_value_p): Likewise.
8350 (reload_adjust_reg_for_temp): Likewise.
8351 (emit_reload_insns): Likewise.
8352 (delete_output_reload): Likewise.
8353 * sel-sched.c (choose_best_reg_1): Likewise.
8354 (choose_best_pseudo_reg): Likewise.
8356 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8357 Alan Hayward <alan.hayward@arm.com>
8358 David Sherwood <david.sherwood@arm.com>
8360 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8361 * target.def (slow_unaligned_access): New hook.
8362 * targhooks.h (default_slow_unaligned_access): Declare.
8363 * targhooks.c (default_slow_unaligned_access): New function.
8364 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8365 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8366 * doc/tm.texi: Regenerate.
8367 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8368 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8369 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8371 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8372 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8374 (rs6000_slow_unaligned_access): New function.
8375 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8376 (expand_block_compare): Likewise.
8377 (expand_strn_compare): Likewise.
8378 (rs6000_rtx_costs): Likewise.
8379 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8380 (riscv_slow_unaligned_access): Likewise.
8381 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8382 (riscv_slow_unaligned_access_p): ...this and make static.
8383 (riscv_option_override): Update accordingly.
8384 (riscv_slow_unaligned_access): New function.
8385 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8386 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8387 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8388 (rs6000_slow_unaligned_access): New function.
8389 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8390 (rs6000_rtx_costs): Likewise.
8391 * config/rs6000/rs6000-string.c (expand_block_compare)
8392 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8393 of SLOW_UNALIGNED_ACCESS.
8394 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8395 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8396 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8397 of SLOW_UNALIGNED_ACCESS.
8398 * expmed.c (simple_mem_bitfield_p): Likewise.
8399 * expr.c (alignment_for_piecewise_move): Likewise.
8400 (emit_group_load_1): Likewise.
8401 (emit_group_store): Likewise.
8402 (copy_blkmode_from_reg): Likewise.
8403 (emit_push_insn): Likewise.
8404 (expand_assignment): Likewise.
8405 (store_field): Likewise.
8406 (expand_expr_real_1): Likewise.
8407 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8408 * lra-constraints.c (simplify_operand_subreg): Likewise.
8409 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8410 * gimple-ssa-store-merging.c: Likewise in block comment at start
8412 * tree-ssa-strlen.c: Include target.h.
8413 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8414 of SLOW_UNALIGNED_ACCESS.
8415 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8417 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8419 PR rtl-optimization/82185
8420 * expmed.c (emit_store_flag_int): Only test tem if it has been
8423 2017-09-12 Richard Biener <rguenther@suse.de>
8426 * match.pd ((FTYPE) N CMP CST): Fix typo.
8428 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8430 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8432 (mips_near_type_p): Add 'short_call' attribute as a synonym
8434 * doc/extend.texi (short_call): Document new function attribute.
8436 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8439 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8440 assertion check that in the condition.
8441 (get_atomic_generic_size): Likewise. Before testing if parameter
8442 has pointer type, if it has array type, call for C++
8443 default_conversion to perform array-to-pointer conversion.
8445 2017-09-12 Richard Biener <rguenther@suse.de>
8447 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8448 for operations we cannot scalarize.
8450 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8452 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8453 vectors heap vectors. Clean up comments.
8454 Make visited_bbs a reference.
8455 (profitable_jump_thread_path): Make GC
8456 vectors heap vectors. Clean up comments.
8458 (convert_and_register_jump_thread_path): Make GC vectors heap
8460 (check_subpath_and_update_thread_path): Same. Clean up comments.
8461 Make visited_bbs a reference.
8462 (handle_phi): Abstract common code to to
8463 register_jump_thread_path_if_profitable.
8464 Rename VAR_BB to DEF_BB.
8466 Make GC vectors heap vectors.
8467 Make visited_bbs a reference.
8468 (handle_assignment): Same.
8469 (register_jump_thread_path_if_profitable): New.
8470 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8472 Make GC vectors heap vectors. Clean up comments.
8473 Make visited_bbs a reference.
8474 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8475 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8478 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8481 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8482 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8484 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8486 Revert r251800 and r251799.
8488 2017-09-11 Martin Jambor <mjambor@suse.cz>
8491 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8492 arguments in advance.
8493 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8494 use it to find predecessor edges.
8495 (naive_outof_ssa): Collect vector of predecessors.
8497 2017-09-08 Jason Merrill <jason@redhat.com>
8499 PR c++/70029 - ICE with ref-qualifier and -flto
8500 * langhooks.h (struct lang_hooks_for_types): Add
8501 copy_lang_qualifiers.
8502 * attribs.c (build_type_attribute_qual_variant): Use it.
8503 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8505 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8506 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8508 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8511 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8512 (*mulsi3_sp64): New instruction.
8513 (mulsi3): New expander.
8515 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8517 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8519 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8521 * sancov.c: Include memmodel.h.
8523 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8526 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8529 2017-09-07 Carl Love <cel@us.ibm.com>
8531 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8532 the sldi instruction.
8534 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8536 * sancov.c: Include tm_p.h.
8538 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8541 * output.h (switch_to_other_text_partition): New declaration.
8542 * varasm.c (switch_to_other_text_partition): New function.
8543 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8544 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8545 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8546 to the other text partition before emitting LCL label and switch back
8547 after emitting the word after it.
8549 2017-09-07 Richard Biener <rguenther@suse.de>
8551 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8552 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8553 critical edges here, after loop init.
8554 (pass_data_pre): Remove PROP_no_crit_edges flags.
8555 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8556 for valueization of call args to avoid leaking VN_TOP.
8557 (visit_use): Assert we do not visit default defs.
8558 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8559 Use error_mark_node to more easily detect leaking VN_TOP.
8560 All default-defs are varying, not VN_TOP. Mark them visited.
8561 (run_scc_vn): Make code match comment.
8563 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8565 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8566 OPTION_MASK_FLOAT128_KEYWORD.
8567 (POWERPC_MASKS): Likewise.
8568 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8569 support for the -mfloat128-type option, and make -mfloat128
8570 default on PowerPC Linux systems. Define or undefine
8571 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8572 Define __float128 to be __ieee128 if IEEE 128-bit support is
8573 enabled, or undefine it.
8574 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8575 Delete defining __FLOAT128_TYPE__.
8576 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8577 -mfloat128-type option and make -mfloat128 default on PowerPC
8579 (TARGET_FLOAT128_TYPE): Likewise.
8580 (-mfloat128-type): Likewise.
8581 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8582 Delete the -mfloat128-type option and make -mfloat128 default on
8583 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8584 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8585 128-bit floating point is enabled. Change tests from using
8586 -mfloat128-type to -mfloat128.
8587 (rs6000_mangle_type): Use the correct mangling for the __float128
8588 type even if normal long double is restricted to 64-bits.
8589 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8591 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8592 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8593 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8595 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8596 documentation for -mfloat128.
8598 2017-09-06 Olivier Hainque <hainque@adacore.com>
8600 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8602 2017-09-06 Wish Wu <wishwu007@gmail.com>
8603 Jakub Jelinek <jakub@redhat.com>
8605 * asan.c (initialize_sanitizer_builtins): Add
8606 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8607 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8608 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8609 BT_FN_VOID_UINT64_PTR variables.
8610 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8611 (BT_FN_VOID_UINT16_UINT16): Likewise.
8612 (BT_FN_VOID_UINT32_UINT32): Likewise.
8613 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8614 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8615 (BT_FN_VOID_UINT64_PTR): Likewise.
8616 * common.opt (flag_sanitize_coverage): New variable.
8617 (fsanitize-coverage=trace-pc): Remove.
8618 (fsanitize-coverage=): Add.
8619 * flag-types.h (enum sanitize_coverage_code): New enum.
8620 * fold-const.c (fold_range_test): Disable non-short-circuit
8621 optimization if flag_sanitize_coverage.
8622 (fold_truth_andor): Likewise.
8623 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8624 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8625 (coverage_sanitizer_opts): New array.
8626 (get_closest_sanitizer_option): Add OPTS argument, handle also
8627 OPT_fsanitize_coverage_.
8628 (parse_sanitizer_options): Adjusted to also handle
8629 OPT_fsanitize_coverage_.
8630 (common_handle_option): Add OPT_fsanitize_coverage_.
8631 * sancov.c (instrument_comparison, instrument_switch): New function.
8632 (sancov_pass): Add trace-cmp support.
8633 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8634 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8635 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8636 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8637 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8638 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8639 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8640 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8641 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8642 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8644 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8646 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8647 error. Only quit immediately if parsing is complete.
8648 (BEGIN): Initialize fatal_err and parse_done.
8649 (begin fpu, end fpu): Check number of arguments.
8650 (begin arch, end arch): Likewise.
8651 (begin cpu, end cpu): Likewise.
8652 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8653 (optalias): Likewise.
8655 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8657 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8658 * config/arm/arm-isa.h: Delete. Move definitions to ...
8659 * arm-cpus.in: ... here. Use new feature and fgroup values.
8660 * config/arm/arm.c (arm_option_override): Use lower case for feature
8662 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8663 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8664 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8666 (print_isa_bits_for): New function.
8667 (gen_isa): New function.
8668 (gen_comm_data): Use print_isa_bits_for.
8669 (define feature): New keyword.
8670 (define fgroup): New keyword.
8671 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8672 (arm-isa.h): Add rule to generate file.
8673 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8674 case for feature bit names.
8676 2017-09-06 Richard Biener <rguenther@suse.de>
8678 * tree-ssa-pre.c (NECESSARY): Remove.
8679 (create_expression_by_pieces): Do not touch pass-local flags.
8680 (insert_into_preds_of_block): Likewise.
8681 (do_pre_regular_insertion): Likewise.
8682 (eliminate_insert): Likewise.
8683 (eliminate_dom_walker::before_dom_children): Likewise.
8684 (fini_eliminate): Do not look at inserted_exprs.
8685 (mark_operand_necessary): Remove.
8686 (remove_dead_inserted_code): Replace with simple work-list
8687 algorithm based on inserted_exprs and SSA uses.
8688 (pass_pre::execute): Re-order fini_eliminate and
8689 remove_dead_inserted_code.
8691 2017-09-06 Olivier Hainque <hainque@adacore.com>
8693 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8694 for VxWorks 7. Adjust surrounding comments.
8696 2017-09-06 Richard Biener <rguenther@suse.de>
8698 * gimple-ssa-strength-reduction.c
8699 (find_candidates_dom_walker::before_dom_children): Also allow
8702 2017-09-06 Richard Biener <rguenther@suse.de>
8704 PR tree-optimization/82108
8705 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8706 for gap in the non-permutation SLP case.
8708 2017-09-06 Martin Jambor <mjambor@suse.cz>
8710 PR tree-optimization/82078
8711 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8712 add_access_to_work_queue...
8713 (build_accesses_from_assign): ...here.
8714 (propagate_all_subaccesses): Make sure racc is the group
8715 representative, if there is one.
8717 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8720 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8723 2017-09-06 Richard Biener <rguenther@suse.de>
8725 * gimple-ssa-strength-reduction.c
8726 (find_candidates_dom_walker::before_doom_children): Use a
8727 type and not a mode check.
8729 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8732 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8734 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8736 2017-09-05 Jeff Law <law@redhat.com>
8738 PR tree-optimization/64910
8739 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8740 cases where we have 3 or more operands.
8742 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8745 * omp-low.c (lower_omp_for): Recompute tree invariant if
8746 gimple_omp_for_initial/final is ADDR_EXPR.
8749 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8750 into gimple val before gimplification fo the COND_EXPR.
8752 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8754 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8755 REGION_COPY argument.
8756 (thread_through_all_blocks): Remove unused argument to
8757 duplicate_thread_path.
8759 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8760 Alan Hayward <alan.hayward@arm.com>
8761 David Sherwood <david.sherwood@arm.com>
8763 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8764 Take a scalar_mode rather than a machine_mode.
8765 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8766 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8767 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8768 (aarch64_gen_adjusted_ldpstp): Likewise.
8769 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8771 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8772 Alan Hayward <alan.hayward@arm.com>
8773 David Sherwood <david.sherwood@arm.com>
8775 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8776 Take a scalar_int_mode instead of a machine_mode.
8777 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8778 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8779 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8780 (aarch64_simd_attr_length_rglist): Delete.
8781 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8782 a scalar_int_mode instead of a machine_mode.
8783 (aarch64_add_offset): Likewise.
8784 (aarch64_internal_mov_immediate): Likewise
8785 (aarch64_add_constant_internal): Likewise.
8786 (aarch64_add_constant): Likewise.
8787 (aarch64_movw_imm): Likewise.
8788 (aarch64_rtx_arith_op_extract_p): Likewise.
8789 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8790 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8791 Remove assert that the mode isn't a vector.
8792 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8793 (aarch64_expand_mov_immediate): Update calls after above changes.
8794 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8795 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8796 (aarch64_move_imm): Likewise.
8797 (aarch64_can_const_movi_rtx_p): Likewise.
8798 (aarch64_strip_extend): Likewise.
8799 (aarch64_extr_rtx_p): Likewise.
8800 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8801 a CONST_INT when the mode parameter is VOIDmode.
8802 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8804 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8806 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8807 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8808 (bitwise_type_for_mode): Update accordingly.
8810 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8812 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8813 * stor-layout.c (mode_for_size_tree): Likewise.
8814 (mode_for_array): Update accordingly.
8815 (layout_decl): Likewise.
8816 (compute_record_mode): Likewise. Only set the mode once.
8818 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8820 * target.def (get_mask_mode): Change return type to opt_mode.
8822 * doc/tm.texi: Regenerate.
8823 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8824 * targhooks.c (default_get_mask_mode): Likewise.
8825 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8826 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8827 targetm.get_mask_mode.
8828 * tree.c (build_truth_vector_type): Likewise.
8830 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8832 * machmode.h (mode_for_vector): Return an opt_mode.
8833 * stor-layout.c (mode_for_vector): Likewise.
8834 (mode_for_int_vector): Update accordingly.
8835 (layout_type): Likewise.
8836 * config/i386/i386.c (emit_memmov): Likewise.
8837 (ix86_expand_set_or_movmem): Likewise.
8838 (ix86_expand_vector_init): Likewise.
8839 (ix86_get_mask_mode): Likewise.
8840 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8842 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8843 * expmed.c (extract_bit_field_1): Likewise.
8844 * expr.c (expand_expr_real_2): Likewise.
8845 * optabs-query.c (can_vec_perm_p): Likewise.
8846 (can_vec_mask_load_store_p): Likewise.
8847 * optabs.c (expand_vec_perm): Likewise.
8848 * targhooks.c (default_get_mask_mode): Likewise.
8849 * tree-vect-stmts.c (vectorizable_store): Likewise.
8850 (vectorizable_load): Likewise.
8851 (get_vectype_for_scalar_type_and_size): Likewise.
8853 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8855 * machmode.h (mode_for_int_vector): New function.
8856 * stor-layout.c (mode_for_int_vector): Likewise.
8857 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8858 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8859 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8860 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8861 (s390_expand_vcond): Likewise.
8863 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8865 * machmode.h (opt_machine_mode): New type.
8866 (opt_mode<T>): Allow construction from anything that can be
8868 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8869 (mode_for_size): Return an opt_machine_mode.
8870 * stor-layout.c (mode_for_size): Likewise.
8871 (mode_for_size_tree): Update call accordingly.
8872 (bitwise_mode_for_mode): Likewise.
8873 (make_fract_type): Likewise.
8874 (make_accum_type): Likewise.
8875 * caller-save.c (replace_reg_with_saved_mem): Update call
8877 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8878 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8879 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8880 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8881 * expmed.c (extract_bit_field_1): Likewise.
8882 * reload.c (get_secondary_mem): Likewise.
8883 * varasm.c (assemble_integer): Likewise.
8884 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8887 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8889 * machmode.h (decimal_float_mode_for_size): New function.
8890 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8891 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8892 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8893 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8894 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8895 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8897 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8899 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8900 (get_builtin_sync_mode): Likewise.
8901 (expand_ifn_atomic_compare_exchange): Likewise.
8902 (expand_builtin_atomic_clear): Likewise.
8903 (expand_builtin_atomic_test_and_set): Likewise.
8904 (fold_builtin_atomic_always_lock_free): Likewise.
8905 * calls.c (compute_argument_addresses): Likewise.
8906 (emit_library_call_value_1): Likewise.
8907 (store_one_arg): Likewise.
8908 * combine.c (combine_instructions): Likewise.
8909 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8910 * config/arm/arm.c (arm_function_value): Likewise.
8911 (aapcs_allocate_return_reg): Likewise.
8912 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8913 * config/i386/i386.c (construct_container): Likewise.
8914 (ix86_gimplify_va_arg): Likewise.
8915 (ix86_expand_sse_cmp): Likewise.
8916 (emit_memmov): Likewise.
8917 (emit_memset): Likewise.
8918 (expand_small_movmem_or_setmem): Likewise.
8919 (ix86_expand_pextr): Likewise.
8920 (ix86_expand_pinsr): Likewise.
8921 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8922 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8924 * config/mips/mips.c (mips_function_value_1) Likewise.
8925 (mips_block_move_straight): Likewise.
8926 (mips_expand_ins_as_unaligned_store): Likewise.
8927 * config/powerpcspe/powerpcspe.c
8928 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8929 (rs6000_darwin64_record_arg_flush): Likewise.
8930 * config/rs6000/rs6000.c
8931 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8932 (rs6000_darwin64_record_arg_flush): Likewise.
8933 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8934 (sparc_function_value_1): Likewise.
8935 * config/spu/spu.c (adjust_operand): Likewise.
8936 (spu_emit_branch_or_set): Likewise.
8937 (arith_immediate_p): Likewise.
8938 * emit-rtl.c (gen_lowpart_common): Likewise.
8939 * expr.c (expand_expr_real_1): Likewise.
8940 * function.c (assign_parm_setup_block): Likewise.
8941 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8942 * reload1.c (alter_reg): Likewise.
8943 * stor-layout.c (mode_for_vector): Likewise.
8944 (layout_type): Likewise.
8946 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8948 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8949 (spu_convert_move): Likewise.
8950 * lower-subreg.c (resolve_simple_move): Likewise.
8952 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8955 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8956 define_insn to a define_expand.
8957 (altivec_vsum2sws_direct): New define_insn.
8958 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8960 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8962 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8965 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8970 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8971 realign stack if stack alignment needed is less than incoming
8974 2017-09-05 Marek Polacek <polacek@redhat.com>
8977 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8980 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8982 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8984 2017-09-05 Richard Biener <rguenther@suse.de>
8986 PR tree-optimization/82084
8987 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8989 2017-09-05 Richard Biener <rguenther@suse.de>
8991 PR tree-optimization/82102
8992 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8994 2017-09-05 Martin Liska <mliska@suse.cz>
8996 PR tree-optimization/82032
8997 * tree-cfg.c (generate_range_test): New function.
8998 * tree-cfg.h (generate_range_test): Declared here.
8999 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9000 (cleanup_control_expr_graph): Use it.
9001 * tree-switch-conversion.c (try_switch_expansion): Remove
9003 (emit_case_nodes): Use generate_range_test.
9005 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9008 * config/i386/i386.md (*<btsc><mode>_mask): Add
9009 TARGET_USE_BT to insn constraint.
9010 (*btr<mode>_mask): Ditto.
9012 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9014 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9015 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9017 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9020 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9021 TARGET_NEON and TARGET_IWMMXT.
9022 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9023 TARGET_NEON and TARGET_IWMMXT.
9024 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9026 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9028 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9029 (ix86_rewrite_tls_address): Ditto.
9030 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9031 (ix86_rewrite_tls_address_1): Ditto.
9032 (ix86_rewrite_tls_address): Ditto.
9033 * config/i386/predicates.md (tls_address_pattern): New predicate.
9034 * config/i386/i386.md (TLS address splitter): New splitter.
9036 2017-09-04 Richard Biener <rguenther@suse.de>
9038 PR tree-optimization/82084
9039 * fold-const.h (can_native_encode_string_p): Declare.
9040 * fold-const.c (can_native_encode_string_p): Factor out from ...
9041 (native_encode_string): ... here.
9042 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9043 vectorizing stores from constants we later cannot handle.
9045 2017-09-04 Marek Polacek <polacek@redhat.com>
9048 * doc/invoke.texi: Update -Wtautological-compare documentation.
9050 2017-09-04 Jeff Law <law@redhat.com>
9052 PR tree-optimization/64910
9053 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9054 swap the first and last operand if the last is a constant.
9056 2017-09-04 Marek Polacek <polacek@redhat.com>
9059 * convert.c (do_narrow): When sanitizing signed integer overflows,
9060 bail out for signed types.
9061 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9063 2017-09-04 Richard Biener <rguenther@suse.de>
9065 PR tree-optimization/82060
9066 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9067 Move devirtualization after stmt folding and before EH/AB/noreturn
9068 cleanup to get the stmt refs canonicalized. Use a bool instead
9069 of gimple_modified_p since that doesn't work for NOPs. Schedule
9070 NOPs generated by folding for removal.
9072 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9073 Alan Hayward <alan.hayward@arm.com>
9074 David Sherwood <david.sherwood@arm.com>
9076 * coretypes.h (pad_direction): New enum.
9077 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9078 (FUNCTION_ARG_PADDING): Likewise.
9079 * target.def (function_arg_padding): New hook.
9080 * targhooks.h (default_function_arg_padding): Declare.
9081 * targhooks.c (default_function_arg_padding): New function.
9082 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9083 (TARGET_FUNCTION_ARG_PADDING): ...this.
9084 * doc/tm.texi: Regenerate.
9085 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9086 instead of direction.
9087 (compute_argument_addresses): Likewise.
9088 (load_register_parameters): Likewise.
9089 (emit_library_call_value_1): Likewise.
9090 (store_one_arg): Use targetm.calls.function_arg_padding instead
9091 of FUNCTION_ARG_PADDING.
9092 (must_pass_in_stack_var_size_or_pad): Likewise.
9093 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9094 (emit_group_store): Likewise.
9095 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9096 instead of FUNCTION_ARG_PADDING.
9097 (emit_push_insn): Likewise, and propagate enum change throughout
9099 * function.h (direction): Delete.
9100 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9102 * function.c (assign_parm_find_stack_rtl): Likewise.
9103 (assign_parm_setup_block_p): Likewise.
9104 (assign_parm_setup_block): Likewise.
9105 (gimplify_parameters): Likewise.
9106 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9107 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9109 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9110 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9111 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9112 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9113 (aarch64_function_arg_padding): ...this new function.
9114 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9115 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9116 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9117 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9118 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9119 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9120 (arm_pad_arg_upward): Replace with...
9121 (arm_function_arg_padding): ...this new function.
9122 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9124 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9125 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9126 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9127 (ia64_hpux_function_arg_padding): Replace with...
9128 (ia64_function_arg_padding): ...this new function. Use pad_direction
9129 instead of direction. Check for TARGET_HPUX.
9130 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9131 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9132 (iq2000_function_arg_padding): New function.
9133 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9134 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9135 (mips_function_arg_padding): ...this new function.
9136 (mips_pad_reg_upward): Update accordingly.
9137 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9138 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9139 targetm.calls.function_arg_padding.
9140 (FUNCTION_ARG_PADDING): Delete.
9141 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9142 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9143 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9144 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9145 (nios2_block_reg_padding): Return pad_direction instead of direction.
9146 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9147 instead of direction.
9148 (nios2_function_arg_padding): Likewise. Make static.
9149 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9150 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9151 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9152 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9153 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9154 (pa_function_arg_padding): Make static. Return pad_direction instead
9156 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9157 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9158 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9159 instead of direction. Use targetm.calls.function_arg_padding.
9160 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9161 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9162 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9163 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9164 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9166 (function_arg_padding): Rename to...
9167 (rs6000_function_arg_padding): ...this. Make static. Return
9168 pad_direction instead of direction.
9169 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9170 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9171 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9172 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9173 instead of direction. Use targetm.calls.function_arg_padding.
9174 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9175 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9176 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9177 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9178 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9179 (function_arg_padding): Rename to...
9180 (rs6000_function_arg_padding): ...this. Make static. Return
9181 pad_direction instead of direction.
9182 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9183 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9184 * config/s390/s390.c (s390_function_arg_padding): New function.
9185 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9186 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9187 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9188 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9189 (function_arg_padding): Rename to...
9190 (sparc_function_arg_padding): ...this. Make static. Return
9191 pad_direction instead of direction.
9192 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9193 * config/spu/spu.c (spu_function_arg_padding): New function.
9194 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9195 * system.h (FUNCTION_ARG_PADDING): Poison.
9197 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9198 Alan Hayward <alan.hayward@arm.com>
9199 David Sherwood <david.sherwood@arm.com>
9201 * target.def (modes_tieable_p): New hook.
9202 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9203 (TARGET_MODES_TIEABLE_P): ...this.
9204 * doc/tm.texi.in: Regenerate.
9205 * hooks.h (hook_bool_mode_mode_true): Declare.
9206 * hooks.c (hook_bool_mode_mode_true): New function.
9207 * combine.c (subst): Use targetm.modes_tieable_p instead of
9209 * dse.c (find_shift_sequence): Likewise.
9210 * expmed.c (extract_low_bits): Likewise.
9211 * lower-subreg.c: Include target.h.
9212 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9214 * rtlanal.c (rtx_cost): Likewise.
9215 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9216 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9217 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9218 (TARGET_MODES_TIEABLE_P): Redefine.
9219 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9220 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9221 (TARGET_MODES_TIEABLE_P): Redefine.
9222 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9223 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9224 (arc_modes_tieable_p): New function.
9225 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9226 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9227 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9228 (arm_modes_tieable_p): Make static.
9229 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9230 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9231 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9232 (TARGET_MODES_TIEABLE_P): Redefine.
9233 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9234 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9235 (TARGET_MODES_TIEABLE_P): Redefine.
9236 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9237 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9238 (cr16_modes_tieable_p): New function.
9239 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9240 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9241 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9242 (TRULY_NOOP_TRUNCATION): Update comment.
9243 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9244 (TRULY_NOOP_TRUNCATION): Update comment.
9245 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9246 (frv_modes_tieable_p): New function.
9247 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9248 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9249 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9250 (TARGET_MODES_TIEABLE_P): Redefine.
9251 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9252 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9253 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9254 (TARGET_MODES_TIEABLE_P): Redefine.
9255 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9256 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9257 (ia64_modes_tieable_p): New function.
9258 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9259 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9260 (iq2000_modes_tieable_p): New function.
9261 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9262 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9263 (lm32_modes_tieable_p): New function.
9264 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9265 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9266 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9267 (TARGET_MODES_TIEABLE_P): Redefine.
9268 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9269 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9270 (m32r_modes_tieable_p): New function.
9271 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9272 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9273 (m68k_modes_tieable_p): New function.
9274 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9275 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9276 (mcore_modes_tieable_p): New function.
9277 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9278 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9280 (TARGET_MODES_TIEABLE_P): Redefine.
9281 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9282 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9283 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9284 (TARGET_MODES_TIEABLE_P): Redefine.
9285 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9286 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9287 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9288 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9289 (mn10300_modes_tieable_p): ...this and make static.
9290 (TARGET_MODES_TIEABLE_P): Redefine.
9291 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9292 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9293 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9294 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9295 (msp430_modes_tieable_p): Make static.
9296 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9297 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9298 (TARGET_MODES_TIEABLE_P): Redefine.
9299 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9300 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9301 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9302 (TARGET_MODES_TIEABLE_P): Redefine.
9303 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9304 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9305 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9306 (TARGET_MODES_TIEABLE_P): Redefine.
9307 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9308 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9309 (pdp11_modes_tieable_p): New function.
9310 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9311 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9312 (rs6000_modes_tieable_p): New function.
9313 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9314 * config/powerpcspe/powerpcspe.md: Update comment.
9315 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9316 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9317 (TARGET_MODES_TIEABLE_P): Redefine.
9318 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9319 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9320 (rl78_modes_tieable_p): New function.
9321 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9322 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9323 (rs6000_modes_tieable_p): New function.
9324 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9325 * config/rs6000/rs6000.md: Update comment.
9326 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9327 * config/rx/rx.c (rx_modes_tieable_p): New function.
9328 (TARGET_MODES_TIEABLE_P): Redefine.
9329 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9330 * config/s390/s390.c (s390_modes_tieable_p): New function.
9331 (TARGET_MODES_TIEABLE_P): Redefine.
9332 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9333 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9334 (sh_modes_tieable_p): New function.
9335 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9336 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9337 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9338 (sparc_modes_tieable_p): Make static.
9339 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9340 * config/spu/spu.c (spu_modes_tieable_p): New function.
9341 (TARGET_MODES_TIEABLE_P): Redefine.
9342 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9343 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9344 (TARGET_MODES_TIEABLE_P): Redefine.
9345 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9346 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9347 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9348 * config/v850/v850.c (v850_modes_tieable_p): New function.
9349 (TARGET_MODES_TIEABLE_P): Redefine.
9350 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9351 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9352 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9353 (visium_modes_tieable_p): New function.
9354 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9355 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9356 (xtensa_modes_tieable_p): New function.
9357 * system.h (MODES_TIEABLE_P): Poison.
9359 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9360 Alan Hayward <alan.hayward@arm.com>
9361 David Sherwood <david.sherwood@arm.com>
9363 * target.def (hard_regno_mode_ok): New hook.
9364 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9365 (TARGET_HARD_REGNO_MODE_OK): ...this.
9366 * doc/tm.texi.in: Regenerate.
9367 * hooks.h (hook_bool_uint_mode_true): Declare.
9368 * hooks.c (hook_bool_uint_mode_true): New function.
9369 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9371 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9372 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9373 instead of HARD_REGNO_MODE_OK.
9374 * caller-save.c: Include target.h.
9375 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9377 * combine.c (can_combine_p): Likewise.
9378 (combinable_i3pat): Likewise.
9379 (can_change_dest_mode): Likewise.
9380 * expr.c (init_expr_target): Likewise.
9381 (convert_move): Likewise.
9382 (convert_modes): Likewise.
9383 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9384 (setup_prohibited_mode_move_regs): Likewise.
9385 * ira.h (target_ira): Likewise.
9386 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9387 * lra-constraints.c (process_alt_operands): Likewise.
9388 (split_reg): Likewise.
9389 * recog.c (peep2_find_free_register): Likewise.
9390 * ree.c (combine_reaching_defs): Likewise.
9391 * regcprop.c (maybe_mode_change): Likewise.
9392 * reginfo.c (init_reg_sets_1): Likewise.
9393 (choose_hard_reg_mode): Likewise.
9394 (simplifiable_subregs): Likewise.
9395 * regrename.c (check_new_reg_p): Likewise.
9396 * reload.c (find_valid_class): Likewise.
9397 (find_valid_class_1): Likewise.
9398 (reload_inner_reg_of_subreg): Likewise.
9399 (push_reload): Likewise.
9400 (combine_reloads): Likewise.
9401 (find_dummy_reload): Likewise.
9402 (find_reloads): Likewise.
9403 * reload1.c (find_reg): Likewise.
9404 (set_reload_reg): Likewise.
9405 (allocate_reload_reg): Likewise.
9406 (choose_reload_regs): Likewise.
9407 (reload_adjust_reg_for_temp): Likewise.
9408 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9409 (simplify_subreg_regno): Likewise.
9410 * sel-sched.c (init_regs_for_mode): Likewise.
9411 * varasm.c (make_decl_rtl): Likewise.
9412 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9413 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9415 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9416 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9417 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9418 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9419 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9420 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9421 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9422 (arc_mode_class): Delete.
9423 (HARD_REGNO_MODE_OK): Delete.
9424 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9425 (arc_hard_regno_mode_ok): Rename old array to...
9426 (arc_hard_regno_mode_ok_modes): ...this.
9427 (arc_conditional_register_usage): Update accordingly.
9428 (arc_mode_class): Make static.
9429 (arc_hard_regno_mode_ok): New function.
9430 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9431 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9432 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9433 (arm_hard_regno_mode_ok): Make static.
9434 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9436 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9437 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9438 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9440 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9441 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9442 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9443 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9444 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9445 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9446 * config/bfin/predicates.md (valid_reg_operand): Use
9447 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9448 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9449 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9450 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9451 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9452 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9453 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9454 (cr16_hard_regno_mode_ok): Make static and return a bool.
9455 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9456 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9457 (cris_hard_regno_mode_ok): New function.
9458 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9459 (epiphany_mode_class): Delete.
9460 (HARD_REGNO_MODE_OK): Delete.
9461 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9462 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9463 (hard_regno_mode_ok): Rename to...
9464 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9465 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9466 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9468 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9469 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9470 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9471 (frv_hard_regno_mode_ok): Make static and return a bool.
9472 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9474 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9475 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9476 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9477 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9479 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9480 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9481 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9482 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9484 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9485 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9486 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9487 (ia64_hard_regno_mode_ok): New function.
9488 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9489 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9490 (iq2000_hard_regno_mode_ok): New function.
9491 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9492 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9493 (lm32_hard_regno_mode_ok): New function.
9494 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9495 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9496 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9497 instead of HARD_REGNO_MODE_OK.
9498 (m32c_hard_regno_ok): Rename to...
9499 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9500 (m32c_cannot_change_mode_class): Update accordingly.
9501 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9502 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9503 (m32r_mode_class): Delete.
9504 (HARD_REGNO_MODE_OK): Delete.
9505 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9506 (m32r_hard_regno_mode_ok): Rename to...
9507 (m32r_hard_regno_modes): ...this.
9508 (m32r_mode_class): Make static.
9509 (m32r_hard_regno_mode_ok): New function.
9510 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9511 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9512 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9513 (m68k_hard_regno_mode_ok): Make static.
9514 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9515 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9516 (mcore_hard_regno_mode_ok): New function.
9517 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9518 (HARD_REGNO_MODE_OK): Delete.
9519 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9521 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9522 (microblaze_hard_regno_mode_ok): New function.
9523 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9524 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9525 (mips_hard_regno_mode_ok): Delete.
9526 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9527 (mips_hard_regno_mode_ok_p): ...this and make static.
9528 (mips_hard_regno_mode_ok_p): Rename to...
9529 (mips_hard_regno_mode_ok_uncached): ...this.
9530 (mips_hard_regno_mode_ok): New function.
9531 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9532 of HARD_REGNO_MODE_OK.
9533 (mips_option_override): Update after above name changes.
9534 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9535 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9536 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9537 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9538 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9539 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9540 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9541 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9542 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9543 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9544 (msp430_hard_regno_mode_ok): Make static and return a bool.
9545 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9546 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9547 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9549 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9550 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9551 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9552 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9553 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9554 (PA_HARD_REGNO_MODE_OK): ...this
9555 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9556 (PA_HARD_REGNO_MODE_OK): ...this.
9557 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9558 (pa_hard_regno_mode_ok): New function.
9559 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9560 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9561 (pdp11_hard_regno_mode_ok): New function.
9562 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9563 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9565 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9567 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9568 (rs6000_hard_regno_mode_ok): Rename to...
9569 (rs6000_hard_regno_mode_ok_uncached): ...this.
9570 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9571 (rs6000_hard_regno_mode_ok): New function.
9572 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9573 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9574 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9575 (riscv_hard_regno_mode_ok): ...this and make static.
9576 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9577 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9578 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9579 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9580 (rl78_hard_regno_mode_ok): Make static and return bool.
9581 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9582 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9584 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9585 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9586 (rs6000_hard_regno_mode_ok): Rename to...
9587 (rs6000_hard_regno_mode_ok_uncached): ...this.
9588 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9589 (rs6000_hard_regno_mode_ok): New function.
9590 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9591 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9592 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9593 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9594 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9595 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9596 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9597 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9598 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9599 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9600 (sh_hard_regno_mode_ok): Make static.
9601 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9602 instead of HARD_REGNO_MODE_OK.
9603 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9604 (sparc_mode_class): Delete.
9605 (HARD_REGNO_MODE_OK): Delete.
9606 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9607 (hard_regno_mode_classes): Make static.
9608 (sparc_mode_class): Likewise.
9609 (sparc_hard_regno_mode_ok): New function.
9610 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9611 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9612 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9614 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9615 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9616 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9617 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9618 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9619 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9620 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9621 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9622 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9623 (visium_hard_regno_mode_ok): New function.
9624 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9625 instead of HARD_REGNO_MODE_OK.
9626 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9627 (HARD_REGNO_MODE_OK): Delete.
9628 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9629 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9630 (xtensa_option_override): Update accordingly.
9631 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9632 (xtensa_hard_regno_mode_ok): New function.
9633 * system.h (HARD_REGNO_MODE_OK): Poison.
9635 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9636 Alan Hayward <alan.hayward@arm.com>
9637 David Sherwood <david.sherwood@arm.com>
9639 * target.def (hard_regno_call_part_clobbered): New hook.
9640 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9641 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9642 * doc/tm.texi: Regenerate.
9643 * hooks.h (hook_bool_uint_mode_false): Declare.
9644 * hooks.c (hook_bool_uint_mode_false): New function.
9645 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9646 * cselib.c (cselib_process_insn): Use
9647 targetm.hard_regno_call_part_clobbered instead of
9648 HARD_REGNO_CALL_PART_CLOBBERED.
9649 * ira-conflicts.c (ira_build_conflicts): Likewise.
9650 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9651 * lra-constraints.c (need_for_call_save_p): Likewise.
9652 * lra-lives.c: Include target.h.
9653 (check_pseudos_live_through_calls): Use
9654 targetm.hard_regno_call_part_clobbered instead of
9655 HARD_REGNO_CALL_PART_CLOBBERED.
9656 * regcprop.c: Include target.h.
9657 (copyprop_hardreg_forward_1): Use
9658 targetm.hard_regno_call_part_clobbered instead of
9659 HARD_REGNO_CALL_PART_CLOBBERED.
9660 * reginfo.c (choose_hard_reg_mode): Likewise.
9661 * regrename.c (check_new_reg_p): Likewise.
9662 * reload.c (find_equiv_reg): Likewise.
9663 * reload1.c (emit_reload_insns): Likewise.
9664 * sched-deps.c (deps_analyze_insn): Likewise.
9665 * sel-sched.c (init_regs_for_mode): Likewise.
9666 (mark_unavailable_hard_regs): Likewise.
9667 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9668 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9669 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9671 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9672 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9673 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9675 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9677 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9678 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9679 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9681 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9682 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9683 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9685 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9686 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9688 * config/powerpcspe/powerpcspe.c
9689 (rs6000_hard_regno_call_part_clobbered): New function.
9690 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9691 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9692 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9694 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9695 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9696 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9698 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9699 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9700 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9702 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9703 Alan Hayward <alan.hayward@arm.com>
9704 David Sherwood <david.sherwood@arm.com>
9706 * rtl.h (subreg_memory_offset): Declare.
9707 * emit-rtl.c (subreg_memory_offset): New function.
9708 * expmed.c (store_bit_field_1): Use it.
9709 * expr.c (undefined_operand_subword_p): Likewise.
9710 * simplify-rtx.c (simplify_subreg): Likewise.
9712 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9714 PR rtl-optimization/57448
9717 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9718 using atomic_load pattern.
9719 (expand_atomic_store): Likewise.
9721 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9724 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9725 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9726 handling. Use replace_call_with_value with NULL instead of
9727 gsi_replace, unlink_stmt_vdef and release_defs.
9729 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9732 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9734 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9737 * rtl.h (emit_library_call_value_1): Declare.
9738 (emit_library_call): Replace declaration with a series of overloads.
9739 Remove the parameter count argument.
9740 (emit_library_call_value): Likewise.
9741 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9742 with an "rtx_mode_t *".
9743 (emit_library_call_value): Delete.
9744 (emit_library_call): Likewise.
9745 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9746 (asan_emit_allocas_unpoison): Likewise.
9747 * builtins.c (expand_builtin_powi): Likewise.
9748 (expand_asan_emit_allocas_unpoison): Likewise.
9749 * cfgexpand.c (expand_main_function): Likewise.
9750 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9751 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9752 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9753 * config/arm/arm.c (arm_trampoline_init): Likewise.
9754 (arm_call_tls_get_addr): Likewise.
9755 (arm_expand_divmod_libfunc): Likewise.
9756 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9757 (smulsi3_highpart): Likewise.
9758 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9759 (c6x_expand_compare): Likewise.
9760 (c6x_expand_movmem): Likewise.
9761 * config/frv/frv.c (frv_trampoline_init): Likewise.
9762 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9763 (ix86_expand_divmod_libfunc): Likewise.
9764 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9765 (ia64_expand_compare): Likewise.
9766 (ia64_profile_hook): Likewise.
9767 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9768 (nonlocal_goto): Likewise.
9769 (restore_stack_nonlocal): Likewise.
9770 * config/m32r/m32r.c (block_move_call): Likewise.
9771 (m32r_trampoline_init): Likewise.
9772 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9773 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9774 (m68k_call_m68k_read_tp): Likewise.
9775 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9776 (microblaze_expand_divide): Likewise.
9777 * config/mips/mips.h (mips_args): Likewise.
9778 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9779 (MIPS_ICACHE_SYNC): Likewise.
9780 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9781 (nios2_trampoline_init): Likewise.
9782 * config/pa/pa.c (hppa_tls_call): Likewise.
9783 (pa_trampoline_init): Likewise.
9784 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9785 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9786 (expand_strn_compare): Likewise.
9787 (rs6000_generate_compare): Likewise.
9788 (rs6000_expand_float128_convert): Likewise.
9789 (output_profile_hook): Likewise.
9790 (rs6000_trampoline_init): Likewise.
9791 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9792 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9793 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9794 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9795 (rs6000_generate_compare): Likewise.
9796 (rs6000_expand_float128_convert): Likewise.
9797 (output_profile_hook): Likewise.
9798 (rs6000_trampoline_init): Likewise.
9799 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9800 * config/sh/sh.c (sh_trampoline_init): Likewise.
9801 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9802 (sparc_emit_float_lib_cmp): Likewise.
9803 (sparc32_initialize_trampoline): Likewise.
9804 (sparc64_initialize_trampoline): Likewise.
9805 (sparc_profile_hook): Likewise.
9806 * config/spu/spu.c (ea_load_store): Likewise.
9807 * config/spu/spu.md (floatunssidf2): Likewise.
9808 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9809 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9810 * config/visium/visium.c (expand_block_move_4): Likewise.
9811 (expand_block_move_2): Likewise.
9812 (expand_block_move_1): Likewise.
9813 (expand_block_set_4): Likewise.
9814 (expand_block_set_2): Likewise.
9815 (expand_block_set_1): Likewise.
9816 (visium_trampoline_init): Likewise.
9817 (visium_profile_hook): Likewise.
9818 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9819 (xtensa_setup_frame_addresses): Likewise.
9820 (xtensa_trampoline_init): Likewise.
9821 * except.c (sjlj_emit_function_enter): Likewise.
9822 (sjlj_emit_function_exit): Likewise.
9823 * explow.c (allocate_dynamic_stack_space): Likewise.
9824 (probe_stack_range): Likewise.
9825 * expr.c (convert_mode_scalar): Likewise.
9826 * optabs.c (expand_binop): Likewise.
9827 (expand_twoval_binop_libfunc): Likewise.
9828 (expand_unop): Likewise.
9829 (prepare_cmp_insn): Likewise.
9830 (prepare_float_lib_cmp): Likewise.
9831 (expand_float): Likewise.
9832 (expand_fix): Likewise.
9833 (expand_fixed_convert): Likewise.
9834 (maybe_emit_sync_lock_test_and_set): Likewise.
9835 (expand_atomic_compare_and_swap): Likewise.
9836 (expand_mem_thread_fence): Likewise.
9837 (expand_atomic_fetch_op): Likewise.
9839 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9841 * doc/generic.texi (OpenACC): Adjust URL.
9842 * doc/invoke.texi (C Dialect Options): Ditto.
9844 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9846 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9847 predicate for operand 1. Add (m,<S>) constraint.
9848 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9849 Prevent memory operand 1 with register operand 2.
9851 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9853 PR rtl-optimization/82024
9854 * combine.c (try_combine): If the combination result is a PARALLEL,
9855 and we only need to retain the SET in there that would be placed
9856 at I2, check that we can place that at I3 instead, before doing so.
9858 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9861 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9863 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9864 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9867 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9868 Jeff Law <law@redhat.com>
9870 * varasm.c (bss_initializer_p): Do not put constants into .bss
9871 (categorize_decl_for_section): Handle bss_initializer_p returning
9872 false when DECL_INITIAL is NULL.
9874 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9877 * config/s390/s390.c (s390_can_inline_p): New function.
9879 2017-09-01 Jeff Law <law@redhat.com>
9881 PR tree-optimization/82052
9882 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9883 Always initialize the returned slot after a hash table miss
9884 when INSERT is true.
9886 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9888 * config/s390/s390.md (mem_signal_fence): Remove.
9889 * doc/md.texi (mem_signal_fence): Remove.
9890 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9892 * target-insns.def (mem_signal_fence): Remove.
9894 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9897 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9900 * asan.c (create_odr_indicator): Strip name encoding from assembler
9901 name before appending it after __odr_asan_.
9903 2017-09-01 Martin Liska <mliska@suse.cz>
9905 PR tree-optimization/82059
9906 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9907 frequency only when an edge is redirected.
9909 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9911 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9912 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9913 (arc_conditional_register_usage): Remove ARC600 lp_count
9915 (arc_file_start): Emit Tag_ARC_CPU_variation.
9916 (arc_can_use_doloop_p): New conditions to use ZOLs.
9917 (hwloop_fail): New function.
9918 (hwloop_optimize): Likewise.
9919 (hwloop_pattern_reg): Likewise.
9920 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9921 (arc_reorg_loops): New function, calls reorg_loops.
9922 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9923 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9925 (arc_loop_hazard): Remove function, functionality moved into
9927 (arc_hazard): Remove arc_loop_hazard call.
9928 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9929 into hwloop_optimize.
9930 (arc_label_align): Remove ZOL handling.
9931 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9932 * config/arc/arc.md (doloop_begin): Remove pattern.
9933 (doloop_begin_i): Likewise.
9934 (doloop_end_i): Likewise.
9935 (doloop_fallback): Likewise.
9936 (doloop_fallback_m): Likewise.
9937 (doloop_end): Reimplement expand.
9938 (arc_lp): New pattern for LP instruction.
9939 (loop_end): New pattern.
9940 (loop_fail): Likewise.
9941 (decrement_and_branch_until_zero): Likewise.
9942 * config/arc/arc.opt (mlpc-width): New option.
9943 * doc/invoke.texi (mlpc-width): Document option.
9945 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9947 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9948 (arc_ccfsm_advance): Fix checking for delay slots.
9949 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9951 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9953 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9955 (movhi_insn): Update store instruction constraint which are saving
9956 6-bit short immediates.
9957 (movsi_insn): Consider also short scaled load operations.
9958 (zero_extendhisi2_i): Use Usd constraint instead of T.
9959 (extendhisi2_i): Add q constraint.
9960 (arc_clzsi2): Add type and length attributes.
9961 (arc_ctzsi2): Likewise.
9962 * config/arc/constraints.md (Usc): Update constraint, the
9963 assembler can parse two relocations for a single instruction.
9965 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9967 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9968 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9970 2017-08-31 Olivier Hainque <hainque@adacore.com>
9972 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9973 match as powerpc-wrs-vxworks*.
9975 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9977 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9978 register constraint for by-element operand.
9979 (aarch64_mls_elt_merge<mode>): Likewise.
9981 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9983 * config/arc/arc.c (arc_can_follow_jump): Check for short
9986 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9988 * config.gcc: Use g.opt for arc.
9989 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9990 functionality moved to ...
9991 (legitimate_scaled_address_p): New function, ...here.
9992 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9993 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9994 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9996 (arc_override_options): Handle G option.
9997 (arc_output_pic_addr_const): Correct function definition.
9998 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9999 (arc_decl_anon_ns_mem_p): Delete.
10000 (arc_in_small_data_p): Overhaul this function to take into
10001 consideration the value given via G option.
10002 (arc_rewrite_small_data_1): Renamed and corrected old
10003 arc_rewrite_small_data function.
10004 (arc_rewrite_small_data): New function.
10005 (small_data_pattern): Don't use pic_offset_table_rtx.
10006 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10007 * config/arc/simdext.md (movmisalignv2hi): Use
10008 prepare_move_operands function.
10010 (movmisalign*): Likewise.
10011 * doc/invoke.texi (ARC options): Document -G option.
10013 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10015 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10017 * config/arc/arc.c (arc_print_operand): Output scalled address for
10018 sdata whenever is possible.
10019 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10020 load/stores are available.
10021 (compact_sda_memory_operand): Check for the alignment required by
10022 code density instructions.
10023 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10025 * config/arc/constraints.md (Usd): Update constraint.
10026 (Us0): New constraint.
10027 (Usc): Update constraint.
10029 2017-08-31 Richard Biener <rguenther@suse.de>
10031 PR middle-end/82054
10032 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10033 function only once.
10035 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10037 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10038 Resize type_signature.
10040 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10041 Alan Hayward <alan.hayward@arm.com>
10042 David Sherwood <david.sherwood@arm.com>
10044 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10045 subregs whose inner modes can be stored in GPRs.
10046 (aarch64_classify_index): Likewise.
10048 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10049 Alan Hayward <alan.hayward@arm.com>
10050 David Sherwood <david.sherwood@arm.com>
10052 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10053 (V_INT_EQUIV): ...this.
10054 (v_cmp_result): Rename to...
10055 (v_int_equiv): ...this.
10056 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10057 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10058 (copysign<mode>3): Likewise.
10059 (aarch64_simd_bsl<mode>_internal): Likewise.
10060 (aarch64_simd_bsl<mode>): Likewise.
10061 (vec_cmp<mode><mode>): Likewise.
10062 (vcond<mode><mode>): Likewise.
10063 (vcond<v_cmp_mixed><mode>): Likewise.
10064 (vcondu<mode><v_cmp_mixed>): Likewise.
10065 (aarch64_cm<optab><mode>): Likewise.
10066 (aarch64_cmtst<mode>): Likewise.
10067 (aarch64_fac<optab><mode>): Likewise.
10068 (vec_perm_const<mode>): Likewise.
10069 (vcond_mask_<mode><v_cmp_result>): Rename to...
10070 (vcond_mask_<mode><v_int_equiv>): ...this.
10071 (vec_cmp<mode><v_cmp_result>): Rename to...
10072 (vec_cmp<mode><v_int_equiv>): ...this.
10074 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10075 Alan Hayward <alan.hayward@arm.com>
10076 David Sherwood <david.sherwood@arm.com>
10078 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10080 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10081 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10082 (UNSPEC_LD4_DREG): New unspecs.
10083 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10084 (aarch64_ld2<mode>_dreg_be): Replace with...
10085 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10087 (aarch64_ld3<mode>_dreg_le)
10088 (aarch64_ld3<mode>_dreg_be): Replace with...
10089 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10091 (aarch64_ld4<mode>_dreg_le)
10092 (aarch64_ld4<mode>_dreg_be): Replace with...
10093 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10096 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10098 PR tree-optimization/81987
10099 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10100 insert an initializer in a location not dominated by the stride
10103 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10105 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10106 on the entire header of the finally block in the fallthru case.
10108 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10110 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10112 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10114 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10115 rs6000_emit_move_from_cr and call renamed function.
10116 (rs6000_emit_prologue): Call renamed functions.
10117 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10118 movesi_from_cr, remove volatile CRs.
10120 2017-08-30 Jon Beniston <jon@beniston.com>
10121 Richard Biener <rguenther@suse.de>
10123 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10124 of VECTOR_MODE_P check.
10125 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10126 element vector types.
10128 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10130 * df.h (df_read_modify_subreg_p): Remove in favor of...
10131 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10132 const_rtx instead of an rtx.
10133 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10134 * df-problems.c (df_word_lr_mark_ref): Likewise.
10135 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10136 (mark_pseudo_reg_dead): Likewise.
10137 (mark_ref_dead): Likewise.
10138 * reginfo.c (init_subregs_of_mode): Likewise.
10139 * sched-deps.c (sched_analyze_1): Likewise.
10140 * df-scan.c (df_def_record_1): Likewise.
10141 (df_uses_record): Likewise.
10142 (df_read_modify_subreg_p): Remove in favor of...
10143 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10144 const_rtx instead of an rtx.
10146 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10147 Alan Hayward <alan.hayward@arm.com>
10148 David Sherwood <david.sherwood@arm.com>
10150 * rtl.h (partial_subreg_p): New function.
10151 * caller-save.c (save_call_clobbered_regs): Use it.
10152 * calls.c (expand_call): Likewise.
10153 * combine.c (combinable_i3pat): Likewise.
10154 (simplify_set): Likewise.
10155 (make_extraction): Likewise.
10156 (make_compound_operation_int): Likewise.
10157 (gen_lowpart_or_truncate): Likewise.
10158 (force_to_mode): Likewise.
10159 (make_field_assignment): Likewise.
10160 (reg_truncated_to_mode): Likewise.
10161 (record_truncated_value): Likewise.
10162 (move_deaths): Likewise.
10163 * cse.c (record_jump_cond): Likewise.
10164 (cse_insn): Likewise.
10165 * cselib.c (cselib_lookup_1): Likewise.
10166 * expmed.c (extract_bit_field_using_extv): Likewise.
10167 * function.c (assign_parm_setup_reg): Likewise.
10168 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10169 * ira-build.c (create_insn_allocnos): Likewise.
10170 * lra-coalesce.c (merge_pseudos): Likewise.
10171 * lra-constraints.c (match_reload): Likewise.
10172 (simplify_operand_subreg): Likewise.
10173 (curr_insn_transform): Likewise.
10174 * lra-lives.c (process_bb_lives): Likewise.
10175 * lra.c (new_insn_reg): Likewise.
10176 (lra_substitute_pseudo): Likewise.
10177 * regcprop.c (mode_change_ok): Likewise.
10178 (maybe_mode_change): Likewise.
10179 (copyprop_hardreg_forward_1): Likewise.
10180 * reload.c (push_reload): Likewise.
10181 (find_reloads): Likewise.
10182 (find_reloads_subreg_address): Likewise.
10183 * reload1.c (alter_reg): Likewise.
10184 (eliminate_regs_1): Likewise.
10185 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10187 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10189 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10190 back to if statements, including unpack.
10192 2017-08-30 Martin Liska <mliska@suse.cz>
10194 PR inline-asm/82001
10195 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10197 (func_checker::compare_asm_inputs_outputs): ... this function.
10198 (func_checker::compare_gimple_asm): Use the function to compare
10199 also ASM constrains.
10200 * ipa-icf-gimple.h: Rename the function.
10202 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10203 Alan Hayward <alan.hayward@arm.com>
10204 David Sherwood <david.sherwood@arm.com>
10206 * coretypes.h (complex_mode): New type.
10207 * gdbhooks.py (build_pretty_printer): Handle it.
10208 * machmode.h (complex_mode): New class.
10209 (complex_mode::includes_p): New function.
10210 (is_complex_int_mode): Likewise.
10211 (is_complex_float_mode): Likewise.
10212 * genmodes.c (get_mode_class): Handle complex mode classes.
10213 * function.c (expand_function_end): Use is_complex_int_mode.
10215 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10216 Alan Hayward <alan.hayward@arm.com>
10217 David Sherwood <david.sherwood@arm.com>
10219 * coretypes.h (scalar_mode_pod): New typedef.
10220 * gdbhooks.py (build_pretty_printer): Handle it.
10221 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10222 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10223 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10224 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10225 as_a <scalar_mode>.
10227 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10228 Alan Hayward <alan.hayward@arm.com>
10229 David Sherwood <david.sherwood@arm.com>
10231 * machmode.h (mode_for_vector): Take a scalar_mode instead
10233 * stor-layout.c (mode_for_vector): Likewise.
10234 * explow.c (promote_mode): Use as_a <scalar_mode>.
10235 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10237 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10238 Alan Hayward <alan.hayward@arm.com>
10239 David Sherwood <david.sherwood@arm.com>
10241 * target.def (preferred_simd_mode): Take a scalar_mode
10242 instead of a machine_mode.
10243 * targhooks.h (default_preferred_simd_mode): Likewise.
10244 * targhooks.c (default_preferred_simd_mode): Likewise.
10245 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10246 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10247 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10248 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10249 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10250 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10251 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10252 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10254 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10255 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10256 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10257 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10258 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10259 * doc/tm.texi: Regenerate.
10260 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10263 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10264 Alan Hayward <alan.hayward@arm.com>
10265 David Sherwood <david.sherwood@arm.com>
10267 * target.def (scalar_mode_supported_p): Take a scalar_mode
10268 instead of a machine_mode.
10269 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10270 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10271 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10272 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10273 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10274 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10275 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10276 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10277 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10278 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10279 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10280 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10281 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10282 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10284 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10285 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10286 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10287 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10288 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10290 * doc/tm.texi: Regenerate.
10292 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10293 Alan Hayward <alan.hayward@arm.com>
10294 David Sherwood <david.sherwood@arm.com>
10296 * coretypes.h (opt_scalar_mode): New typedef.
10297 * gdbhooks.py (build_pretty_printers): Handle it.
10298 * machmode.h (mode_iterator::get_2xwider): Add overload for
10300 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10302 * expr.c (convert_mode_scalar): Likewise.
10303 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10304 * optabs.c (expand_float): Likewise.
10305 (expand_fix): Likewise.
10306 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10308 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10309 Alan Hayward <alan.hayward@arm.com>
10310 David Sherwood <david.sherwood@arm.com>
10312 * optabs.c (expand_float): Explicitly check for scalars before
10313 using a branching expansion.
10314 (expand_fix): Likewise.
10316 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10317 Alan Hayward <alan.hayward@arm.com>
10318 David Sherwood <david.sherwood@arm.com>
10320 * expr.c (convert_mode): Split scalar handling out into...
10321 (convert_mode_scalar): ...this new function. Treat the modes
10324 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10325 Alan Hayward <alan.hayward@arm.com>
10326 David Sherwood <david.sherwood@arm.com>
10328 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10330 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10332 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10333 Alan Hayward <alan.hayward@arm.com>
10334 David Sherwood <david.sherwood@arm.com>
10336 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10337 rather than a machine_mode.
10338 (fixed_from_string): Likewise.
10339 (fixed_convert): Likewise.
10340 (fixed_convert_from_int): Likewise.
10341 (fixed_convert_from_real): Likewise.
10342 (real_convert_from_fixed): Likewise.
10343 * fixed-value.c (fixed_from_double_int): Likewise.
10344 (fixed_from_string): Likewise.
10345 (fixed_convert): Likewise.
10346 (fixed_convert_from_int): Likewise.
10347 (fixed_convert_from_real): Likewise.
10348 (real_convert_from_fixed): Likewise.
10349 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10351 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10352 Alan Hayward <alan.hayward@arm.com>
10353 David Sherwood <david.sherwood@arm.com>
10355 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10356 of separate mode class checks. Do not allow vector modes here.
10357 (immed_wide_int_const): Use as_a <scalar_mode>.
10358 * explow.c (trunc_int_for_mode): Likewise.
10359 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10360 (wi::shwi): Likewise.
10361 (wi::min_value): Likewise.
10362 (wi::max_value): Likewise.
10363 * dwarf2out.c (loc_descriptor): Likewise.
10364 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10365 for CONST_WIDE_INT.
10367 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10368 Alan Hayward <alan.hayward@arm.com>
10369 David Sherwood <david.sherwood@arm.com>
10371 * tree.h (SCALAR_TYPE_MODE): New macro.
10372 * expr.c (expand_expr_addr_expr_1): Use it.
10373 (expand_expr_real_2): Likewise.
10374 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10375 (fold_convert_const_fixed_from_int): Likewise.
10376 (fold_convert_const_fixed_from_real): Likewise.
10377 (native_encode_fixed): Likewise
10378 (native_encode_complex): Likewise
10379 (native_encode_vector): Likewise.
10380 (native_interpret_fixed): Likewise.
10381 (native_interpret_real): Likewise.
10382 (native_interpret_complex): Likewise.
10383 (native_interpret_vector): Likewise.
10384 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10385 (simd_clone_adjust_argument_types): Likewise.
10386 (simd_clone_init_simd_arrays): Likewise.
10387 (simd_clone_adjust): Likewise.
10388 * stor-layout.c (layout_type): Likewise.
10389 * tree.c (build_minus_one_cst): Likewise.
10390 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10391 * tree-inline.c (estimate_move_cost): Likewise.
10392 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10393 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10394 (vectorizable_reduction): Likewise.
10395 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10396 (vect_recog_mixed_size_cond_pattern): Likewise.
10397 (check_bool_pattern): Likewise.
10398 (adjust_bool_pattern): Likewise.
10399 (search_type_for_mask_1): Likewise.
10400 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10401 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10402 (vectorizable_load): Likewise.
10403 (vectorizable_store): Likewise.
10404 * ubsan.c (ubsan_encode_value): Likewise.
10405 * varasm.c (output_constant): Likewise.
10407 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10408 Alan Hayward <alan.hayward@arm.com>
10409 David Sherwood <david.sherwood@arm.com>
10411 * coretypes.h (scalar_mode): New class.
10412 * machmode.h (scalar_mode): Likewise.
10413 (scalar_mode::includes_p): New function.
10414 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10415 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10416 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10417 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10418 * expmed.c (store_bit_field_1): Likewise.
10419 (extract_bit_field_1): Likewise.
10420 * expr.c (write_complex_part): Likewise.
10421 (read_complex_part): Likewise.
10422 (emit_move_complex_push): Likewise.
10423 (expand_expr_real_2): Likewise.
10424 * function.c (assign_parm_setup_reg): Likewise.
10425 (assign_parms_unsplit_complex): Likewise.
10426 * optabs.c (expand_binop): Likewise.
10427 * rtlanal.c (subreg_get_info): Likewise.
10428 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10429 * varasm.c (output_constant_pool_2): Likewise.
10431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10432 Alan Hayward <alan.hayward@arm.com>
10433 David Sherwood <david.sherwood@arm.com>
10435 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10437 (expmed_mult_highpart_optab): Likewise.
10438 (expmed_mult_highpart): Likewise.
10440 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10441 Alan Hayward <alan.hayward@arm.com>
10442 David Sherwood <david.sherwood@arm.com>
10444 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10445 instead of a machine_mode.
10446 (builtin_memset_read_str): Likewise.
10447 * builtins.c (c_readstr): Likewise.
10448 (builtin_memcpy_read_str): Likewise.
10449 (builtin_strncpy_read_str): Likewise.
10450 (builtin_memset_read_str): Likewise.
10451 (builtin_memset_gen_str): Likewise.
10452 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10453 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10454 instead of a machine_mode.
10455 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10457 (make_extraction): Likewise.
10458 (try_widen_shift_mode): Take and return scalar_int_modes instead
10460 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10461 a scalar_int_mode instead of a machine_mode.
10462 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10463 (avr_addr_space_pointer_mode): Likewise.
10464 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10465 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10466 (msp430_unwind_word_mode): Likewise.
10467 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10468 (spu_addr_space_pointer_mode): Likewise.
10469 (spu_addr_space_address_mode): Likewise.
10470 (spu_libgcc_cmp_return_mode): Likewise.
10471 (spu_libgcc_shift_count_mode): Likewise.
10472 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10473 (rl78_addr_space_pointer_mode): Likewise.
10474 (fl78_unwind_word_mode): Likewise.
10475 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10477 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10478 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10479 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10480 (mips_valid_pointer_mode): Likewise.
10481 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10482 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10483 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10485 (ft32_addr_space_address_mode): Likewise.
10486 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10487 scalar_int_mode instead of a machine_mode.
10488 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10490 (m32c_addr_space_address_mode): Likewise.
10491 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10492 (rs6000_eh_return_filter_mode): Likewise.
10493 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10494 (rs6000_eh_return_filter_mode): Likewise.
10495 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10496 (s390_libgcc_shift_count_mode): Likewise.
10497 (s390_unwind_word_mode): Likewise.
10498 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10500 * target.def (mode_rep_extended): Likewise.
10501 (valid_pointer_mode): Likewise.
10502 (addr_space.valid_pointer_mode): Likewise.
10503 (eh_return_filter_mode): Return a scalar_int_mode rather than
10505 (libgcc_cmp_return_mode): Likewise.
10506 (libgcc_shift_count_mode): Likewise.
10507 (unwind_word_mode): Likewise.
10508 (addr_space.pointer_mode): Likewise.
10509 (addr_space.address_mode): Likewise.
10510 * doc/tm.texi: Regenerate.
10511 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10513 (do_jump): Use scalar_int_mode for local variables.
10514 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10515 rather than a machine_mode.
10516 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10517 (scompare_loc_descriptor_wide): Likewise.
10518 (scompare_loc_descriptor_narrow): Likewise.
10519 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10521 * except.c (sjlj_emit_dispatch_table): Likewise.
10522 (expand_builtin_eh_copy_values): Likewise.
10523 * explow.c (convert_memory_address_addr_space_1): Likewise.
10524 Take a scalar_int_mode rather than a machine_mode.
10525 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10526 than a machine_mode.
10527 (memory_address_addr_space): Use scalar_int_mode for local variables.
10528 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10529 rather than a machine_mode.
10530 * expmed.c (mask_rtx): Likewise.
10531 (init_expmed_one_conv): Likewise.
10532 (expand_mult_highpart_adjust): Likewise.
10533 (extract_high_half): Likewise.
10534 (expmed_mult_highpart_optab): Likewise.
10535 (expmed_mult_highpart): Likewise.
10536 (expand_smod_pow2): Likewise.
10537 (expand_sdiv_pow2): Likewise.
10538 (emit_store_flag_int): Likewise.
10539 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10541 (extract_low_bits): Likewise.
10542 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10544 * expr.c (pieces_addr::adjust): Likewise.
10545 (can_store_by_pieces): Likewise.
10546 (store_by_pieces): Likewise.
10547 (clear_by_pieces_1): Likewise.
10548 (expand_expr_addr_expr_1): Likewise.
10549 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10550 (expand_expr_real_1): Likewise.
10551 (try_casesi): Likewise.
10552 * final.c (shorten_branches): Likewise.
10553 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10554 type of "mode" to machine_mode.
10555 * internal-fn.c (expand_arith_overflow_result_store): Take a
10556 scalar_int_mode rather than a machine_mode.
10557 (expand_mul_overflow): Use scalar_int_mode for local variables.
10558 * loop-doloop.c (doloop_modify): Likewise.
10559 (doloop_optimize): Likewise.
10560 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10561 than a machine_mode.
10562 (expand_doubleword_shift_condmove): Likewise.
10563 (expand_doubleword_shift): Likewise.
10564 (expand_doubleword_clz): Likewise.
10565 (expand_doubleword_popcount): Likewise.
10566 (expand_doubleword_parity): Likewise.
10567 (expand_absneg_bit): Use scalar_int_mode for local variables.
10568 (prepare_float_lib_cmp): Likewise.
10569 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10570 rather than a machine_mode.
10571 (convert_memory_address_addr_space): Likewise.
10572 (get_mode_bounds): Likewise.
10573 (get_address_mode): Return a scalar_int_mode rather than a
10575 * rtlanal.c (get_address_mode): Likewise.
10576 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10577 than a machine_mode.
10578 * targhooks.c (default_mode_rep_extended): Likewise.
10579 (default_valid_pointer_mode): Likewise.
10580 (default_addr_space_valid_pointer_mode): Likewise.
10581 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10582 than a machine_mode.
10583 (default_libgcc_cmp_return_mode): Likewise.
10584 (default_libgcc_shift_count_mode): Likewise.
10585 (default_unwind_word_mode): Likewise.
10586 (default_addr_space_pointer_mode): Likewise.
10587 (default_addr_space_address_mode): Likewise.
10588 * targhooks.h (default_eh_return_filter_mode): Likewise.
10589 (default_libgcc_cmp_return_mode): Likewise.
10590 (default_libgcc_shift_count_mode): Likewise.
10591 (default_unwind_word_mode): Likewise.
10592 (default_addr_space_pointer_mode): Likewise.
10593 (default_addr_space_address_mode): Likewise.
10594 (default_mode_rep_extended): Take a scalar_int_mode rather than
10596 (default_valid_pointer_mode): Likewise.
10597 (default_addr_space_valid_pointer_mode): Likewise.
10598 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10600 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10601 rather than a machine_mode.
10602 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10603 for local variables.
10604 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10605 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10606 than a machine_mode.
10608 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10609 Alan Hayward <alan.hayward@arm.com>
10610 David Sherwood <david.sherwood@arm.com>
10612 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10613 the mode argument to scalar_int_mode.
10614 (do_jump_by_parts_zero_rtx): Likewise.
10615 (do_jump_by_parts_equality_rtx): Likewise.
10616 (do_jump_by_parts_greater): Take a mode argument.
10617 (do_jump_by_parts_equality): Likewise.
10618 (do_jump_1): Update calls accordingly.
10620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10621 Alan Hayward <alan.hayward@arm.com>
10622 David Sherwood <david.sherwood@arm.com>
10624 * is-a.h (safe_dyn_cast): New function.
10625 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10626 (jump_table_for_label): Likewise.
10627 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10628 instead of an rtx_insn *.
10629 (shorten_branches): Use dyn_cast instead of LABEL_P and
10630 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10631 rtx_jump_table_data::get_data_mode.
10632 (final_scan_insn): Likewise.
10634 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10635 Alan Hayward <alan.hayward@arm.com>
10636 David Sherwood <david.sherwood@arm.com>
10638 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10639 trying to combine a full-register integer set with a subreg
10642 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10643 Alan Hayward <alan.hayward@arm.com>
10644 David Sherwood <david.sherwood@arm.com>
10646 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10647 that is always either address_mode or pointer_mode.
10649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10650 Alan Hayward <alan.hayward@arm.com>
10651 David Sherwood <david.sherwood@arm.com>
10653 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10654 when the two are known to be equal.
10656 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10657 Alan Hayward <alan.hayward@arm.com>
10658 David Sherwood <david.sherwood@arm.com>
10660 * simplify-rtx.c (simplify_const_unary_operation): Use
10661 is_a <scalar_int_mode> instead of checking for a nonzero
10662 precision. Forcibly convert op_mode to a scalar_int_mode
10663 in that case. More clearly differentiate the operand and
10664 result modes and use the former when deciding what the value
10665 of a count-bits operation should be. Use is_int_mode instead
10666 of checking for a MODE_INT. Remove redundant check for whether
10667 this mode has a zero precision.
10669 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10670 Alan Hayward <alan.hayward@arm.com>
10671 David Sherwood <david.sherwood@arm.com>
10673 * optabs.c (widen_leading): Change the type of the mode argument
10674 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10675 (widen_bswap): Likewise.
10676 (expand_parity): Likewise.
10677 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10678 (expand_ffs): Likewise.
10679 (epand_unop): Check for scalar integer modes before calling the
10682 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10683 Alan Hayward <alan.hayward@arm.com>
10684 David Sherwood <david.sherwood@arm.com>
10686 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10688 (expand_expr_real_1): Update call accordingly.
10690 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10691 Alan Hayward <alan.hayward@arm.com>
10692 David Sherwood <david.sherwood@arm.com>
10694 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10695 value_mode arguments. Use scalar_int_mode internally.
10696 (store_bit_field_1): Rename the new integer mode from imode
10697 to op0_mode and use it instead of GET_MODE (op0). Update calls
10698 to store_split_bit_field, store_bit_field_using_insv and
10699 store_fixed_bit_field.
10700 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10701 Use scalar_int_mode internally. Use a bit count rather than a mode
10702 when calculating the largest bit size for get_best_mode.
10703 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10704 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10705 Remove assertion that OP0 has a scalar integer mode.
10706 (store_split_bit_field): Add op0_mode and value_mode arguments.
10707 Update calls to extract_fixed_bit_field.
10708 (extract_bit_field_using_extv): Add an op0_mode argument.
10709 Use scalar_int_mode internally.
10710 (extract_bit_field_1): Rename the new integer mode from imode to
10711 op0_mode and use it instead of GET_MODE (op0). Update calls to
10712 extract_split_bit_field, extract_bit_field_using_extv and
10713 extract_fixed_bit_field.
10714 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10715 to extract_split_bit_field and extract_fixed_bit_field_1.
10716 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10717 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10718 on the target mode.
10719 (extract_split_bit_field): Add an op0_mode argument. Update call
10720 to extract_fixed_bit_field.
10722 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10723 Alan Hayward <alan.hayward@arm.com>
10724 David Sherwood <david.sherwood@arm.com>
10726 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10727 * explow.c (hard_function_value): Likewise.
10728 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10729 convert_to_mode call outside the loop.
10730 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10731 for the mode iterator. Require the mode specified by max_pieces
10733 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10735 (copy_blkmode_to_reg): Likewise.
10736 (set_storage_via_setmem): Likewise.
10737 * optabs.c (prepare_cmp_insn): Likewise.
10738 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10739 * stor-layout.c (finish_bitfield_representative): Likewise.
10741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10742 Alan Hayward <alan.hayward@arm.com>
10743 David Sherwood <david.sherwood@arm.com>
10745 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10746 * expr.c (convert_move): Use them.
10747 (convert_modes): Likewise.
10748 (store_expr_with_bounds): Likewise.
10750 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10751 Alan Hayward <alan.hayward@arm.com>
10752 David Sherwood <david.sherwood@arm.com>
10754 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10755 parameter for the mode of "x". Remove the "known_x", "known_mode"
10756 and "known_ret" arguments. Change the type of the mode argument
10757 to scalar_int_mode.
10758 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10759 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10760 (reg_num_sign_bit_copies_for_combine): Likewise.
10761 * rtlanal.c (nonzero_bits1): Likewise.
10762 (num_sign_bit_copies1): Likewise.
10763 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10764 (reg_num_sign_bit_copies_general): Likewise.
10765 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10766 (reg_nonzero_bits_general): Likewise.
10768 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10769 Alan Hayward <alan.hayward@arm.com>
10770 David Sherwood <david.sherwood@arm.com>
10772 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10773 than in subroutines. Return 1 for non-integer modes.
10774 (cached_num_sign_bit_copies): Change the type of the mode parameter
10775 to scalar_int_mode.
10776 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10777 classes. Handle CONST_INT_P first and then check whether X also
10778 has a scalar integer mode. Check the same thing for inner registers
10779 of a SUBREG and for values that are being extended or truncated.
10781 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10782 Alan Hayward <alan.hayward@arm.com>
10783 David Sherwood <david.sherwood@arm.com>
10785 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10786 in subroutines. Return the mode mask for non-integer modes.
10787 (cached_nonzero_bits): Change the type of the mode parameter
10788 to scalar_int_mode.
10789 (nonzero_bits1): Likewise. Remove early exit for other mode
10790 classes. Handle CONST_INT_P first and then check whether X
10791 also has a scalar integer mode.
10793 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10794 Alan Hayward <alan.hayward@arm.com>
10795 David Sherwood <david.sherwood@arm.com>
10797 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10798 Return a scalar_int_mode and assert that the size is greater than
10800 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10801 (op_by_pieces_d::op_by_pieces_d): Likewise.
10802 (op_by_pieces_d::run): Likewise.
10803 (can_store_by_pieces): Likewise.
10805 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10806 Alan Hayward <alan.hayward@arm.com>
10807 David Sherwood <david.sherwood@arm.com>
10809 * combine.c (extract_left_shift): Add a mode argument and update
10811 (make_compound_operation_int): Change the type of the mode parameter
10812 to scalar_int_mode and update the call to extract_left_shift.
10814 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10815 Alan Hayward <alan.hayward@arm.com>
10816 David Sherwood <david.sherwood@arm.com>
10818 * combine.c (simplify_and_const_int): Change the type of the mode
10819 parameter to scalar_int_mode.
10820 (simplify_and_const_int_1): Likewise. Update recursive call.
10822 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10823 Alan Hayward <alan.hayward@arm.com>
10824 David Sherwood <david.sherwood@arm.com>
10826 * combine.c (simplify_compare_const): Check that the mode is a
10827 scalar_int_mode (rather than VOIDmode) before testing its
10829 (simplify_comparison): Move COMPARISON_P handling out of the
10830 loop and restrict the latter part of the loop to scalar_int_modes.
10831 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10832 and when considering SUBREG_REGs. Use is_int_mode instead of
10833 checking GET_MODE_CLASS against MODE_INT.
10835 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10836 Alan Hayward <alan.hayward@arm.com>
10837 David Sherwood <david.sherwood@arm.com>
10839 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10840 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10841 shift_unit_mode and for modes involved in scalar shifts.
10843 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10844 Alan Hayward <alan.hayward@arm.com>
10845 David Sherwood <david.sherwood@arm.com>
10847 * combine.c (force_int_to_mode): New function, split out from...
10848 (force_to_mode): ...here. Keep xmode up-to-date and use it
10849 instead of GET_MODE (x).
10851 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10852 Alan Hayward <alan.hayward@arm.com>
10853 David Sherwood <david.sherwood@arm.com>
10855 * optabs-query.h (extraction_insn::struct_mode): Change type to
10856 opt_scalar_int_mode and update comment.
10857 (extraction_insn::field_mode): Change type to scalar_int_mode.
10858 (extraction_insn::pos_mode): Likewise.
10859 * combine.c (make_extraction): Update accordingly.
10860 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10861 (get_optab_extraction_insn): Likewise.
10862 * recog.c (simplify_while_replacing): Likewise.
10863 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10864 parameter to opt_scalar_int_mode.
10866 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10867 Alan Hayward <alan.hayward@arm.com>
10868 David Sherwood <david.sherwood@arm.com>
10870 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10871 to a scalar_int_mode instead of a machine_mode.
10872 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10873 (get_best_mode): Return a boolean and use a pointer argument to store
10874 the selected mode. Replace the limit mode parameter with a bit limit.
10875 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10876 for the values returned by bit_field_mode_iterator::next_mode.
10877 (store_bit_field): Update call to get_best_mode.
10878 (store_fixed_bit_field): Likewise.
10879 (extract_fixed_bit_field): Likewise.
10880 * expr.c (optimize_bitfield_assignment_op): Likewise.
10881 * fold-const.c (optimize_bit_field_compare): Likewise.
10882 (fold_truth_andor_1): Likewise.
10883 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10884 Update for new type of m_mode.
10885 (get_best_mode): As above.
10887 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10888 Alan Hayward <alan.hayward@arm.com>
10889 David Sherwood <david.sherwood@arm.com>
10891 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10892 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10893 (store_bit_field): Check is_a <scalar_int_mode> before calling
10894 strict_volatile_bitfield_p.
10895 (extract_bit_field): Likewise.
10897 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10898 Alan Hayward <alan.hayward@arm.com>
10899 David Sherwood <david.sherwood@arm.com>
10901 * target.def (cstore_mode): Return a scalar_int_mode.
10902 * doc/tm.texi: Regenerate.
10903 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10904 * targhooks.h (default_cstore_mode): Likewise.
10905 * targhooks.c (default_cstore_mode): Likewise, using a forced
10907 * expmed.c (emit_cstore): Expect the target of the cstore to be
10910 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10911 Alan Hayward <alan.hayward@arm.com>
10912 David Sherwood <david.sherwood@arm.com>
10914 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10916 (niter_desc): Likewise mode.
10917 (iv_analyze): Add a mode parameter.
10919 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10920 and change its type to scalar_int_mode.
10921 * loop-iv.c: Update commentary at head of file.
10922 (iv_constant): Pass the mode paraeter before the rtx it describes
10923 and change its type to scalar_int_mode. Remove VOIDmode handling.
10924 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10925 (iv_extend): Likewise.
10926 (shorten_into_mode): Likewise.
10927 (iv_add): Use scalar_int_mode.
10928 (iv_mult): Likewise.
10929 (iv_shift): Likewise.
10930 (canonicalize_iv_subregs): Likewise.
10931 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10932 it describes and change its mode to scalar_int_mode. Also change
10933 the type of the returned inner_mode to scalar_int_mode.
10934 (get_biv_step): Likewise, turning outer_mode from a pointer
10935 into a direct parameter. Update call to get_biv_step_1.
10936 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10937 iv_constant and get_biv_step.
10938 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10939 and change its type to scalar_int_mode. Don't initialise iv->mode
10940 to VOIDmode and remove later checks for its still being VOIDmode.
10941 Update calls to iv_analyze_op and iv_analyze_expr. Check
10942 is_a <scalar_int_mode> when changing the mode under consideration.
10943 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10944 Update call to iv_analyze_expr.
10945 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10946 inner register is not also a scalar_int_mode. Update call to
10948 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10949 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10950 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10951 separate mode class checks. Update calls to iv_analyze. Remove
10952 fix-up of VOIDmodes after iv_analyze_biv.
10953 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10954 don't have a scalar_int_mode. Update call to biv_p.
10956 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10957 Alan Hayward <alan.hayward@arm.com>
10958 David Sherwood <david.sherwood@arm.com>
10960 * cfgexpand.c (convert_debug_memory_address): Use
10961 as_a <scalar_int_mode>.
10962 * combine.c (expand_compound_operation): Likewise.
10963 (make_extraction): Likewise.
10964 (change_zero_ext): Likewise.
10965 (simplify_comparison): Likewise.
10966 * cse.c (cse_insn): Likewise.
10967 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10968 (mem_loc_descriptor): Likewise.
10969 (loc_descriptor): Likewise.
10970 * expmed.c (init_expmed_one_mode): Likewise.
10971 (synth_mult): Likewise.
10972 (emit_store_flag_1): Likewise.
10973 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10974 of a comparison with size.
10975 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10976 (reduce_to_bit_field_precision): Likewise.
10977 * function.c (expand_function_end): Likewise.
10978 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10979 * loop-doloop.c (doloop_modify): Likewise.
10980 * optabs.c (expand_binop): Likewise.
10981 (expand_unop): Likewise.
10982 (expand_copysign_absneg): Likewise.
10983 (prepare_cmp_insn): Likewise.
10984 (maybe_legitimize_operand): Likewise.
10985 * recog.c (const_scalar_int_operand): Likewise.
10986 * rtlanal.c (get_address_mode): Likewise.
10987 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10988 (simplify_cond_clz_ctz): Likewise.
10989 * tree-nested.c (get_nl_goto_field): Likewise.
10990 * tree.c (build_vector_type_for_mode): Likewise.
10991 * var-tracking.c (use_narrower_mode): Likewise.
10993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10994 Alan Hayward <alan.hayward@arm.com>
10995 David Sherwood <david.sherwood@arm.com>
10997 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10998 * builtins.c (expand_builtin_signbit): Use it.
10999 * cfgexpand.c (expand_debug_expr): Likewise.
11000 * dojump.c (do_jump): Likewise.
11001 (do_compare_and_jump): Likewise.
11002 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11003 * expmed.c (make_tree): Likewise.
11004 * expr.c (expand_expr_real_2): Likewise.
11005 (expand_expr_real_1): Likewise.
11006 (try_casesi): Likewise.
11007 * fold-const-call.c (fold_const_call_ss): Likewise.
11008 * fold-const.c (unextend): Likewise.
11009 (extract_muldiv_1): Likewise.
11010 (fold_single_bit_test): Likewise.
11011 (native_encode_int): Likewise.
11012 (native_encode_string): Likewise.
11013 (native_interpret_int): Likewise.
11014 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11015 * internal-fn.c (expand_addsub_overflow): Likewise.
11016 (expand_neg_overflow): Likewise.
11017 (expand_mul_overflow): Likewise.
11018 (expand_arith_overflow): Likewise.
11019 * match.pd: Likewise.
11020 * stor-layout.c (layout_type): Likewise.
11021 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11022 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11023 * tree-ssanames.c (get_range_info): Likewise.
11024 * tree-switch-conversion.c (array_value_type) Likewise.
11025 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11026 (vect_recog_divmod_pattern): Likewise.
11027 (vect_recog_mixed_size_cond_pattern): Likewise.
11028 * tree-vrp.c (extract_range_basic): Likewise.
11029 (simplify_float_conversion_using_ranges): Likewise.
11030 * tree.c (int_fits_type_p): Likewise.
11031 * ubsan.c (instrument_bool_enum_load): Likewise.
11032 * varasm.c (mergeable_string_section): Likewise.
11033 (narrowing_initializer_constant_valid_p): Likewise.
11034 (output_constant): Likewise.
11036 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11037 Alan Hayward <alan.hayward@arm.com>
11038 David Sherwood <david.sherwood@arm.com>
11040 * machmode.h (NARROWEST_INT_MODE): New macro.
11041 * expr.c (alignment_for_piecewise_move): Use it instead of
11042 GET_CLASS_NARROWEST_MODE (MODE_INT).
11043 (push_block): Likewise.
11044 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11046 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11049 Alan Hayward <alan.hayward@arm.com>
11050 David Sherwood <david.sherwood@arm.com>
11052 * postreload.c (move2add_valid_value_p): Change the type of the
11053 mode parameter to scalar_int_mode.
11054 (move2add_use_add2_insn): Add a mode parameter and use it instead
11056 (move2add_use_add3_insn): Likewise.
11057 (reload_cse_move2add): Update accordingly.
11059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11060 Alan Hayward <alan.hayward@arm.com>
11061 David Sherwood <david.sherwood@arm.com>
11063 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11065 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11066 * optabs.c (expand_unop): Likewise.
11068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11069 Alan Hayward <alan.hayward@arm.com>
11070 David Sherwood <david.sherwood@arm.com>
11072 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11073 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11074 (popcount_loc_descriptor): Likewise.
11075 (bswap_loc_descriptor): Likewise.
11076 (rotate_loc_descriptor): Likewise.
11077 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11078 calling the functions above.
11080 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11081 Alan Hayward <alan.hayward@arm.com>
11082 David Sherwood <david.sherwood@arm.com>
11084 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11086 (try_combine): Likewise.
11087 (simplify_if_then_else): Likewise.
11088 * cse.c (cse_insn): Likewise.
11089 * dwarf2out.c (mem_loc_descriptor): Likewise.
11090 * emit-rtl.c (gen_lowpart_common): Likewise.
11091 * simplify-rtx.c (simplify_truncation): Likewise.
11092 (simplify_binary_operation_1): Likewise.
11093 (simplify_const_relational_operation): Likewise.
11094 (simplify_ternary_operation): Likewise.
11095 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11098 Alan Hayward <alan.hayward@arm.com>
11099 David Sherwood <david.sherwood@arm.com>
11101 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11102 * reload.c (push_reload): Likewise.
11103 (find_reloads): Likewise.
11105 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11106 Alan Hayward <alan.hayward@arm.com>
11107 David Sherwood <david.sherwood@arm.com>
11109 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11110 (make_compound_operation_int): Likewise.
11111 (change_zero_ext): Likewise.
11112 * expr.c (convert_move): Likewise.
11113 (convert_modes): Likewise.
11114 * fwprop.c (forward_propagate_subreg): Likewise.
11115 * loop-iv.c (get_biv_step_1): Likewise.
11116 * optabs.c (widen_operand): Likewise.
11117 * postreload.c (move2add_valid_value_p): Likewise.
11118 * recog.c (simplify_while_replacing): Likewise.
11119 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11120 (simplify_binary_operation_1): Likewise. Remove redundant
11121 mode equality check.
11123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11124 Alan Hayward <alan.hayward@arm.com>
11125 David Sherwood <david.sherwood@arm.com>
11127 * combine.c (combine_simplify_rtx): Add checks for
11128 is_a <scalar_int_mode>.
11129 (simplify_if_then_else): Likewise.
11130 (make_field_assignment): Likewise.
11131 (simplify_comparison): Likewise.
11132 * ifcvt.c (noce_try_bitop): Likewise.
11133 * loop-invariant.c (canonicalize_address_mult): Likewise.
11134 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11137 Alan Hayward <alan.hayward@arm.com>
11138 David Sherwood <david.sherwood@arm.com>
11140 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11141 is_a <scalar_int_mode> instead of != BLKmode.
11143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11144 Alan Hayward <alan.hayward@arm.com>
11145 David Sherwood <david.sherwood@arm.com>
11147 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11148 instead of != VOIDmode.
11149 * combine.c (if_then_else_cond): Likewise.
11150 (change_zero_ext): Likewise.
11151 * dwarf2out.c (mem_loc_descriptor): Likewise.
11152 (loc_descriptor): Likewise.
11153 * rtlanal.c (canonicalize_condition): Likewise.
11154 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11157 Alan Hayward <alan.hayward@arm.com>
11158 David Sherwood <david.sherwood@arm.com>
11160 * simplify-rtx.c (simplify_binary_operation_1): Use
11161 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11163 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11164 Alan Hayward <alan.hayward@arm.com>
11165 David Sherwood <david.sherwood@arm.com>
11167 * wide-int.h (int_traits<unsigned char>) New class.
11168 (int_traits<unsigned short>) Likewise.
11169 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11170 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11172 * combine.c (set_nonzero_bits_and_sign_copies): Use
11173 is_a <scalar_int_mode>.
11174 (find_split_point): Likewise.
11175 (combine_simplify_rtx): Likewise.
11176 (simplify_logical): Likewise.
11177 (expand_compound_operation): Likewise.
11178 (expand_field_assignment): Likewise.
11179 (make_compound_operation): Likewise.
11180 (extended_count): Likewise.
11181 (change_zero_ext): Likewise.
11182 (simplify_comparison): Likewise.
11183 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11184 (ucompare_loc_descriptor): Likewise.
11185 (minmax_loc_descriptor): Likewise.
11186 (mem_loc_descriptor): Likewise.
11187 (loc_descriptor): Likewise.
11188 * expmed.c (init_expmed_one_mode): Likewise.
11189 * lra-constraints.c (lra_constraint_offset): Likewise.
11190 * optabs.c (prepare_libcall_arg): Likewise.
11191 * postreload.c (move2add_note_store): Likewise.
11192 * reload.c (operands_match_p): Likewise.
11193 * rtl.h (load_extend_op): Likewise.
11194 * rtlhooks.c (gen_lowpart_general): Likewise.
11195 * simplify-rtx.c (simplify_truncation): Likewise.
11196 (simplify_unary_operation_1): Likewise.
11197 (simplify_binary_operation_1): Likewise.
11198 (simplify_const_binary_operation): Likewise.
11199 (simplify_const_relational_operation): Likewise.
11200 (simplify_subreg): Likewise.
11201 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11202 * var-tracking.c (adjust_mems): Likewise.
11203 (prepare_call_arguments): Likewise.
11205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11206 Alan Hayward <alan.hayward@arm.com>
11207 David Sherwood <david.sherwood@arm.com>
11209 * machmode.h (is_int_mode): New fuction.
11210 * combine.c (find_split_point): Use it.
11211 (combine_simplify_rtx): Likewise.
11212 (simplify_if_then_else): Likewise.
11213 (simplify_set): Likewise.
11214 (simplify_shift_const_1): Likewise.
11215 (simplify_comparison): Likewise.
11216 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11217 * cse.c (notreg_cost): Likewise.
11218 (cse_insn): Likewise.
11219 * cselib.c (cselib_lookup_1): Likewise.
11220 * dojump.c (do_jump_1): Likewise.
11221 (do_compare_rtx_and_jump): Likewise.
11222 * dse.c (get_call_args): Likewise.
11223 * dwarf2out.c (rtl_for_decl_init): Likewise.
11224 (native_encode_initializer): Likewise.
11225 * expmed.c (emit_store_flag_1): Likewise.
11226 (emit_store_flag): Likewise.
11227 * expr.c (convert_modes): Likewise.
11228 (store_field): Likewise.
11229 (expand_expr_real_1): Likewise.
11230 * fold-const.c (fold_read_from_constant_string): Likewise.
11231 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11232 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11233 * optabs.c (expand_binop): Likewise.
11234 (expand_unop): Likewise.
11235 (expand_abs_nojump): Likewise.
11236 (expand_one_cmpl_abs_nojump): Likewise.
11237 * simplify-rtx.c (mode_signbit_p): Likewise.
11238 (val_signbit_p): Likewise.
11239 (val_signbit_known_set_p): Likewise.
11240 (val_signbit_known_clear_p): Likewise.
11241 (simplify_relational_operation_1): Likewise.
11242 * tree.c (vector_type_mode): Likewise.
11244 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11245 Alan Hayward <alan.hayward@arm.com>
11246 David Sherwood <david.sherwood@arm.com>
11248 * machmode.h (smallest_mode_for_size): Fix formatting.
11249 (smallest_int_mode_for_size): New function.
11250 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11251 instead of smallest_mode_for_size.
11252 * combine.c (make_extraction): Likewise.
11253 * config/arc/arc.c (arc_expand_movmem): Likewise.
11254 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11255 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11256 * config/s390/s390.c (s390_expand_insv): Likewise.
11257 * config/sparc/sparc.c (assign_int_registers): Likewise.
11258 * config/spu/spu.c (spu_function_value): Likewise.
11259 (spu_function_arg): Likewise.
11260 * coverage.c (get_gcov_type): Likewise.
11261 (get_gcov_unsigned_t): Likewise.
11262 * dse.c (find_shift_sequence): Likewise.
11263 * expmed.c (store_bit_field_1): Likewise.
11264 * expr.c (convert_move): Likewise.
11265 (store_field): Likewise.
11266 * internal-fn.c (expand_arith_overflow): Likewise.
11267 * optabs-query.c (get_best_extraction_insn): Likewise.
11268 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11269 * stor-layout.c (layout_type): Likewise.
11270 (initialize_sizetypes): Likewise.
11271 * targhooks.c (default_get_mask_mode): Likewise.
11272 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11275 Alan Hayward <alan.hayward@arm.com>
11276 David Sherwood <david.sherwood@arm.com>
11278 * machmode.h (opt_mode::else_blk): New function.
11279 (int_mode_for_mode): Declare.
11280 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11281 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11283 * cfgexpand.c (expand_debug_expr): Likewise.
11284 * combine.c (gen_lowpart_or_truncate): Likewise.
11285 (gen_lowpart_for_combine): Likewise.
11286 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11287 * config/avr/avr.c (avr_to_int_mode): Likewise.
11288 (avr_out_plus_1): Likewise.
11289 (avr_out_plus): Likewise.
11290 (avr_out_round): Likewise.
11291 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11292 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11293 (s390_expand_vcond): Likewise.
11294 * config/spu/spu.c (spu_split_immediate): Likewise.
11295 (spu_expand_mov): Likewise.
11296 * dse.c (get_stored_val): Likewise.
11297 * expmed.c (store_bit_field_1): Likewise.
11298 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11300 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11301 (extract_low_bits): Likewise.
11302 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11303 handling rather than repeating the check.
11304 (emit_group_store): Likewise.
11305 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11306 * optabs.c (expand_absneg_bit): Likewise.
11307 (expand_copysign_absneg): Likewise.
11308 (expand_copysign_bit): Likewise.
11309 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11310 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11311 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11312 * var-tracking.c (prepare_call_arguments): Likewise.
11313 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11314 int_mode_for_mode instead of mode_for_size.
11315 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11317 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11318 Alan Hayward <alan.hayward@arm.com>
11319 David Sherwood <david.sherwood@arm.com>
11321 * machmode.h (int_mode_for_size): New function.
11322 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11323 instead of mode_for_size.
11324 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11326 * combine.c (expand_field_assignment): Use int_mode_for_size
11327 instead of mode_for_size.
11328 (make_extraction): Likewise.
11329 (simplify_shift_const_1): Likewise.
11330 (simplify_comparison): Likewise.
11331 * dojump.c (do_jump): Likewise.
11332 * dwarf2out.c (mem_loc_descriptor): Likewise.
11333 * emit-rtl.c (init_derived_machine_modes): Likewise.
11334 * expmed.c (flip_storage_order): Likewise.
11335 (convert_extracted_bit_field): Likewise.
11336 * expr.c (copy_blkmode_from_reg): Likewise.
11337 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11338 * internal-fn.c (expand_mul_overflow): Likewise.
11339 * lower-subreg.c (simple_move): Likewise.
11340 * optabs-libfuncs.c (init_optabs): Likewise.
11341 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11342 * tree.c (vector_type_mode): Likewise.
11343 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11344 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11345 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11346 * tree-vect-stmts.c (vectorizable_load): Likewise.
11347 (vectorizable_store): Likewise.
11349 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11350 Alan Hayward <alan.hayward@arm.com>
11351 David Sherwood <david.sherwood@arm.com>
11353 * coretypes.h (pod_mode): New type.
11354 (scalar_int_mode_pod): New typedef.
11355 * machmode.h (pod_mode): New class.
11356 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11357 * genmodes.c (emit_mode_int_n): Update accordingly.
11358 * lower-subreg.h (target_lower_subreg): Change type to
11359 scalar_int_mode_pod.
11360 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11361 scalar_int_mode_pod.
11363 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11364 Alan Hayward <alan.hayward@arm.com>
11365 David Sherwood <david.sherwood@arm.com>
11367 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11368 machine_mode to scalar_int_mode.
11369 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11370 (rs6000_option_override_internal): Remove cast to int.
11371 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11372 machine_mode to scalar_int_mode.
11373 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11374 (rs6000_option_override_internal): Remove cast to int.
11375 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11376 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11378 * config/s390/s390.c (s390_expand_builtin): Likewise.
11379 * coretypes.h (scalar_int_mode): New type.
11380 (opt_scalar_int_mode): New typedef.
11381 * machmode.h (scalar_int_mode): New class.
11382 (scalar_int_mode::includes_p): New function.
11383 (byte_mode): Change type to scalar_int_mode.
11384 (word_mode): Likewise.
11385 (ptr_mode): Likewise.
11386 * emit-rtl.c (byte_mode): Likewise.
11387 (word_mode): Likewise.
11388 (ptr_mode): Likewise.
11389 (init_derived_machine_modes): Update accordingly.
11390 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11391 and MODE_PARTIAL_INT.
11392 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11393 opt_scalar_int_mode.
11395 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11396 Alan Hayward <alan.hayward@arm.com>
11397 David Sherwood <david.sherwood@arm.com>
11399 * target.def (libgcc_floating_mode_supported_p): Take a
11401 * doc/tm.texi: Regenerate.
11402 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11404 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11405 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11408 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11409 Alan Hayward <alan.hayward@arm.com>
11410 David Sherwood <david.sherwood@arm.com>
11412 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11413 * doc/tm.texi: Regenerate.
11414 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11415 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11416 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11417 * targhooks.h (default_floatn_mode): Likewise.
11418 * targhooks.c (default_floatn_mode): Likewise.
11419 * tree.c (build_common_tree_nodes): Update accordingly.
11421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11422 Alan Hayward <alan.hayward@arm.com>
11423 David Sherwood <david.sherwood@arm.com>
11425 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11426 (mode_iterator::iterate_p): Likewise.
11427 (mode_iterator::get_wider): Likewise.
11428 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11430 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11431 Alan Hayward <alan.hayward@arm.com>
11432 David Sherwood <david.sherwood@arm.com>
11434 * coretypes.h (opt_scalar_float_mode): New typedef.
11435 * machmode.h (float_mode_for_size): New function.
11436 * emit-rtl.c (double_mode): Delete.
11437 (init_emit_once): Use float_mode_for_size.
11438 * stor-layout.c (layout_type): Likewise.
11439 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11441 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11442 Alan Hayward <alan.hayward@arm.com>
11443 David Sherwood <david.sherwood@arm.com>
11445 * output.h (assemble_real): Take a scalar_float_mode.
11446 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11447 * config/arm/arm.md (consttable_4): Likewise.
11448 (consttable_8): Likewise.
11449 (consttable_16): Likewise.
11450 * config/mips/mips.md (consttable_float): Likewise.
11451 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11452 * varasm.c (assemble_real): Take a scalar_float_mode.
11453 (output_constant_pool_2): Update accordingly.
11454 (output_constant): Likewise.
11456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11457 Alan Hayward <alan.hayward@arm.com>
11458 David Sherwood <david.sherwood@arm.com>
11460 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11461 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11462 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11463 (native_encode_real): Likewise.
11464 (native_interpret_real): Likewise.
11465 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11466 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11468 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11469 Alan Hayward <alan.hayward@arm.com>
11470 David Sherwood <david.sherwood@arm.com>
11472 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11473 <scalar_float_mode>. Simplify.
11474 (gen_extend_conv_libfunc): Likewise.
11476 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11477 Alan Hayward <alan.hayward@arm.com>
11478 David Sherwood <david.sherwood@arm.com>
11480 * coretypes.h (scalar_float_mode): New type.
11481 * machmode.h (mode_traits::from_int): Use machine_mode if
11482 USE_ENUM_MODES is defined.
11483 (is_a): New function.
11485 (dyn_cast): Likewise.
11486 (scalar_float_mode): New class.
11487 (scalar_float_mode::includes_p): New function.
11488 (is_float_mode): Likewise.
11489 * gdbhooks.py (MachineModePrinter): New class.
11490 (build_pretty_printer): Use it for scalar_float_mode.
11491 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11492 (format_helper::format_helper): Turn into a template.
11493 * genmodes.c (get_mode_class): New function.
11494 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11495 or machine_mode if none.
11496 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11497 as_a <scalar_float_mode>.
11498 * dwarf2out.c (mem_loc_descriptor): Likewise.
11499 (insert_float): Likewise.
11500 (add_const_value_attribute): Likewise.
11501 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11502 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11503 (expand_unop): Update accordingly.
11504 (expand_abs_nojump): Likewise.
11505 (expand_copysign_absneg): Take a scalar_float_mode.
11506 (expand_copysign_bit): Likewise.
11507 (expand_copysign): Update accordingly.
11509 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11510 Alan Hayward <alan.hayward@arm.com>
11511 David Sherwood <david.sherwood@arm.com>
11513 * coretypes.h (opt_mode): New class.
11514 * machmode.h (opt_mode): Likewise.
11515 (opt_mode::else_void): New function.
11516 (opt_mode::require): Likewise.
11517 (opt_mode::exists): Likewise.
11518 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11519 (GET_MODE_2XWIDER_MODE): Likewise.
11520 (mode_iterator::get_wider): Update accordingly.
11521 (mode_iterator::get_2xwider): Likewise.
11522 (mode_iterator::get_known_wider): Likewise, turning into a template.
11523 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11524 forcing a wider mode to exist.
11525 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11526 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11527 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11528 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11529 * lower-subreg.c (init_lower_subreg): Likewise.
11530 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11531 on the final iteration.
11532 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11533 a wider mode exists before asking for a move pattern.
11534 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11535 forcing a wider mode to exist.
11536 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11537 returning false if no such mode exists.
11538 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11539 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11540 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11541 Avoid checking for a MODE_INT if we already know the mode is not a
11543 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11544 forcing a wider mode to exist.
11545 (expmed_mult_highpart_optab): Likewise.
11546 (expmed_mult_highpart): Likewise.
11547 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11549 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11550 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11551 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11552 * internal-fn.c (expand_mul_overflow): Update use of
11553 GET_MODE_2XWIDER_MODE.
11554 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11555 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11556 GET_MODE_WIDER_MODE.
11557 (convert_plusminus_to_widen): Likewise.
11558 * tree-switch-conversion.c (array_value_type): Likewise.
11559 * var-tracking.c (emit_note_insn_var_location): Likewise.
11560 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11561 Return false inside rather than outside the loop if no wider mode
11563 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11564 and GET_MODE_2XWIDER_MODE
11565 (can_compare_p): Use else_void.
11566 * gdbhooks.py (OptMachineModePrinter): New class.
11567 (build_pretty_printer): Use it for opt_mode.
11569 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11570 Alan Hayward <alan.hayward@arm.com>
11571 David Sherwood <david.sherwood@arm.com>
11573 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11574 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11576 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11577 Alan Hayward <alan.hayward@arm.com>
11578 David Sherwood <david.sherwood@arm.com>
11580 * machmode.h (mode_traits): New structure.
11581 (get_narrowest_mode): New function.
11582 (mode_iterator::start): Likewise.
11583 (mode_iterator::iterate_p): Likewise.
11584 (mode_iterator::get_wider): Likewise.
11585 (mode_iterator::get_known_wider): Likewise.
11586 (mode_iterator::get_2xwider): Likewise.
11587 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11588 (FOR_EACH_MODE): Likewise.
11589 (FOR_EACH_MODE_FROM): Likewise.
11590 (FOR_EACH_MODE_UNTIL): Likewise.
11591 (FOR_EACH_WIDER_MODE): Likewise.
11592 (FOR_EACH_2XWIDER_MODE): Likewise.
11593 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11594 * combine.c (simplify_comparison): Likewise
11595 * config/i386/i386.c (type_natural_mode): Likewise.
11596 * cse.c (cse_insn): Likewise.
11597 * dse.c (find_shift_sequence): Likewise.
11598 * emit-rtl.c (init_derived_machine_modes): Likewise.
11599 (init_emit_once): Likewise.
11600 * explow.c (hard_function_value): Likewise.
11601 * expmed.c (extract_fixed_bit_field_1): Likewise.
11602 (extract_bit_field_1): Likewise.
11603 (expand_divmod): Likewise.
11604 (emit_store_flag_1): Likewise.
11605 * expr.c (init_expr_target): Likewise.
11606 (convert_move): Likewise.
11607 (alignment_for_piecewise_move): Likewise.
11608 (widest_int_mode_for_size): Likewise.
11609 (emit_block_move_via_movmem): Likewise.
11610 (copy_blkmode_to_reg): Likewise.
11611 (set_storage_via_setmem): Likewise.
11612 (compress_float_constant): Likewise.
11613 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11614 * optabs-query.c (get_best_extraction_insn): Likewise.
11615 * optabs.c (expand_binop): Likewise.
11616 (expand_twoval_unop): Likewise.
11617 (expand_twoval_binop): Likewise.
11618 (widen_leading): Likewise.
11619 (widen_bswap): Likewise.
11620 (expand_parity): Likewise.
11621 (expand_unop): Likewise.
11622 (prepare_cmp_insn): Likewise.
11623 (prepare_float_lib_cmp): Likewise.
11624 (expand_float): Likewise.
11625 (expand_fix): Likewise.
11626 (expand_sfix_optab): Likewise.
11627 * postreload.c (move2add_use_add2_insn): Likewise.
11628 * reg-stack.c (reg_to_stack): Likewise.
11629 * reginfo.c (choose_hard_reg_mode): Likewise.
11630 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11631 * stor-layout.c (mode_for_size): Likewise.
11632 (smallest_mode_for_size): Likewise.
11633 (mode_for_vector): Likewise.
11634 (finish_bitfield_representative): Likewise.
11635 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11636 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11637 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11638 * var-tracking.c (prepare_call_arguments): Likewise.
11640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11641 Alan Hayward <alan.hayward@arm.com>
11642 David Sherwood <david.sherwood@arm.com>
11644 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11645 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11646 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11647 * machmode.h (mode_size): Move earlier in file.
11648 (mode_precision): Likewise.
11649 (mode_inner): Likewise.
11650 (mode_nunits): Likewise.
11651 (mode_unit_size): Likewise.
11652 (unit_unit_precision): Likewise.
11653 (mode_wider): Likewise.
11654 (mode_2xwider): Likewise.
11655 (machine_mode): New class.
11656 (mode_to_bytes): New function.
11657 (mode_to_bits): Likewise.
11658 (mode_to_precision): Likewise.
11659 (mode_to_inner): Likewise.
11660 (mode_to_unit_size): Likewise.
11661 (mode_to_unit_precision): Likewise.
11662 (mode_to_nunits): Likewise.
11663 (GET_MODE_SIZE): Use mode_to_bytes.
11664 (GET_MODE_BITSIZE): Use mode_to_bits.
11665 (GET_MODE_PRECISION): Use mode_to_precision.
11666 (GET_MODE_INNER): Use mode_to_inner.
11667 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11668 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11669 (GET_MODE_NUNITS): Use mode_to_nunits.
11670 * system.h (ALWAYS_INLINE): New macro.
11671 * config/powerpcspe/powerpcspe-c.c
11672 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11673 int for arg1_mode and arg2_mode.
11675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11676 Alan Hayward <alan.hayward@arm.com>
11677 David Sherwood <david.sherwood@arm.com>
11679 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11680 Prefix mode names with E_ in case statements.
11681 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11682 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11683 (aarch64_split_simd_move): Likewise.
11684 (aarch64_gen_storewb_pair): Likewise.
11685 (aarch64_gen_loadwb_pair): Likewise.
11686 (aarch64_gen_store_pair): Likewise.
11687 (aarch64_gen_load_pair): Likewise.
11688 (aarch64_get_condition_code_1): Likewise.
11689 (aarch64_constant_pool_reload_icode): Likewise.
11690 (get_rsqrte_type): Likewise.
11691 (get_rsqrts_type): Likewise.
11692 (get_recpe_type): Likewise.
11693 (get_recps_type): Likewise.
11694 (aarch64_gimplify_va_arg_expr): Likewise.
11695 (aarch64_simd_container_mode): Likewise.
11696 (aarch64_emit_load_exclusive): Likewise.
11697 (aarch64_emit_store_exclusive): Likewise.
11698 (aarch64_expand_compare_and_swap): Likewise.
11699 (aarch64_gen_atomic_cas): Likewise.
11700 (aarch64_emit_bic): Likewise.
11701 (aarch64_emit_atomic_swap): Likewise.
11702 (aarch64_emit_atomic_load_op): Likewise.
11703 (aarch64_evpc_trn): Likewise.
11704 (aarch64_evpc_uzp): Likewise.
11705 (aarch64_evpc_zip): Likewise.
11706 (aarch64_evpc_ext): Likewise.
11707 (aarch64_evpc_rev): Likewise.
11708 (aarch64_evpc_dup): Likewise.
11709 (aarch64_gen_ccmp_first): Likewise.
11710 (aarch64_gen_ccmp_next): Likewise.
11711 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11712 (alpha_emit_xfloating_libcall): Likewise.
11713 (emit_insxl): Likewise.
11714 (alpha_arg_type): Likewise.
11715 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11716 (arc_preferred_simd_mode): Likewise.
11717 (arc_secondary_reload): Likewise.
11718 (get_arc_condition_code): Likewise.
11719 (arc_print_operand): Likewise.
11720 (arc_legitimate_constant_p): Likewise.
11721 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11722 * config/arc/arc.md (casesi_load): Likewise.
11723 (casesi_compact_jump): Likewise.
11724 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11725 (cc_use_register): Likewise.
11726 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11727 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11728 (arm_init_iwmmxt_builtins): Likewise.
11729 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11730 (neon_expand_vector_init): Likewise.
11731 (arm_attr_length_move_neon): Likewise.
11732 (maybe_get_arm_condition_code): Likewise.
11733 (arm_emit_vector_const): Likewise.
11734 (arm_preferred_simd_mode): Likewise.
11735 (arm_output_iwmmxt_tinsr): Likewise.
11736 (thumb1_output_casesi): Likewise.
11737 (thumb2_output_casesi): Likewise.
11738 (arm_emit_load_exclusive): Likewise.
11739 (arm_emit_store_exclusive): Likewise.
11740 (arm_expand_compare_and_swap): Likewise.
11741 (arm_evpc_neon_vuzp): Likewise.
11742 (arm_evpc_neon_vzip): Likewise.
11743 (arm_evpc_neon_vrev): Likewise.
11744 (arm_evpc_neon_vtrn): Likewise.
11745 (arm_evpc_neon_vext): Likewise.
11746 (arm_validize_comparison): Likewise.
11747 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11748 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11749 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11750 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11751 (c6x_preferred_simd_mode): Likewise.
11752 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11753 (epiphany_rtx_costs): Likewise.
11754 * config/epiphany/predicates.md (proper_comparison_operator):
11756 * config/frv/frv.c (condexec_memory_operand): Likewise.
11757 (frv_emit_move): Likewise.
11758 (output_move_single): Likewise.
11759 (output_condmove_single): Likewise.
11760 (frv_hard_regno_mode_ok): Likewise.
11761 (frv_matching_accg_mode): Likewise.
11762 * config/h8300/h8300.c (split_adds_subs): Likewise.
11763 (h8300_rtx_costs): Likewise.
11764 (h8300_print_operand): Likewise.
11765 (compute_mov_length): Likewise.
11766 (output_logical_op): Likewise.
11767 (compute_logical_op_length): Likewise.
11768 (compute_logical_op_cc): Likewise.
11769 (h8300_shift_needs_scratch_p): Likewise.
11770 (output_a_shift): Likewise.
11771 (compute_a_shift_length): Likewise.
11772 (compute_a_shift_cc): Likewise.
11773 (expand_a_rotate): Likewise.
11774 (output_a_rotate): Likewise.
11775 * config/i386/i386.c (classify_argument): Likewise.
11776 (function_arg_advance_32): Likewise.
11777 (function_arg_32): Likewise.
11778 (function_arg_64): Likewise.
11779 (function_value_64): Likewise.
11780 (ix86_gimplify_va_arg): Likewise.
11781 (ix86_legitimate_constant_p): Likewise.
11782 (put_condition_code): Likewise.
11783 (split_double_mode): Likewise.
11784 (ix86_avx256_split_vector_move_misalign): Likewise.
11785 (ix86_expand_vector_logical_operator): Likewise.
11786 (ix86_split_idivmod): Likewise.
11787 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11788 (ix86_build_const_vector): Likewise.
11789 (ix86_build_signbit_mask): Likewise.
11790 (ix86_match_ccmode): Likewise.
11791 (ix86_cc_modes_compatible): Likewise.
11792 (ix86_expand_branch): Likewise.
11793 (ix86_expand_sse_cmp): Likewise.
11794 (ix86_expand_sse_movcc): Likewise.
11795 (ix86_expand_int_sse_cmp): Likewise.
11796 (ix86_expand_vec_perm_vpermi2): Likewise.
11797 (ix86_expand_vec_perm): Likewise.
11798 (ix86_expand_sse_unpack): Likewise.
11799 (ix86_expand_int_addcc): Likewise.
11800 (ix86_split_to_parts): Likewise.
11801 (ix86_vectorize_builtin_gather): Likewise.
11802 (ix86_vectorize_builtin_scatter): Likewise.
11803 (avx_vpermilp_parallel): Likewise.
11804 (inline_memory_move_cost): Likewise.
11805 (ix86_tieable_integer_mode_p): Likewise.
11806 (x86_maybe_negate_const_int): Likewise.
11807 (ix86_expand_vector_init_duplicate): Likewise.
11808 (ix86_expand_vector_init_one_nonzero): Likewise.
11809 (ix86_expand_vector_init_one_var): Likewise.
11810 (ix86_expand_vector_init_concat): Likewise.
11811 (ix86_expand_vector_init_interleave): Likewise.
11812 (ix86_expand_vector_init_general): Likewise.
11813 (ix86_expand_vector_set): Likewise.
11814 (ix86_expand_vector_extract): Likewise.
11815 (emit_reduc_half): Likewise.
11816 (ix86_emit_i387_round): Likewise.
11817 (ix86_mangle_type): Likewise.
11818 (ix86_expand_round_sse4): Likewise.
11819 (expand_vec_perm_blend): Likewise.
11820 (canonicalize_vector_int_perm): Likewise.
11821 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11822 (expand_vec_perm_1): Likewise.
11823 (expand_vec_perm_interleave3): Likewise.
11824 (expand_vec_perm_even_odd_pack): Likewise.
11825 (expand_vec_perm_even_odd_1): Likewise.
11826 (expand_vec_perm_broadcast_1): Likewise.
11827 (ix86_vectorize_vec_perm_const_ok): Likewise.
11828 (ix86_expand_vecop_qihi): Likewise.
11829 (ix86_expand_mul_widen_hilo): Likewise.
11830 (ix86_expand_sse2_abs): Likewise.
11831 (ix86_expand_pextr): Likewise.
11832 (ix86_expand_pinsr): Likewise.
11833 (ix86_preferred_simd_mode): Likewise.
11834 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11835 * config/i386/sse.md (*andnot<mode>3): Likewise.
11836 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11837 (*<code><mode>3): Likewise.
11838 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11839 (ia64_expand_atomic_op): Likewise.
11840 (ia64_arg_type): Likewise.
11841 (ia64_mode_to_int): Likewise.
11842 (ia64_scalar_mode_supported_p): Likewise.
11843 (ia64_vector_mode_supported_p): Likewise.
11844 (expand_vec_perm_broadcast): Likewise.
11845 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11846 (iq2000_function_arg_advance): Likewise.
11847 (iq2000_function_arg): Likewise.
11848 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11849 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11850 (m68k_libcall_value): Likewise.
11851 (m68k_function_value): Likewise.
11852 (sched_attr_op_type): Likewise.
11853 * config/mcore/mcore.c (mcore_output_move): Likewise.
11854 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11856 (microblaze_function_arg): Likewise.
11857 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11858 (mips_print_operand): Likewise.
11859 (mips_mode_ok_for_mov_fmt_p): Likewise.
11860 (mips_vector_mode_supported_p): Likewise.
11861 (mips_preferred_simd_mode): Likewise.
11862 (mips_expand_vpc_loongson_even_odd): Likewise.
11863 (mips_expand_vec_unpack): Likewise.
11864 (mips_expand_vi_broadcast): Likewise.
11865 (mips_expand_vector_init): Likewise.
11866 (mips_expand_vec_reduc): Likewise.
11867 (mips_expand_msa_cmp): Likewise.
11868 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11869 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11870 (cc_flags_for_mode): Likewise.
11871 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11872 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11873 (nds32_output_casesi_pc_relative): Likewise.
11874 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11875 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11876 (nvptx_gen_unpack): Likewise.
11877 (nvptx_gen_pack): Likewise.
11878 (nvptx_gen_shuffle): Likewise.
11879 (nvptx_gen_wcast): Likewise.
11880 (nvptx_preferred_simd_mode): Likewise.
11881 * config/pa/pa.c (pa_secondary_reload): Likewise.
11882 * config/pa/predicates.md (base14_operand): Likewise.
11883 * config/powerpcspe/powerpcspe-c.c
11884 (altivec_resolve_overloaded_builtin): Likewise.
11885 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11887 (rs6000_preferred_simd_mode): Likewise.
11888 (output_vec_const_move): Likewise.
11889 (rs6000_expand_vector_extract): Likewise.
11890 (rs6000_split_vec_extract_var): Likewise.
11891 (reg_offset_addressing_ok_p): Likewise.
11892 (rs6000_legitimate_offset_address_p): Likewise.
11893 (rs6000_legitimize_address): Likewise.
11894 (rs6000_emit_set_const): Likewise.
11895 (rs6000_const_vec): Likewise.
11896 (rs6000_emit_move): Likewise.
11897 (spe_build_register_parallel): Likewise.
11898 (rs6000_darwin64_record_arg_recurse): Likewise.
11899 (swap_selector_for_mode): Likewise.
11900 (spe_init_builtins): Likewise.
11901 (paired_init_builtins): Likewise.
11902 (altivec_init_builtins): Likewise.
11903 (do_load_for_compare): Likewise.
11904 (rs6000_generate_compare): Likewise.
11905 (rs6000_expand_float128_convert): Likewise.
11906 (emit_load_locked): Likewise.
11907 (emit_store_conditional): Likewise.
11908 (rs6000_output_function_epilogue): Likewise.
11909 (rs6000_handle_altivec_attribute): Likewise.
11910 (rs6000_function_value): Likewise.
11911 (emit_fusion_gpr_load): Likewise.
11912 (emit_fusion_p9_load): Likewise.
11913 (emit_fusion_p9_store): Likewise.
11914 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11915 (fusion_gpr_mem_load): Likewise.
11916 (fusion_addis_mem_combo_load): Likewise.
11917 (fusion_addis_mem_combo_store): Likewise.
11918 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11919 (fusion_gpr_mem_load): Likewise.
11920 (fusion_addis_mem_combo_load): Likewise.
11921 (fusion_addis_mem_combo_store): Likewise.
11922 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11924 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11925 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11926 (rs6000_preferred_simd_mode): Likewise.
11927 (output_vec_const_move): Likewise.
11928 (rs6000_expand_vector_extract): Likewise.
11929 (rs6000_split_vec_extract_var): Likewise.
11930 (reg_offset_addressing_ok_p): Likewise.
11931 (rs6000_legitimate_offset_address_p): Likewise.
11932 (rs6000_legitimize_address): Likewise.
11933 (rs6000_emit_set_const): Likewise.
11934 (rs6000_const_vec): Likewise.
11935 (rs6000_emit_move): Likewise.
11936 (rs6000_darwin64_record_arg_recurse): Likewise.
11937 (swap_selector_for_mode): Likewise.
11938 (paired_init_builtins): Likewise.
11939 (altivec_init_builtins): Likewise.
11940 (rs6000_expand_float128_convert): Likewise.
11941 (emit_load_locked): Likewise.
11942 (emit_store_conditional): Likewise.
11943 (rs6000_output_function_epilogue): Likewise.
11944 (rs6000_handle_altivec_attribute): Likewise.
11945 (rs6000_function_value): Likewise.
11946 (emit_fusion_gpr_load): Likewise.
11947 (emit_fusion_p9_load): Likewise.
11948 (emit_fusion_p9_store): Likewise.
11949 * config/rx/rx.c (rx_gen_move_template): Likewise.
11950 (flags_from_mode): Likewise.
11951 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11952 (s390_slb_comparison): Likewise.
11953 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11954 (s390_vector_mode_supported_p): Likewise.
11955 (s390_cc_modes_compatible): Likewise.
11956 (s390_match_ccmode_set): Likewise.
11957 (s390_canonicalize_comparison): Likewise.
11958 (s390_emit_compare_and_swap): Likewise.
11959 (s390_branch_condition_mask): Likewise.
11960 (s390_rtx_costs): Likewise.
11961 (s390_secondary_reload): Likewise.
11962 (__SECONDARY_RELOAD_CASE): Likewise.
11963 (s390_expand_cs): Likewise.
11964 (s390_preferred_simd_mode): Likewise.
11965 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11966 * config/sh/sh.c (sh_print_operand): Likewise.
11967 (dump_table): Likewise.
11968 (sh_secondary_reload): Likewise.
11969 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11970 * config/sh/sh.md (casesi_worker_1): Likewise.
11971 (casesi_worker_2): Likewise.
11972 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11973 (fcc_comparison_operator): Likewise.
11974 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11975 (emit_soft_tfmode_cvt): Likewise.
11976 (sparc_preferred_simd_mode): Likewise.
11977 (output_cbranch): Likewise.
11978 (sparc_print_operand): Likewise.
11979 (sparc_expand_vec_perm_bmask): Likewise.
11980 (vector_init_bshuffle): Likewise.
11981 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11982 (spu_vector_mode_supported_p): Likewise.
11983 (spu_expand_insv): Likewise.
11984 (spu_emit_branch_or_set): Likewise.
11985 (spu_handle_vector_attribute): Likewise.
11986 (spu_builtin_splats): Likewise.
11987 (spu_builtin_extract): Likewise.
11988 (spu_builtin_promote): Likewise.
11989 (spu_expand_sign_extend): Likewise.
11990 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11991 (tilegx_simd_int): Likewise.
11992 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11993 (tilepro_simd_int): Likewise.
11994 * config/v850/v850.c (const_double_split): Likewise.
11995 (v850_print_operand): Likewise.
11996 (ep_memory_offset): Likewise.
11997 * config/vax/vax.c (vax_rtx_costs): Likewise.
11998 (vax_output_int_move): Likewise.
11999 (vax_output_int_add): Likewise.
12000 (vax_output_int_subtract): Likewise.
12001 * config/visium/predicates.md (visium_branch_operator): Likewise.
12002 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12003 (visium_print_operand_address): Likewise.
12004 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12005 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12006 (xtensa_expand_conditional_branch): Likewise.
12007 (xtensa_copy_incoming_a7): Likewise.
12008 (xtensa_output_literal): Likewise.
12009 * dfp.c (decimal_real_maxval): Likewise.
12010 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12013 Alan Hayward <alan.hayward@arm.com>
12014 David Sherwood <david.sherwood@arm.com>
12016 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12017 (mode_nunits_inline): Likewise.
12018 (mode_inner_inline): Likewise.
12019 (mode_unit_size_inline): Likewise.
12020 (mode_unit_precision_inline): Likewise.
12021 (emit_insn_modes_h): Likewise. Also emit a #define of the
12023 (emit_mode_wider): Add an E_ prefix to mode names.
12024 (emit_mode_complex): Likewise.
12025 (emit_mode_inner): Likewise.
12026 (emit_mode_adjustments): Likewise.
12027 (emit_mode_int_n): Likewise.
12028 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12029 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12030 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12031 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12032 (CRC32_BUILTIN, ENTRY): Likewise.
12033 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12034 (aarch64_pop_regs): Likewise.
12035 (aarch64_process_components): Likewise.
12036 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12037 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12038 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12039 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12040 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12041 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12043 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12044 (pcmpistr): Likewise.
12045 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12046 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12047 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12049 * config/rl78/rl78.c (mduc_regs): Likewise.
12050 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12051 (htm_expand_builtin): Likewise.
12052 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12053 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12054 * config/xtensa/xtensa.c (print_operand): Likewise.
12055 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12056 (NUM_MODE_VECTOR_INT): Likewise.
12057 * genoutput.c (null_operand): Likewise.
12058 (output_operand_data): Likewise.
12059 * genrecog.c (print_parameter_value): Likewise.
12060 * lra.c (debug_operand_data): Likewise.
12062 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12064 * dwarf2out.c (scompare_loc_descriptor_wide)
12065 (scompare_loc_descriptor_narrow): New functions, split out from...
12066 (scompare_loc_descriptor): ...here.
12067 * expmed.c (emit_store_flag_int): New function, split out from...
12068 (emit_store_flag): ...here.
12070 2017-08-30 Richard Biener <rguenther@suse.de>
12072 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12073 (dwarf2out_early_finish): Move setting of AT_pubnames from
12074 early debug output to early finish.
12076 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12078 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12079 and -mdata-region to the assembler.
12081 2017-08-30 Richard Biener <rguenther@suse.de>
12083 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12085 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12086 (dwarf2out_early_global_decl): Only generate a DIE for the
12087 abstract origin if it doesn't already exist or is a declaration DIE.
12088 (resolve_addr): Do not add the linkage name twice when
12089 generating a stub DIE for the DW_TAG_GNU_call_site target.
12091 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12093 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12094 Use machine_mode rather than int for arg1_mode.
12096 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12099 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12100 that the second argument of the built-in functions to unpack
12101 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12102 switch statement instead a lot of if statements.
12103 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12104 Allow 64-bit values to be in Altivec registers as well as
12105 traditional floating point registers.
12106 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12108 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12110 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12111 MAX_REGS_PER_ADDRESS == 1.
12113 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12115 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12116 * config/i386/i386.c (ix86_option_override_internal): Simplify
12117 setting of opts->x_flag_entry.
12119 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12120 Jakub Jelinek <jakub@redhat.com>
12121 Richard Biener <rguenther@suse.de>
12123 PR tree-optimization/81503
12124 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12125 folded constant fits in the target type; reorder tests for clarity.
12127 2017-08-29 Martin Liska <mliska@suse.cz>
12129 * passes.def: Include pass_lower_switch.
12130 * stmt.c (dump_case_nodes): Remove and move to
12131 tree-switch-conversion.
12132 (case_values_threshold): Likewise.
12133 (expand_switch_as_decision_tree_p): Likewise.
12134 (emit_case_decision_tree): Likewise.
12135 (expand_case): Likewise.
12136 (balance_case_nodes): Likewise.
12137 (node_has_low_bound): Likewise.
12138 (node_has_high_bound): Likewise.
12139 (node_is_bounded): Likewise.
12140 (emit_case_nodes): Likewise.
12141 (struct simple_case_node): New struct.
12142 (add_case_node): Remove.
12143 (emit_case_dispatch_table): Use vector instead of case_list.
12144 (reset_out_edges_aux): Remove.
12145 (compute_cases_per_edge): Likewise.
12146 (expand_case): Build list of simple_case_node.
12147 (expand_sjlj_dispatch_table): Use it.
12148 * tree-switch-conversion.c (struct case_node): Moved from
12149 stmt.c and adjusted.
12150 (emit_case_nodes): Likewise.
12151 (node_has_low_bound): Likewise.
12152 (node_has_high_bound): Likewise.
12153 (node_is_bounded): Likewise.
12154 (case_values_threshold): Likewise.
12155 (reset_out_edges_aux): Likewise.
12156 (compute_cases_per_edge): Likewise.
12157 (add_case_node): Likewise.
12158 (dump_case_nodes): Likewise.
12159 (balance_case_nodes): Likewise.
12160 (expand_switch_as_decision_tree_p): Likewise.
12161 (emit_jump): Likewise.
12162 (emit_case_decision_tree): Likewise.
12163 (try_switch_expansion): Likewise.
12164 (do_jump_if_equal): Likewise.
12165 (emit_cmp_and_jump_insns): Likewise.
12166 (fix_phi_operands_for_edge): New function.
12167 (record_phi_operand_mapping): Likewise.
12168 (class pass_lower_switch): New pass.
12169 (pass_lower_switch::execute): New function.
12170 (make_pass_lower_switch): Likewise.
12171 (conditional_probability):
12172 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12173 * tree-pass.h: Add make_pass_lower_switch.
12175 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12178 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12181 2017-08-29 Richard Biener <rguenther@suse.de>
12183 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12184 we do not add a DW_AT_inline attribute twice.
12185 (gen_subprogram_die): Remove code setting DW_AT_inline on
12186 DECL_ABSTRACT_P nodes.
12188 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12190 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12191 calls to internal functions.
12192 (gimplify_modify_expr): Likewise.
12193 * tree-call-cdce.c (use_internal_fn): Likewise.
12194 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12195 (convert_to_divmod): Set the nothrow flag.
12196 * tree-if-conv.c (predicate_mem_writes): Likewise.
12197 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12198 (vectorizable_call): Likewise.
12199 (vectorizable_store): Likewise.
12200 (vectorizable_load): Likewise.
12201 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12202 (vect_recog_mask_conversion_pattern): Likewise.
12204 2017-08-29 Martin Liska <mliska@suse.cz>
12207 * gcc.c (driver_handle_option): Add new argument.
12208 * opts-common.c (handle_option): Pass
12209 target_option_override_hook.
12210 * opts-global.c (lang_handle_option): Add new option.
12211 (set_default_handlers): Add new argument.
12212 (decode_options): Likewise.
12213 * opts.c (target_handle_option): Likewise.
12214 (common_handle_option): Call target_option_override_hook.
12215 * opts.h (struct cl_option_handler_func): Add hook for
12216 target option override.
12217 (struct cl_option_handlers): Likewise.
12218 (set_default_handlers): Add new argument.
12219 (decode_options): Likewise.
12220 (common_handle_option): Likewise.
12221 (target_handle_option): Likewise.
12222 * toplev.c (toplev::main): Pass targetm.target_option.override
12225 2017-08-29 Richard Biener <rguenther@suse.de>
12226 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12228 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12229 life to the active subtree.
12231 2017-08-28 Jeff Law <law@redhat.com>
12233 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12234 derive_equivalences.
12235 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12236 Code moved into....
12237 (edge_info::derive_equivalences): New private member function
12239 * tree-ssa-dom.c (class edge_info): Changed from a struct
12240 to a class. Add ctor/dtor, methods and data members.
12241 (edge_info::edge_info): Renamed from allocate_edge_info.
12242 Initialize additional members.
12243 (edge_info::~edge_info): New.
12244 (free_dom_edge_info): Delete the edge info.
12245 (record_edge_info): Use new class & associated member functions.
12246 Tighten forms for testing for edge equivalences.
12247 (record_temporary_equivalences): Iterate over the simple
12248 equivalences rather than assuming there's only one per edge.
12249 (cprop_into_successor_phis): Iterate over the simple
12250 equivalences rather than assuming there's only one per edge.
12251 (optimize_stmt): Use operand_equal_p rather than pointer
12252 equality for mini-DSE code.
12254 2017-08-28 Nathan Sidwell <nathan@acm.org>
12256 * gcc.c (execute): Fold SIGPIPE handling into switch
12257 statement. Adjust internal error message.
12259 2017-08-28 Richard Biener <rguenther@suse.de>
12262 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12263 Do nothing for removed DIEs.
12265 2017-08-28 Richard Biener <rguenther@suse.de>
12267 PR tree-optimization/81977
12268 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12271 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12274 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12275 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12276 using targetm.gen_mem_thread_fence.
12278 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12281 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12282 predicate to register_operand. Reorder operands.
12283 (*btr<mode>): Ditto.
12284 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12285 (*btr<mode>_mask): Ditto.
12287 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12289 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12290 * config/rs6000/xmmintrin.h: New file.
12291 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12293 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12296 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12297 parameter and_insn and return it.
12298 (recombine_lvx_pattern): Insert a copy to ensure availability of
12299 the base register of the copied masking operation at the point of
12300 the instruction replacement.
12301 (recombine_stvx_pattern): Likewise.
12303 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12305 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12306 undocumented switches.
12307 (-mpower9-dform-vector): Likewise.
12308 (-mpower9-dform): Likewise.
12309 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12310 comments to delete references to -mpower9-dform* switches.
12311 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12312 Delete reference to -mpower9-dform* switches, test for
12313 -mpower9-vector instead.
12314 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12315 (OTHER_P9_VECTOR_MASKS): Likewise.
12316 (POWERPC_MASKS): Likewise.
12317 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12318 tests against -mpower9-dform* to -mpower9-vector. Delete code
12319 that checked for -mpower9-dform* consistancy with other options.
12320 Add test for -mpower9-misc to enable other power9 switches.
12321 (rs6000_init_hard_regno_mode_ok): Likewise.
12322 (rs6000_option_override_internal): Likewise.
12323 (rs6000_emit_prologue): Likewise.
12324 (rs6000_emit_epilogue): Likewise.
12325 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12326 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12327 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12329 (emit_fusion_p9_store): Likewise.
12330 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12331 resetting these macros if the assembler does not support ISA 3.0
12333 (TARGET_P9_DFORM_VECTOR): Likewise.
12334 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12335 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12337 2017-08-25 Alan Modra <amodra@gmail.com>
12339 PR rtl-optimization/81747
12340 * cse.c (cse_extended_basic_block): Don't attempt to record
12341 equivalences for degenerate conditional branches that branch
12342 to their fall-through.
12344 2017-08-24 Martin Sebor <msebor@redhat.com>
12346 PR middle-end/81908
12347 * gimple-fold.c (size_must_be_zero_p): New function.
12348 (gimple_fold_builtin_memory_op): Call it.
12350 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12352 * config/rs6000/mm_malloc.h: New file.
12354 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12356 PR tree-optimization/81913
12357 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12358 analysis when either IVs in condition can wrap.
12360 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12362 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12363 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12365 2017-08-24 Richard Biener <rguenther@suse.de>
12368 * targhooks.c (default_target_can_inline_p): Properly
12369 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12370 is present and always compare.
12371 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12372 infer -mfpmath=sse from TARGET_SSE_P.
12373 (ix86_can_inline_p): Properly use target_option_default_node when
12374 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12376 2017-08-24 Richard Biener <rguenther@suse.de>
12379 * dwarf2out.c (output_die): Handle flag_generate_offload like
12381 (output_comp_unit): Likewise.
12382 (gen_array_type_die): Likewise.
12383 (dwarf2out_early_finish): Likewise.
12384 (note_variable_value_in_expr): Likewise.
12385 (dwarf2out_finish): Likewise. Adjust assert.
12386 * cgraphunit.c (symbol_table::compile): Move setting of
12387 flag_generate_offload earlier ...
12388 (symbol_table::finalize_compilation_unit): ... here, before
12389 early debug finalization.
12391 2017-08-24 Richard Biener <rguenther@suse.de>
12393 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12394 and ipa-fnsummary.h.
12395 (ix86_can_inline_p): When ix86_fpmath flags do not match
12396 check whether the callee uses FP math at all.
12398 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12400 PR middle-end/81931
12401 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12402 instead of TYPE_PRECISION.
12404 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12405 Alan Hayward <alan.hayward@arm.com>
12406 David Sherwood <david.sherwood@arm.com>
12408 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12410 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12412 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12414 (ix86_frame::stack_realign_allocate): New field.
12415 (struct machine_frame_state): Modify comments.
12416 (machine_frame_state::sp_realigned_fp_end): New field.
12417 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12418 layout calculation.
12419 (sp_valid_at): Add assertion to assure no attempt to access invalid
12420 offset of a realigned stack.
12421 (fp_valid_at): Likewise.
12422 (choose_baseaddr): Modify comments.
12423 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12424 ix86_expand_prologue.
12425 (ix86_expand_prologue): Modify stack realignment and allocation.
12426 (ix86_expand_epilogue): Modify comments.
12427 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12428 avx2_runtime, avx512f, and avx512f_runtime.
12430 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12432 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12433 (mstackrealign): Do not init to -1.
12434 * config/i386/i386.c (ix86_option_override_internal):
12435 Check opts_set, not opts when setting default value of
12436 opts->x_ix86_force_align_arg_pointer.
12438 2017-08-23 Richard Biener <rguenther@suse.de>
12440 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12441 lang_hooks.decl_printable_name.
12442 * print-rtl-function.c (print_rtx_function): Likewise.
12443 * tree-pretty-print.c (dump_function_header): Likewise.
12445 2017-08-23 Richard Biener <rguenther@suse.de>
12448 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12449 -g0 at compile-time.
12451 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12453 PR middle-end/19706
12454 * doc/sourcebuild.texi (Other hardware attributes):
12457 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12459 PR middle-end/19706
12460 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12461 Fix single-use check.
12463 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12465 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12467 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12469 * doc/install.texi: Modify to add more details on running selected
12472 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12474 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12475 is combined with -mabi=ms.
12476 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12479 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12481 PR tree-optimization/81488
12482 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12483 and cached_basis fields.
12484 (MAX_SPREAD): New constant.
12485 (alloc_cand_and_find_basis): Initialize new fields.
12486 (clear_visited): New function.
12487 (create_phi_basis_1): Rename from create_phi_basis, set visited
12488 and cached_basis fields.
12489 (create_phi_basis): New wrapper function.
12490 (phi_add_costs_1): Rename from phi_add_costs, add spread
12491 parameter, set visited field, short-circuit when limits reached.
12492 (phi_add_costs): New wrapper function.
12493 (record_phi_increments_1): Rename from record_phi_increments, set
12495 (record_phi_increments): New wrapper function.
12496 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12497 (phi_incr_cost): New wrapper function.
12498 (all_phi_incrs_profitable_1): Rename from
12499 all_phi_incrs_profitable, set visited field.
12500 (all_phi_incrs_profitable): New wrapper function.
12502 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12503 Alan Hayward <alan.hayward@arm.com>
12504 David Sherwood <david.sherwood@arm.com>
12506 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12507 that takes the outer and inner modes.
12508 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12509 comparison as the canonical test for a paradoxical subreg.
12510 * combine.c (simplify_set): Use paradoxical_subreg_p.
12511 (make_extraction): Likewise.
12512 (force_to_mode): Likewise.
12513 (rtx_equal_for_field_assignment_p): Likewise.
12514 (gen_lowpart_for_combine): Likewise.
12515 (simplify_comparison): Likewise.
12516 * cse.c (equiv_constant): Likewise.
12517 * expmed.c (store_bit_field_1): Likewise.
12518 * final.c (alter_subreg): Likewise.
12519 * fwprop.c (propagate_rtx): Likewise.
12520 (forward_propagate_subreg): Likewise.
12521 * ira-conflicts.c (ira_build_conflicts): Likewise.
12522 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12523 * lra-constraints.c (curr_insn_transform): Likewise.
12524 (split_reg): Likewise.
12525 * lra-eliminations.c (move_plus_up): Likewise.
12526 (lra_eliminate_regs_1): Likewise.
12527 * recog.c (general_operand): Likewise.
12528 * ree.c (combine_reaching_defs): Likewise.
12529 * reload.c (push_reload): Likewise.
12530 (find_reloads): Likewise.
12531 * reload1.c (elimination_effects): Likewise.
12532 (compute_reload_subreg_offset): Likewise.
12533 (choose_reload_regs): Likewise.
12534 * rtlanal.c (subreg_lsb_1): Likewise.
12535 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12536 (simplify_subreg): Likewise.
12537 * var-tracking.c (track_loc_p): Likewise.
12538 * emit-rtl.c (byte_lowpart_offset): Likewise.
12539 (paradoxical_subreg_p): Delete out-of-line definition.
12541 2017-08-22 Jeff Law <law@redhat.com>
12543 PR tree-optimization/81741
12544 PR tree-optimization/71947
12545 * tree-ssa-dom.c: Include tree-inline.h.
12546 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12547 equivalences if one is more expensive to compute than the other.
12548 * tree-ssa-scopedtables.h (class const_or_copies): Make
12549 record_const_or_copy_raw method private.
12550 (class avail_exprs_stack): New method simplify_binary_operation.
12551 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12552 avail_exprs_stack::simplify_binary_operation as needed.
12553 (avail_exprs_stack::simplify_binary_operation): New function.
12555 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12557 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12558 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12559 (DOT_SYMBOLS): Likewise.
12560 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12561 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12562 (RS6000_ABI_NAME): Likewise.
12563 (TARGET_CMODEL): Likewise.
12564 (TOC_SECTION_ASM_OP): Likewise.
12565 (SET_CMODEL): New macro.
12566 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12568 2017-08-22 Richard Biener <rguenther@suse.de>
12570 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12571 to allow for free-lang-data replacements similar to verify_type_variant.
12573 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12574 Alan Hayward <alan.hayward@arm.com>
12575 David Sherwood <david.sherwood@arm.com>
12577 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12578 VOIDmode for the LABEL_REF.
12580 2017-08-22 Richard Biener <rguenther@suse.de>
12582 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12584 2017-08-22 Richard Biener <rguenther@suse.de>
12586 * common.opt (feliminate-dwarf2-dups): Ignore.
12587 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12588 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12589 same_die_p_wrap, compute_section_prefix,
12590 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12591 (comdat_symbol_id, comdat_symbol_number): Likewise.
12592 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12594 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12595 (output_die): Mark unreachable path unreachable.
12596 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12597 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12598 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12599 (dwarf2out_early_finish): Likewise.
12601 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12603 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12605 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12608 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12609 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12610 (avr_attribute_table) <io, io_low, address>: Initialize
12611 .decl_required with true.
12613 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12615 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12616 undocumented debugging options.
12617 (-mvsx-scalar-double): Likewise.
12618 (-mallow-df-permute): Likewise.
12619 (-mvectorize-builtins): Likewise.
12620 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12621 (rs6000_builtin_vectorized_function): Likewise.
12622 (rs6000_builtin_md_vectorized_function): Likewise.
12623 (rs6000_opt_vars): Likewise.
12625 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12628 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12629 (*btrq_imm): Rename from *btrq.
12630 (*btcq_imm): Rename from *btcq.
12631 (btsc): New code attribute.
12632 (*<btsc><mode>): New insn pattern.
12633 (*btr<mode>): Ditto.
12634 (*<btsc><mode>_mask): New insn_and_split pattern.
12635 (*btr<mode>_mask): Ditto.
12637 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12638 Alan Hayward <alan.hayward@arm.com>
12639 David Sherwood <david.sherwood@arm.com>
12641 * function.c (pad_below): Simplify padding calculation.
12643 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12644 Alan Hayward <alan.hayward@arm.com>
12645 David Sherwood <david.sherwood@arm.com>
12647 * target.def (function_prologue): Remove frame size argument.
12648 (function_epilogue): Likewise.
12649 * doc/tm.texi: Regenerate.
12650 * final.c (final_start_function): Update call to function_prologue.
12651 (final_end_function): Update call to function_epilogue.
12652 (default_function_pro_epilogue): Remove frame size argument.
12653 * output.h (default_function_pro_epilogue): Likewise.
12654 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12655 (arm_output_function_prologue): Likewise.
12656 * config/frv/frv.c (frv_function_prologue): Likewise.
12657 (frv_function_epilogue): Likewise.
12658 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12659 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12660 (ia64_output_function_epilogue): Likewise.
12661 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12662 (m32r_output_function_epilogue): Likewise.
12663 * config/microblaze/microblaze.c (microblaze_function_prologue)
12664 (microblaze_function_epilogue): Likewise.
12665 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12666 (mips_output_function_epilogue): Likewise.
12667 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12668 (mmix_target_asm_function_epilogue): Likewise.
12669 * config/msp430/msp430.c (msp430_start_function): Likewise.
12670 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12671 (nds32_asm_function_epilogue): Likewise.
12672 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12673 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12674 (pa_output_function_epilogue): Likewise.
12675 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12676 (rs6000_output_function_epilogue): Likewise.
12677 * config/rl78/rl78.c (rl78_start_function): Likewise.
12678 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12679 (rs6000_output_function_epilogue): Likewise.
12680 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12681 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12682 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12683 (sparc_asm_function_epilogue): Likewise.
12685 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12687 * tree.h (type_has_mode_precision_p): New function.
12688 * convert.c (convert_to_integer_1): Use it.
12689 * expr.c (expand_expr_real_2): Likewise.
12690 (expand_expr_real_1): Likewise.
12691 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12692 * match.pd: Likewise.
12693 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12694 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12695 * tree-tailcall.c (process_assignment): Likewise.
12696 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12697 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12698 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12699 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12700 (vectorizable_assignment): Likewise.
12701 (vectorizable_shift): Likewise.
12702 (vectorizable_operation): Likewise.
12703 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12705 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12707 * match.pd: Add pow (C, x) simplification.
12709 2017-08-21 Richard Biener <rguenther@suse.de>
12711 PR tree-optimization/81900
12712 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12713 for blocks with abnormal predecessors.
12714 (compute_antic): Do not set visited flag prematurely.
12716 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12719 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12721 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12723 * stor-layout.h (vector_type_mode): Move to...
12724 * tree.h (vector_type_mode): ...here.
12725 * stor-layout.c (vector_type_mode): Move to...
12726 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12728 2017-08-21 Richard Biener <rguenther@suse.de>
12730 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12731 register_external_die hooks.
12732 (debug_false_tree_charstarstar_uhwistar): Declare.
12733 (debug_nothing_tree_charstar_uhwi): Likewise.
12734 * debug.c (do_nothing_debug_hooks): Adjust.
12735 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12736 (debug_nothing_tree_charstar_uhwi): Likewise.
12737 * dbxout.c (dbx_debug_hooks): Adjust.
12738 (xcoff_debug_hooks): Likewise.
12739 * sdbout.c (sdb_debug_hooks): Likewise.
12740 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12741 * dwarf2out.c (macinfo_label_base): New global.
12742 (dwarf2out_register_external_die): New function for the
12743 register_external_die hook.
12744 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12745 (dwarf2_debug_hooks): Use them.
12746 (dwarf2_lineno_debug_hooks): Adjust.
12747 (struct die_struct): Add with_offset flag.
12748 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12749 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12750 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12751 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12752 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12753 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12754 defining section names for the early LTO debug variants.
12755 (reset_indirect_string): New helper.
12756 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12757 (print_dw_val): Add support for offsetted symbol references.
12758 (get_ultimate_context): Split out from is_cxx.
12759 (is_cxx): Use get_ultimate_context.
12760 (is_fortran): Add decl overload.
12761 (compute_comp_unit_symbol): Split out worker from
12762 compute_section_prefix.
12763 (compute_section_prefix): Call compute_comp_unit_symbol and
12764 set comdat_type_p here.
12765 (output_die): Skip DIE symbol output for the LTO added one.
12766 Handle DIE symbol references with offset.
12767 (output_comp_unit): Guard section name mangling properly.
12768 For LTO debug sections emit a symbol at the section beginning
12769 which we use to refer to its DIEs.
12770 (add_abstract_origin_attribute): For DIEs registered via
12771 dwarf2out_register_external_die directly refer to the early
12772 DIE rather than indirectly through the shadow one we created.
12773 Remove obsolete call to dwarf2out_abstract_function for
12774 non-function/block origins.
12775 (gen_array_type_die): When generating early LTO debug do
12776 not emit DW_AT_string_length.
12777 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12778 late when in LTO. As suggested place a gcc_unreachable for
12779 the DECL_ABSTRACT_P case.
12780 (gen_subprogram_die): Avoid another specification DIE
12781 for early built declarations/definitions for the late LTO case.
12782 (gen_variable_die): Add type references for late duplicated VLA dies
12784 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12785 we have the abstract instance already.
12786 (process_scope_var): Adjust decl DIE contexts in LTO which
12787 first puts them in limbo.
12788 (gen_decl_die): Do not generate type DIEs late apart from
12789 types for VLAs or for decls we do not yet have a DIE. Do not
12790 call dwarf2out_abstract_function late.
12791 (dwarf2out_early_global_decl): Make sure to create DIEs
12792 for abstract instances of a decl first.
12793 (dwarf2out_late_global_decl): Adjust comment.
12794 (output_macinfo_op): With multiple macro sections use
12795 macinfo_label_base to distinguish labels.
12796 (output_macinfo): Likewise. Update macinfo_label_base.
12797 Pass in the line info label.
12798 (note_variable_value_in_expr): When generating LTO resolve
12799 all variable values here by generating DIEs as needed.
12800 (init_sections_and_labels): Add early LTO debug flag parameter
12801 and generate different sections and names if set. Add generation
12802 counter for the labels so we can have multiple of them.
12803 (reset_dies): Helper to allow DIEs to be output multiple times.
12804 (dwarf2out_finish): When outputting DIEs to the fat part of an
12805 LTO object first reset DIEs.
12806 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12807 (modified_type_die): Check for decl_ultimate_origin being self
12809 (gen_type_die_with_usage): Likewise.
12810 (gen_typedef_die): Allow decl_ultimate_origin being self.
12811 (set_decl_abstract_flags): Remove.
12812 (set_block_abstract_flags): Likewise.
12813 (dwarf2out_abstract_function): Treat the early generated DIEs
12814 as the abstract copy and only add DW_AT_inline and
12815 DW_AT_artificial here and call set_decl_origin_self.
12816 If the DIE has an abstract origin don't do anything.
12817 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12818 if we have none yet (Go fails to build one, PR78628).
12819 (variably_modified_type_p): Prevent endless recursion for Ada
12820 cyclic pointer types.
12821 * lto-streamer-in.c: Include debug.h.
12822 (dref_queue): New global.
12823 (lto_read_tree_1): Stream in DIE references.
12824 (lto_input_tree): Register DIE references.
12825 (input_function): Stream DECL_DEBUG_ARGS.
12826 * lto-streamer-out.c: Include debug.h.
12827 (lto_write_tree_1): Output DIE references.
12828 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12829 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12830 (output_function): Stream DECL_DEBUG_ARGS.
12831 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12832 Stream DECL_ABSTRACT_ORIGIN.
12833 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12834 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12835 DECL_CONTEXT for file-scope decls.
12836 * lto-streamer.h (struct dref_entry): Declare.
12837 (dref_queue): Likewise.
12838 * cfgexpand.c (pass_expand::execute): Do not call the
12839 outlining_inline_function hook here.
12840 * lto-wrapper.c (debug_obj): New global.
12841 (tool_cleanup): Unlink it if required.
12842 (debug_objcopy): New function.
12843 (run_gcc): Handle early debug sections in the IL files by
12844 extracting them to separate files, partially linkin them and
12845 feeding the result back as result to the linker.
12846 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12847 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12848 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12849 sections into a separate segment.
12850 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12852 (darwin_asm_dwarf_section): Likewise.
12853 (darwin_asm_output_dwarf_offset): Likewise.
12854 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12856 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12857 Alan Hayward <alan.hayward@arm.com>
12858 David Sherwood <david.sherwood@arm.com>
12860 * read-md.h (md_reader::record_potential_iterator_use): Replace
12861 pointer argument with an rtx and an index.
12862 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12863 (apply_mode_iterator): Likewise.
12864 (apply_code_iterator): Likewise.
12865 (apply_int_iterator): Likewise.
12866 (apply_subst_iterator): Likewise.
12867 (record_iterator_use): Likewise.
12868 (record_attribute_use): Likewise.
12869 (md_reader::record_potential_iterator_use): Likewise. Update calls
12870 to record_iterator_use and apply_iterator.
12871 (iterator_use): Replace ptr with x and index.
12872 (attribute_use): Likewise.
12873 (apply_attribute_uses): Update calls to apply_iterator.
12874 (apply_iterators): Likewise. Update initialization of iterator_use.
12875 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12876 and record_potential_iterator_use.
12877 (rtx_reader::read_rtx_operand): Likewise.
12879 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12880 Alan Hayward <alan.hayward@arm.com>
12881 David Sherwood <david.sherwood@arm.com>
12883 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12886 2017-08-21 Richard Biener <rguenther@suse.de>
12888 PR middle-end/81884
12889 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12890 at struct end conservatively when comparing common bases.
12892 2017-08-21 Richard Biener <rguenther@suse.de>
12894 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12895 (mem_ref_in_stmt): Remove.
12896 (determine_max_movement): Use ref index to get at the reference.
12897 (invariantness_dom_walker::before_dom_children): Deal with
12898 lim data already initialized.
12899 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12901 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12903 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12904 (TARGET_ISA_ROUND): Ditto.
12905 (TARGET_ROUND): Ditto.
12906 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12907 * config/i386/i386.md: Ditto.
12908 * config/i386/sse.md: Ditto.
12909 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12910 with OPTION_MASK_ISA_SSE4_1.
12912 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12915 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12916 __builtin_ia32_lzcnt_u16.
12918 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12921 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12922 (rs6000_set_current_function): Rewrite function to use it.
12924 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12927 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12928 and TYPE_WARN_IF_NOT_ALIGN.
12929 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12930 (handle_warn_if_not_align): New.
12931 (place_union_field): Call handle_warn_if_not_align.
12932 (place_field): Call handle_warn_if_not_align.
12933 Copy TYPE_WARN_IF_NOT_ALIGN.
12934 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12935 (layout_type): Likewise.
12936 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12938 (tree_decl_common): Add warn_if_not_align.
12939 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12940 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12941 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12942 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12943 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12944 * doc/extend.texi: Document warn_if_not_aligned attribute.
12945 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12947 2017-08-17 Martin Liska <mliska@suse.cz>
12950 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12951 (get_data_dependence): Use it as pointer type.
12952 (distribute_loop): Likewise.
12954 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12956 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12957 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12958 (p8_vmrgow_<mode>_direct): New define_insn.
12959 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12960 handle endianness for vmrgew and vmrgow permute patterns.
12962 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12964 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12965 * config/rs6000/rs6000-cpus.def: Remove comment.
12966 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12967 (POWERPC_MASKS): Likewise.
12968 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12969 use of TARGET_VSX_TIMODE.
12970 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12971 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12972 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12973 (rs6000_option_override_internal): Remove dead code.
12974 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12975 (rs6000_legitimize_reload_address): Likewise.
12976 (rs6000_legitimate_address_p): Likewise.
12977 (rs6000_opt_masks): Delete "vsx-timode".
12978 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12979 from function comment.
12980 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12981 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12982 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12984 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12985 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12986 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12987 (VSX_TI): Likewise.
12989 (define_peephole2): Likewise.
12991 2017-08-17 Martin Sebor <msebor@redhat.com>
12994 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12995 past the end of an array.
12996 (test_pp_format): Add test cases.
12998 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13000 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13001 missing ECF_NOTHROW flags.
13003 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13006 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13007 operands residing in integer registers.
13008 (*vsx_le_perm_load_<mode>): Likewise.
13009 (*vsx_le_perm_store_<mode>): Likewise.
13010 (define_peephole2): Add peepholes to optimize the above.
13012 2017-08-17 Marek Polacek <polacek@redhat.com>
13014 PR middle-end/81814
13015 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13016 to mimic what shorten_compare did. Change the return type to bool.
13017 (fold_cond_expr_with_comparison): Update call to
13018 operand_equal_for_comparison_p.
13019 (fold_ternary_loc): Likewise.
13021 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13023 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13025 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13027 2017-08-17 Richard Biener <rguenther@suse.de>
13029 * tree-ssa-structalias.c (solve_graph): When propagating
13030 to successors update the graphs succ edges and avoid duplicate work.
13032 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13035 * config/i386/i386.c (ix86_option_override_internal): Save target
13036 specific options after ix86_stack_protector_guard_reg was changed.
13038 2017-08-17 Richard Biener <rguenther@suse.de>
13040 PR tree-optimization/81827
13041 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13042 (new_var_info): Initialize it conservatively.
13043 (get_call_vi): Mark register vars.
13044 (new_scalar_tmp_constraint_exp): Likewise.
13045 (handle_rhs_call): Likewise.
13046 (handle_const_call): Likewise.
13047 (create_function_info_for): Likewise.
13048 (solve_constraints): Sort varinfos to separate register from
13049 non-register vars to pack points-to solution bitmaps during
13052 2017-08-17 Marek Polacek <polacek@redhat.com>
13054 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13056 2017-08-17 Richard Biener <rguenther@suse.de>
13058 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13059 to true when overflow is undefined and we saturated the result.
13061 2017-08-17 Alan Modra <amodra@gmail.com>
13064 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13065 Don't use store multiple if only one reg needs saving.
13066 (interesting_frame_related_regno): New function.
13067 (rs6000_frame_related): Don't emit frame info for regs that
13069 (rs6000_emit_epilogue): Likewise.
13071 2017-08-16 Nathan Sidwell <nathan@acm.org>
13073 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13074 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13075 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13076 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13077 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13078 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13079 (verify_type): Adjust for TYPE_BINFO move.
13080 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13081 process TYPE_BINFO directly.
13082 (hash_tree): Likewise.
13083 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13085 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13088 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13090 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13092 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13095 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13096 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13097 Add ix86_binary_operator_ok to insn constraint.
13098 (*iordi_1_bts): Ditto.
13099 (*xordi_1_btc): Ditto.
13100 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13101 Update corresponding peephole2 pattern.
13105 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13107 PR tree-optimization/81832
13108 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13109 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13111 2017-08-16 Marek Polacek <polacek@redhat.com>
13114 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13115 regarding size_zero_node.
13117 2017-08-16 Martin Liska <mliska@suse.cz>
13120 * config.gcc: Respect previously set extra_objs in case
13123 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13125 PR tree-optimization/81835
13126 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13127 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13128 not depend on the phi.
13130 2017-08-16 Alan Modra <amodra@gmail.com>
13132 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13135 2017-08-16 Alan Modra <amodra@gmail.com>
13137 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13138 (save_reg_p): ..into this. Update all callers.
13139 (first_reg_to_save): Simplify.
13141 2017-08-16 Alan Modra <amodra@gmail.com>
13143 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13146 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13150 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13151 constant count if that count is less than 32.
13153 2017-08-15 Nathan Sidwell <nathan@acm.org>
13155 * gcc.c (execute): Emit friendlier message if inferior is killed
13156 by an external cause.
13158 2017-08-15 Richard Biener <rguenther@suse.de>
13160 PR tree-optimization/81790
13161 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13162 CONSTRUCTORs from simplifying and VN.
13164 2017-08-14 Martin Sebor <msebor@redhat.com>
13166 * builtin-attrs.def: Add comments.
13168 2017-08-14 Martin Sebor <msebor@redhat.com>
13171 * doc/extend.texi (attribute nonstring): Document new attribute.
13173 2017-08-14 Martin Sebor <msebor@redhat.com>
13176 * tree-diagnostic.c (default_tree_printer): Handle %G.
13177 * gimple-pretty-print.h (percent_G_format): Declare new function.
13178 * gimple-pretty-print.c (percent_G_format): Define.
13179 * tree-pretty-print.c (percent_K_format): Add argument.
13181 2017-08-14 Martin Sebor <msebor@redhat.com>
13183 PR translation/79998
13184 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13185 Remove a stray space.
13187 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13190 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13191 (*iordi_1_bts): Ditto.
13192 (*xordi_1_btc): Ditto.
13194 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13197 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13198 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13200 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13201 quoted strings, and make more translator-friendly.
13202 (darwin_rs6000_override_options): Likewise.
13203 (rs6000_option_override_internal): Likewise.
13204 (rs6000_return_in_memory): Fix overlong line.
13205 (init_cmulative_args): Use quoted strings, and make more
13206 translator-friendly.
13207 (rs6000_pass_by_reference): Fix overlong line.
13208 (def_builtin): Use quoted strings.
13209 (altivec_expand_predicate_builtin): Use quoted strings, and make
13210 more translator-friendly.
13211 (htm_expand_builtin): Use quoted strings.
13212 (cpu_expand_builtin): Use quoted strings, and make more
13213 translator-friendly.
13214 (altivec_expand_builtin): Likewise.
13215 (paired_expand_predicate_builtin): Likewise.
13216 (rs6000_invalid_builtin): Likewise.
13217 (builtin_function_type): Use quoted strings.
13218 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13219 more translator-friendly.
13220 (rs6000_trampoline_init): Likewise.
13221 (rs6000_handle_altivec_attribute): Likewise.
13222 (rs6000_inner_target_options): Use quoted strings.
13223 (rs6000_disable_incompatible_switches): Likewise.
13224 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13225 strings, and make more translator-friendly.
13226 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13228 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13230 PR tree-optimization/81799
13231 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13232 cond_expr to simple gimple operand.
13234 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13236 PR middle-end/46932
13237 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13239 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13243 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13244 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13245 TARGET_GASISR_PROLOGUES.
13246 * config/avr/avr.c (avr_option_override): Same.
13247 (avr_pass_pre_proep::execute): Same.
13249 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13252 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13253 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13255 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13257 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13258 define_insn_and_split. Split to a memory load from 0 in
13259 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13260 using PTR mode iterator.
13261 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13262 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13263 (*add_tp_<mode>): Redefine as define_insn_and_split.
13264 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13265 address space. Merge with *add_tp_x32 using PTR mode iterator.
13266 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13267 Split to an add with a memory load from 0 in
13268 DEFAULT_TLS_SEG_REG address space.
13270 2017-08-12 Andrew Pinski <apinski@cavium.com>
13272 * config/aarch64/aarch64-option-extensions.def (rdma):
13273 Fix feature string to what Linux prints out in /proc/cpuinfo.
13275 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13278 * dwarf2out.c (modified_type_die): For C typedef types that have
13279 an ultimate origin, process the ultimate origin instead of the
13281 (gen_typedef_die): Assert that input DECLs have no ultimate
13283 (gen_type_die_with_usage): For typedef variants that have an
13284 ultimate origin, just call gen_decl_die on the original DECL.
13285 (process_scope_var): Avoid creating DIEs for local typedefs and
13286 concrete static variables.
13288 2017-08-12 Alan Modra <amodra@gmail.com>
13292 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13293 match gnu-user.h startfile.
13294 (ENDFILE_LINUX_SPEC): Similarly.
13296 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13299 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13301 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13303 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13304 * config/aarch64/aarch64.md (mov<mode>): Change.
13305 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13306 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13307 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13309 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13311 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13312 for storage order barriers.
13314 2017-08-11 Martin Liska <mliska@suse.cz>
13317 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13318 variables of void type.
13320 2017-08-11 Martin Liska <mliska@suse.cz>
13322 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13323 TARGET_SUPPORTS_ALIASES.
13324 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13325 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13326 (optimize_weakref): Likewise.
13327 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13328 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13329 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13331 2017-08-11 Martin Liska <mliska@suse.cz>
13334 * config/i386/i386.c (make_resolver_func): Do complete
13335 refactoring of the function.
13337 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13340 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13341 * config/i386/i386.c (ix86_stack_protect_guard): Use
13342 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13343 * doc/invoke.texi (x86 Options): Document
13344 -mstack-protector-guard-symbol= option.
13346 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13348 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13349 * config/i386/i386.c (ix86_split_stack_guard): New function.
13350 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13351 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13352 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13353 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13354 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13355 (split_stack_space_check): Call ix86_split_stack_guard.
13357 2017-08-10 Martin Sebor <msebor@redhat.com>
13359 * print-tree.c (print_node): Print location using the established
13361 Replace spaces with colons.
13362 (debug_raw, debug): Ditto.
13364 2017-08-10 Martin Sebor <msebor@redhat.com>
13367 * pretty-print.c (pp_format): Correct the handling of %s precision.
13369 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13372 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13374 (ix86_finalize_stack_frame_flags): This. Also clear
13375 frame_pointer_needed if -fno-omit-frame-pointer is used without
13377 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13378 with ix86_finalize_stack_frame_flags.
13379 (ix86_expand_epilogue): Likewise.
13380 (ix86_expand_split_stack_prologue): Likewise.
13381 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13383 2017-08-10 Martin Liska <mliska@suse.cz>
13386 * c-attribs.c (handle_target_attribute):
13387 Report warning for an empty string argument of target attribute.
13389 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13392 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13394 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13395 or DECL_NONLOCAL labels.
13396 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13397 or DECL_NONLOCAL labels here.
13399 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13401 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13402 to indicate when early gimple folding has been disabled.
13403 (rs6000_gimple_fold_builtin): Add debug content.
13404 (rs6000_invalid_builtin): Fix whitespace.
13405 (rs6000_expand_builtin): Fix whitespace.
13406 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13408 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13411 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13412 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13414 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13416 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13417 (qdf24xx): Likewise.
13418 * config/aarch64/aarch64.md: Include falkor.md.
13419 * config/aarch64/falkor.md: New.
13421 2017-08-09 Marek Polacek <polacek@redhat.com>
13424 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13425 * diagnostic.c (emit_diagnostic): Add a comment.
13426 (emit_diagnostic_valist): New function.
13428 2017-08-09 Marek Polacek <polacek@redhat.com>
13431 * input.c (make_location): New overload.
13432 * input.h (make_location): Declare.
13434 2017-08-08 Alan Modra <amodra@gmail.com>
13435 H.J. Lu <hongjiu.lu@intel.com>
13438 * gcc.c (NO_PIE_SPEC): Delete.
13439 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13441 (LINK_PIE_SPEC): ..to here.
13442 (LINK_COMMAND_SPEC): Support -no-pie.
13443 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13444 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13446 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13447 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13448 (ENDFILE_CRTEND_SPEC): Similarly.
13450 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13453 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13454 (mstack-protector-guard-offset=): Ditto.
13455 * config/i386/i386.c (ix86_option_override): Handle
13456 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13458 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13459 ix86_stack_protect_guard_offset variables.
13460 (TARGET_STACK_PROTECT_GUARD): Always define.
13461 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13462 and -mstack-protector-guard-offset= options.
13464 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13466 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13467 boundary case for the last candidate.
13469 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13471 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13472 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13474 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13476 PR middle-end/19706
13477 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13478 * config/aarch64/aarch64-builtins.c
13479 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13480 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13481 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13483 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13484 Andrew Pinski <pinskia@gmail.com>
13486 PR middle-end/19706
13487 * internal-fn.def (XORSIGN): New.
13488 * optabs.def (xorsign_optab): New.
13489 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13490 (convert_expand_mult_copysign): New.
13491 (pass_optimize_widening_mul::execute): Call
13492 convert_expand_mult_copysign.
13494 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13496 PR tree-optimization/81354
13497 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13498 Insert on edges rather than explicitly creating landing pads.
13499 (analyze_candidates_and_replace): Commit edge inserts.
13501 2017-08-08 Richard Biener <rguenther@suse.de>
13503 PR middle-end/81719
13504 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13505 (expand_simple_operations): Also look through ADDR_EXPRs with
13506 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13508 2017-08-08 Richard Biener <rguenther@suse.de>
13510 PR tree-optimization/81723
13511 * tree-vect-slp.c (struct bst_traits): New hash traits.
13512 (bst_fail): New global.
13513 (vect_build_slp_tree_2): New worker, split out from ...
13514 (vect_build_slp_tree): ... this now wrapping it with using
13515 bst_fail set to cache SLP tree build fails. Properly handle
13517 (vect_analyze_slp_instance): Allocate and free bst_fail.
13519 2017-08-08 Martin Liska <mliska@suse.cz>
13522 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13523 LABEL_DECLs that can be from a different function.
13525 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13527 PR tree-optimization/81744
13528 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13529 loop's number of iterations.
13531 2017-08-08 Martin Liska <mliska@suse.cz>
13533 * asan.c: Include header files.
13534 * attribs.c (build_decl_attribute_variant): New function moved
13536 (build_type_attribute_qual_variant): Likewise.
13537 (cmp_attrib_identifiers): Likewise.
13538 (simple_cst_list_equal): Likewise.
13539 (omp_declare_simd_clauses_equal): Likewise.
13540 (attribute_value_equal): Likewise.
13541 (comp_type_attributes): Likewise.
13542 (build_type_attribute_variant): Likewise.
13543 (lookup_ident_attribute): Likewise.
13544 (remove_attribute): Likewise.
13545 (merge_attributes): Likewise.
13546 (merge_type_attributes): Likewise.
13547 (merge_decl_attributes): Likewise.
13548 (merge_dllimport_decl_attributes): Likewise.
13549 (handle_dll_attribute): Likewise.
13550 (attribute_list_equal): Likewise.
13551 (attribute_list_contained): Likewise.
13552 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13553 (lookup_attribute_by_prefix): Likewise.
13554 * bb-reorder.c: Include header files.
13555 * builtins.c: Likewise.
13556 * calls.c: Likewise.
13557 * cfgexpand.c: Likewise.
13558 * cgraph.c: Likewise.
13559 * cgraphunit.c: Likewise.
13560 * convert.c: Likewise.
13561 * dwarf2out.c: Likewise.
13562 * final.c: Likewise.
13563 * fold-const.c: Likewise.
13564 * function.c: Likewise.
13565 * gimple-expr.c: Likewise.
13566 * gimple-fold.c: Likewise.
13567 * gimple-pretty-print.c: Likewise.
13568 * gimple.c: Likewise.
13569 * gimplify.c: Likewise.
13570 * hsa-common.c: Likewise.
13571 * hsa-gen.c: Likewise.
13572 * internal-fn.c: Likewise.
13573 * ipa-chkp.c: Likewise.
13574 * ipa-cp.c: Likewise.
13575 * ipa-devirt.c: Likewise.
13576 * ipa-fnsummary.c: Likewise.
13577 * ipa-inline.c: Likewise.
13578 * ipa-visibility.c: Likewise.
13580 * lto-cgraph.c: Likewise.
13581 * omp-expand.c: Likewise.
13582 * omp-general.c: Likewise.
13583 * omp-low.c: Likewise.
13584 * omp-offload.c: Likewise.
13585 * omp-simd-clone.c: Likewise.
13586 * opts-global.c: Likewise.
13587 * passes.c: Likewise.
13588 * predict.c: Likewise.
13589 * sancov.c: Likewise.
13590 * sanopt.c: Likewise.
13591 * symtab.c: Likewise.
13592 * toplev.c: Likewise.
13593 * trans-mem.c: Likewise.
13594 * tree-chkp.c: Likewise.
13595 * tree-eh.c: Likewise.
13596 * tree-into-ssa.c: Likewise.
13597 * tree-object-size.c: Likewise.
13598 * tree-parloops.c: Likewise.
13599 * tree-profile.c: Likewise.
13600 * tree-ssa-ccp.c: Likewise.
13601 * tree-ssa-live.c: Likewise.
13602 * tree-ssa-loop.c: Likewise.
13603 * tree-ssa-sccvn.c: Likewise.
13604 * tree-ssa-structalias.c: Likewise.
13605 * tree-ssa.c: Likewise.
13606 * tree-streamer-in.c: Likewise.
13607 * tree-vectorizer.c: Likewise.
13608 * tree-vrp.c: Likewise.
13609 * tsan.c: Likewise.
13610 * ubsan.c: Likewise.
13611 * varasm.c: Likewise.
13612 * varpool.c: Likewise.
13613 * tree.c: Remove functions moved to attribs.[ch].
13614 * tree.h: Likewise.
13615 * config/aarch64/aarch64.c: Add attrs.h header file.
13616 * config/alpha/alpha.c: Likewise.
13617 * config/arc/arc.c: Likewise.
13618 * config/arm/arm.c: Likewise.
13619 * config/avr/avr.c: Likewise.
13620 * config/bfin/bfin.c: Likewise.
13621 * config/c6x/c6x.c: Likewise.
13622 * config/cr16/cr16.c: Likewise.
13623 * config/cris/cris.c: Likewise.
13624 * config/darwin.c: Likewise.
13625 * config/epiphany/epiphany.c: Likewise.
13626 * config/fr30/fr30.c: Likewise.
13627 * config/frv/frv.c: Likewise.
13628 * config/ft32/ft32.c: Likewise.
13629 * config/h8300/h8300.c: Likewise.
13630 * config/i386/winnt.c: Likewise.
13631 * config/ia64/ia64.c: Likewise.
13632 * config/iq2000/iq2000.c: Likewise.
13633 * config/lm32/lm32.c: Likewise.
13634 * config/m32c/m32c.c: Likewise.
13635 * config/m32r/m32r.c: Likewise.
13636 * config/m68k/m68k.c: Likewise.
13637 * config/mcore/mcore.c: Likewise.
13638 * config/microblaze/microblaze.c: Likewise.
13639 * config/mips/mips.c: Likewise.
13640 * config/mmix/mmix.c: Likewise.
13641 * config/mn10300/mn10300.c: Likewise.
13642 * config/moxie/moxie.c: Likewise.
13643 * config/msp430/msp430.c: Likewise.
13644 * config/nds32/nds32-isr.c: Likewise.
13645 * config/nds32/nds32.c: Likewise.
13646 * config/nios2/nios2.c: Likewise.
13647 * config/nvptx/nvptx.c: Likewise.
13648 * config/pa/pa.c: Likewise.
13649 * config/pdp11/pdp11.c: Likewise.
13650 * config/powerpcspe/powerpcspe.c: Likewise.
13651 * config/riscv/riscv.c: Likewise.
13652 * config/rl78/rl78.c: Likewise.
13653 * config/rx/rx.c: Likewise.
13654 * config/s390/s390.c: Likewise.
13655 * config/sh/sh.c: Likewise.
13656 * config/sol2.c: Likewise.
13657 * config/sparc/sparc.c: Likewise.
13658 * config/spu/spu.c: Likewise.
13659 * config/stormy16/stormy16.c: Likewise.
13660 * config/tilegx/tilegx.c: Likewise.
13661 * config/tilepro/tilepro.c: Likewise.
13662 * config/v850/v850.c: Likewise.
13663 * config/vax/vax.c: Likewise.
13664 * config/visium/visium.c: Likewise.
13665 * config/xtensa/xtensa.c: Likewise.
13667 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13670 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13671 constraints since the -mupper-regs-* switches have been
13673 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13674 into a vector from a double word element that was extracted from
13675 another vector, and eliminate extra XXPERMDI instructions.
13676 (vsx_concat_<mode>_2): Likewise.
13677 (vsx_concat_<mode>_3): Likewise.
13678 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13679 concat to allow optimizing inserts from previous extracts.
13681 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13683 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13684 memory reference to a SSP offset in TLS address space.
13685 (ix86_print_operand) <case '@'>: Remove.
13686 (ix86_print_operand_punct_valid_p): Remove '@' code.
13687 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13688 UNSPEC_SP_TLS_TEST.
13689 (stack_tls_protect_set_<mode>): Remove.
13690 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13691 (stack_tls_protect_test_<mode>): Remove.
13692 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13694 2017-08-07 Olivier Hainque <hainque@adacore.com>
13697 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13699 2017-08-07 Douglas Rupp <rupp@adacore.com>
13701 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13702 variable was referenced as multidir in command.
13704 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13707 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13710 2017-08-07 Martin Liska <mliska@suse.cz>
13712 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13713 * config/rl78/rl78.c: Add include of attribs.h.
13714 * config/sh/sh.c: Likewise.
13715 * config/v850/v850.c: Likewise.
13717 2017-08-07 Tom de Vries <tom@codesourcery.com>
13719 PR middle-end/78266
13720 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13722 2017-08-07 Martin Liska <mliska@suse.cz>
13724 * config/mips/mips.c: Include attribs.h.
13726 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13729 * doc/invoke.texi: Document change in behvaior for -Ofast for
13732 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13734 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13736 (aarch64_pop_regs): Likewise.
13737 (aarch64_gen_load_pair): Likewise.
13738 (aarch64_save_callee_saves): Likewise.
13739 (aarch64_restore_callee_saves): Likewise.
13741 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13743 * config/i386/i386.c: Revert the last change.
13745 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13748 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13750 (ix86_finalize_stack_frame_flags): This. Also clear
13751 frame_pointer_needed if -fno-omit-frame-pointer is used without
13753 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13754 with ix86_finalize_stack_frame_flags.
13755 (ix86_expand_epilogue): Likewise.
13756 (ix86_expand_split_stack_prologue): Likewise.
13758 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13761 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13762 to P_AES for Westmere.
13764 2017-08-07 Jonathan Yong <10walls@gmail.com>
13766 * config/i386/mingw.opt (fset-stack-executable): Removed.
13767 * config/i386/cygming.opt (fset-stack-executable): Moved
13769 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13771 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13773 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13775 2017-08-07 Marek Polacek <polacek@redhat.com>
13777 PR middle-end/81737
13778 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13780 2017-08-07 Martin Liska <mliska@suse.cz>
13782 * attribs.h (canonicalize_attr_name): New function.
13783 (cmp_attribs): Move from c-format.c and adjusted.
13784 (is_attribute_p): Moved from tree.h.
13785 * tree-inline.c: Add new includes.
13786 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13787 (private_is_attribute_p): Remove.
13788 (private_lookup_attribute): Likewise.
13789 (private_lookup_attribute_by_prefix): Simplify.
13790 (remove_attribute): Use is_attribute_p.
13791 * tree.h: Remove removed declarations.
13793 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13795 PR middle-end/81698
13796 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13797 instead of computing it in the function. Formatting fix.
13798 (expand_case): Don't rely on default_edge being the first edge,
13799 clear it if removing it, pass default_edge to
13800 emit_case_dispatch_table.
13801 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13804 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13806 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13807 insn in the function, emit NOP after the insn.
13809 2017-08-06 Tom de Vries <tom@codesourcery.com>
13811 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13814 2017-08-06 Tom de Vries <tom@codesourcery.com>
13816 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13819 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13821 PR tree-optimization/57371
13822 * match.pd: New pattern.
13824 2017-08-04 Marek Polacek <polacek@redhat.com>
13826 PR middle-end/81695
13827 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13828 perform the computation in offset_int.
13830 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13832 PR tree-optimization/81136
13833 * tree-vectorizer.h: Include tree-hash-traits.h.
13834 (vec_base_alignments): New typedef.
13835 (vec_info): Add a base_alignments field.
13836 (vect_record_base_alignments): Declare.
13837 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13839 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13840 (create_data_ref): Add an is_conditional_in_stmt argument.
13841 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13842 the is_conditional_in_stmt field.
13843 (data_ref_loc): Add an is_conditional_in_stmt field.
13844 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13845 (find_data_references_in_stmt): Update call to create_data_ref.
13846 (graphite_find_data_references_in_stmt): Likewise.
13847 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13848 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13849 (vect_record_base_alignment): New function.
13850 (vect_record_base_alignments): Likewise.
13851 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13852 for nested statements even if we fail to compute a misalignment.
13853 Use pooled base alignments for unconditional references.
13854 (vect_find_same_alignment_drs): Compare base addresses instead
13856 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13857 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13859 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13861 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13862 Add an explicit name for the enum. Use auto_vec for slp_instances
13863 and grouped_stores.
13864 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13866 (_bb_vec_info): Add a constructor and destructor.
13867 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13868 (destroy_loop_vec_info): Delete.
13869 (vect_destroy_datarefs): Likewise.
13870 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13871 (vec_info::vec_info): New function.
13872 (vec_info::~vec_info): Likewise.
13873 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13874 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13875 destroy_loop_vec_info.
13876 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13877 (_loop_vec_info::_loop_vec_info): ...this.
13878 (destroy_loop_vec_info): Replace with...
13879 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13880 the stmt_vec_infos. Leave handling of vec_info information to its
13881 destructor. Remove explicit vector releases.
13882 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13883 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13884 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13885 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13886 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13887 (destroy_bb_vec_info): Replace with...
13888 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13889 information to its destructor.
13890 (vect_slp_analyze_bb_1): Use new and delete instead of
13891 new_bb_vec_info and destroy_bb_vec_info.
13892 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13895 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13897 * tree-data-ref.h (subscript): Add access_fn field.
13898 (data_dependence_relation): Add could_be_independent_p.
13899 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13900 (same_access_functions): Move to tree-data-ref.c.
13901 * tree-data-ref.c (ref_contains_union_access_p): New function.
13902 (access_fn_component_p): Likewise.
13903 (access_fn_components_comparable_p): Likewise.
13904 (dr_analyze_indices): Add a reference to access_fn_component_p.
13905 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13907 (constant_access_functions): Likewise.
13908 (add_other_self_distances): Likewise.
13909 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13910 (initialize_data_dependence_relation): Use XCNEW and remove
13911 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13912 of access functions that have the same type. Allow the
13913 subsequence to end with different bases in some circumstances.
13914 Record the chosen access functions in SUB_ACCESS_FN.
13915 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13916 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13917 (subscript_dependence_tester_1): Likewise dra and drb.
13918 (build_classic_dist_vector): Update calls accordingly.
13919 (subscript_dependence_tester): Likewise.
13920 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13921 DDR_COULD_BE_INDEPENDENT_P.
13922 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13923 comp_alias_ddrs instead of may_alias_ddrs.
13924 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13926 (vect_analyze_data_ref_dependence): Use it if
13927 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13928 distance vectors if that fails.
13929 (dependence_distance_ge_vf): New function.
13930 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13931 LOOP_VINFO_MAY_ALIAS_DDRS.
13933 2017-08-04 Richard Biener <rguenther@suse.de>
13935 PR middle-end/81705
13936 * fold-const.c (fold_binary_loc): Properly restrict
13937 minus_var0 && minus_var1 case when associating undefined overflow
13940 2017-08-04 Tom de Vries <tom@codesourcery.com>
13942 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13944 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13946 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13947 Don't start diagnostic messages with a capital letter.
13948 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13950 (rs6000_invalid_builtin): Likewise.
13951 (rs6000_trampoline_init): Likewise.
13953 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13956 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13957 after setting changeable df flags.
13959 2017-08-03 Richard Biener <rguenther@suse.de>
13961 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13962 up if the use is in USE - X.
13964 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13966 * toplev.c (dumpfile.h): New include.
13967 (internal_error_reentered): New static function. Use it...
13968 (internal_error_function): ...here to handle reentered internal_error.
13970 2017-08-03 Richard Biener <rguenther@suse.de>
13972 PR middle-end/81148
13973 * fold-const.c (split_tree): Add minus_var and minus_con
13974 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13975 here but always use minus_*.
13976 (associate_trees): Assert we never associate with MINUS_EXPR
13977 and NULL first operand. Do not recurse for PLUS_EXPR operands
13978 when associating as MINUS_EXPR either.
13979 (fold_binary_loc): Track minus_var and minus_con.
13981 2017-08-03 Tom de Vries <tom@codesourcery.com>
13984 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13985 ACCEL_COMPILER, apply finish_options on
13986 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13988 2017-08-03 Tom de Vries <tom@codesourcery.com>
13991 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13992 function_entry_patch_area_size > 0.
13994 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13997 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13998 instead of 10??LU, perform unit multiplication in wide_int,
13999 don't change alloc_object_size_limit if the limit is larger
14002 PR tree-optimization/81655
14003 PR tree-optimization/81588
14004 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14005 the case when ranges[i].low and high are 1 for unsigned type with
14008 PR middle-end/81052
14009 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14010 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14012 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14014 * tree-vrp.h: Add include guard.
14016 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14019 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14020 (ud2): New insn pattern.
14021 * config/i386/i386.c (ix86_expand_epilogue):
14022 For naked functions, generate ud2 instead of trap insn.
14024 2017-08-02 Marek Polacek <polacek@redhat.com>
14027 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14029 2017-08-02 Tom de Vries <tom@codesourcery.com>
14030 Cesar Philippidis <cesar@codesourcery.com>
14032 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14033 Add missing edge probabilities.
14035 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14037 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14038 Correct endianness.
14040 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14042 PR middle-end/79499
14043 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14044 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14045 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14046 prologue_seq sequences - if any.
14048 2017-08-02 Richard Biener <rguenther@suse.de>
14050 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14051 via vectors if supported, integer extracts via punning if supported
14052 or otherwise vector extracts.
14054 2017-08-02 Richard Biener <rguenther@suse.de>
14056 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14058 (bitmap_insert_into_set): ... this.
14060 2017-08-02 Richard Biener <rguenther@suse.de>
14062 PR tree-optimization/81633
14064 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14066 PR tree-optimization/71752
14067 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14069 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14071 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14072 (machine_function::call_ms2sysv_pad_out): Remove field.
14073 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14074 (ix86_compute_frame_layout): Likewise.
14076 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14079 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14080 attribute with interrupt attribute.
14082 2017-08-01 Andrew Pinski <apinski@cavium.com>
14084 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14085 BIT_INSERT_EXPR's operand 1
14086 to see if the types precision matches.
14088 2017-08-01 Martin Liska <mliska@suse.cz>
14090 PR middle-end/70140
14091 * builtins.c (expand_builtin_memcpy_args): Remove.
14092 (expand_builtin_memcpy): Call newly added function
14093 expand_builtin_memory_copy_args.
14094 (expand_builtin_memcpy_with_bounds): Likewise.
14095 (expand_builtin_mempcpy): Remove last argument.
14096 (expand_builtin_mempcpy_with_bounds): Likewise.
14097 (expand_builtin_memory_copy_args): New function created from
14098 expand_builtin_mempcpy_args with small modifications.
14099 (expand_builtin_mempcpy_args): Remove.
14100 (expand_builtin_stpcpy): Remove unused argument.
14101 (expand_builtin): Likewise.
14102 (expand_builtin_with_bounds): Likewise.
14104 2017-08-01 Martin Liska <mliska@suse.cz>
14107 Make mempcpy more optimal (PR middle-end/70140).
14109 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14112 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14113 __builtin_vec_cmpne verify both arguments are compatible vectors
14114 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14115 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14116 move computation of aligned to after checking the argument types.
14120 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14121 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14124 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14125 Doug Rupp <rupp@adacore.com>
14126 Olivier Hainque <hainque@adacore.com>
14128 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14129 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14131 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14132 for TARGET_OS_CPP_BUILTIN.
14133 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14134 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14136 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14137 passing required abi options to the assembler for EABI configurations.
14138 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14139 of .text.hot and .text.unlikely sections for kernel modules when
14140 using ARM style exceptions.
14141 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14142 options. Add EXTRA_CC1_SPEC.
14143 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14145 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14147 (ARM_TARGET2_DWARF_FORMAT): Define.
14148 * config/arm/t-vxworks: Adjust multilib control to removal of the
14149 Diab command line options.
14151 2017-08-01 Martin Liska <mliska@suse.cz>
14153 PR gcov-profile/81561
14154 * gcov.c (unblock): Make unblocking safe as we need to preserve
14155 index correspondence of blocks and block_lists.
14157 2017-08-01 Richard Biener <rguenther@suse.de>
14159 PR tree-optimization/81181
14160 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14161 (compute_antic): ... end of iteration here.
14163 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14165 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14166 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14167 (ftree-slp-vectorize): Likewise.
14168 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14169 can no longer be set independent of flag_tree_loop_vectorize.
14170 * omp-general.c (emp_max_vf): Likewise.
14171 * opts.c (enable_fdo_optimizations): Remove references to
14172 flag_tree_vectorize, these are now implicit.
14173 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14174 and leave it for the options machinery.
14176 2017-08-01 Martin Liska <mliska@suse.cz>
14178 PR middle-end/70140
14179 * builtins.c (expand_builtin_memcpy_args): Remove.
14180 (expand_builtin_memcpy): Call newly added function
14181 expand_builtin_memory_copy_args.
14182 (expand_builtin_memcpy_with_bounds): Likewise.
14183 (expand_builtin_mempcpy): Remove last argument.
14184 (expand_builtin_mempcpy_with_bounds): Likewise.
14185 (expand_builtin_memory_copy_args): New function created from
14186 expand_builtin_mempcpy_args with small modifications.
14187 (expand_builtin_mempcpy_args): Remove.
14188 (expand_builtin_stpcpy): Remove unused argument.
14189 (expand_builtin): Likewise.
14190 (expand_builtin_with_bounds): Likewise.
14192 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14195 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14196 print "ds:" only for immediates in generic address space.
14198 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14201 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14202 (ix86_function_ok_for_sibcall): Return false for naked functions.
14204 2017-08-01 Richard Biener <rguenther@suse.de>
14206 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14207 (compute_antic): Seed worklist with exit block predecessors.
14208 * cfganal.c (dfs_find_deadend): For a cycle return the source
14209 of the edge closing it.
14211 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14213 * config/aarch64/aarch64.c
14214 (aarch64_can_const_movi_rtx_p): Move 0 check.
14216 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14218 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14219 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14221 * match.pd: Ditto in address comparison pattern.
14223 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14225 PR tree-optimization/81627
14226 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14227 closed ssa form for store-store chain.
14229 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14231 PR tree-optimization/81620
14232 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14233 for store-store chain.
14235 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14237 PR tree-optimization/81588
14238 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14239 ranges[i].in_p, invert comparison code ccode. For >/>=,
14240 swap rhs1 and rhs2 and comparison code unconditionally,
14241 for </<= don't do that. Don't swap rhs1/rhs2 again if
14242 ranges[i].in_p, instead invert comparison code ccode if
14243 opcode or oe->rank is BIT_IOR_EXPR.
14246 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14247 a direct optab to conversion optab.
14248 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14249 with GET_MODE_INNER as last argument instead of optab_handler.
14250 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14251 vector extraction if possible and optab is available.
14252 * expr.c (store_constructor): Use convert_optab_handler instead
14253 of optab_handler. Use vector initialization from smaller
14254 vectors if possible and optab is available.
14255 * tree-vect-stmts.c (vectorizable_load): Likewise.
14256 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14257 now have two modes.
14258 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14259 of vec_init from half-sized vectors with the same element mode.
14260 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14261 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14262 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14263 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14264 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14265 after mode in gen_vec_extract* calls.
14266 (vec_extract<mode>): Renamed to ...
14267 (vec_extract<mode><ssescalarmodelower>): ... this.
14268 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14269 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14270 element mode after mode in gen_vec_init* calls.
14271 (VEC_INIT_HALF_MODE): New mode iterator.
14272 (vec_init<mode>): Renamed to ...
14273 (vec_init<mode><ssescalarmodelower>): ... this.
14274 (vec_init<mode><ssehalfvecmodelower>): New expander.
14275 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14276 (vec_extractv2sfsf): ... this.
14277 (vec_initv2sf): Renamed to ...
14278 (vec_initv2sfsf): ... this.
14279 (vec_extractv2si): Renamed to ...
14280 (vec_extractv2sisi): ... this.
14281 (vec_initv2si): Renamed to ...
14282 (vec_initv2sisi): ... this.
14283 (vec_extractv4hi): Renamed to ...
14284 (vec_extractv4hihi): ... this.
14285 (vec_initv4hi): Renamed to ...
14286 (vec_initv4hihi): ... this.
14287 (vec_extractv8qi): Renamed to ...
14288 (vec_extractv8qiqi): ... this.
14289 (vec_initv8qi): Renamed to ...
14290 (vec_initv8qiqi): ... this.
14291 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14292 (vec_init<mode>): Renamed to ...
14293 (vec_init<mode><VEC_base_l>): ... this.
14294 (vec_extract<mode>): Renamed to ...
14295 (vec_extract<mode><VEC_base_l>): ... this.
14296 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14297 (vec_initv2sfsf): ... this.
14298 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14299 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14300 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14301 element mode after mode in gen_vec_init* calls.
14302 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14303 (vec_init<mode><Vel>): ... this.
14304 (vec_extract<mode>): Renamed to ...
14305 (vec_extract<mode><Vel>): ... this.
14306 * config/aarch64/iterators.md (Vel): New mode attribute.
14307 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14308 Add element mode after mode in gen_vec_extract* calls.
14309 * config/s390/vector.md (non_vec_l): New mode attribute.
14310 (vec_extract<mode>): Renamed to ...
14311 (vec_extract<mode><non_vec_l>): ... this.
14312 (vec_init<mode>): Renamed to ...
14313 (vec_init<mode><non_vec_l>): ... this.
14314 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14315 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14317 * config/arm/iterators.md (V_elem_l): New mode attribute.
14318 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14319 (vec_extract<mode><V_elem_l>): ... this.
14320 (vec_extractv2di): Renamed to ...
14321 (vec_extractv2didi): ... this.
14322 (vec_init<mode>): Renamed to ...
14323 (vec_init<mode><V_elem_l>): ... this.
14324 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14325 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14326 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14327 Add element mode after gen_vec_extract* calls.
14328 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14329 (vec_init<mode><unitmode>): ... this.
14330 (vec_extract<mode>): Renamed to ...
14331 (vec_extract<mode><unitmode>): ... this.
14332 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14333 (vec_init<mode><unitmode>): ... this.
14334 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14335 (vec_initv2sfsf): ... this.
14336 (vec_extractv2sf): Renamed to ...
14337 (vec_extractv2sfsf): ... this.
14338 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14339 Add element mode after gen_vec_extract* calls.
14340 * config/mips/mips.md (unitmode): New mode iterator.
14341 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14342 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14343 * config/spu/spu.md (inner_l): New mode attribute.
14344 (vec_init<mode>): Renamed to ...
14345 (vec_init<mode><inner_l>): ... this.
14346 (vec_extract<mode>): Renamed to ...
14347 (vec_extract<mode><inner_l>): ... this.
14348 * config/sparc/sparc.md (veltmode): New mode iterator.
14349 (vec_init<VMALL:mode>): Renamed to ...
14350 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14351 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14352 (vec_initv2sisi): ... this.
14353 (vec_initv2sf): Renamed to ...
14354 (vec_initv2sfsf): ... this.
14355 (vec_extractv2sf): Renamed to ...
14356 (vec_extractv2sfsf): ... this.
14357 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14358 (vec_init<mode>): Renamed to ...
14359 (vec_init<mode><VEC_base_l>): ... this.
14360 (vec_extract<mode>): Renamed to ...
14361 (vec_extract<mode><VEC_base_l>): ... this.
14362 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14363 (vec_initv2sfsf): ... this.
14364 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14365 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14366 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14367 gen_vec_init* calls.
14369 2017-08-01 Richard Biener <rguenther@suse.de>
14371 PR tree-optimization/81297
14372 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14373 TREE_OVERFLOW from INTEGER_CSTs.
14375 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14377 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14379 2017-07-31 Carl Love <cel@us.ibm.com>
14381 * config/rs6000/rs6000-c: Add support for built-in functions
14382 vector signed char vec_xl_be (signed long long, signed char *);
14383 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14384 vector signed int vec_xl_be (signed long long, signed int *);
14385 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14386 vector signed long long vec_xl_be (signed long long, signed long long *);
14387 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14388 vector signed short vec_xl_be (signed long long, signed short *);
14389 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14390 vector double vec_xl_be (signed long long, double *);
14391 vector float vec_xl_be (signed long long, float *);
14392 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14393 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14394 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14396 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14397 (altivec_expand_builtin): Add switch statement to call
14398 altivec_expand_xl_be for each builtin.
14399 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14400 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14401 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14402 __builtin_vsx_le_be_v16qi.
14403 * doc/extend.texi: Update the built-in documentation file for the
14404 new built-in functions.
14406 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14409 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14411 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14413 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14415 * config.gcc: Add z14.
14416 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14417 CPU model numbers for z13s and z14.
14418 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14420 * config/s390/s390-opts.h (enum processor_type): Rename
14421 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14422 * config/s390/s390.c (processor_table): Add field for CPU name to
14423 be passed to Binutils.
14424 (s390_asm_output_machine_for_arch): Use the new field in
14425 processor_table for Binutils.
14426 (s390_expand_builtin): Replace arch12 with z14.
14427 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14428 (s390_get_sched_attrmask): Likewise.
14429 (s390_get_unit_mask): Likewise.
14430 * config/s390/s390.opt: Add z14 to processor_type enum.
14432 2017-07-31 Martin Jambor <mjambor@suse.cz>
14435 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14436 regardless of optimization level.
14438 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14439 Martin Liska <mliska@suse.cz>
14441 * predict.def: Remove old comment and adjust probability.
14442 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14443 PREDICT statements.
14445 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14448 * config/i386/i386.c (ix86_function_naked): New function.
14449 (ix86_can_use_return_insn_p): Return false for naked functions.
14450 (ix86_expand_prologue): Skip prologue for naked functions.
14451 (ix86_expand_epilogue): Skip epilogue for naked functions
14452 and emit trap instruction.
14453 (ix86_warn_func_return): New function.
14454 (ix86_attribute_table): Add "naked" attribute specification.
14455 (TARGET_WARN_FUNC_RETURN): Define.
14456 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14458 2017-07-31 Martin Liska <mliska@suse.cz>
14460 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14461 (dump_gimple_bb_header): Always dump BB info.
14462 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14464 2017-07-31 Martin Liska <mliska@suse.cz>
14467 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14468 also with current_function_decl non-null equality.
14470 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14473 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14474 change type to the element type, instead add eltype variable and
14475 use it where we are interested in the element type.
14477 PR tree-optimization/81603
14478 * ipa-polymorphic-call.c
14479 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14480 offset arithmetic in offset_int, bail out if the resulting bit offset
14481 doesn't fit into shwi.
14483 2017-07-31 Martin Liska <mliska@suse.cz>
14485 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14486 (gimplify_save_expr): Fix comment.
14488 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14491 * config/i386/i386.c (ix86_function_arg): Update arguments for
14493 (ix86_compute_frame_layout): Set the initial stack offset to
14494 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14495 INCOMING_FRAME_SP_OFFSET.
14496 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14497 stack before exception handler returns.
14498 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14499 the 'ERROR_CODE' for exception handler.
14501 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14503 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14504 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14505 (ASM_OUTPUT_REG_POP): Ditto.
14506 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14507 instead of asm_fprintf to output pure string.
14509 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14511 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14512 to imported_module_or_decl hook.
14513 (debug_nothing_tree_tree_tree_bool): Remove.
14514 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14515 * debug.c (do_nothing_debug_hooks): Use
14516 debug_nothing_tree_tree_tree_bool_bool instead of
14517 debug_nothing_tree_tree_tree_bool.
14518 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14519 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14520 * sdbout.c (sdb_debug_hooks): Likewise.
14521 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14522 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14524 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14525 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14526 attribute, don't add anything.
14528 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14530 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14531 (fold_build2_stat_loc): Likewise.
14532 (fold_build3_stat_loc): Likewise.
14533 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14534 (fold_build1_loc): Remove macro.
14535 (fold_build2_loc): Likewise.
14536 (fold_build3_loc): Likewise.
14538 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14540 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14541 (gimple_build_debug_bind_source_stat): Likewise.
14542 * gimple.h (gimple_build_debug_bind): Remove macro.
14543 (gimple_build_debug_bind_source): Likewise.
14545 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14547 * bitmap.c (bitmap_alloc): Adjust.
14548 (bitmap_gc_alloc): Likewise.
14549 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14551 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14553 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14554 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14555 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14556 (bitmap_gc_alloc_stat): Likewise.
14557 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14559 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14561 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14562 * rtl.h (shallow_copy_rtx): Remove macro.
14564 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14566 * emit-rtl.c (gen_raw_REG): Adjust.
14567 * gengenrtl.c (gendef): Likewise.
14568 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14569 * rtl.h (rtx_alloc): Remove macro.
14571 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14573 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14574 (build_tree_list_stat): Likewise.
14575 * tree.h (build_tree_list): Remove macro.
14576 (build_tree_list_vec): Likewise.
14578 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14580 * tree.c (make_vector_stat): Remove _stat from name.
14581 (build_vector_stat): Likewise.
14582 * tree.h (make_vector_stat): Remove macro.
14583 (build_vector_stat): Likewise.
14585 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14587 * tree.h (build_var_debug_value): Remove prototype.
14589 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14591 * tree.c (tree_cons_stat): Remove _stat from name.
14592 * tree.h (tree_cons): Remove macro.
14594 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14596 * tree.c (build_vl_exp_stat): Remove _stat from name.
14597 * tree.h (build_vl_exp): Remove macro.
14599 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14601 * tree.c (build_decl_stat): Remove _stat from name.
14602 * tree.h (build_decl): Remove macro.
14604 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14606 * gimple.c (gimple_build_with_ops_stat): Adjust.
14607 (gimple_alloc_stat): Remove _stat from name.
14608 * gimple.h (gimple_alloc): Remove macro.
14610 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14612 * tree.c (make_tree_vec_stat): Remove _stat from name.
14613 (grow_tree_vec_stat): Likewise.
14614 * tree.h (make_tree_vec_stat): Adjust prototype.
14615 (grow_tree_vec_stat): Likewise.
14616 (make_tree_vec): Remove macro.
14617 (grow_tree_vec): Likewise.
14619 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14621 * fold-const.c (fold_build1_stat_loc): Adjust.
14622 (fold_build2_stat_loc): Likewise.
14623 (fold_build3_stat_loc): Likewise.
14624 * tree.c (build0_stat): Remove _stat from name.
14625 (build1_stat): Likewise.
14626 (build2_stat): Likewise.
14627 (build3_stat): Likewise.
14628 (build4_stat): Likewise.
14629 (build5_stat): Likewise.
14630 * tree.h (build1_loc): Remove macro, and rename _stat function
14632 (build2_loc): Likewise.
14633 (build3_loc): Likewise.
14634 (build4_loc): Likewise.
14635 (build5_loc): Likewise.
14637 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14639 * tree.c (make_int_cst_stat): Remove _stat from name.
14640 * tree.h (make_int_cst_stat): Adjust prototype.
14641 (make_int_cst): Remove macro.
14643 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14645 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14646 * tree.h (make_tree_binfo_stat): Adjust prototype.
14647 (make_tree_binfo): Remove.
14649 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14651 * tree.c (copy_node_stat): Rename to copy_node.
14652 (build_distinct_type_copy): Adjust.
14653 * tree.h (copy_node_stat): Adjust prototype.
14654 (copy_node): Remove macro.
14656 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14658 * tree.c (make_node_stat): rename to make_node.
14659 (build_tree_list_stat): Adjust.
14660 (build0_stat): Likewise.
14661 (build2_stat): Likewise.
14662 (build3_stat): Likewise.
14663 (build4_stat): Likewise.
14664 (build5_stat): Likewise.
14665 (build_decl_stat): Likewise.
14666 * tree.h (make_node_stat): Adjust prototype.
14667 (make_node): remove macro.
14669 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14671 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14672 (PPC_FEATURE2_SCV): Likewise.
14673 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14675 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14677 * config/aarch64/aarch64.c
14678 (aarch64_internal_mov_immediate): Add new special pattern.
14679 * config/aarch64/aarch64.md (*movdi_aarch64):
14680 Add reg/32bit const mov case.
14682 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14683 Richard Sandiford <richard.sandiford@linaro.org>
14685 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14686 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14687 Add integer and movi cases.
14688 (movi-split-hf-df-sf split, fp16): New.
14689 (enabled): Added TARGET_FP_F16INST.
14690 * config/aarch64/iterators.md (GPF_HF): New.
14691 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14693 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14695 * config/aarch64/aarch64.c
14696 (aarch64_simd_container_mode): Add prototype.
14697 (aarch64_expand_mov_immediate): Add HI support.
14698 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14699 (aarch64_can_const_movi_rtx_p): New.
14700 (aarch64_preferred_reload_class):
14701 Remove restrictions of using FP registers for certain SIMD operations.
14702 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14703 (aarch64_valid_floating_const): Add integer move validation.
14704 (aarch64_simd_imm_scalar_p): Remove.
14705 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14706 (aarch64_legitimate_constant_p): Expand list of supported cases.
14707 * config/aarch64/aarch64-protos.h
14708 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14709 (aarch64_reinterpret_float_as_int): New.
14710 (aarch64_simd_imm_scalar_p): Remove.
14711 * config/aarch64/constraints.md (Uvi): New.
14712 (Dd): Split into Ds and new Dd.
14713 * config/aarch64/aarch64.md (*movsi_aarch64):
14715 (*movdi_aarch64): Add SIMD mov case.
14717 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14719 * tree-predcom.c: (struct chain): Handle store-store chain in which
14720 stores for elimination only store loop invariant values.
14721 (execute_pred_commoning_chain): Ditto.
14722 (prepare_initializers_chain_store_elim): Ditto.
14723 (prepare_finalizers): Ditto.
14724 (is_inv_store_elimination_chain): New function.
14725 (initialize_root_vars_store_elim_1): New function.
14727 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14729 * tree-predcom.c: Revise general description of the pass.
14730 (enum chain_type): New enum type for store elimination.
14731 (struct chain): New field supporting store elimination.
14732 (struct component): Ditto.
14733 (dump_chain): Dump store-stores chain.
14734 (release_chain): Release resources.
14735 (split_data_refs_to_components): Compute and create component
14736 contains only stores for elimination.
14737 (get_chain_last_ref_at): New function.
14738 (make_invariant_chain): Initialization.
14739 (make_rooted_chain): Specify chain type in parameter and record it.
14740 (add_looparound_copies): Skip for store-stores chain.
14741 (determine_roots_comp): Compute type of chain and pass it to
14743 (initialize_root_vars_store_elim_2): New function.
14744 (finalize_eliminated_stores): New function.
14745 (remove_stmt): Handle store for elimination.
14746 (execute_pred_commoning_chain): Execute predictive commoning on
14747 store-store chains.
14748 (determine_unroll_factor): Skip unroll for store-stores chain.
14749 (prepare_initializers_chain_store_elim): New function.
14750 (prepare_initializers_chain): Hanlde store-store chain.
14751 (prepare_finalizers_chain, prepare_finalizers): New function.
14752 (tree_predictive_commoning_loop): Return integer value indicating
14753 if loop is unrolled or lcssa form is corrupted.
14754 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14756 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14758 * tree-predcom.c (initialize_root): Delete.
14759 (execute_pred_commoning_chain): Initialize root vars and replace
14760 reference of non-combined chain directly, rather than call above
14763 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14765 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14766 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14768 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14770 * tree-predcom.c (struct chain): New field init_seq.
14771 (release_chain): Release init_seq.
14772 (prepare_initializers_chain): Record intialization stmts in above
14774 (insert_init_seqs): New function.
14775 (tree_predictive_commoning_loop): Call insert_init_seqs.
14777 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14779 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14781 2017-07-28 Richard Biener <rguenther@suse.de>
14783 * match.pd: Remove superfluous :c.
14784 * genmatch.c (simplify::id): Add member.
14785 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14787 (current_id): New global.
14788 (dt_node::parent): Move from ...
14789 (dt_operand::parent): ... here. Add for_id member.
14790 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14791 (decision_tree::find_node): Relax order requirement when
14792 merging DT_TRUE nodes to ones inbetween the current simplify
14793 and the one we try to merge with. Add diagnostic whenever
14794 we need to enforce pattern order by not merging.
14795 (decision_tree::insert): Set current_id.
14796 (decision_tree::print_node): Dump parent node and for_id.
14797 (parser::last_id): Add member.
14798 (parser::push_simplify): Assign unique id.
14799 (parser::parser): Initialize last_id.
14801 2017-07-28 Martin Liska <mliska@suse.cz>
14804 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14805 gimple_build_debug_bind.
14807 2017-07-28 Richard Biener <rguenther@suse.de>
14809 PR tree-optimization/81502
14810 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14812 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14814 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14815 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14816 for BIT_FIELD_REF args.
14817 * fold-const.c (make_bit_field_ref): Likewise.
14818 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14820 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14823 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14824 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14825 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14826 Or it into SANITIZER_UNDEFINED.
14827 * ubsan.c: Include gimple-fold.h and varasm.h.
14828 (ubsan_expand_ptr_ifn): New function.
14829 (instrument_pointer_overflow): New function.
14830 (maybe_instrument_pointer_overflow): New function.
14831 (instrument_object_size): Formatting fix.
14832 (pass_ubsan::execute): Call instrument_pointer_overflow
14833 and maybe_instrument_pointer_overflow.
14834 * internal-fn.c (expand_UBSAN_PTR): New function.
14835 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14836 * sanitizer.def (__ubsan_handle_pointer_overflow,
14837 __ubsan_handle_pointer_overflow_abort): New builtins.
14838 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14839 * internal-fn.def (UBSAN_PTR): New internal function.
14840 * opts.c (sanitizer_opts): Add pointer-overflow.
14841 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14842 * fold-const.c (build_range_check): Compute pointer range check in
14843 integral type if pointer arithmetics would be needed. Formatting
14846 2017-07-28 Martin Liska <mliska@suse.cz>
14849 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14850 parameters that are of a variable-length.
14852 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14854 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14856 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14857 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14858 (CRT_CALL_STATIC_FUNCTION): Likewise.
14859 (ASM_DEFAULT_SPEC): New define.
14860 (ASM_SPEC32): Likewise.
14861 (ASM_SPEC64): Likewise.
14862 (ASM_SPEC_COMMON): Likewise.
14863 (ASM_SPEC): Likewise.
14864 (INVALID_64BIT): Likewise.
14865 (LINK_OS_DEFAULT_SPEC): Likewise.
14866 (LINK_OS_SPEC32): Likewise.
14867 (LINK_OS_SPEC64): Likewise.
14868 (POWERPC_LINUX): Likewise.
14869 (PTRDIFF_TYPE): Likewise.
14870 (RESTORE_FP_PREFIX): Likewise.
14871 (RESTORE_FP_SUFFIX): Likewise.
14872 (SAVE_FP_PREFIX): Likewise.
14873 (SAVE_FP_SUFFIX): Likewise.
14874 (SIZE_TYPE): Likewise.
14875 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14876 (TARGET_64BIT): Likewise.
14877 (TARGET_64BIT): Likewise.
14878 (TARGET_AIX): Likewise.
14879 (WCHAR_TYPE_SIZE): Likewise.
14880 (WCHAR_TYPE): Undefine.
14881 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14882 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14883 (CPP_OS_RTEMS_SPEC): Delete.
14884 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14885 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14887 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14889 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14891 PR tree-optimization/81578
14892 * tree-parloops.c (build_new_reduction): Bail out if
14893 reduction_code isn't one of the standard OpenMP reductions.
14894 Move the details printing after that decision.
14896 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14898 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14899 related to reload_in_progress.
14900 (splat_input_operand): Likewise.
14901 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14903 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14905 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14906 (TARGET_INSTANTIATE_DECLS): Likewise.
14907 (legitimate_indexed_address_p): Delete reload_in_progress code.
14908 (rs6000_debug_legitimate_address_p): Likewise.
14909 (rs6000_eliminate_indexed_memrefs): Likewise.
14910 (rs6000_emit_le_vsx_store): Likewise.
14911 (rs6000_emit_move_si_sf_subreg): Likewise.
14912 (rs6000_emit_move): Likewise.
14913 (register_to_reg_type): Likewise.
14914 (rs6000_pre_atomic_barrier): Likewise.
14915 (rs6000_machopic_legitimize_pic_address): Likewise.
14916 (rs6000_allocate_stack_temp): Likewise.
14917 (rs6000_address_for_fpconvert): Likewise.
14918 (rs6000_address_for_altivec): Likewise.
14919 (rs6000_secondary_memory_needed_rtx): Delete function.
14920 (rs6000_check_sdmode): Likewise.
14921 (rs6000_alloc_sdmode_stack_slot): Likewise.
14922 (rs6000_instantiate_decls): Likewise.
14923 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14924 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14925 Delete reload_in_progress.
14926 (*vec_reload_and_plus_<mptrsize>): Likewise.
14927 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14928 (vsx_div_v2di): Likewise.
14929 (vsx_udiv_v2di): Likewise.
14931 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14933 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14934 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14935 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14936 (rs6000_debug_reg_global): Delete print of LRA status.
14937 (rs6000_option_override_internal): Delete dead LRA related code.
14938 (rs6000_lra_p): Delete function.
14939 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14941 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14943 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14944 * config/riscv/rtems.h: New file.
14946 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14947 Sudakshina Das <sudi.das@arm.com>
14949 * config/aarch64/aarch64.md
14950 (define_split for and<mode>3nr_compare): Move
14951 non aarch64_logical_operand to a register.
14952 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14953 register immediate operand to a register.
14954 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14956 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14958 PR middle-end/81564
14959 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14961 2017-07-27 Richard Biener <rguenther@suse.de>
14963 PR tree-optimization/81573
14964 PR tree-optimization/81494
14965 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14966 multi defuse cycle case.
14968 2017-07-27 Richard Biener <rguenther@suse.de>
14970 PR tree-optimization/81571
14971 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14974 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14976 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14977 earlier and only if MASK_FPU is set. Adjust formatting.
14979 2017-07-27 Martin Liska <mliska@suse.cz>
14981 * opt-functions.awk: Add validation of value of Init.
14982 * optc-gen.awk: Pass new argument.
14984 2017-07-27 Martin Liska <mliska@suse.cz>
14986 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14987 Fix wrong condition.
14989 2017-07-27 Martin Liska <mliska@suse.cz>
14991 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14992 BBs and edges seen by autoFDO.
14994 2017-07-27 Richard Biener <rguenther@suse.de>
14996 PR tree-optimization/81502
14997 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14998 with incompatible but same sized type.
14999 (execute_update_addresses_taken): Likewise.
15001 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15003 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15004 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15006 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15009 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15010 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15011 Change s_operand to memory_operand.
15013 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15015 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15016 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15017 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15018 Emit instructions rather than returning an expression. Handle TFmode
15019 and KFmode by casting to TImode.
15020 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15021 (rs6000_emit_le_vsx_store): Likewise.
15022 * config/rs6000/vsx.md (VSX_TI): New iterator.
15023 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15024 (*vsx_le_undo_permute_<mode>): Likewise.
15025 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15026 emit the split sequence.
15027 (*vsx_le_perm_store_<mode>): Likewise.
15029 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15031 PR tree-optimization/81555
15032 PR tree-optimization/81556
15033 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15034 if true, force CHANGED for the recursive invocation.
15035 (reassociate_bb): Remember original length of ops array, pass
15036 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15038 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15039 attributes for noipa attribute. For naked attribute use
15040 lookup_attribute first before lookup_attribute_spec.
15041 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15043 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15044 for functions with noipa attribute.
15045 (cgraph_externally_visible_p): Return true for functions with noipa
15047 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15048 for functions with noipa attribute.
15049 * doc/extend.texi: Document noipa function attribute.
15050 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15051 also for functions with noipa attribute.
15052 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15054 2017-07-26 Andrew Pinski <apinski@cavium.com>
15056 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15057 vec_unalign_load_cost and vec_unalign_store_cost.
15059 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15061 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15062 -mvsx-small-integer option.
15063 (ISA_3_0_MASKS_IEEE): Likewise.
15064 (OTHER_VSX_VECTOR_MASKS): Likewise.
15065 (POWERPC_MASKS): Likewise.
15066 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15067 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15068 code, only testing for DImode being allowed in non-VSX floating
15070 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15071 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15073 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15074 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15075 TARGET_P8_VECTOR test.
15076 (rs6000_secondary_reload_simple_move): Likewise.
15077 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15078 since TARGET_P9_VECTOR was already tested.
15079 (rs6000_opt_masks): Remove -mvsx-small-integer.
15080 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15081 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15083 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15084 test for TARGET_VEXTRACTUB was used, and that uses
15086 (p9 extract splitter): Likewise.
15087 (vsx_extract_<mode>_di_p9): Likewise.
15088 (vsx_extract_<mode>_store_p9): Likewise.
15089 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15090 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15091 the elimination of TARGET_VSX_SMALL_INTEGER.
15092 (vsx_extract_<mode>_p8): Likewise.
15093 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15094 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15095 (vsx_set_<mode>_p9): Likewise.
15096 (vsx_set_v4sf_p9): Likewise.
15097 (vsx_set_v4sf_p9_zero): Likewise.
15098 (vsx_insert_extract_v4sf_p9): Likewise.
15099 (vsx_insert_extract_v4sf_p9_2): Likewise.
15100 * config/rs6000/rs6000.md (sign extend splitter): Change
15101 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15102 (floatsi<mode>2_lfiwax_mem): Likewise.
15103 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15104 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15105 since a test for TARGET_P9_VECTOR was used.
15106 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15107 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15108 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15109 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15110 TARGET_P8_VECTOR test.
15111 (fix_trunc<mode>si2_stfiwx): Likewise.
15112 (fix_trunc<mode>si2_internal): Likewise.
15113 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15114 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15116 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15117 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15118 TARGET_P8_VECTOR test.
15119 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15120 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15121 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15123 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15124 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15125 since a test for TARGET_P9_VECTOR was used.
15126 (splitter for loading small constants): Likewise.
15128 2017-07-26 Andrew Pinski <apinski@cavium.com>
15130 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15133 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15136 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15137 (fp_valid_at): Likewise.
15139 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15141 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15142 (qdf24xx_addrcost_table): Likewise.
15143 (cortexa57_tunings): Update to use generic_branch_cost.
15144 (cortexa72_tunings): Likewise.
15145 (cortexa73_tunings): Likewise.
15146 (qdf24xx_tunings): Likewise.
15148 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15150 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15151 (thunderx2t99_branch_cost): Likewise.
15152 (cortexa35_tunings): Update to use generic_branch_cost.
15153 (cortexa53_tunings): Likewise.
15154 (cortexa57_tunings): Likewise.
15155 (cortexa72_tunings): Likewise.
15156 (cortexa73_tunings): Likewise.
15157 (thunderx2t99_tunings): Likewise.
15159 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15161 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15162 (sparc_option_override): Honour MASK_FSMULD.
15163 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15164 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15165 * config/sparc/sparc.opt (mfsmuld): New option.
15166 * doc/invoke.texi (mfsmuld): Document option.
15168 2017-07-26 Marek Polacek <polacek@redhat.com>
15170 PR middle-end/70992
15171 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15173 2017-07-26 Richard Biener <rguenther@suse.de>
15175 * gimple-match-head.c (do_valueize): Return OP if valueize
15177 (get_def): New helper to get at the def stmt of a SSA name
15178 if valueize allows.
15179 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15180 do_valueize to get at the def stmt.
15181 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15183 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15185 PR middle-end/46932
15186 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15188 2017-07-26 Martin Liska <mliska@suse.cz>
15191 * function.c (expand_function_start): Make expansion of
15192 nonlocal_goto_save_area after parm_birth_insn.
15194 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15196 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15197 from all CPU target flags enable members.
15199 2017-07-26 Richard Biener <rguenther@suse.de>
15201 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15202 (decision_tree::gen): Make 'type' const.
15203 (write_predicate): Likewise.
15205 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15207 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15208 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15209 (rs6000_option_override_internal): Likewise.
15210 (rs6000_expand_vector_set): Likewise.
15211 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15212 (TARGET_UPPER_REGS_SF): Likewise.
15213 (TARGET_UPPER_REGS_DI): Likewise.
15214 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15215 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15216 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15217 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15218 (Splitters for DI constants in Altivec registers): Likewise.
15219 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15220 (vsx_set_v4sf_p9): Likewise.
15221 (vsx_set_v4sf_p9_zero): Likewise.
15222 (vsx_insert_extract_v4sf_p9): Likewise.
15223 (vsx_insert_extract_v4sf_p9_2): Likewise.
15225 2017-07-25 Carl Love <cel@us.ibm.com>
15227 * doc/extend.texi: Update the built-in documentation file for the
15228 existing built-in functions
15229 vector signed char vec_cnttz (vector signed char);
15230 vector unsigned char vec_cnttz (vector unsigned char);
15231 vector signed short vec_cnttz (vector signed short);
15232 vector unsigned short vec_cnttz (vector unsigned short);
15233 vector signed int vec_cnttz (vector signed int);
15234 vector unsigned int vec_cnttz (vector unsigned int);
15235 vector signed long long vec_cnttz (vector signed long long);
15236 vector unsigned long long vec_cnttz (vector unsigned long long);
15238 2017-07-25 Andrew Pinski <apinski@cavium.com>
15240 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15241 accesses where the use is for the first operand of a BIT_INSERT.
15243 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15246 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15247 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15249 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15251 * config/i386/gstabs.h: Delete.
15252 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15254 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15256 * config/i386/i386.c (ix86_decompose_address): Do not check for
15257 register RTX when looking at index_reg or base_reg.
15258 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15260 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15262 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15263 to update EH info here.
15265 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15267 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15269 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15271 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15273 2017-07-25 Torsten Duwe <duwe@suse.de>
15275 * common.opt: Introduce -fpatchable-function-entry
15276 command line option, and its variables function_entry_patch_area_size
15277 and function_entry_patch_area_start.
15278 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15279 including a two-value parser.
15280 * target.def (print_patchable_function_entry): New target hook.
15281 * targhooks.h (default_print_patchable_function_entry): New function.
15282 * targhooks.c (default_print_patchable_function_entry): Likewise.
15283 * toplev.c (process_options): Switch off IPA-RA if
15284 patchable function entries are being generated.
15285 * varasm.c (assemble_start_function): Look at the
15286 patchable-function-entry command line switch and current
15287 function attributes and maybe generate NOP instructions by
15288 calling the print_patchable_function_entry hook.
15289 * doc/extend.texi: Document patchable_function_entry attribute.
15290 * doc/invoke.texi: Document -fpatchable_function_entry
15291 command line option.
15292 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15294 * doc/tm.texi: Re-generate.
15296 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15299 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15300 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15302 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15304 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15306 2017-07-25 Richard Biener <rguenther@suse.de>
15308 PR tree-optimization/81455
15309 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15310 not walk in cycles when looking for guards.
15312 2017-07-25 Richard Biener <rguenther@suse.de>
15314 PR tree-optimization/81529
15315 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15316 when optimizing backedge uses.
15318 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15320 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15322 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15323 to dl_section_ref. On AIX, append an expression to subtract
15324 the size of the section length to dl_section_ref.
15326 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15328 * configure.ac: If any of the config.* scripts fail, exit 1.
15329 * configure: Regenerate.
15331 2017-07-25 Richard Biener <rguenther@suse.de>
15333 PR middle-end/81546
15334 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15335 of immediate uses, be more verbose on errors.
15337 2017-07-25 Richard Biener <rguenther@suse.de>
15339 PR tree-optimization/81510
15340 * tree-vect-loop.c (vect_is_simple_reduction): When the
15341 reduction stmt is not inside the loop bail out.
15343 2017-07-25 Richard Biener <rguenther@suse.de>
15345 PR tree-optimization/81303
15346 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15347 profitability check against LOOP_VINFO_NITERSM1.
15349 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15351 * domwalk.c (cmp_bb_postorder): Simplify.
15352 (sort_bbs_postorder): New function. Use it...
15353 (dom_walker::walk): ...here to optimize common cases.
15355 2017-07-25 Martin Liska <mliska@suse.cz>
15358 * ipa-visibility.c (function_and_variable_visibility): Make the
15359 redirection just on target that supports aliasing.
15360 Fix GNU coding style.
15362 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15365 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15367 * config/arm/rtems.h: Likewise.
15368 * config/bfin/rtems.h: Likewise.
15369 * config/i386/rtemself.h: Likewise.
15370 * config/lm32/rtems.h: Likewise.
15371 * config/m32c/rtems.h: Likewise.
15372 * config/m68k/rtemself.h: Likewise.
15373 * config/microblaze/rtems.h: Likewise.
15374 * config/mips/rtems.h: Likewise.
15375 * config/moxie/rtems.h: Likewise.
15376 * config/nios2/rtems.h: Likewise.
15377 * config/powerpcspe/rtems.h: Likewise.
15378 * config/rs6000/rtems.h: Likewise.
15379 * config/rtems.h: Likewise.
15380 * config/sh/rtems.h: Likewise.
15381 * config/sh/rtemself.h: Likewise.
15382 * config/sparc/rtemself.h: Likewise.
15384 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15387 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15388 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15389 * tree-ssa-structalias.c (alias_get_name): Same.
15391 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15394 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15395 instructions if no du chain is found.
15397 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15399 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15401 2017-07-25 Richard Biener <rguenther@suse.de>
15403 PR middle-end/81505
15404 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15407 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15409 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15410 upper-regs options.
15411 (ISA_2_7_MASKS_SERVER): Likewise.
15412 (ISA_3_0_MASKS_IEEE): Likewise.
15413 (OTHER_P8_VECTOR_MASKS): Likewise.
15414 (OTHER_VSX_VECTOR_MASKS): Likewise.
15415 (POWERPC_MASKS): Likewise.
15416 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15417 duplicate list of options.
15418 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15419 explicit -mupper-regs options.
15420 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15421 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15422 alias for -mupper-regs-df.
15423 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15424 (rs6000_init_hard_regno_mode_ok): Likewise.
15425 (rs6000_option_override_internal): Likewise.
15426 (rs6000_opt_masks): Likewise.
15427 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15428 options in terms of whether -mvsx or -mpower8-vector was used.
15429 (TARGET_UPPER_REGS_DI): Likewise.
15430 (TARGET_UPPER_REGS_SF): Likewise.
15431 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15432 -mupper-regs-* options.
15434 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15436 * passes.c (emergency_dump_function): Print some empty lines and a
15437 header before the RTL dump.
15439 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15441 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15443 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15446 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15447 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15449 2017-07-24 Carl Love <cel@us.ibm.com>
15451 * config/rs6000/rs6000-c.c: Add support for built-in functions
15452 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15453 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15454 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15455 vec_extract_fp_from_shortl): Add defines for the two builtins.
15456 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15457 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15459 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15460 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15461 * doc/extend.texi: Update the built-in documentation file for the
15462 new built-in function.
15464 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15467 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15469 * doc/generic.texi: Likewise.
15470 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15471 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15473 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15475 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15476 (aarch64_mls_elt_merge<mode>): Likewise.
15478 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15480 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15481 having __cxa_atexit.
15483 2017-07-23 Michael Collison <michael.collison@arm.com>
15485 * config/arm/arm.c (arm_option_override): Deprecate
15486 use of -mstructure-size-boundary.
15487 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15488 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15490 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15493 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15494 Reduce cost estimate for direct moves.
15496 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15499 * config/i386/i386.c (ix86_option_override_internal): Disable
15500 BMI, BMI2 and TBM instructions for -m16.
15502 2017-07-21 Carl Love <cel@us.ibm.com>
15504 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15505 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15506 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15507 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15508 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15509 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15510 VMULOSW): New enum "unspec" values.
15511 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15512 altivec_vmulosw): New patterns.
15513 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15514 VMULOSW): Add definitions.
15516 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15518 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15519 (qdf24xx): Likewise.
15520 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15521 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15522 (AARCH64_FL_V8_1): Renumber.
15523 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15524 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15525 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15526 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15527 rdma to feature modifiers list.
15529 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15531 PR middle-end/56727
15532 * ipa-visibility (function_and_variable_visibility): Convert
15533 recursive PLT call to direct call if appropriate.
15535 2017-07-21 Andrew Pinski <apinski@cavium.com>
15537 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15538 operand 1 to see if the types precision matches.
15539 * fold-const.c (operand_equal_p): Likewise.
15541 2017-07-21 Richard Biener <rguenther@suse.de>
15543 PR tree-optimization/81303
15544 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15545 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15546 (vect_peeling_hash_get_lowest_cost): Adjust.
15547 (vect_enhance_data_refs_alignment): Likewise. Use
15548 vect_get_peeling_costs_all_drs to compute the penalty for no
15549 peeling to match up costs.
15551 2017-07-21 Richard Biener <rguenther@suse.de>
15553 PR tree-optimization/81500
15554 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15555 we didn't identify a reduction path.
15557 2017-07-21 Tom de Vries <tom@codesourcery.com>
15558 Cesar Philippidis <cesar@codesourcery.com>
15560 PR gcov-profile/81442
15561 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15564 2017-07-21 Tom de Vries <tom@codesourcery.com>
15567 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15569 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15570 nvptx_override_options_after_change.
15572 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15574 * dwarf2out.c (output_file_names): Avoid double testing for
15575 dwarf_version >= 5.
15577 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15579 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15581 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15583 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15585 (try_crossjump_to_edge): Do not punt on partitioned functions.
15587 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15589 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15590 Put all BBs reachable only via paths crossing cold region to cold
15592 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15594 2016-07-21 Richard Biener <rguenther@suse.de>
15596 PR tree-optimization/81303
15597 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15598 into account prologue and epilogue iterations when raising
15599 min_profitable_iters to sth at least covering one vector iteration.
15601 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15603 * config/arm/arm.c (arm_test_cpu_arch_dat):
15606 2017-07-20 Nathan Sidwell <nathan@acm.org>
15608 Remove TYPE_METHODS.
15609 * tree.h (TYPE_METHODS): Delete.
15610 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15611 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15612 (dbxout_type_methods): Scan TYPE_FIELDS.
15613 (dbxout_type): Don't check TYPE_METHODS here.
15614 * function.c (use_register_for_decl): Always ignore register for
15615 class types when not optimizing.
15616 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15617 * tree.c (free_lang_data_in_type): Stitch out member functions and
15618 templates from TYPE_FIELDS.
15619 (build_distinct_type_copy, verify_type_variant,
15620 verify_type): Member fns are on TYPE_FIELDS.
15621 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15622 * tree-pretty-print.c (dump_generic_node): Likewise.
15624 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15627 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15628 V2TImode and V4TImode.
15629 (ix86_expand_vector_extract): Likewise.
15630 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15631 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15632 (ssescalarmode): Handle V4TImode and V2TImode.
15633 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15634 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15635 (VEXTRACTI128_MODE): New mode iterator.
15636 (splitter for *vec_extractv?ti first element): New.
15637 (VEC_INIT_MODE): New mode iterator.
15638 (vec_init<mode>): Consolidate 3 expanders into one using
15639 VEC_INIT_MODE mode iterator.
15641 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15643 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15644 non_spilled_static_chain_regno_p.
15646 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15648 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15650 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15652 * bb-reorder.c (connect_traces): Allow copying of blocks within
15655 2017-07-20 Richard Biener <rguenther@suse.de>
15657 * gimple.h (gimple_phi_result): Add gphi * overload.
15658 (gimple_phi_result_ptr): Likewise.
15659 (gimple_phi_arg): Likewise. Adjust index assert to only
15660 allow actual argument accesses rather than all slots available
15662 (gimple_phi_arg_def): Add gphi * overload.
15663 * tree-phinodes.c (make_phi_node): Initialize only actual
15665 (resize_phi_node): Clear memory not covered by old node,
15666 do not initialize excess argument slots.
15667 (reserve_phi_args_for_new_edge): Initialize new argument slot
15670 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15672 PR tree-optimization/81388
15674 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15676 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15677 by removing computation of may_be_zero.
15679 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15680 Tom de Vries <tom@codesourcery.com>
15682 PR middle-end/81030
15683 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15684 when gimple level profile disagrees with what RTL expander did.
15686 2017-07-20 Richard Biener <rguenther@suse.de>
15688 PR tree-optimization/61171
15689 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15690 (vect_analyze_stmt): Add slp instance parameter.
15691 (vectorizable_reduction): Likewise.
15692 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15693 (vect_is_simple_reduction): Deal with chains not detected
15694 as SLP reduction chain, specifically not properly associated
15695 chains containing a mix of plus/minus.
15696 (get_reduction_op): Remove.
15697 (get_initial_defs_for_reduction): Simplify, pass in whether
15698 this is a reduction chain, pass in the SLP node for the PHIs.
15699 (vect_create_epilog_for_reduction): Get the SLP instance as
15701 (vectorizable_reduction): Get the SLP instance as arg.
15702 During analysis remember the SLP node with the PHIs in the
15703 instance. Simplify getting at the vectorized reduction PHIs.
15704 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15705 through SLP instance.
15706 (vect_slp_analyze_operations): Likewise.
15707 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15708 (vect_transform_stmt): Likewise.
15710 2017-07-20 Tom de Vries <tom@codesourcery.com>
15712 PR tree-optimization/81489
15713 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15714 read of phi arg location to before loop that modifies phi.
15716 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15718 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15721 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15723 PR middle-end/81331
15724 * except.c (execute): Fix ordering issue.
15726 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15728 PR rtl-optimization/81423
15729 * combine.c (make_compound_operation_int): Don't try to optimize
15730 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15732 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15734 PR rtl-optimization/81423
15735 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15736 with a constant that is -1 in the truncated to mode.
15738 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15740 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15741 (determine_unlikely_bbs): ... here.
15742 * predict.h (propagate_unlikely_bbs_forward): Declare.
15743 * cfgexpand.c (pass_expand::execute): Use it.
15744 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15746 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15747 propagate_unlikely_bbs_forward.
15749 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15751 PR middle-end/81331
15752 * except.c (maybe_add_nop_after_section_switch): New function.
15755 2017-07-19 Tom de Vries <tom@codesourcery.com>
15757 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15759 2017-07-19 Tom de Vries <tom@codesourcery.com>
15761 * gimple.h (gimple_phi_arg): Make assert more strict.
15763 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15765 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15766 * config/rs6000/mmintrin.h: New file.
15767 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15769 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15771 PR tree-optimization/81346
15772 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15774 2017-07-19 Tom de Vries <tom@codesourcery.com>
15776 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15778 2017-07-19 Tom de Vries <tom@codesourcery.com>
15780 * config/nvptx/nvptx-modes.def: Add V2DImode.
15781 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15782 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15783 (nvptx_output_mov_insn): Handle lack of mov.b128.
15784 (nvptx_print_operand): Handle 'H' and 'L' codes.
15785 (nvptx_vector_mode_supported): Allow V2DImode.
15786 (nvptx_preferred_simd_mode): New function.
15787 (nvptx_data_alignment): New function.
15788 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15789 nvptx_preferred_simd_mode.
15790 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15792 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15794 2017-07-19 Tom de Vries <tom@codesourcery.com>
15796 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15797 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15798 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15799 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15800 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15801 (mov<VECIM>_insn): New define_insn.
15802 (define_expand "mov<VECIM>): New define_expand.
15804 2017-07-19 Tom de Vries <tom@codesourcery.com>
15806 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15808 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15810 PR tree-optimization/81346
15811 * fold-const.h (fold_div_compare, range_check_type): Declare.
15812 * fold-const.c (range_check_type): New function.
15813 (build_range_check): Use range_check_type.
15814 (fold_div_compare): No longer static, rewritten into
15815 a match.pd helper function.
15816 (fold_comparison): Don't call fold_div_compare here.
15817 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15818 as helper function.
15820 2017-07-19 Nathan Sidwell <nathan@acm.org>
15822 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15823 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15824 * tree.c (find_decls_types_r, verify_type): Use
15825 TYPE_{MIN,MAX}_VALUE_RAW.
15826 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15827 (hash_tree): Likewise.
15828 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15830 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15833 2017-07-18 Tom de Vries <tom@codesourcery.com>
15835 PR middle-end/81464
15836 * omp-expand.c (expand_omp_for_static_chunk): Handle
15837 equal-argument loop exit phi.
15839 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15842 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15843 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15844 operand 2 predicate.
15845 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15846 operand 2 predicate.
15847 (ror,rol -> rorx splitters): Use const_int_operand as
15848 operand 2 predicate.
15850 2017-06-18 Richard Biener <rguenther@suse.de>
15852 PR tree-optimization/81410
15853 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15854 the gap in the ! slp_perm SLP case after each group.
15856 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15858 PR middle-end/81463
15859 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15862 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15864 PR middle-end/81462
15865 * predict.c (set_even_probabilities): Cleanup; do not affect
15866 probabilities that are already known.
15867 (combine_predictions_for_bb): Call even when count is set.
15869 2017-07-18 Nathan Sidwell <nathan@acm.org>
15871 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15874 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15877 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15878 optimization for loop niter analysis.
15880 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15883 * config/avr/avr.c (avr_optimize_casesi): Don't use
15884 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15886 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15888 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15889 body_cost_vec from _vect_peel_extended_info.
15890 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15891 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15894 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15896 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15898 2017-07-18 Richard Biener <rguenther@suse.de>
15900 PR tree-optimization/80620
15901 PR tree-optimization/81403
15902 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15903 info when re-using a VN table entry.
15905 2017-07-18 Richard Biener <rguenther@suse.de>
15907 PR tree-optimization/81418
15908 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15909 vectype_in. Verify that with lane-reducing reduction operations
15910 we have a single def-use cycle.
15912 2017-07-17 Carl Love <cel@us.ibm.com>
15914 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15916 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15917 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15918 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15919 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15920 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15921 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15922 VMULOSW): New enum "unspec" values.
15923 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15924 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15925 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15926 altivec_vmulosw): New patterns.
15927 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15928 VMULOSW): Add definitions.
15929 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15931 * config/alpha/alpha.c: Include predict.h.
15933 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15935 * tree-vrp.c (compare_assert_loc): Fix comparison function
15936 to return predictable results.
15938 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15940 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15942 (subdi3): Likewise.
15943 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15944 * doc/invoke.texi (mexpand-adddi): Update text.
15946 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15948 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15949 that also clobbers the CC register. The old expand code is moved
15951 (*arc_clzsi2): ... here.
15952 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15953 the CC register. The old expand code is moved to ...
15954 (arc_ctzsi2): ... here.
15956 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15958 * config/arc/arc.opt (mindexed-loads): Use initial value
15959 TARGET_INDEXED_LOADS_DEFAULT.
15960 (mauto-modify-reg): Use initial value
15961 TARGET_AUTO_MODIFY_REG_DEFAULT.
15962 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15963 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15964 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15965 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15967 2017-07-17 Martin Liska <mliska@suse.cz>
15970 * opts.c (finish_options): Do not allow -fgnu-tm
15971 w/ -fsanitize={kernel-,}address. Say sorry.
15973 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15976 * tree-loop-distribution.c (classify_partition): Only assert on
15977 numer of iterations.
15978 (merge_dep_scc_partitions): Delete prameter. Update function call.
15979 (distribute_loop): Remove code handling loop with unknown niters.
15980 (pass_loop_distribution::execute): Skip loop with unknown niters.
15982 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15985 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15986 function sort_partitions_by_post_order.
15988 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15990 PR tree-optimization/81374
15991 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15992 the max index of basic blocks, rather than number of basic blocks.
15994 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15996 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15998 (arc_legitimate_pic_operand_p): Likewise.
15999 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16001 (arc_needs_pcl_p): Likewise.
16002 (arc_legitimate_pc_offset_p): Likewise.
16003 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16004 function is also used in constrains.md.
16005 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16006 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16007 PLUS. Only return true/false in known cases, otherwise assert.
16008 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16009 is already called in arc_legitimate_constant_p.
16010 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16012 (LEGITIMATE_PIC_OPERAND_P): Use
16013 arc_raw_symbolic_reference_mentioned_p function.
16014 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16019 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16020 Andrew Burgess <andrew.burgess@embecosm.com>
16022 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16023 (arc_return_address_register): New function.
16024 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16025 (arc_handle_fndecl_attribute): Add naked attribute.
16026 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16027 (TARGET_WARN_FUNC_RETURN): Likewise.
16028 (arc_allocate_stack_slots_for_args): New function.
16029 (arc_warn_func_return): Likewise.
16030 (machine_function): Change type fn_type.
16031 (arc_compute_function_type): Consider new naked function type,
16032 change function return type.
16033 (arc_must_save_register): Adapt to handle new
16034 arc_compute_function_type's return type.
16035 (arc_expand_prologue): Likewise.
16036 (arc_expand_epilogue): Likewise.
16037 (arc_return_address_regs): Delete.
16038 (arc_return_address_register): New function.
16039 (arc_epilogue_uses): Use above function.
16040 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16041 (arc_function_type): Change encoding, add naked type.
16042 (ARC_INTERRUPT_P): Change to handle the new encoding.
16043 (ARC_FAST_INTERRUPT_P): Likewise.
16044 (ARC_NORMAL_P): Define.
16045 (ARC_NAKED_P): Likewise.
16046 (arc_compute_function_type): Delete prototype.
16047 * config/arc/arc.md (in_ret_delay_slot): Use
16048 arc_return_address_register function.
16049 (simple_return): Likewise.
16050 (p_return_i): Likewise.
16052 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16054 PR tree-optimization/81428
16055 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16056 can't be built for those types.
16058 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16060 Remove stuff dead since r239246.
16062 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16063 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16064 (avr_inform_devices): Remove dead stuff.
16066 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16068 * config/arm/arm_neon.h: Fix softp typo.
16070 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16072 PR tree-optimization/81365
16073 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16074 aggregate moves onto bb predecessor edges, make sure there are no
16075 loads that could alias the lhs in between the start of bb and the
16078 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16081 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16082 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16083 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16085 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16087 PR tree-optimization/81396
16088 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16089 (init_symbolic_number): Initialize it to 1.
16090 (perform_symbolic_merge): Add n_ops from both operands into the new
16092 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16093 without base_addr as useless if they need more than one operation.
16094 (bswap_replace): Handle !bswap case for NULL base_addr.
16096 2017-07-17 Tom de Vries <tom@codesourcery.com>
16099 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16102 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16104 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16105 conditional builtin define __FIX_LEON3FT_B2BST.
16107 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16109 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16110 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16113 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16115 PR rtl-optimization/81424
16116 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16117 to remove potential trapping from operands if -fnon-call-exceptions.
16119 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16121 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16122 profile_proability for scalling.
16123 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16125 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16127 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16129 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16131 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16132 fixpoint arithmetics.
16134 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16136 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16137 fixpoint arithmetics.
16139 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16141 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16142 fixpoint arithmetics.
16144 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16146 * profile-count.h (profile_probability::from_reg_br_prob_note,
16147 profile_probability::to_reg_br_prob_note): New functions.
16148 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16149 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16150 * predict.c (probability_reliable_p): Update.
16151 (edge_probability_reliable_p): Update.
16152 (br_prob_note_reliable_p): Update.
16153 (invert_br_probabilities): Update.
16154 (add_reg_br_prob_note): New function.
16155 (combine_predictions_for_insn): Update.
16156 * asan.c (asan_clear_shadow): Update.
16157 * cfgbuild.c (compute_outgoing_frequencies): Update.
16158 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16159 (update_br_prob_note): Update.
16160 (rtl_verify_edges): Update.
16161 (purge_dead_edges): Update.
16162 (fixup_reorder_chain): Update.
16163 * emit-rtl.c (try_split): Update.
16164 * ifcvt.c (cond_exec_process_insns): Update.
16165 (cond_exec_process_if_block): Update.
16166 (dead_or_predicable): Update.
16167 * internal-fn.c (expand_addsub_overflow): Update.
16168 (expand_neg_overflow): Update.
16169 (expand_mul_overflow): Update.
16170 * loop-doloop.c (doloop_modify): Update.
16171 * loop-unroll.c (compare_and_jump_seq): Update.
16172 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16173 * predict.h: Update.
16174 * reorg.c (mostly_true_jump): Update.
16176 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16177 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16178 * config/arc/arc.c: (emit_unlikely_jump): Update.
16179 * config/arm/arm.c: (emit_unlikely_jump): Update.
16180 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16181 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16182 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16183 (ix86_print_operand): Update.
16184 (ix86_split_fp_branch): Update.
16185 (predict_jump): Update.
16186 * config/ia64/ia64.c (ia64_print_operand): Update.
16187 * config/mmix/mmix.c (mmix_print_operand): Update.
16188 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16189 (rs6000_expand_split_stack_prologue): Update.
16190 * config/rs6000/rs6000.c: Update.
16191 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16192 (s390_expand_vec_movstr): Update.
16193 (s390_expand_cs_tdsi): Update.
16194 (s390_expand_split_stack_prologue): Update.
16195 * config/sh/sh.c (sh_print_operand): Update.
16196 (expand_cbranchsi4): Update.
16197 (expand_cbranchdi4): Update.
16198 * config/sparc/sparc.c (output_v9branch): Update.
16199 * config/spu/spu.c (get_branch_target): Update.
16200 (ea_load_store_inline): Update.
16201 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16202 * config/tilepro/tilepro.c: Update.
16204 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16206 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16207 (gimplify_save_expr): Likewise.
16209 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16211 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16213 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16215 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16217 * timevar.def (TV_IPA_FNSUMMARY): Define.
16219 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16221 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16222 to back store errata sensitive sequence from being generated.
16223 (sqrtdf2_fix): Likewise.
16225 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16227 * tree-ssa-threadupdate.c (compute_path_counts,
16228 update_joiner_offpath_counts): Use profile_probability.
16230 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16233 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16235 * config/arm/arm-c.c (arm_cpu_builtins): Define
16236 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16238 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16240 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16241 array entries to represent __ieee128 versions of the
16242 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16243 scalar_extract_sig, and scalar_insert_exp built-in functions.
16244 (altivec_resolve_overloaded_builtin): Add special case handling
16245 for the __builtin_scalar_insert_exp function, as represented by
16246 the P9V_BUILTIN_VEC_VSIEDP constant.
16247 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16248 exponent support for __ieee128 argument.
16249 (VSESQP): Add scalar extract signature support for __ieee128
16251 (VSTDCNQP): Add scalar test negative support for __ieee128
16253 (VSIEQP): Add scalar insert exponent support for __int128 argument
16254 with __ieee128 result.
16255 (VSIEQPF): Add scalar insert exponent support for __ieee128
16256 argument with __ieee128 result.
16257 (VSTDCQP): Add scalar test data class support for __ieee128
16259 (VSTDCNQP): Add overload support for scalar test negative with
16260 __ieee128 argument.
16261 (VSTDCQP): Add overload support for scalar test data class
16262 __ieee128 argument.
16263 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16264 UNSPEC_VSX_SXSIGDP.
16265 (UNSPEC_VSX_SIEXPQP): New constant.
16266 (xsxexpqp): New insn for VSX scalar extract exponent quad
16268 (xsxsigqp): New insn for VSX scalar extract significand quad
16270 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16271 precision with floating point argument.
16272 (xststdcqp): New expand for VSX scalar test data class quad
16274 (xststdcnegqp): New expand for VSX scalar test negative quad
16276 (xststdcqp): New insn to match expansions for VSX scalar test data
16277 class quad precision and VSX scalar test negative quad precision.
16278 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16279 special case operand checking to enforce that second operand of
16280 VSX scalar test data class with quad precision argument is a 7-bit
16282 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16283 prototypes and descriptions of __ieee128 versions of
16284 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16285 scalar_test_data_class, and scalar_test_neg built-in functions.
16287 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16289 PR tree-optimization/81162
16290 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16291 replace a negate with an add.
16293 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16295 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16297 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16299 * config/arm/arm-c.c (arm_cpu_builtins): Define
16300 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16302 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16304 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16305 (armv8-r): Set ARM Cortex-R52 as default CPU.
16306 * config/arm/arm-tables.opt: Regenerate.
16307 * config/arm/arm-tune.md: Regenerate.
16308 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16310 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16311 extension for -mcpu=cortex-r52.
16313 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16315 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16316 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16317 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16318 (fp-armv8): Define it as FP_ARMv8 only.
16319 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16320 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16322 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16323 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16324 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16325 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16326 than TARGET_FPU_ARMV8.
16327 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16328 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16329 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16331 * config/arm/neon.md (neon_vrint): Likewise.
16332 (neon_vcvt): Likewise.
16333 (neon_<fmaxmin_op><mode>): Likewise.
16334 (<fmaxmin><mode>3): Likewise.
16335 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16336 * config/arm/predicates.md (arm_cond_move_operator): Check against
16337 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16339 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16341 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16342 to top of function.
16344 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16346 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16347 loop in comment with memset.
16349 2017-07-14 Martin Liska <mliska@suse.cz>
16351 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16352 * dwarf2out.c (is_java): Remove the function.
16353 (output_pubname): Remove usage of the function.
16354 (lower_bound_default): Remove usage of DW_LANG_Java.
16355 (gen_compile_unit_die): Likewise.
16356 * gcc.c: Remove compiler defaults for .java and .zip files.
16357 * gimple-expr.c (remove_suffix): Change as there's no longer
16358 extension than 4-letter one.
16359 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16360 (gimplify_save_expr): Likewise.
16361 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16362 as it's possible even for other languages than Java.
16363 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16364 * lto-opts.c (lto_write_options): Remove reference to Java.
16365 * opts.c (strip_off_ending): Update file extension handling.
16366 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16367 * tree-eh.c (lower_resx): Likewise.
16368 * tree.c (free_lang_data_in_type): Remove dead code.
16369 (find_decls_types_r): Likewise.
16370 (build_common_builtin_nodes): Remove Java from a comment.
16371 (verify_type): Remove dead code.
16372 * varasm.c (assemble_external): Remove Java from a comment.
16374 2017-07-14 Martin Liska <mliska@suse.cz>
16376 * opts.c (finish_options): Add quotes.
16377 (common_handle_option): Likewise.
16379 2017-07-14 Martin Liska <mliska@suse.cz>
16381 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16382 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16383 Remove N_SO_PASCAL.
16384 * dwarf2out.c (lower_bound_default): Do not handle
16386 (gen_compile_unit_die): Likewise.
16387 * gcc.c: Remove default extension binding for GNU Pascal.
16388 * stmt.c: Remove Pascal language from a comment.
16389 * xcoffout.c: Likewise.
16391 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16394 * diagnostic-show-locus.c (fixit_cmp): New function.
16395 (layout::layout): Sort m_fixit_hints.
16396 (column_range::column_range): Assert that the values are valid.
16397 (struct char_span): New struct.
16398 (correction::overwrite): New method.
16399 (struct source_line): New struct.
16400 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16401 calls in terms of classes source_line and char_span, and
16402 correction::overwrite.
16403 (selftest::test_overlapped_fixit_printing_2): New function.
16404 (selftest::diagnostic_show_locus_c_tests): Call it.
16406 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16408 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16409 early if there is no lhs.
16411 2017-07-13 Martin Liska <mliska@suse.cz>
16413 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16414 (gen_reference_type_die): Likewise.
16415 * stor-layout.c: Remove Pascal-related comment.
16417 2017-07-13 Martin Liska <mliska@suse.cz>
16419 * opts.c (finish_options): Add quotes to error messages.
16420 (parse_sanitizer_options): Likewise.
16422 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16424 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16426 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16428 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16430 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16432 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16434 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16436 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16439 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16440 provides the hardware capability bits, define the macro
16441 __BUILTIN_CPU_SUPPORTS__.
16442 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16443 if GLIBC does not provide the hardware capability bits. Add a
16444 gcc_unreachable call if the built-in cpu function is neither
16445 __builtin_cpu_is nor __builtin_cpu_supports.
16446 (rs6000_get_function_versions_dispatcher): Change the warning
16447 that an old GLIBC is used which does not export the capability
16448 bits to be an error.
16449 * doc/extend.texi (target_clones attribute): Document the
16450 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16451 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16452 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16453 the macros defined by GCC if the newer GLIBC is available.
16455 2017-07-12 Jeff Law <law@redhat.com>
16457 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16458 remaining includes slightly.
16459 * config/riscv/riscv-builtins.c: Include profile-count.h.
16461 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16464 * config/avr/avr.c (avr_set_current_function): In diagnostic
16465 messages: Quote keywords and (parts of) identifiers.
16466 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16469 2017-07-12 Carl Love <cel@us.ibm.com>
16471 * config/rs6000/rs6000-c.c: Add support for built-in functions
16472 vector bool char vec_revb (vector bool char);
16473 vector bool short vec_revb (vector short char);
16474 vector bool int vec_revb (vector bool int);
16475 vector bool long long vec_revb (vector bool long long);
16476 * doc/extend.texi: Update the built-in documentation file for the
16477 new built-in functions.
16479 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16481 * config/s390/s390.md: Remove movcc splitter.
16483 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16485 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16486 load/store on condition.
16488 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16491 * config/avr/avr.c (avr_encode_section_info)
16492 [progmem && !TREE_READONLY]: Error if progmem object needs
16495 2017-07-11 Michael Collison <michael.collison@arm.com>
16497 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16500 2017-07-11 Carl Love <cel@us.ibm.com>
16502 * config/rs6000/rs6000-c.c: Add support for builtins
16503 vector unsigned int vec_parity_lsbb (vector signed int);
16504 vector unsigned int vec_parity_lsbb (vector unsigned int);
16505 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16506 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16507 vector unsigned long long vec_parity_lsbb (vector signed long long);
16508 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16509 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16510 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16511 * doc/extend.texi: Update the built-in documentation file for the
16512 new built-in functions.
16514 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16516 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16517 (layout::m_primary_loc): New field.
16518 (layout::layout): Initialize new field. Move location filtering
16519 logic from here to...
16520 (layout::maybe_add_location_range): ...this new method. Add
16521 support for filtering to just the lines already specified by other
16523 (layout::will_show_line_p): New method.
16524 (gcc_rich_location::add_location_if_nearby): New method.
16525 (selftest::test_add_location_if_nearby): New test function.
16526 (selftest::diagnostic_show_locus_c_tests): Call it.
16527 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16530 2017-07-11 Tom de Vries <tom@codesourcery.com>
16532 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16533 (bb_first_real_insn): New function.
16534 (nvptx_single): Add extra initialization of broadcasted condition
16537 2017-07-11 Nathan Sidwell <nathan@acm.org>
16539 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16541 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16543 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16544 Binutils doc as TEXI will mess them up.
16545 * doc/invoke.texi (AVR Options): Same here.
16547 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16549 * config/sparc/sparc.opt (mfix-ut700): New option.
16550 (mfix-gr712rc): Likewise.
16551 (sparc_fix_b2bst): New variable.
16552 * doc/invoke.texi (SPARC options): Document them.
16553 (ARM options): Fix warnings.
16554 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16555 instructions to prevent sequences that can trigger the store-store
16556 errata for certain LEON3FT processors.
16557 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16558 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16559 * config/sparc/sparc.md (fix_b2bst): New attribute.
16560 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16562 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16565 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16567 (*rsqrtsf2_sse): Ditto.
16569 (div<mode>3): Macroize insn from divdf3 and divsf3
16570 using MODEF mode iterator.
16572 2017-07-10 Martin Sebor <msebor@redhat.com>
16574 PR tree-optimization/80397
16575 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16576 instead of testing for equality to INTEGER_TYPE.
16578 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16580 * config.gcc: Remove uclibc from arc target spec.
16582 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16584 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16586 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16589 * lto-wrapper.c (remove_option): New function.
16590 (merge_and_complain): Merge PIC/PIE options more realistically.
16592 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16594 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16598 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16599 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16600 * config.in: Regenerate.
16601 * configure: Regenerate.
16602 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16603 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16604 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16605 (TARGET_GASISR_PROLOGUES): ...target mask.
16606 * common/config/avr/avr-common.c
16607 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16608 Set -mgas-isr-prologues.
16609 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16610 INSERT_PASS_BEFORE for it.
16611 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16612 * config/avr/avr.c (avr_option_override)
16613 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16614 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16615 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16616 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16617 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16619 (make_avr_pass_pre_proep): New function.
16620 (emit_push_sfr) <treg>: Add argument to function and use it
16621 instead of TMP_REG.
16622 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16623 and set machine->gasisr.yes.
16624 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16625 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16626 __gcc_isr.n_pushed to .L__stack_usage.
16627 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16628 (avr_asm_final_postscan_insn): ...this new static function.
16629 * config/avr/avr.h (machine_function)
16630 <is_no_gccisr, use_L__stack_usage>: New fields.
16631 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16632 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16633 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16634 (gasisr, *gasisr): New expander and insn.
16635 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16636 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16637 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16639 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16641 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16644 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16646 Move jump-tables out of .text again.
16649 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16650 (ASM_OUTPUT_ADDR_VEC): New function.
16651 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16652 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16653 INSN_ADDRESSes as asm comment.
16654 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16655 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16656 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16657 * config/avr/avr.md (*tablejump): Adjust comment.
16658 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16659 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16661 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16662 (avr_output_addr_vec): New proto.
16663 (avr_log_t) <insn_addresses>: New field.
16665 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16668 * config/i386/i386.c (ix86_function_arg_advance): Set
16669 outgoing_args_on_stack to true if there are outgoing arguments
16671 (ix86_function_arg): Likewise.
16672 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16673 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16674 * config/i386/i386.h (machine_function): Add
16675 outgoing_args_on_stack.
16677 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16679 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16680 supporting pthreds.
16681 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16683 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16685 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16686 (REAL_H): Remove $(MACHMODE_H).
16687 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16689 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16690 $(MACHMODE_H) and double-int.h.
16691 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16693 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16696 2017-07-07 Andrew Pinski <apinski@cavium.com>
16698 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16699 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16701 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16703 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16704 Add warning if GCC was not configured to link against a GLIBC that
16705 exports the hardware capability bits.
16706 (make_resolver_func): Make resolver function private and not a
16707 COMDAT function. Create the name with clone_function_name instead
16708 of make_unique_name.
16711 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16712 correct operand in doing the split.
16714 2017-07-07 Carl Love <cel@us.ibm.com>
16716 * config/rs6000/rs6000-c: Add support for built-in function
16717 vector unsigned short vec_pack_to_short_fp32 (vector float,
16719 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16720 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16721 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16722 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16723 (convert_4f32_8i16): Add define_expand.
16724 * doc/extend.texi: Update the built-in documentation file for the
16725 new built-in function.
16727 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16729 * config/sparc/m8.md: New file.
16730 * config/sparc/sparc.md: Include m8.md.
16732 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16734 * config/sparc/sparc.opt: New option -mvis4b.
16735 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16736 (sparc_option_override): Handle VIS4B.
16737 (enum sparc_builtins): Define
16738 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16739 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16740 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16741 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16742 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16743 (check_constant_argument): New function.
16744 (sparc_vis_init_builtins): Define builtins
16745 __builtin_vis_dictunpack{8,16,32},
16746 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16747 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16748 __builtin_vis_fpcmpde{8,16,32}shl and
16749 __builtin_vis_fpcmpur{8,16,32}shl.
16750 (sparc_expand_builtin): Check that the constant operands to
16751 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16752 constant and in range.
16753 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16755 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16756 (SPARC_IMM5_P): Likewise.
16757 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16758 (enabled): Handle vis4b.
16759 (UNSPEC_DICTUNPACK): New unspec.
16760 (UNSPEC_FPCMPSHL): Likewise.
16761 (UNSPEC_FPUCMPSHL): Likewise.
16762 (UNSPEC_FPCMPDESHL): Likewise.
16763 (UNSPEC_FPCMPURSHL): Likewise.
16764 (cpu_feature): New CPU feature `vis4b'.
16765 (dictunpack{8,16,32}): New insns.
16766 (FPCSMODE): New mode iterator.
16767 (fpcscond): New code iterator.
16768 (fpcsucond): Likewise.
16769 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16770 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16771 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16772 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16773 * config/sparc/constraints.md: Define constraints `q' for unsigned
16774 2-bit integer constants and `t' for unsigned 5-bit integer
16776 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16778 (imm5_operand_dictunpack16): Likewise.
16779 (imm5_operand_dictunpack32): Likewise.
16780 (imm2_operand): Likewise.
16781 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16782 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16783 ditunpack* and fpcmp*shl builtins.
16785 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16787 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16788 * config.in: Add HAVE_AS_SPARC6 define.
16789 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16791 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16793 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16794 (CPP_CPU_SPEC): Handle m8.
16795 (ASM_CPU_SPEC): Likewise.
16796 * config/sparc/sparc-opts.h (enum processor_type): Add
16798 * config/sparc/sparc.c (m8_costs): New struct.
16799 (sparc_option_override): Handle TARGET_CPU_m8.
16800 (sparc32_initialize_trampoline): Likewise.
16801 (sparc64_initialize_trampoline): Likewise.
16802 (sparc_issue_rate): Likewise.
16803 (sparc_register_move_cost): Likewise.
16804 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16805 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16806 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16807 (CPP_CPU_SPEC): Handle M8.
16808 (ASM_CPU_SPEC): Likewise.
16809 (AS_M8_FLAG): Define.
16810 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16811 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16812 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16814 * configure: Regenerate.
16815 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16818 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16820 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16822 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16823 ("*movdi_insn_sp32"): Do not set v3pipe.
16824 ("*movsi_insn"): Likewise.
16825 ("*movdi_insn_sp64"): Likewise.
16826 ("*movsf_insn"): Likewise.
16827 ("*movdf_insn_sp32"): Likewise.
16828 ("*movdf_insn_sp64"): Likewise.
16829 ("*zero_extendsidi2_insn_sp64"): Likewise.
16830 ("*sign_extendsidi2_insn"): Likewise.
16831 ("*mov<VM32:mode>_insn"): Likewise.
16832 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16833 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16834 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16835 ("<vlop:code><VL:mode>3"): Likewise.
16836 ("*not_<vlop:code><VL:mode>3"): Likewise.
16837 ("*nand<VL:mode>_vis"): Likewise.
16838 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16839 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16840 ("one_cmpl<VL:mode>2"): Likewise.
16841 ("faligndata<VM64:mode>_vis"): Likewise.
16842 ("alignaddrsi_vis"): Likewise.
16843 ("alignaddrdi_vis"): Likweise.
16844 ("alignaddrlsi_vis"): Likewise.
16845 ("alignaddrldi_vis"): Likewise.
16846 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16847 ("bmaskdi_vis"): Likewise.
16848 ("bmasksi_vis"): Likewise.
16849 ("bshuffle<VM64:mode>_vis"): Likewise.
16850 ("cmask8<P:mode>_vis"): Likewise.
16851 ("cmask16<P:mode>_vis"): Likewise.
16852 ("cmask32<P:mode>_vis"): Likewise.
16853 ("pdistn<P:mode>_vis"): Likewise.
16854 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16856 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16858 * config/sparc/sparc.md ("subtype"): New insn attribute.
16859 ("*wrgsr_sp64"): Set insn subtype.
16860 ("*rdgsr_sp64"): Likewise.
16861 ("alignaddrsi_vis"): Likewise.
16862 ("alignaddrdi_vis"): Likewise.
16863 ("alignaddrlsi_vis"): Likewise.
16864 ("alignaddrldi_vis"): Likewise.
16865 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16866 ("fexpand_vis"): Likewise.
16867 ("fpmerge_vis"): Likewise.
16868 ("faligndata<VM64:mode>_vis"): Likewise.
16869 ("bshuffle<VM64:mode>_vis"): Likewise.
16870 ("cmask8<P:mode>_vis"): Likewise.
16871 ("cmask16<P:mode>_vis"): Likewise.
16872 ("cmask32<P:mode>_vis"): Likewise.
16873 ("fchksm16_vis"): Likewise.
16874 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16875 ("fmean16_vis"): Likewise.
16876 ("fp<plusminus_insn>64_vis"): Likewise.
16877 ("<plusminus_insn>v8qi3"): Likewise.
16878 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16879 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16880 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16881 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16882 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16883 ("*movqi_insn"): Likewise.
16884 ("*movhi_insn"): Likewise.
16885 ("*movsi_insn"): Likewise.
16886 ("movsi_pic_gotdata_op"): Likewise.
16887 ("*movdi_insn_sp32"): Likewise.
16888 ("*movdi_insn_sp64"): Likewise.
16889 ("movdi_pic_gotdata_op"): Likewise.
16890 ("*movsf_insn"): Likewise.
16891 ("*movdf_insn_sp32"): Likewise.
16892 ("*movdf_insn_sp64"): Likewise.
16893 ("*zero_extendhisi2_insn"): Likewise.
16894 ("*zero_extendqihi2_insn"): Likewise.
16895 ("*zero_extendqisi2_insn"): Likewise.
16896 ("*zero_extendqidi2_insn"): Likewise.
16897 ("*zero_extendhidi2_insn"): Likewise.
16898 ("*zero_extendsidi2_insn_sp64"): Likewise.
16899 ("ldfsr"): Likewise.
16900 ("prefetch_64"): Likewise.
16901 ("prefetch_32"): Likewise.
16902 ("tie_ld32"): Likewise.
16903 ("tie_ld64"): Likewise.
16904 ("*tldo_ldub_sp32"): Likewise.
16905 ("*tldo_ldub1_sp32"): Likewise.
16906 ("*tldo_ldub2_sp32"): Likewise.
16907 ("*tldo_ldub_sp64"): Likewise.
16908 ("*tldo_ldub1_sp64"): Likewise.
16909 ("*tldo_ldub2_sp64"): Likewise.
16910 ("*tldo_ldub3_sp64"): Likewise.
16911 ("*tldo_lduh_sp32"): Likewise.
16912 ("*tldo_lduh1_sp32"): Likewise.
16913 ("*tldo_lduh_sp64"): Likewise.
16914 ("*tldo_lduh1_sp64"): Likewise.
16915 ("*tldo_lduh2_sp64"): Likewise.
16916 ("*tldo_lduw_sp32"): Likewise.
16917 ("*tldo_lduw_sp64"): Likewise.
16918 ("*tldo_lduw1_sp64"): Likewise.
16919 ("*tldo_ldx_sp64"): Likewise.
16920 ("*mov<VM32:mode>_insn"): Likewise.
16921 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16922 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16924 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16926 * config/sparc/sparc.md ("type"): New insn type viscmp.
16927 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16929 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16930 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16931 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16932 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16934 ("n7_vis_logical_11cycle"): Likewise.
16935 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16936 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16937 * config/sparc/niagara.md ("niag_vis"): Likewise.
16938 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16939 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16941 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16943 * config/sparc/sparc.md: New instruction type `bmask'.
16944 (bmaskdi_vis): Use the `bmask' type.
16945 (bmasksi_vis): Likewise.
16946 * config/sparc/ultra3.md (us3_array): Likewise.
16947 * config/sparc/niagara7.md (n7_array): Likewise.
16948 * config/sparc/niagara4.md (n4_array): Likewise.
16949 * config/sparc/niagara2.md (niag2_vis): Likewise.
16950 (niag3_vis): Likewise.
16951 * config/sparc/niagara.md (niag_vis): Likewise.
16953 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16955 * ipa-comdats.c: Remove optimize check from gate.
16956 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16957 for functions not optimized.
16958 (ipa_fn_summary_read): Skip optimize check.
16959 (ipa_fn_summary_write): Likewise.
16960 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16962 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16964 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16965 (check_callers): Give up on caller which is not optimized.
16966 (inline_small_functions): Likewise.
16967 (ipa_inline): Do not give up when not optimizing.
16968 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16969 away unoptimizes cdtors.
16970 (whole_program_function_and_variable_visibility): Do
16971 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16972 * ipa.c (process_references): Do not check optimize.
16973 (symbol_table::remove_unreachable_nodes): Update optimize check.
16974 (set_writeonly_bit): Update optimize check.
16975 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16976 (pass_ipa_single_use::gate): Remove.
16978 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16980 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16981 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16982 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16983 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16984 permute_load, permute_store, adjust_extract, adjust_splat,
16985 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16986 replace_swap_with_copy, dump_swap_insn_table,
16987 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16988 recombine_lvx_pattern, recombine_stvx_pattern,
16989 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16990 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16991 to file rs6000-p8swap.c.
16992 * config/rs6000/rs6000-p8swap.c: New file.
16993 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16994 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16995 and rs6000*-*-* targets.
16997 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16999 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17001 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17003 * lto-wrapper.c (merge_and_complain): Do not merge
17004 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17005 fsigned_zeros, ftrapping_math, fwrapv.
17006 (append_compiler_options): Do not track these options.
17007 (append_linker_options): Likewie
17009 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17011 * cgraphunit.c (cgraph_node::finalize_function): When
17012 !flag_toplevel_reorde set no_reorder flag.
17013 (varpool_node::finalize_decl): Likewise.
17014 (symbol_table::compile): Drop no toplevel reorder path.
17016 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17018 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17019 edges; zero probability is not better than uninitialized.
17021 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17023 * asan.h (asan_sanitize_allocas_p): Declare.
17024 * asan.c (asan_sanitize_allocas_p): New function.
17025 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17026 (handle_builtin_alloca): Likewise.
17027 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17028 if !asan_sanitize_allocas_p.
17029 * params.def (asan-instrument-allocas): Add new option.
17030 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17031 * opts.c (common_handle_option): Disable allocas sanitization for
17034 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17036 * asan.c: Include gimple-fold.h.
17037 (get_last_alloca_addr): New function.
17038 (handle_builtin_stackrestore): Likewise.
17039 (handle_builtin_alloca): Likewise.
17040 (asan_emit_allocas_unpoison): Likewise.
17041 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17042 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17043 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17044 (instrument_builtin_call): Pass gimple iterator to
17045 get_mem_refs_of_builtin_call.
17046 (last_alloca_addr): New global.
17047 * asan.h (asan_emit_allocas_unpoison): Declare.
17048 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17049 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17050 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17051 if function calls alloca.
17052 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17053 * gimple-fold.h (replace_call_with_value): Declare.
17054 * internal-fn.c: Include asan.h.
17055 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17056 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17058 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17060 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17061 (C_SELFTEST_FLAGS): New.
17062 (CPP_SELFTEST_FLAGS): New.
17063 (SELFTEST_DEPS): New, from deps of s-selftest.
17064 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17065 (CPP_SELFTEST_DEPS): New.
17066 (selftest): Add dependency on s-selftest-c++.
17067 (s-selftest): Rename to...
17068 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17069 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17070 than SELFTEST_FLAGS.
17071 (selftest-gdb): Rename to...
17072 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17074 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17075 (selftest-valgrind): Rename to...
17076 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17078 (selftest-valgrind): Reintroduce as an alias for
17079 selftest-c-valgrind.
17080 (s-selftest-c++): New.
17081 (selftest-c++-gdb): New.
17082 (selftest-c++-valgrind): New.
17084 2017-07-06 Olivier Hainque <hainque@adacore.com>
17086 * gcc.c (process_command): When deciding if undefined variables
17087 should be ignored when processing specs, accept "gcc -v" as well.
17089 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17091 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17092 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17094 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17096 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17097 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17098 * config/arm/arm-tables.opt: Regenerate.
17099 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17101 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17103 2017-07-06 Carl Love <cel@us.ibm.com>
17105 * ChangeLog: Clean up from mid air collision
17107 2017-07-06 Carl Love <cel@us.ibm.com>
17109 * config/rs6000/rs6000-c.c: Add support for built-in functions
17110 vector signed int vec_subc (vector signed int, vector signed int);
17111 vector signed __int128 vec_subc (vector signed __int128,
17112 vector signed __int128);
17113 vector unsigned __int128 vec_subc (vector unsigned __int128,
17114 vector unsigned __int128);
17115 vector signed int vec_sube (vector signed int, vector signed int,
17116 vector signed int);
17117 vector unsigned int vec_sube (vector unsigned int,
17118 vector unsigned int,
17119 vector unsigned int);
17120 vector signed __int128 vec_sube (vector signed __int128,
17121 vector signed __int128,
17122 vector signed__int128);
17123 vector unsigned __int128 vec_sube (vector unsigned __int128,
17124 vector unsigned __int128,
17125 vector unsigned __int128);
17126 vector signed int vec_subec (vector signed int, vector signed int,
17127 vector signed int);
17128 vector unsigned int vec_subec (vector unsigned int,
17129 vector unsigned int,
17130 vector unsigned int);
17131 vector signed __int128 vec_subec (vector signed __int128,
17132 vector signed __int128,
17133 vector signed__int128);
17134 vector unsigned __int128 vec_subec (vector unsigned __int128,
17135 vector unsigned __int128,
17136 vector unsigned __int128);
17137 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17138 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17139 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17140 BU_ALTIVEC_OVERLOAD_X definitions.
17141 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17142 * doc/extend.texi: Update the built-in documentation file for the new
17143 built-in functions.
17145 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17148 * diagnostic-show-locus.c (layout::layout): Use start and finish
17149 spelling location for the start and finish of each range.
17150 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17151 Add unused aspect param.
17152 * input.c (expand_location_1): Add "aspect" param, and use it
17153 to access the correct part of the location.
17154 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17156 (expand_location_to_spelling_point): Likewise.
17157 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17158 param, and pass it to expand_location_1.
17160 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17162 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17163 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17164 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17165 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17166 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17167 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17168 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17169 _mm_maskz_getmant_ss): New intrinsics.
17170 (__builtin_ia32_getexpss128_mask): Changed to ...
17171 __builtin_ia32_getexpss128_round ... this.
17172 (__builtin_ia32_getexpsd128_mask): Changed to ...
17173 __builtin_ia32_getexpsd128_round ... this.
17174 * config/i386/i386-builtin-types.def
17175 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17176 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17177 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17178 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17179 __builtin_ia32_getmantss_mask_round): New builtins.
17180 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17181 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17182 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17183 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17184 * config/i386/sse.md
17185 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17186 avx512f_sgetexp<mode><mask_scalar_name>
17187 <round_saeonly_scalar_name> ... this.
17188 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17189 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17190 vgetexp<ssescalarmodesuffix>
17191 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17192 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17193 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17194 avx512f_vgetmant<mode><mask_scalar_name>
17195 <round_saeonly_scalar_name> ... this.
17196 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17197 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17198 vgetmant<ssescalarmodesuffix>
17199 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17200 %0<mask_scalar_operand4>, %1, %2
17201 <round_saeonly_scalar_mask_op4>, %3} ... this.
17202 * config/i386/subst.md (mask_scalar_operand4,
17203 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17204 round_saeonly_scalar_nimm_predicate): New subst attributes.
17206 2017-07-06 Julia Koval <julia.koval@intel.com>
17208 * config/i386/i386.c (ix86_erase_embedded_rounding):
17209 Remove code for old rounding pattern.
17211 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17213 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17215 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17217 * doc/sourcebuild.texi (Test Directives, Variants of
17218 dg-require-support): Add documentation for dg-require-stack-check.
17220 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17222 * config/i386/subst.md (mask_scalar, round_scalar,
17223 round_saeonly_scalar): New meta-templates.
17224 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17225 round_scalar_mask_operand3, round_scalar_mask_op3,
17226 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17227 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17228 round_saeonly_scalar_constraint,
17229 round_saeonly_scalar_prefix): New subst attribute.
17230 * config/i386/sse.md
17231 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17232 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17233 <round_scalar_name> ... this.
17234 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17235 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17236 <round_scalar_name> ... this.
17237 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17238 <sse>_vm<code><mode>3<mask_scalar_name>
17239 <round_saeonly_scalar_name> ... this.
17240 (v<plusminus_mnemonic><ssescalarmodesuffix>
17241 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17242 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17243 v<plusminus_mnemonic><ssescalarmodesuffix>
17244 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17245 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17246 (v<multdiv_mnemonic><ssescalarmodesuffix>
17247 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17248 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17249 v<multdiv_mnemonic><ssescalarmodesuffix>
17250 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17251 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17252 (v<maxmin_float><ssescalarmodesuffix>
17253 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17254 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17255 v<maxmin_float><ssescalarmodesuffix>
17256 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17257 %0<mask_scalar_operand3>, %1, %<iptr>2
17258 <round_saeonly_scalar_mask_op3>} ... this.
17260 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17262 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17263 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17265 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17266 Alan Hayward <alan.hayward@arm.com>
17267 David Sherwood <david.sherwood@arm.com>
17269 * combine.c (simplify_if_then_else): Remove "enum" before
17271 * compare-elim.c (can_eliminate_compare): Likewise.
17272 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17274 (aarch64_lookup_simd_builtin_type): Likewise.
17275 (aarch64_simd_builtin_type): Likewise.
17276 (aarch64_init_simd_builtin_types): Likewise.
17277 (aarch64_simd_expand_args): Likewise.
17278 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17280 (aarch64_reverse_mask): Likewise.
17281 (aarch64_simd_emit_reg_reg_move): Likewise.
17282 (aarch64_gen_adjusted_ldpstp): Likewise.
17283 (aarch64_ccmp_mode_to_code): Likewise.
17284 (aarch64_operands_ok_for_ldpstp): Likewise.
17285 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17286 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17288 (aarch64_min_divisions_for_recip_mul): Likewise.
17289 (aarch64_reassociation_width): Likewise.
17290 (aarch64_get_condition_code_1): Likewise.
17291 (aarch64_simd_emit_reg_reg_move): Likewise.
17292 (aarch64_simd_attr_length_rglist): Likewise.
17293 (aarch64_reverse_mask): Likewise.
17294 (aarch64_operands_ok_for_ldpstp): Likewise.
17295 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17296 (aarch64_gen_adjusted_ldpstp): Likewise.
17297 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17299 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17300 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17301 (arm_lookup_simd_builtin_type): Likewise.
17302 (arm_simd_builtin_type): Likewise.
17303 (arm_init_simd_builtin_types): Likewise.
17304 (arm_expand_builtin_args): Likewise.
17305 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17306 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17307 (ft32_setup_incoming_varargs): Likewise.
17308 (ft32_function_arg): Likewise.
17309 (ft32_function_arg_advance): Likewise.
17310 (ft32_pass_by_reference): Likewise.
17311 (ft32_arg_partial_bytes): Likewise.
17312 (ft32_valid_pointer_mode): Likewise.
17313 (ft32_addr_space_pointer_mode): Likewise.
17314 (ft32_addr_space_legitimate_address_p): Likewise.
17315 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17317 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17318 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17319 (iamcu_alignment): Likewise.
17320 (canonicalize_vector_int_perm): Likewise.
17321 (ix86_noce_conversion_profitable_p): Likewise.
17322 (ix86_mpx_bound_mode): Likewise.
17323 (ix86_operands_ok_for_move_multiple): Likewise.
17324 * config/microblaze/microblaze-protos.h
17325 (microblaze_expand_conditional_branch_reg): Likewise.
17326 * config/microblaze/microblaze.c
17327 (microblaze_expand_conditional_branch_reg): Likewise.
17328 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17330 (rs6000_reassociation_width): Likewise.
17331 (rs6000_invalid_binary_op): Likewise.
17332 (fusion_p9_p): Likewise.
17333 (emit_fusion_p9_load): Likewise.
17334 (emit_fusion_p9_store): Likewise.
17335 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17337 (riscv_hard_regno_mode_ok_p): Likewise.
17338 (riscv_address_insns): Likewise.
17339 (riscv_split_symbol): Likewise.
17340 (riscv_legitimize_move): Likewise.
17341 (riscv_function_value): Likewise.
17342 (riscv_hard_regno_nregs): Likewise.
17343 (riscv_expand_builtin): Likewise.
17344 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17345 (riscv_build_integer): Likewise.
17346 (riscv_split_integer): Likewise.
17347 (riscv_legitimate_constant_p): Likewise.
17348 (riscv_cannot_force_const_mem): Likewise.
17349 (riscv_regno_mode_ok_for_base_p): Likewise.
17350 (riscv_valid_base_register_p): Likewise.
17351 (riscv_valid_offset_p): Likewise.
17352 (riscv_valid_lo_sum_p): Likewise.
17353 (riscv_classify_address): Likewise.
17354 (riscv_legitimate_address_p): Likewise.
17355 (riscv_address_insns): Likewise.
17356 (riscv_load_store_insns): Likewise.
17357 (riscv_force_binary): Likewise.
17358 (riscv_split_symbol): Likewise.
17359 (riscv_force_address): Likewise.
17360 (riscv_legitimize_address): Likewise.
17361 (riscv_move_integer): Likewise.
17362 (riscv_legitimize_const_move): Likewise.
17363 (riscv_legitimize_move): Likewise.
17364 (riscv_address_cost): Likewise.
17365 (riscv_subword): Likewise.
17366 (riscv_output_move): Likewise.
17367 (riscv_canonicalize_int_order_test): Likewise.
17368 (riscv_emit_int_order_test): Likewise.
17369 (riscv_function_arg_boundary): Likewise.
17370 (riscv_pass_mode_in_fpr_p): Likewise.
17371 (riscv_pass_fpr_single): Likewise.
17372 (riscv_pass_fpr_pair): Likewise.
17373 (riscv_get_arg_info): Likewise.
17374 (riscv_function_arg): Likewise.
17375 (riscv_function_arg_advance): Likewise.
17376 (riscv_arg_partial_bytes): Likewise.
17377 (riscv_function_value): Likewise.
17378 (riscv_pass_by_reference): Likewise.
17379 (riscv_setup_incoming_varargs): Likewise.
17380 (riscv_print_operand): Likewise.
17381 (riscv_elf_select_rtx_section): Likewise.
17382 (riscv_save_restore_reg): Likewise.
17383 (riscv_for_each_saved_reg): Likewise.
17384 (riscv_register_move_cost): Likewise.
17385 (riscv_hard_regno_mode_ok_p): Likewise.
17386 (riscv_hard_regno_nregs): Likewise.
17387 (riscv_class_max_nregs): Likewise.
17388 (riscv_memory_move_cost): Likewise.
17389 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17390 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17391 (rl78_addr_space_address_mode): Likewise.
17392 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17394 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17395 (rs6000_reassociation_width): Likewise.
17396 (rs6000_invalid_binary_op): Likewise.
17397 (fusion_p9_p): Likewise.
17398 (emit_fusion_p9_load): Likewise.
17399 (emit_fusion_p9_store): Likewise.
17400 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17401 (ok_for_simple_move_operands): Likewise.
17402 (ok_for_simple_move_strict_operands): Likewise.
17403 (ok_for_simple_arith_logic_operands): Likewise.
17404 (visium_legitimize_reload_address): Likewise.
17405 (visium_select_cc_mode): Likewise.
17406 (output_cbranch): Likewise.
17407 (visium_split_double_move): Likewise.
17408 (visium_expand_copysign): Likewise.
17409 (visium_expand_int_cstore): Likewise.
17410 (visium_expand_fp_cstore): Likewise.
17411 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17412 (visium_function_arg): Likewise.
17413 (visium_function_arg_advance): Likewise.
17414 (visium_libcall_value): Likewise.
17415 (visium_setup_incoming_varargs): Likewise.
17416 (visium_legitimate_constant_p): Likewise.
17417 (visium_legitimate_address_p): Likewise.
17418 (visium_legitimize_address): Likewise.
17419 (visium_secondary_reload): Likewise.
17420 (visium_register_move_cost): Likewise.
17421 (visium_memory_move_cost): Likewise.
17422 (prepare_move_operands): Likewise.
17423 (ok_for_simple_move_operands): Likewise.
17424 (ok_for_simple_move_strict_operands): Likewise.
17425 (ok_for_simple_arith_logic_operands): Likewise.
17426 (visium_function_value_1): Likewise.
17427 (rtx_ok_for_offset_p): Likewise.
17428 (visium_legitimize_reload_address): Likewise.
17429 (visium_split_double_move): Likewise.
17430 (visium_expand_copysign): Likewise.
17431 (visium_expand_int_cstore): Likewise.
17432 (visium_expand_fp_cstore): Likewise.
17433 (visium_split_cstore): Likewise.
17434 (visium_select_cc_mode): Likewise.
17435 (visium_split_cbranch): Likewise.
17436 (output_cbranch): Likewise.
17437 (visium_print_operand_address): Likewise.
17438 * expmed.c (flip_storage_order): Likewise.
17439 * expmed.h (emit_cstore): Likewise.
17440 (flip_storage_order): Likewise.
17441 * genrecog.c (validate_pattern): Likewise.
17442 * hsa-gen.c (gen_hsa_addr): Likewise.
17443 * internal-fn.c (expand_arith_overflow): Likewise.
17444 * ira-color.c (allocno_copy_cost_saving): Likewise.
17445 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17446 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17447 (process_invariant_for_inheritance): Likewise.
17448 * lra-eliminations.c (move_plus_up): Likewise.
17449 * omp-low.c (lower_oacc_reductions): Likewise.
17450 * simplify-rtx.c (simplify_subreg): Likewise.
17451 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17452 (TARGET_CHKP_BOUND_MODE): Likewise..
17453 * targhooks.c (default_chkp_bound_mode): Likewise.
17454 (default_setup_incoming_vararg_bounds): Likewise.
17455 * targhooks.h (default_chkp_bound_mode): Likewise.
17456 (default_setup_incoming_vararg_bounds): Likewise.
17457 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17458 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17459 (have_whole_vector_shift): Likewise.
17460 * tree-vect-stmts.c (vectorizable_load): Likewise.
17461 * doc/tm.texi: Regenerate.
17463 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17465 Graceful degrade if Binutils PR21472 is not available.
17468 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17469 .rodata in flash test fails.
17470 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17471 * confgure: Regenerate.
17472 * config.in: Regenerate.
17473 * config/avr/avr.c (avr_asm_named_section)
17474 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17475 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17476 (avr_asm_init_sections): Same.
17478 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17480 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17481 (fma<VH:mode>4_intrinsic): Likewise.
17482 (*fmsub<VCVTF:mode>4): Likewise.
17483 (*fmsub<VH:mode>4_intrinsic): Likewise.
17485 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17488 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17489 Don't depend on "optimize > 0".
17490 (out_movhi_r_mr, out_movqi_mr_r): Same.
17491 (out_movhi_mr_r, out_movqi_r_mr): Same.
17492 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17493 io_address_operand on "optimize > 0".
17495 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17497 * tree-loop-distribution.c: Add general explanantion on the pass.
17498 (generate_loops_for_partition): Mark distributed loop.
17499 (pg_add_dependence_edges): New parameter. Handle alias data
17500 dependence specially and record it in the parameter if asked.
17501 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17502 (init_partition_graph_vertices, add_partition_graph_edge): New.
17503 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17504 (free_partition_graph_vdata, build_partition_graph): New.
17505 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17506 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17507 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17508 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17509 (version_for_distribution_p, finalize_partitions): New.
17510 (distribute_loop): Handle alias data dependence specially. Factor
17511 out loop fusion code as functions and call these functions.
17513 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17515 * tree-loop-distribution.c (classify_partition): New parameter and
17516 better handle reduction statement.
17517 (rdg_build_partitions): Revise comment.
17518 (distribute_loop): Compute statements in all partitions and pass it
17519 to classify_partition.
17521 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17523 * tree-loop-distribution.c (enum partition_type): New.
17524 (struct partition): New field type.
17525 (partition_merge_into): Add parameter. Update partition type.
17526 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17527 (build_rdg_partition_for_vertex): Compute partition type.
17528 (rdg_build_partitions): Dump partition type.
17529 (distribute_loop): Update calls to partition_merge_into.
17531 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17533 * tree-loop-distribution.c (struct ddr_hasher): New.
17534 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17536 (classify_partition): Call get_data_dependence.
17537 (pg_add_dependence_edges): Ditto.
17538 (distribute_loop): Release data dependence hash table.
17540 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17542 * tree-loop-distribution.c (ref_base_address): Delete.
17543 (similar_memory_accesses): Rename ...
17544 (share_memory_accesses): ... to this. Check if partitions access
17545 the same memory reference.
17546 (distribute_loop): Call share_memory_accesses.
17548 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17550 * tree-loop-distribution.c (struct partition): New field recording
17551 its data reference.
17552 (partition_alloc, partition_free): Init and release data refs.
17553 (partition_merge_into): Merge data refs.
17554 (build_rdg_partition_for_vertex): Collect data refs for partition.
17555 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17557 (distribute_loop): Remve data refs from vertice data of partition
17560 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17562 * tree-loop-distribution.c (params.h): Include header file.
17563 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17564 (datarefs_vec): New global var.
17565 (create_rdg_vertices): Use datarefs_vec directly.
17566 (free_rdg): Don't free data references.
17567 (build_rdg): Update use. Don't free data references.
17568 (distribute_loop): Compute global variable for data references.
17569 Bail out if there are too many data references.
17571 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17573 * tree-loop-distribution.c (loop_nest): New global var.
17574 (build_rdg): Use loop directly, rather than loop nest.
17575 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17577 (distribute_loop): Compute global variable loop nest. Update use.
17579 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17581 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17582 (partition_merge_into): New parameter. Dump reason for fusion.
17583 (distribute_loop): Update use of partition_merge_into.
17585 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17587 * tree-loop-distribution.c (bb_top_order_index): New.
17588 (bb_top_order_index_size, bb_top_order_cmp): New.
17589 (stmts_from_loop): Use topological order.
17590 (pass_loop_distribution::execute): Compute and release topological
17591 order for basic blocks.
17593 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17595 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17598 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17600 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17601 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17602 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17603 * internal-fn.def (LOOP_DIST_ALIAS): New.
17604 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17605 (fold_loop_internal_call): ... this.
17606 (vect_loop_dist_alias_call): New function.
17607 (set_uid_loop_bbs): Call fold_loop_internal_call.
17608 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17611 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17614 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17615 Require dead FLAGS_REG at the beginning of a peephole.
17617 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17620 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17621 arguments in the call to __builtin_ia32_sbb_u32.
17622 (_subborrow_u64): Swap _X and _Y arguments in the call to
17623 __builtin_ia32_sbb_u64.
17625 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17628 * tree-vrp.c (compare_assert_loc): Turn into a function template
17629 with stable template parameter. Only test if a->e is NULL,
17630 !a->e == !b->e has been verified already. Use e == NULL or
17631 e != NULL instead of e or ! e tests. If stable is true, don't use
17632 iterative_hash_expr, on the other side allow a or b or both NULL
17633 and sort the NULLs last.
17634 (process_assert_insertions): Sort using compare_assert_loc<false>
17635 instead of compare_assert_loc, later sort using
17636 compare_assert_loc<true> before calling process_assert_insertions_for
17637 in a loop. Use break instead of continue once seen NULL pointer.
17639 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17641 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17642 Cortex-R7 and Cortex-R8 processors.
17644 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17646 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17647 uninitialized while src is not.
17649 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17651 * common/config/arm/arm-common.c: Adjust include path for
17653 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17654 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17655 (arm-cpu-data.h): Likewise.
17656 (arm-cpu-cdata.h): Likewise.
17657 * config/arm/arm-cpu.h: Delete.
17658 * config/arm/arm-cpu-cdata.h: Delete.
17659 * config/arm/arm-cpu-data.h: Delete.
17661 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17663 * config/arm/arm-cpus.in (cortex-a55): New.
17664 (cortex-a75): Likewise.
17665 (cortex-a75.cortex-a55): Likewise.
17666 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17668 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17669 * config/arm/arm-cpu-cdata.h: Regenerate.
17670 * config/arm/arm-cpu-data.h: Regenerate.
17671 * config/arm/arm-cpu.h: Regenerate.
17672 * config/arm/arm-tables.opt: Regenerate.
17673 * config/arm/arm-tune.md: Regenerate.
17675 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17677 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17679 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17681 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17684 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17686 PR tree-optimization/81292
17687 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17688 full_string_p, also call adjust_related_strinfos if the adjustment
17689 is simple, otherwise invalidate related strinfos.
17691 2017-07-04 Martin Liska <mliska@suse.cz>
17694 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17695 newly created variable as DECL_IGNORED_P.
17697 2017-07-04 Martin Liska <mliska@suse.cz>
17700 * ipa-inline.c (inline_small_functions):
17701 Use xstrdup_for_dump.
17703 2017-07-04 Tom de Vries <tom@codesourcery.com>
17705 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17707 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17710 * config/darwin.c (darwin_function_switched_text_sections):
17713 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17715 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17717 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17719 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17721 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17723 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17724 min_profitable_iters, and th as inclusive lower bounds.
17725 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17726 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17727 for min_profitable_iters and min_profitable_estimate.
17728 (vect_transform_loop): Treat th as an inclusive lower bound.
17729 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17731 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17734 * config/darwin.c (darwin_function_switched_text_sections):
17735 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17736 in two pieces, and suppress the use of buf.
17738 2017-07-03 Nathan Sidwell <nathan@acm.org>
17740 * hash-table.h (hash_table_mod1): Fix indentation.
17742 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17744 PR middle-end/81290
17745 * predict.c (force_edge_cold): Be more careful about propagation
17747 * profile-count.h (profile_probability::guessed,
17748 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17750 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17752 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17754 * doc/invoke.texi (rcpc architecture extension): Document it.
17756 2017-07-03 Richard Biener <rguenther@suse.de>
17758 PR tree-optimization/60510
17759 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17760 the scalar reduction PHI and use it.
17761 (vectorizable_reduction): Properly guard the single_defuse_cycle
17762 path for non-SLP reduction chains where we cannot use it.
17763 Rework reduc_def/index and vector type deduction. Rework
17764 vector operand gathering during reduction op code-gen.
17765 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17766 chains dissolve the chain and leave it to non-SLP reduction
17769 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17771 * tree-data-ref.h (dr_alignment): Declare.
17772 * tree-data-ref.c (dr_alignment): New function.
17773 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17774 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17776 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17778 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17780 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17781 and base_misalignment fields.
17782 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17783 * tree-data-ref.c: Include builtins.h.
17784 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17785 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17786 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17787 * tree-vect-data-refs.c: Include tree-cfg.h.
17788 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17789 fields instead of calculating an alignment here.
17790 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17791 innermost_loop_behavior fields.
17793 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17795 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17797 (DR_STEP_ALIGNMENT): New macro.
17798 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17799 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17800 (create_data_ref): Print it.
17801 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17802 to tell whether the step preserves vector (mis)alignment.
17803 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17804 Move the check for an integer step and generalise to all INTEGER_CST.
17805 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17806 Print the outer step alignment.
17808 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17810 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17811 with offset_alignment.
17812 (DR_ALIGNED_TO): Delete.
17813 (DR_OFFSET_ALIGNMENT): New macro.
17814 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17815 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17816 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17817 (create_data_ref): Likewise.
17818 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17819 (vect_analyze_data_refs): Likewise.
17820 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17821 creating dummy innermost behavior.
17823 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17825 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17826 with a "innermost_loop_behavior *" and refeence tree.
17827 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17828 (create_data_ref): Update call accordingly.
17829 * tree-predcom.c (find_looparound_phi): Likewise.
17831 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17833 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17834 fields with dr_wrt_vec_loop.
17835 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17836 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17837 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17838 (vect_dr_behavior): New function.
17839 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17840 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17841 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17842 track whether the step preserves the misalignment.
17843 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17844 Use vect_dr_behavior.
17845 (vect_setup_realignment): Update call accordingly.
17846 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17847 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17848 call to vect_create_addr_base_for_vector_ref.
17849 (vect_create_cond_for_align_checks): Likewise.
17850 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17851 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17852 (vect_recog_mask_conversion_pattern): Likewise.
17853 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17854 (new_stmt_vec_info): Remove redundant zeroing.
17856 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17858 * common/config/arm/arm-common.c (arm_be8_option): New function.
17859 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17860 (ISA_ARMv6): Add isa_bit_be8.
17861 * config/arm/arm.h (arm_be8_option): Add prototype.
17862 (BE8_SPEC_FUNCTION): New define.
17863 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17864 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17865 (mlittle-endian): Similarly.
17866 (mbe8, mbe32): New options.
17867 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17868 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17870 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17872 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17874 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17876 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17877 (cleanup_tree_cfg_bb): Use it.
17878 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17880 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17882 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17885 * loop-doloop.c (add_test): Update profile.
17887 2017-07-03 Martin Liska <mliska@suse.cz>
17890 * sanopt.c (rewrite_usage_of_param): New function.
17891 (sanitize_rewrite_addressable_params): Likewise.
17892 (pass_sanopt::execute): Call rewrite_usage_of_param.
17894 2017-07-03 Richard Biener <rguenther@suse.de>
17896 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17897 back to using VIEW_CONVERT_EXPR.
17899 2017-07-03 Martin Liska <mliska@suse.cz>
17902 * doc/extend.texi: Document when a resolver function is
17903 generated for target_clones.
17905 2017-07-03 Martin Liska <mliska@suse.cz>
17907 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17908 and shadow memory of auto variables which are subject of
17909 use-after-scope sanitization.
17910 (asan_expand_mark_ifn): Add do set only when is_poison.
17912 2016-07-03 Richard Biener <rguenther@suse.de>
17914 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17916 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17917 (vectorizable_reduction): Perform reduction PHI creation when
17918 visiting a reduction PHI and adjust and simplify code generation
17919 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17920 (vect_transform_loop): Visit reduction PHIs.
17921 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17922 defs into the SLP tree.
17923 (vect_build_slp_tree): Reduction defs terminate the recursion.
17924 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17926 (vect_get_vec_defs_for_stmt_copy): Export.
17927 (vect_get_vec_defs): Likewise.
17928 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17930 (vect_get_vec_defs_for_stmt_copy): Declare.
17931 (vect_get_vec_defs): Likewise.
17933 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17935 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17936 parameter with a "loop" parameter and use it instead of the
17937 loop containing DR_STMT. Don't check simple_iv when doing
17938 BB analysis. Describe the two analysis modes in the comment.
17940 2017-07-03 Tom de Vries <tom@codesourcery.com>
17942 PR tree-optimization/69468
17943 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17944 (find_same_succ_bb): Handle ignore_edge_flags.
17946 2017-07-03 Tom de Vries <tom@codesourcery.com>
17948 PR tree-optimization/81192
17949 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17951 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17953 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17955 2017-07-03 Tom de Vries <tom@codesourcery.com>
17957 PR tree-optimization/81192
17958 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17959 BB_SAME_SUCC (bb) == NULL.
17961 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17963 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17966 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17968 * dumpfile.c: Include profile-count.h
17969 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17971 (insert_cond_bb): Update profile.
17972 * tree-cfg.h (insert_cond_bb): Update prototype.
17973 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17974 * tree-dump.c: Do not include tree-cfg.
17976 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17978 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17980 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17982 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17985 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17987 * tree-complex.c (expand_complex_div_wide): update profile.
17989 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17990 Alan Hayward <alan.hayward@arm.com>
17991 David Sherwood <david.sherwood@arm.com>
17993 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17994 (CORETYPES_H): New define.
17995 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17996 (insn-modes-inline.h, s-modes-inline-h): New rules.
17997 (generated_files): Add insn-modes-inline.h.
17998 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17999 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18000 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18001 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18002 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18003 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18004 (build/gencodes.o, build/genconditions.o): Likewise.
18005 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18006 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18007 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18008 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18009 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18010 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18011 * coretypes.h: Include everything up to real.h for generators.
18012 Include insn-modes.h first. Include wide-int-print.h after
18013 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18014 * machmode.h: Don't include insn-modes.h here.
18015 * function-tests.c: Remove includes of signop.h, machmode.h,
18016 double-int.h and wide-int.h.
18018 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18020 * optc-save-gen.awk: Likewise.
18021 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18022 * godump.c: Remove include of wide-int-print.h.
18023 * pretty-print.h: Likewise.
18024 * wide-int-print.cc: Likewise.
18025 * wide-int.cc: Likewise.
18026 * hash-map-tests.c: Remove include of signop.h.
18027 * hash-set-tests.c: Likewise.
18028 * rtl-tests.c: Likewise.
18029 * mkconfig.sh: Remove include of machmode.h.
18030 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18032 (emit_insn_modes_inline_h): ...this new function. Emit the code
18033 into an insn-modes-inline.h header file, adding appropriate
18034 include guards and end comments.
18035 (emit_insn_modes_c_header): Remove include of machmode.h.
18036 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18038 (main): Handle -i flag and call emit_insn_modes_inline_h when
18041 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18043 * tree-ssa-strlen.c (strinfo): Rename the length field to
18044 nonzero_chars. Add a full_string_p field.
18045 (compare_nonzero_chars, zero_length_string_p): New functions.
18046 (get_addr_stridx): Add an offset_out parameter.
18047 Use compare_nonzero_chars.
18048 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18049 (new_strinfo): Update after above changes to strinfo.
18050 (set_endptr_and_length): Set full_string_p.
18051 (get_string_length): Update after above changes to strinfo.
18052 (unshare_strinfo): Update call to new_strinfo.
18053 (maybe_invalidate): Likewise.
18054 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18055 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18056 as a uhwi instead of an shwi. Update after above changes to
18057 strinfo and new_strinfo.
18058 (zero_length_string): Assert that chainsi contains full strings.
18059 Use zero_length_string_p. Update call to new_strinfo.
18060 (adjust_related_strinfos): Update after above changes to strinfo.
18061 Copy full_string_p from origsi.
18062 (adjust_last_stmt): Use zero_length_string_p.
18063 (handle_builtin_strlen): Update after above changes to strinfo and
18064 new_strinfo. Install the lhs as the string length if the previous
18065 entry didn't describe a full string.
18066 (handle_builtin_strchr): Update after above changes to strinfo
18068 (handle_builtin_strcpy): Likewise.
18069 (handle_builtin_strcat): Likewise.
18070 (handle_builtin_malloc): Likewise.
18071 (handle_pointer_plus): Likewise.
18072 (handle_builtin_memcpy): Likewise. Track nonzero characters
18073 that aren't necessarily followed by a nul terminator.
18074 (handle_char_store): Likewise.
18076 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18078 PR tree-optimization/80769
18079 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18080 for malloc and calloc. Document the new invariant that all related
18081 strinfos have delayed lengths or none do.
18082 (verify_related_strinfos): Move earlier in file.
18083 (set_endptr_and_length): New function, split out from...
18084 (get_string_length): ...here. Also set the lengths of related
18086 (zero_length_string): Assert that chainsi has known (rather than
18088 (adjust_related_strinfos): Likewise.
18090 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18092 PR tree-optimization/81136
18093 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18094 assert that two references with the same misalignment have the same
18095 compile-time misalignment if those compile-time misalignments
18098 2017-07-01 Andi Kleen <ak@linux.intel.com>
18100 * print-tree.c (print_node): Print all attributes.
18102 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18104 * cfg.c (scale_bbs_frequencies): New function.
18105 * cfg.h (scale_bbs_frequencies): Declare it.
18106 * cfgloopanal.c (single_likely_exit): Cleanup.
18107 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18109 (scale_loop_profile): Likewise.
18110 (loop_version): Likewise.
18111 (create_empty_loop_on_edge): Update.
18112 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18113 scale_loop_frequencies, scale_loop_profile, loopify,
18114 loop_version): Update prototypes.
18115 * modulo-sched.c (sms_schedule): Update.
18116 * predict.c (unlikely_executed_edge_p): Also check probability.
18117 (probably_never_executed_edge_p): Fix typo.
18118 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18119 * tree-parloops.c (gen_parallel_loop): Update.
18120 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18121 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18122 * tree-ssa-loop-split.c (split_loop): Update.
18123 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18124 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18125 (vect_loop_versioning): Update.
18126 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18128 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18130 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18132 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18134 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18135 to keep profile consistent.
18137 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18139 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18140 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18141 * profile-count.h (max_safe_multiplier): Make unsigned.
18142 (profile_count::guessed_zero): New.
18144 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18146 * bb-reorder.c (fix_up_crossing_landing_pad,
18147 fix_crossing_conditional_branches): Use make_single_succ_edge
18148 to keep profile consistent.
18150 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18152 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18155 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18158 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18159 the right scopes, make sure cond_jump isn't preserved between multiple
18160 iterations. Search for fallthru edge whenever there are 3+ edges and
18161 use find_fallthru_edge for it.
18163 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18165 Patch by Alexander Monakov <amonakov@ispras.ru>
18166 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18167 probabilities consistently.
18169 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18171 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18172 emit_cmp_and_jump_insns.
18174 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18177 * tree-inline.c (expand_call_inline): Combine profile statuses.
18179 2017-06-30 Andrew Pinski <apinski@cavium.com>
18181 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18182 fold_stmt returned true.
18184 2017-06-30 Nathan Sidwell <nathan@acm.org>
18186 * ggc.h (empty_string): Delete.
18187 * cfgexpand.c (expand_asm_stmt): Use plain "".
18188 * optabs.c (expand_asm_memory_barrier): Likewise.
18189 * stringpool.c (empty_string): Delete.
18190 (digit_vector, digit_string): Delete.
18191 (ggc_alloc_string): Use plain "", don't optimize single digit
18192 strings. Use ggc_alloc_atomic.
18194 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18196 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18197 comparison set and one other set, use the cost of the non-comparison
18200 2017-06-30 Nathan Sidwell <nathan@acm.org>
18202 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18205 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18207 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18208 loops. Remove now unneeded calls to gimple_switch_set_label() that
18209 just set removed labels to NULL_TREE.
18211 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18213 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18214 (set_range_info): ...here. Only call set_range_info_raw if domain
18216 (set_nonzero_bits): Call set_range_info_raw.
18217 * tree-ssanames.h (set_range_info_raw): New.
18219 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18222 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18223 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18224 of nonimmediate_operand and <store_mask_constraint> instead of m
18225 for the input operand. For V8FI iterator, always split if input
18226 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18227 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18228 <store_mask_predicate> instead of register_operand and
18229 <store_mask_constraint> instead of v for the input operand. Make
18230 sure both operands aren't MEMs for if not <mask_applied>.
18232 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18234 * lto-wrapper.c (copy_file) Close both file descriptors before
18237 2017-06-30 Martin Liska <mliska@suse.cz>
18240 * multiple_target.c (create_dispatcher_calls): Make ifunc
18241 also for function that don't have calls or are not referenced.
18243 2017-06-30 Richard Biener <rguenther@suse.de>
18245 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18246 analyze the first scalar stmt. Move vector type computation
18247 for the BB case here from ...
18248 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18249 live operation processing in the SLP case properly.
18251 2017-06-30 Richard Biener <rguenther@suse.de>
18253 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18255 2017-06-30 Martin Liska <mliska@suse.cz>
18258 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18259 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18261 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18263 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18266 2017-06-30 Richard Biener <rguenther@suse.de>
18268 PR tree-optimization/81249
18269 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18270 condition reduction result to original scalar type.
18272 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18274 * profile-count.h (enum profile_quality): Fix typos and whitespace
18277 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18279 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18280 type for branch probabilities.
18282 2017-06-29 Julian Brown <julian@codesourcery.com>
18283 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18285 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18286 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18287 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18288 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18290 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18292 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18293 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18294 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18295 CC usage from generic code to here.
18296 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18297 CC usage into the target macros.
18299 2017-06-29 Maya Rashish <coypu@sdf.org>
18301 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18304 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18306 * arm/arm-builtins.c: Include profile-count.h
18307 * except.c (sjlj_emit_function_enter): Use
18308 profile_probability::unlikely.
18310 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18312 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18313 and tocrel_offset be pointer args rather than implicitly using
18315 (legitimate_constant_pool_address_p, rs6000_emit_move,
18316 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18317 tocrel_offset and use in toc_relative_expr_p call.
18318 (print_operand, print_operand_address): Use static tocrel_base_oac
18319 and tocrel_offset_oac.
18320 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18323 2017-06-29 Maya Rashish <coypu@sdf.org>
18325 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18327 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18329 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18330 objects, take into account only the alignment of 'op0' and 'mode1' if
18333 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18335 * ccmp.c (ccmp_tree_comparison_p): New function.
18336 (ccmp_candidate_p): Update to use above function.
18337 (get_compare_parts): New function.
18338 (expand_ccmp_next): Update to use new functions.
18339 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18341 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18342 take mode as argument.
18343 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18344 * expr.c (expand_expr_real_1): Pass mode as argument.
18346 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18348 * combine.c (combine_instructions): Print insns to dump_file, together
18351 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18353 * asan.c (asan_emit_stack_protection): Update.
18354 (create_cond_insert_point): Update.
18355 * auto-profile.c (afdo_propagate_circuit): Update.
18356 * basic-block.h (struct edge_def): Turn probability to
18357 profile_probability.
18358 (EDGE_FREQUENCY): Update.
18359 * bb-reorder.c (find_traces_1_round): Update.
18360 (better_edge_p): Update.
18361 (sanitize_hot_paths): Update.
18362 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18363 (make_single_succ_edge): Update.
18364 (check_bb_profile): Update.
18365 (dump_edge_info): Update.
18366 (update_bb_profile_for_threading): Update.
18367 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18368 probabilitycount to 0.
18369 * cfgbuild.c (compute_outgoing_frequencies): Update.
18370 * cfgcleanup.c (try_forward_edges): Update.
18371 (outgoing_edges_match): Update.
18372 (try_crossjump_to_edge): Update.
18373 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18374 (expand_gimple_tailcall): Update.
18375 (construct_init_block): Use make_single_succ_edge.
18376 (construct_exit_block): Use make_single_succ_edge.
18377 * cfghooks.c (verify_flow_info): Update.
18378 (redirect_edge_succ_nodup): Update.
18379 (split_edge): Update.
18380 (account_profile_record): Update.
18381 * cfgloopanal.c (single_likely_exit): Update.
18382 * cfgloopmanip.c (scale_loop_profile): Update.
18383 (set_zero_probability): Remove.
18384 (duplicate_loop_to_header_edge): Update.
18385 * cfgloopmanip.h (loop_version): Update prototype.
18386 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18387 (force_nonfallthru_and_redirect): Update.
18388 (update_br_prob_note): Update.
18389 (rtl_verify_edges): Update.
18390 (purge_dead_edges): Update.
18391 (rtl_lv_add_condition_to_bb): Update.
18392 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18393 * cgraphunit.c (init_lowered_empty_function): Update.
18394 (cgraph_node::expand_thunk): Update.
18395 * cilk-common.c: Include profile-count.h
18396 * dojump.c (inv): Remove.
18397 (jumpifnot): Update.
18398 (jumpifnot_1): Update.
18399 (do_jump_1): Update.
18401 (do_jump_by_parts_greater_rtx): Update.
18402 (do_compare_rtx_and_jump): Update.
18403 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18404 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18405 * dwarf2cfi.c: Include profile-count.h
18406 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18407 (sjlj_emit_dispatch_table): Likewise.
18408 * explow.c: Include profile-count.h
18409 * expmed.c (emit_store_flag_force): Update.
18410 (do_cmp_and_jump): Update.
18411 * expr.c (compare_by_pieces_d::generate): Update.
18412 (compare_by_pieces_d::finish_mode): Update.
18413 (emit_block_move_via_loop): Update.
18414 (store_expr_with_bounds): Update.
18415 (store_constructor): Update.
18416 (expand_expr_real_2): Update.
18417 (expand_expr_real_1): Update.
18418 * expr.h (try_casesi, try_tablejump): Update prototypes.
18419 * gimple-pretty-print.c (dump_probability): Update.
18420 (dump_profile): New.
18421 (dump_gimple_label): Update.
18422 (dump_gimple_bb_header): Update.
18423 * graph.c (draw_cfg_node_succ_edges): Update.
18424 * hsa-gen.c (convert_switch_statements): Update.
18425 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18426 (find_if_case_1): Update.
18427 (find_if_case_2): Update.
18428 * internal-fn.c (expand_arith_overflow_result_store): Update.
18429 (expand_addsub_overflow): Update.
18430 (expand_neg_overflow): Update.
18431 (expand_mul_overflow): Update.
18432 (expand_vector_ubsan_overflow): Update.
18433 * ipa-cp.c (good_cloning_opportunity_p): Update.
18434 * ipa-split.c (split_function): Use make_single_succ_edge.
18435 * ipa-utils.c (ipa_merge_profiles): Update.
18436 * loop-doloop.c (add_test): Update.
18437 (doloop_modify): Update.
18438 * loop-unroll.c (compare_and_jump_seq): Update.
18439 (unroll_loop_runtime_iterations): Update.
18440 * lra-constraints.c (lra_inheritance): Update.
18441 * lto-streamer-in.c (input_cfg): Update.
18442 * lto-streamer-out.c (output_cfg): Update.
18443 * mcf.c (adjust_cfg_counts): Update.
18444 * modulo-sched.c (sms_schedule): Update.
18445 * omp-expand.c (expand_omp_for_init_counts): Update.
18446 (extract_omp_for_update_vars): Update.
18447 (expand_omp_ordered_sink): Update.
18448 (expand_omp_for_ordered_loops): Update.
18449 (expand_omp_for_generic): Update.
18450 (expand_omp_for_static_nochunk): Update.
18451 (expand_omp_for_static_chunk): Update.
18452 (expand_cilk_for): Update.
18453 (expand_omp_simd): Update.
18454 (expand_omp_taskloop_for_outer): Update.
18455 (expand_omp_taskloop_for_inner): Update.
18456 * omp-simd-clone.c (simd_clone_adjust): Update.
18457 * optabs.c (expand_doubleword_shift): Update.
18458 (expand_abs): Update.
18459 (emit_cmp_and_jump_insn_1): Update.
18460 (expand_compare_and_swap_loop): Update.
18461 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18462 * predict.c (predictable_edge_p): Update.
18463 (edge_probability_reliable_p): Update.
18464 (set_even_probabilities): Update.
18465 (combine_predictions_for_insn): Update.
18466 (combine_predictions_for_bb): Update.
18467 (propagate_freq): Update.
18468 (estimate_bb_frequencies): Update.
18469 (force_edge_cold): Update.
18470 * profile-count.c (profile_count::dump): Add missing space into dump.
18471 (profile_count::debug): Add newline.
18472 (profile_count::differs_from_p): Explicitly convert to unsigned.
18473 (profile_count::stream_in): Update.
18474 (profile_probability::dump): New member function.
18475 (profile_probability::debug): New member function.
18476 (profile_probability::differs_from_p): New member function.
18477 (profile_probability::differs_lot_from_p): New member function.
18478 (profile_probability::stream_in): New member function.
18479 (profile_probability::stream_out): New member function.
18480 * profile-count.h (profile_count_quality): Rename to ...
18481 (profile_quality): ... this one.
18482 (profile_probability): New.
18483 (profile_count): Update.
18484 * profile.c (compute_branch_probabilities): Update.
18485 * recog.c (peep2_attempt): Update.
18486 * sched-ebb.c (schedule_ebbs): Update.
18487 * sched-rgn.c (find_single_block_region): Update.
18488 (compute_dom_prob_ps): Update.
18489 (schedule_region): Update.
18490 * sel-sched-ir.c (compute_succs_info): Update.
18491 * stmt.c (struct case_node): Update.
18492 (do_jump_if_equal): Update.
18493 (get_outgoing_edge_probs): Update.
18494 (conditional_probability): Update.
18495 (emit_case_dispatch_table): Update.
18496 (expand_case): Update.
18497 (expand_sjlj_dispatch_table): Update.
18498 (emit_case_nodes): Update.
18499 * targhooks.c: Update.
18500 * tracer.c (better_p): Update.
18501 (find_best_successor): Update.
18502 * trans-mem.c (expand_transaction): Update.
18503 * tree-call-cdce.c: Update.
18504 * tree-cfg.c (gimple_split_edge): Upate.
18505 (move_sese_region_to_fn): Upate.
18506 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18507 * tree-eh.c (lower_resx): Upate.
18508 (cleanup_empty_eh_move_lp): Upate.
18509 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18510 * tree-inline.c (copy_edges_for_bb): Update.
18511 (copy_cfg_body): Update.
18512 * tree-parloops.c (gen_parallel_loop): Update.
18513 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18514 (gimple_gen_time_profiler): Update.
18515 * tree-ssa-dce.c (remove_dead_stmt): Update.
18516 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18517 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18518 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18519 (unloop_loops): Update.
18520 (try_peel_loop): Update.
18521 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18522 * tree-ssa-loop-split.c (connect_loops): Update.
18523 (split_loop): Update.
18524 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18525 (hoist_guard): Update.
18526 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18527 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18528 (value_replacement): Update.
18529 * tree-ssa-reassoc.c (branch_fixup): Update.
18530 * tree-ssa-tail-merge.c (replace_block_by): Update.
18531 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18532 (create_edge_and_update_destination_phis): Update.
18533 (compute_path_counts): Update.
18534 (recompute_probabilities): Update.
18535 (update_joiner_offpath_counts): Update.
18536 (freqs_to_counts_path): Update.
18537 (duplicate_thread_path): Update.
18538 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18539 (struct switch_conv_info): Update.
18540 (gen_inbound_check): Update.
18541 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18542 (vect_do_peeling): Update.
18543 (vect_loop_versioning): Update.
18544 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18545 (optimize_mask_stores): Update.
18546 * ubsan.c (ubsan_expand_null_ifn): Update.
18547 * value-prof.c (gimple_divmod_fixed_value): Update.
18548 (gimple_divmod_fixed_value_transform): Update.
18549 (gimple_mod_pow2): Update.
18550 (gimple_mod_pow2_value_transform): Update.
18551 (gimple_mod_subtract): Update.
18552 (gimple_mod_subtract_transform): Update.
18553 (gimple_ic): Update.
18554 (gimple_stringop_fixed_value): Update.
18555 (gimple_stringops_transform): Update.
18556 * value-prof.h: Update.
18558 2017-06-29 Carl Love <cel@us.ibm.com>
18560 * config/rs6000/rs6000-c.c: Add support for built-in functions
18561 vector signed int vec_signed (vector float);
18562 vector signed long long vec_signed (vector double);
18563 vector signed int vec_signed2 (vector double, vector double);
18564 vector signed int vec_signede (vector double);
18565 vector signed int vec_signedo (vector double);
18566 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18567 instruction generator.
18568 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18569 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18570 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18572 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18573 vunsignede_v2df): Add define_expands.
18574 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18575 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18576 VEC_UNSIGNEDO): Add definitions.
18577 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18578 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18579 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18580 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18581 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18582 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18583 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18584 vec_unsignede, vec_unsignedo): Add builtin defines.
18585 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18587 * doc/extend.texi: Update the built-in documentation file for the
18588 new built-in functions.
18590 2017-06-29 Richard Biener <rguenther@suse.de>
18592 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18593 reduction chains to LOOP_VINFO_REDUCTIONS.
18594 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18595 SLP reductions after processing reduction chains.
18597 2017-06-29 Nathan Sidwell <nathan@acm.org>
18599 * builtins.c (fold_builtin_FUNCTION): Use
18600 lang_hooks.decl_printable_name.
18602 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18604 PR middle-end/81194
18605 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18606 with only one label.
18607 * stmt.c (expand_case): Assert NCASES is greater than one.
18609 2017-06-29 Richard Biener <rguenther@suse.de>
18611 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18613 (group_case_labels): Likewise.
18614 (find_taken_edge): Push sanity checking on val to workers...
18615 (find_taken_edge_cond_expr): ... here
18616 (find_taken_edge_switch_expr): ... and here, handle cases
18617 with just a default label.
18618 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18619 (group_case_labels): Likewise.
18620 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18621 group_case_labels does anything cleanup the CFG again.
18623 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18625 PR tree-optimization/81196
18626 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18627 exit condition comparing two IVs.
18629 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18631 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18632 profile to the dummy entry at the end of the list of architectures.
18633 * config/arm/arm-cpu-cdata.h: Regenerated.
18635 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18636 Michael Collison <michael.collison@arm.com>
18639 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18641 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18642 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18643 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18644 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18645 of shift when the shift amount is masked with constant equal to
18646 the size of the mode.
18647 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18650 2017-06-29 Martin Liska <mliska@suse.cz>
18652 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18654 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18657 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18660 (edge_growth_cache_entry::edge_growth_cache_entry): New
18662 (reset_edge_growth_cache): Update to use constructor.
18664 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18666 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18667 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18668 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18670 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18672 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18673 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18675 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18677 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18678 (*-linux-uclibc*): Add t-uclibc tmake_file.
18679 * config/t-musl: New.
18680 * config/t-uclibc: New.
18682 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18684 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18686 (gen_comm_data): Emit architectural setting of arch_prof.
18687 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18689 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18690 (armv8-m.base, armv8-m.main): Likewise.
18691 * arm-protos.h (arm_build_target): Add profile field.
18692 (arch_option): Likewise.
18693 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18695 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18696 arm_active_target.profile.
18698 2017-06-28 Richard Biener <rguenther@suse.de>
18700 PR middle-end/81227
18701 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18702 TYPE_OVERFLOW_WRAPS.
18703 * match.pd (negate_expr_p): Likewise.
18704 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18705 fold_build2, not fold_binary.
18707 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18709 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18710 Convert memory address to Pmode.
18711 (aarch64_print_operand): Assert MEM operands are always Pmode.
18713 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18716 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18717 Remove redundant if.
18718 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18719 * config/arm/aarch-common-protos.h
18720 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18721 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18723 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18726 * multiple_target.c (create_dispatcher_calls): Set the default
18727 clone to be static, not public.
18729 2017-06-28 Richard Biener <rguenther@suse.de>
18731 * tree-vect-loop.c (vectorizable_reduction): Move special
18732 cond reduction IV var creation ...
18733 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18734 parameter. Use STMT_VINFO_VECTYPE.
18735 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18738 2017-06-28 Martin Liska <mliska@suse.cz>
18741 * ipa-visibility.c (non_local_p): Handle visibility.
18743 2017-06-28 Martin Liska <mliska@suse.cz>
18746 * common.opt: Add IntegerRange to various options.
18747 * opt-functions.awk (integer_range_info): New function.
18748 * optc-gen.awk: Add integer_range_info to cl_options struct.
18749 * opts-common.c (decode_cmdline_option): Handle
18750 CL_ERR_INT_RANGE_ARG.
18751 (cmdline_handle_error): Likewise.
18752 * opts.c (print_filtered_help): Show valid interval in
18753 when --help is provided.
18754 * opts.h (struct cl_option): Add range_min and range_max fields.
18755 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18757 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18759 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18760 (x * C EQ/NE y * C): New transformation.
18762 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18764 * genmultilib (combination_space): Accept '+' in option names.
18766 2017-06-28 Martin Liska <mliska@suse.cz>
18769 * asan.c (instrument_derefs): Bail out inner references
18770 that are hard register variables.
18772 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18775 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18776 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18778 2017-06-28 Richard Biener <rguenther@suse.de>
18780 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18781 (vect_get_slp_defs): Adjust.
18782 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18784 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18786 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18787 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18788 (vectorizable_reduction): Adjust.
18789 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18791 (vect_get_slp_defs): Likewise.
18792 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18793 (vectorizable_bswap): Adjust.
18794 (vectorizable_call): Likewise.
18795 (vectorizable_conversion): Likewise.
18796 (vectorizable_assignment): Likewise.
18797 (vectorizable_shift): Likewise.
18798 (vectorizable_operation): Likewise.
18799 (vectorizable_store): Likewise.
18800 (vectorizable_condition): Likewise.
18801 (vectorizable_comparison): Likewise.
18803 2017-06-28 Michael Collison <michael.collison@arm.com>
18806 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18808 (arm_gen_movmemqi): Removed unused variable 'i'.
18809 Convert 'for' loop into 'while' loop.
18810 (arm_expand_prologue): Remove last unnecessary set of insn.
18811 (thumb_pop): Remove unused variable 'pushed_words'.
18812 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18814 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18816 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18817 * config/s390/s390-protos.h: Add prototype of
18818 s390_rel_address_ok_p.
18819 * config/s390/s390.c (s390_got_symbol): New function.
18820 (s390_rel_address_ok_p): New function.
18821 (legitimize_pic_address): Use s390_rel_address_ok_p.
18822 (s390_load_got): Use s390_got_symbol.
18823 (s390_option_override): Issue error if
18824 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18825 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18827 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18829 2017-06-27 Andrew Pinski <apinski@cavium.com>
18831 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18832 (X * copysign (1.0, X)): New pattern.
18833 (X * copysign (1.0, -X)): New pattern.
18834 (copysign (-1.0, CST)): New pattern.
18836 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18838 * genmultilib (combination_space): Remove variable.
18839 Validate reuse rules against regular expression for any sequence
18840 of multilib options in any order.
18842 2017-06-27 Michael Collison <michael.collison@arm.com>
18844 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18845 call aarch64_split_simd_combine.
18846 * (aarch64_combine_internal<mode>): Delete pattern.
18847 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18848 Allow register and subreg operands.
18850 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18852 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18853 specific need, just fallback on defaults.
18854 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18856 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18857 Olivier Hainque <hainque@adacore.com>
18859 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18861 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18862 targets. Pick a default if no particular attempt applied.
18863 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18866 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18868 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18869 (x86_64-wrs-vxworks7): Likewise.
18871 2017-06-27 Marek Polacek <polacek@redhat.com>
18874 * ubsan.c (instrument_null): Check get_base_address's result for null.
18876 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18878 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18880 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18882 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18883 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18884 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18885 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18886 New function types.
18887 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18888 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18889 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18890 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18891 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18892 BUILT_IN_FEUPDATEENV): New builtins.
18893 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18894 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18895 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18896 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18898 (builtin_structptr_types): Adjust size.
18899 * tree.c (builtin_structptr_types): Add four entries.
18901 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18902 Olivier Hainque <hainque@adacore.com>
18904 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18905 (TLS_SYM): New local macro, forcing reference to __tls__ on
18906 link command lines for VxWorks 7 RTPs, triggering initialization
18908 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18909 OS features TLS support, true for RTPs on VxWorks 7.
18910 * config/vxworks.c (vxworks_override_options): Setup emutls
18913 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18915 * predict.c (test_prediction_value_range): Use -1U instead of -1
18916 to avoid narrowing conversion warning.
18917 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18918 to avoid narrowing conversion warning.
18919 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18921 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18923 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18925 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18926 64bit configurations.
18927 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18928 (SIZE_TYPE): Likewise.
18929 * config/vxworks.c (vxworks_emutls_var_fields): Use
18930 long_unsigned_type_node instead of unsigned_type_node as the offset
18931 field type, which is "pointer" mode in emutls.c.
18933 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18936 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18938 PR middle-end/81207
18939 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18940 gimple_vuse copying separately from gimple_vdef copying.
18942 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18944 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18947 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18948 Olivier Hainque <hainque@adacore.com>
18950 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18951 for all vxworks7 targets.
18952 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18953 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18954 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18955 variations for VX6/VX7 and 32/64bits later on in ...
18956 (VXWORKS_LIB_SPEC): Leverage new macros.
18957 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18958 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18960 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18962 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18963 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18965 2017-06-26 Carl Love <cel@us.ibm.com>
18967 * config/rs6000/rs6000-c.c: Add support for built-in functions
18968 vector bool char vec_reve (vector bool char);
18969 vector signed char vec_reve (vector signed char);
18970 vector unsigned char vec_reve (vector unsigned char);
18971 vector bool int vec_reve (vector bool int);
18972 vector signed int vec_reve (vector signed int);
18973 vector unsigned int vec_reve (vector unsigned int);
18974 vector bool long long vec_reve (vector bool long long);
18975 vector signed long long vec_reve (vector signed long long);
18976 vector unsigned long long vec_reve (vector unsigned long long);
18977 vector bool short vec_reve (vector bool short);
18978 vector signed short vec_reve (vector signed short);
18979 vector double vec_reve (vector double);
18980 vector float vec_reve (vector float);
18981 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18982 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18983 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18984 (altivec_vreve): New pattern.
18985 * config/rs6000/altivec.h (vec_reve): New define.
18986 * doc/extend.texi (vec_rev): Update the built-in documentation file
18987 for the new built-in functions.
18989 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18991 PR tree-optimization/71815
18992 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18994 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18996 (slsr_process_phi): Likewise.
18997 (replace_uncond_cands_and_profitable_phis): Don't replace a
18998 multiply candidate with a stride of 1 (copy or cast).
18999 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19001 (lowest_cost_path): Likewise.
19002 (total_savings): Likewise.
19004 2017-06-26 Richard Biener <rguenther@suse.de>
19007 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19008 Use def_builtin_pure for all gather builtins.
19010 2017-06-26 Richard Biener <rguenther@suse.de>
19012 PR tree-optimization/81203
19013 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19014 non-dominating BBs.
19016 2017-06-26 Marek Polacek <polacek@redhat.com>
19019 * doc/invoke.texi: Document -Wmultistatement-macros.
19021 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19023 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19024 arm_neon_ok_no_float_abi effective target.
19026 2017-06-26 Richard Biener <rguenther@suse.de>
19028 PR tree-optimization/80928
19029 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19030 (copy_bbs): Set BB_DUPLICATED flag early.
19031 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19033 (execute_on_shrinking_pred): Likewise.
19034 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19035 BB_DUPLICATED blocks.
19036 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19037 iterate over all PHIs considering removal of *gsi.
19039 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19041 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19044 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19046 * config/rs6000/rs6000-string.c: (expand_block_clear,
19047 do_load_for_compare, select_block_compare_mode,
19048 compute_current_alignment, expand_block_compare,
19049 expand_strncmp_align_check, expand_strn_compare,
19050 expand_block_move, rs6000_output_load_multiple)
19051 Move functions related to string/block move/compare
19052 to a separate file.
19053 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19054 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19055 for this function which is now used in two files.
19056 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19057 * config.gcc: Add rs6000-string.o to extra_objs for
19058 targets powerpc*-*-* and rs6000*-*-*.
19060 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19063 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19064 32-bit, since indexed is not valid for DImode.
19065 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19066 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19067 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19068 (define_peephole2 for Altivec d-form store): Likewise.
19071 * multiple_target.c (create_dispatcher_calls): Only create the
19072 dispatcher call if the function is the default clone of a
19073 versioned function.
19075 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19077 PR middle-end/80902
19078 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19079 a call, force the call to not be a tail call.
19081 2017-06-23 Jeff Law <law@redhat.com>
19083 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19086 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19088 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19089 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19090 handling for early expansion of vector shifts (sl,sr,sra,rl).
19091 (builtin_function_type): Add vector shift right instructions
19092 to the unsigned argument list.
19094 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19096 rtl-optimizatoin/79286
19097 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19098 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19099 trap. PIC register plus a const unspec without offset can never trap.
19101 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19103 * tree.h (builtin_structptr_type): New type.
19104 (builtin_structptr_types): Declare new array.
19105 * tree.c (builtin_structptr_types): New array.
19106 (free_lang_data, build_common_tree_nodes): Use it.
19108 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19111 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19114 2017-06-22 Matt Turner <mattst88@gmail.com>
19116 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19117 Lake models to skylake case. Assume skylake for unknown
19118 models with clflushopt.
19120 2017-06-22 Jeff Law <law@redhat.com>
19122 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19123 frame sizes that do not satisfy aarch64_uimm12_shift.
19125 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19127 * profile-count.h (apply_probability,
19128 apply_scale, probability_in): Fix checks for zero.
19130 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19132 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19133 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19135 2016-06-22 Richard Biener <rguenther@suse.de>
19137 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19138 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19139 REDUC_MAX_EXPR support.
19140 (vectorizable_reduction): Likewise.
19141 (vect_create_epilog_for_reduction): Likewise.
19143 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19145 * match.pd (A / (1 << B) -> A >> B): New.
19146 * generic-match-head.c: Include optabs-tree.h.
19147 * gimple-match-head.c: Likewise.
19148 * optabs-tree.h (target_supports_op_p): New.
19149 * optabs-tree.c (target_supports_op_p): New.
19151 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19153 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19154 $gcc_cv_ld --help output.
19155 (gcc_cv_ld_demangle): Likewise.
19156 (gcc_cv_ld_eh_frame_hdr): Likewise.
19157 (gcc_cv_ld_pie): Likewise.
19158 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19159 (gcc_cv_ld_buildid): Likewise.
19160 (gcc_cv_ld_sysroot): Likewise.
19161 (ld_bndplt_support): Likewise.
19162 (ld_pushpopstate_support): Likewise.
19163 * configure: Regenerate.
19164 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19166 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19169 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19170 operands indexes to avoid gap between operands and match_dups.
19172 2017-06-21 Andrew Pinski <apinski@cavium.com>
19174 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19175 Increment Arith_shift and Arith_shift_reg by 1.
19176 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19178 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19179 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19180 (aarch64_strip_extend): Add new argument and test for it.
19181 (aarch64_cheap_mult_shift_p): New function.
19182 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19183 add a cost if it is true.
19184 Update calls to aarch64_strip_extend.
19185 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19187 2017-06-21 Andrew Pinski <apinski@cavium.com>
19189 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19191 (thunderxt88): Likewise.
19192 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19193 (thunderx_prefetch_tune): New variable.
19194 (thunderx2t99_prefetch_tune): Update for the correct values.
19195 (thunderxt88_tunings): New variable.
19196 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19197 generic_prefetch_tune.
19198 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19200 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19202 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19203 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19204 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19205 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19206 (aarch64_atomic_cas<mode>, GPI): Likewise.
19208 2017-06-21 Martin Liska <mliska@suse.cz>
19210 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19211 statements on cold and hot labels.
19212 * predict.c (tree_estimate_probability_bb): Remove the
19213 prediction from this place.
19215 2017-06-21 Martin Liska <mliska@suse.cz>
19217 PR tree-optimization/79489
19218 * gimplify.c (maybe_add_early_return_predict_stmt): New
19220 (gimplify_return_expr): Call the function.
19221 * predict.c (tree_estimate_probability_bb): Remove handling
19223 * predict.def: Update comment about early return predictor.
19224 * gimple-predict.h (is_gimple_predict): New function.
19225 * predict.def: Change default value of early return to 66.
19226 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19228 * passes.def: Put pass_strip_predict_hints to the beginning of
19231 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19233 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19234 FUNCTION_DECL declarations.
19235 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19237 (dwaf2out_decl): Likewise.
19238 * godump.c (go_early_global_decl): Skip call to the real debug hook
19239 for FUNCTION_DECL declarations.
19240 * passes.c (rest_of_decl_compilation): Skip call to the
19241 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19242 -fdump-go-spec is passed.
19244 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19246 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19247 Reorder for compactness.
19248 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19249 (def_builtin_pure, def_builtin_pure2): New functions.
19250 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19252 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19254 * match.pd (nop_convert): New predicate.
19255 ((A +- CST1) +- CST2): Allow some NOP conversions.
19257 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19260 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19261 with ctors/dtors if GOVD_SHARED is set.
19263 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19265 * config/aarch64/aarch64.md (movti_aarch64):
19266 Emit mov rather than orr.
19267 (movtf_aarch64): Likewise.
19268 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19269 Emit mov rather than orr.
19271 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19273 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19274 Swap alternatives, make integer dup more expensive.
19276 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19278 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19279 Return true for non-tls symbols.
19281 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19283 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19284 (cortex-a75): Likewise.
19285 (cortex-a75.cortex-a55): Likewise.
19286 * config/aarch64/aarch64-tune.md: Regenerate.
19287 * doc/invoke.texi (-mtune): Document new values for -mtune.
19289 2017-06-21 Tom de Vries <tom@codesourcery.com>
19291 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19292 stack_size feature.
19293 (Effective-Target Keywords, Other attributes): Suggest using
19294 dg-add-options stack_size feature to get stack limit in stack_size
19295 effective target documentation.
19297 2017-06-21 Julian Brown <julian@codesourcery.com>
19298 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19300 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19301 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19302 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19304 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19305 attribute type list for neon_multiply.
19306 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19307 attribute type list for neon_multiply.
19308 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19309 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19310 attribute type list for neon_multiply.
19311 * config/arm/types.md (crypto_pmull): Add.
19312 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19313 attribute type list.
19315 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19317 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19320 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19322 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19323 to make sure not to dereference a NULL cost_classes_ptr pointer.
19325 2017-06-20 Carl Love <cel@us.ibm.com>
19327 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19328 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19329 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19330 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19331 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19332 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19333 VMULOSW): New enum "unspec" values.
19334 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19335 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19336 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19337 altivec_vmulosw): New patterns.
19338 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19339 VMULOSW): Add definitions.
19341 2017-06-20 Julia Koval <julia.koval@intel.com>
19343 * config/i386/i386.c: Fix rounding expand for new pattern.
19344 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19346 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19348 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19349 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19351 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19353 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19356 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19358 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19359 architecture, then by implementer ID.
19360 * config/aarch64/aarch64-tune.md: Regenerate.
19362 2017-06-20 Richard Biener <rguenther@suse.de>
19364 PR middle-end/81097
19365 * fold-const.c (split_tree): Fold to type before negating.
19367 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19369 * diagnostic-show-locus.c
19370 (selftest::test_fixit_deletion_affecting_newline): New function.
19371 (selftest::diagnostic_show_locus_c_tests): Call it.
19373 2017-06-20 Andreas Schwab <schwab@suse.de>
19376 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19379 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19381 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19382 __ARM_FEATURE_COPROC according to support.
19384 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19386 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19387 Rewritten to avoid overflow for > 32-bit pointers.
19390 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19391 by removing enum keyword.
19392 (ubsan_type_descriptor): Likewise. Formatting fix.
19395 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19396 splitter): Require TARGET_SSE2 in the condition.
19398 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19401 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19402 for doing vector set of SFmode on ISA 3.0.
19403 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19404 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19406 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19407 SFmode value into a V4SF variable that was extracted from another
19408 V4SF variable without converting the element to double precision
19409 and back to single precision vector format.
19410 (vsx_insert_extract_v4sf_p9_2): Likewise.
19412 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19414 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19415 in UWHI to avoid undefined overflow.
19418 * ubsan.h (enum ubsan_encode_value_phase): New.
19419 (ubsan_encode_value): Change second argument to
19420 enum ubsan_encode_value_phase with default value of
19421 UBSAN_ENCODE_VALUE_GENERIC.
19422 * ubsan.c (ubsan_encode_value): Change second argument to
19423 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19424 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19425 create_tmp_var_raw instead of create_tmp_var and use a
19427 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19428 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19429 ubsan_encode_value callers.
19432 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19433 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19434 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19436 2017-06-19 Richard Biener <rguenther@suse.de>
19438 PR middle-end/81118
19439 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19440 estimates if we changed anything.
19442 2017-06-19 Richard Biener <rguenther@suse.de>
19444 PR tree-optimization/80887
19445 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19446 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19447 simplified lookups, then reset mprts_hook.
19448 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19450 (try_to_simplify): Likewise.
19452 2017-06-19 Martin Liska <mliska@suse.cz>
19455 * gimplify.c (gimplify_switch_expr):
19456 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19458 2017-06-19 Martin Liska <mliska@suse.cz>
19460 * doc/install.texi: Document that PGO runs in 4 stages.
19462 2017-06-19 Martin Liska <mliska@suse.cz>
19465 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19466 to dispatcher function name.
19467 * multiple_target.c (replace_function_decl): New function.
19468 (create_dispatcher_calls): Redirect both edges and references.
19470 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19472 * profile-count.c (profile_count::dump): Dump quality.
19473 (profile_count::differs_from_p): Update for unsigned val.
19474 * profile-count.h (profile_count_quality): New enum.
19475 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19477 2017-06-19 Richard Biener <rguenther@suse.de>
19479 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19480 struct function as arg.
19481 (estimate_numbers_of_iterations): Export overload with loop arg.
19482 (free_numbers_of_iterations_estimates_loop): Use an overload of
19483 free_numbers_of_iterations_estimates instead.
19484 * tree-cfg.c (remove_bb): Adjust.
19485 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19486 * tree-parloops.c (gen_parallel_loop): Likewise.
19487 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19489 (tree_unroll_loops_completely): Likewise.
19490 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19491 Use an overload instead and export.
19492 (estimated_loop_iterations): Adjust.
19493 (max_loop_iterations): Likewise.
19494 (likely_max_loop_iterations): Likewise.
19495 (estimate_numbers_of_iterations): Take struct function as arg
19497 (loop_exits_before_overflow): Adjust.
19498 (free_numbers_of_iterations_estimates_loop): Use an overload.
19499 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19500 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19502 2017-06-19 Richard Biener <rguenther@suse.de>
19505 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19506 RANGE_EXPR conservatively.
19508 2017-06-16 Carl Love <cel@us.ibm.com>
19510 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19511 definitions for vec_float, vec_float2, vec_floato,
19512 vec_floate built-ins.
19513 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19514 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19516 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19517 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19518 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19519 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19520 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19521 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19522 vec_floato): Add builtin defines.
19523 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19524 Update the built-in documentation file for the new built-in
19527 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19529 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19530 (mthumb): Mark as the negative of -marm.
19532 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19534 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19536 (ARM Options, -mtune): Document that this accepts the same
19537 extension options as -mcpu.
19538 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19540 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19542 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19543 permitted extensions.
19545 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19547 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19548 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19549 (armv8-m.main): Add option +nodsp.
19550 * config/arm/arm-cpu-cdata.h: Regenerated.
19552 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19554 * config/arm/t-fuchsia: New file.
19555 * config.gcc (arm*-*-fuchsia*): Use it.
19557 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19559 * config/arm/t-symbian: Rewrite for new option infrastructure.
19561 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19563 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19564 (MULTILIB_REQUIRED): Likewise.
19566 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19568 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19569 (MULTILIB_RESUE): Likewise.
19570 (MULTILIB_MATCHES): Likewise.
19571 (MULTLIB_REQUIRED): Likewise.
19573 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19575 * config/arm/t-rtems: Rewrite for new option framework.
19577 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19579 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19580 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19581 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19582 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19583 * config/arm/t-multilib: ... here.
19584 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19585 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19586 armv7-a and armv8*-a when A-profile libraries have not been built.
19587 * config/arm/t-rmprofile: Rewrite.
19589 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19591 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19592 with a backslash. Remove the backslash after substituting unescaped
19594 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19598 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19599 the driver to pass through the default setting of -mfloat-abi.
19600 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19602 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19603 (all_feat_combs): New rule.
19604 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19606 * config/arm/t-aprofile: Rewrite.
19608 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19610 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19611 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19612 fpu is not specified by the user/command-line.
19613 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19614 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19615 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19616 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19617 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19618 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19620 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19622 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19623 * config/arm/t-arm-elf: Rewritten.
19625 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19627 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19628 have some floating-point instructions.
19629 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19630 (TARGET_MAYBE_HARD_FLOAT): New macro.
19631 * config/arm/arm-builtins.c (arm_init_builtins): Use
19632 TARGET_MAYBE_HARD_FLOAT.
19633 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19635 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19637 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19638 (configargs.h): Include it.
19639 (arm_print_hint_for_fpu_option): New function.
19640 (arm_parse_fpu_option): New function.
19641 (candidate_extension): New class.
19642 (arm_canon_for_multilib): New function.
19643 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19644 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19645 (ARCH_CANONICAL_SPECS): New macro.
19646 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19648 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19650 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19651 are set after handling multilib fragments. Set target_cpu_default2
19654 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19656 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19658 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19659 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19661 (check_arch): Likewise.
19662 * config/arm/arm.c (arm_configure_build_target): Handle
19663 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19664 options in the default.
19666 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19668 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19669 when an option is an alias of another.
19670 * config/arm/parsecpu.awk (optalias): New parser token.
19671 (gen_comm_data): Mark non-alias options as such. Emit entries
19672 for extension aliases.
19673 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19674 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19675 (armv6kz, armv6zk, armv6t2): Likewise.
19676 (armv7): Make vfpv3-d16 an alias.
19677 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19679 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19680 Sort in canonical order.
19681 (armv8-a): Sort in canonical order.
19682 (armv8.1-a, armv8.2-a): Likewise.
19683 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19685 (cortex-a9): Sort in canonical order.
19686 * config/arm/arm.c (selftests.h): Include it.
19687 (arm_test_cpu_arch_data): New function.
19688 (arm_run_self_tests): New function.
19689 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19690 (targetm): Move declaration to the end of the file.
19691 * arm-cpu-cdata.h: Regenerated.
19693 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19695 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19696 call to target_mode_check describing the type of option passed.
19697 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19698 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19699 arm_parse_cpu_option_name to match parameters against list of
19701 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19702 arm_arch_core_flags data structure.
19703 * config/arm/arm-cpu_cdata.h: Regenerated.
19705 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19707 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19709 (arm_print_hint_for_cpu_option): Likewise.
19710 (arm_print_hint_for_arch_option): Likewise.
19711 (arm_parse_cpu_option_name): Likewise.
19712 (arm_parse_arch_option_name): Likewise.
19713 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19714 of entries in the all_fpus list.
19715 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19716 (arm_parse_cpu_option_name): Declare.
19717 (arm_parse_arch_option_name): Declare.
19718 (arm_parse_option_features): Declare.
19719 (arm_intialize_isa): Declare.
19720 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19722 (gen_comm_data): ... here. Make definitions non-static.
19723 * config/arm/arm-cpu-data.h: Regenerated.
19724 * config/arm/arm-cpu-cdata.h: Regenerated.
19726 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19728 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19729 (cpu_arch_extension): New structure.
19730 (cpu_arch_option, arch_option, cpu_option): New structures.
19731 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19732 architecture types.
19733 (gen_data): Generate new format data tables.
19734 * config/arm/arm.c (cpu_tune): New structure.
19735 (cpu_option, processors): Delete.
19736 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19737 (arm_print_hint_for_cpu_option): ... this and ...
19738 (arm_print_hint_for_arch_option): ... this.
19739 (arm_parse_arch_cpu_name): Delete. Replace with ...
19740 (arm_parse_cpu_option_name): ... this and ...
19741 (arm_parse_arch_option_name): ... this.
19742 (arm_unrecognized_feature): Change type of target parameter to
19744 (arm_parse_arch_cpu_features): Delete. Replace with ...
19745 (arm_parse_option_features): ... this.
19746 (arm_configure_build_target): Rework to use new configuration data
19748 (arm_print_tune_info): Rework for new configuration data tables.
19749 * config/arm/arm-cpu-data.h: Regenerated.
19750 * config/arm/arm-cpu.h: Regenerated.
19752 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19754 * Makefile.in (OBJS): Move sbitmap.o from here ...
19755 (OBJS-libcommon): ... to here.
19757 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19759 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19760 (ISA_ALL_CRYPTO): New macro.
19761 (ISA_ALL_SIMD): New macro
19762 (ISA_ALL_FP): New macro.
19763 * config/arm/arm.c (fpu_bitlist): Update initializer.
19764 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19766 (arm9e): Add fpu. Add option for nofp
19767 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19768 (arm926ej-s, arm1026ej-s): Likewise.
19769 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19770 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19771 neon-fp16, neon-vfpv4, nofp and nosimd.
19772 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19773 (cortex-a8): Add fpu. Add option for nofp.
19774 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19775 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19776 (cortex-r4f): Add fpu.
19777 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19778 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19780 (cortex-r8): Likewise.
19781 (cortex-m4): Add fpu. Add option for nofp.
19782 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19783 (cortex-a17.cortex-a7): Likewise.
19784 (cortex-a32): Add fpu. Add options for crypto and nofp.
19785 (cortex-a35, cortex-a53): Likewise.
19786 (cortex-a57): Add fpu. Add option for crypto.
19787 (cortex-a72, cortex-a73): Likewise.
19788 (exynos-m1): Likewise.
19789 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19790 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19791 (cortex-m33): Add fpu. Add option for nofp.
19792 * config/arm/arm-cpu-cdata.h: Regenerated
19793 * config/arm/arm-cpu-data.h: Regenerated.
19795 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19797 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19798 (armv5te, armv5tej): Likewise.
19799 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19800 (armv7): Add options fp and vfpv3-d16.
19801 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19802 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19804 (armv7ve): Likewise.
19805 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19806 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19807 (armv8-a): Add nocrypto option.
19808 (armv8.1-a, armv8.2-a): Likewise.
19809 (armv8-m.main): add options fp, fp.dp and nofp.
19811 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19813 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19815 (armv8-a+crc): Delete.
19816 (armv8.1-a): Add options simd, crypto and nofp.
19817 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19818 (armv8.2-a+fp16): Delete.
19819 (armv8-m.main): Add option dsp.
19820 (armv8-m.main+dsp): Delete.
19821 (cortex-a8): Add fpu. Add nofp option.
19822 (cortex-a9): Add fpu. Add nofp and nosimd options.
19823 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19824 link to main cpu and architecture data structures.
19825 (gen_comm_data): Only put isa attributes from the main architecture
19827 (option): New statement for architecture and CPU entries.
19828 * arm.c (struct cpu_option): New structure.
19829 (struct processors): Add entry for options.
19830 (arm_unrecognized_feature): New function.
19831 (arm_parse_arch_cpu_name): Ignore any characters after the first
19833 (arm_parse_arch_cpu_feature): New function.
19834 (arm_configure_build_target): Separate out any CPU and architecture
19835 features and parse separately. Don't error out if -mfpu=auto is
19836 used with only an architecture string.
19837 (arm_print_asm_arch_directives): New function.
19838 (arm_file_start): Call it.
19839 * config/arm/arm-cpu-cdata.h: Regenerated.
19840 * config/arm/arm-cpu-data.h: Likewise.
19841 * config/arm/arm-tables.opt: Likewise.
19843 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19845 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19846 assembler when it is not -mfpu=auto.
19848 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19850 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19851 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19852 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19853 (ASM_CPU_SPEC): Rewrite.
19854 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19855 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19856 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19857 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19858 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19859 copied string is NUL-terminated. Also strip any characters prefixed
19861 (arm_rewrite_selected_arch): New function.
19862 (arm_rewrite_march): New function.
19864 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19866 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19867 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19868 (march, mcpu, mtune): Convert to string-based options.
19869 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19870 (arm_parse_arch_cpu_name): New function.
19871 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19872 identify selected architecture or CPU.
19873 (arm_option_save): New function.
19874 (TARGET_OPTION_SAVE): Redefine.
19875 (arm_option_restore): Restore string options.
19876 (arm_option_print): Print string options.
19878 2017-06-16 Martin Sebor <msebor@redhat.com>
19880 PR tree-optimization/80933
19881 PR tree-optimization/80934
19882 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19883 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19884 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19885 (gimple_fold_builtin): Call them.
19887 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19889 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19890 as unlikely; update profile.
19892 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19894 * predict.c (force_edge_cold): Handle declaring edges impossible
19897 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19899 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19901 (try_unroll_loop_completely): Fix reporting.
19903 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19905 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19907 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19910 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19911 if given a non-constant argument for an intrinsic which requires a
19914 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19916 * profile.c (compare_freqs): New function.
19917 (branch_prob): Sort edge list.
19918 (find_spanning_tree): Assume that the list is priority sorted.
19920 2017-06-16 Richard Biener <rguenther@suse.de>
19922 PR tree-optimization/81090
19923 * passes.def (pass_record_bounds): Remove.
19924 * tree-pass.h (make_pass_record_bounds): Likewise.
19925 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19926 make_pass_record_bounds): Likewise.
19927 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19928 not free niter estimates at the beginning but at the end.
19929 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19931 2017-06-16 Richard Biener <rguenther@suse.de>
19933 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19934 initializer to workaround ICE in host GCC 4.8.
19936 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19938 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19940 (clone_inlined_nodes): Update.
19942 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19944 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19945 prefetch settings, and enable prefetching by default at -O3.
19947 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19949 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19950 Set flag_prefetch_loop_arrays according to tuning data.
19952 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19954 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19955 New tune structure.
19956 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19957 [Unrelated to main purpose of the patch] Place the pointer field last
19958 to enable type checking errors when tune structure are wrongly merged.
19959 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19960 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19961 (thunderx2t99_prefetch_tune): New tune constants.
19962 (tune_params *_tunings): Update all tunings (no functional change).
19963 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19964 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19965 from tunings structures.
19967 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19970 * ubsan.c (instrument_null): Add T argument, use it instead
19971 of computing it based on IS_LHS.
19972 (instrument_object_size): Likewise.
19973 (pass_ubsan::execute): Adjust instrument_null and
19974 instrument_object_size callers to pass gimple_get_lhs or
19975 gimple_assign_rhs1 result to it. Use instrument_null instead of
19976 calling get_base_address and instrument_mem_ref. Handle
19977 aggregate call arguments for object-size sanitization.
19979 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19981 PR tree-optimization/81089
19982 * tree-vrp.c (is_masked_range_test): Validate operands of
19985 2017-06-15 Martin Sebor <msebor@redhat.com>
19988 * dumpfile.c (dump_register): Avoid calling memset to initialize
19989 a class with a default ctor.
19990 * gcc.c (struct compiler): Remove const qualification.
19991 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19992 * hash-table.h: Ditto.
19993 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19995 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19996 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19998 * params.h (struct param_info): Make struct members non-const.
19999 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20000 with default initialization.
20001 * vec.h (vec_copy_construct, vec_default_construct): New helper
20003 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20004 with vec_copy_construct.
20005 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20006 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20007 * doc/invoke.texi (-Wclass-memaccess): Document.
20009 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20011 * emit-rtl.h (is_leaf): Update comment about local
20012 register allocator.
20014 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20017 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20018 for a variable to have a section before checking if the section has a
20020 Set section to.persistent if persistent attribute is set.
20021 Warn if .persistent attribute is used on an automatic variable.
20023 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20025 PR rtl-optimization/80474
20026 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20028 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20030 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20033 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20035 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20036 (casesi): Emit bounds checking as RTL.
20037 (casesi_internal_mips16_<mode>): Remove bounds checking.
20039 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20041 * config/xtensa/xtensa.c (xtensa_option_override): Append
20042 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20043 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20044 xtensa_doloop_hooks): Define unconditionally.
20045 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20047 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20048 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20049 for it in xtensa_option_override.
20050 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20051 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20053 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20055 * doc/cppopts.texi: Document '-' special value to -MF.
20057 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20059 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20060 (cortex_a53_fconst): Likewise.
20061 (cortex_a53_fpmul): Likewise.
20062 (cortex_a53_f_load_64): Likewise.
20063 (cortex_a53_f_load_many): Likewise.
20064 (cortex_a53_advsimd_alu): Likewise.
20065 (cortex_a53_advsimd_alu_q): Likewise.
20066 (cortex_a53_advsimd_mul): Likewise.
20067 (cortex_a53_advsimd_mul_q): Likewise.
20068 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20069 Add missing fmul, r2f_cvt and fconst cases.
20071 2017-06-14 Richard Biener <rguenther@suse.de>
20073 PR middle-end/81088
20074 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20076 (fold_binary_loc): When associating do not treat pre-existing
20077 TREE_OVERFLOW on literal constants as a reason to allow
20078 TREE_OVERFLOW on associated literal constants.
20080 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20082 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20083 (MASK_FEATURES): New macro.
20084 * config/sparc/sparc.c (sparc_option_override): Remove the special
20085 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20087 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20089 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20090 a division of 0 if non-call exceptions are enabled.
20092 2017-06-14 Andrew Pinski <apinski@cavium.com>
20093 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20096 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20097 Improve vector initialization code gen for only variable case.
20099 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20101 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20103 2017-06-14 Richard Biener <rguenther@suse.de>
20105 PR tree-optimization/81083
20106 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20109 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20111 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20112 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20113 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20114 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20115 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20116 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20118 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20120 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20121 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20123 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20125 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20127 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20129 * config/rs6000/t-rtems: Don't handle SPE.
20131 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20133 * config/rs6000/t-linux: Don't handle SPE.
20135 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20137 * config/rs6000/eabispe.h: Delete file.
20139 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20141 * config/rs6000/t-spe: Delete file.
20143 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20145 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20146 (rs6000_legitimate_offset_address_p): Return false for anything in
20147 V2SImode or V2SFmode.
20149 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20151 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20152 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20153 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20155 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20156 (rs6000_legitimate_offset_address_p): Ditto.
20157 (rs6000_emit_move): Ditto.
20158 (rs6000_init_builtins): Remove V4HI_type_node.
20160 2017-06-13 Martin Liska <mliska@suse.cz>
20163 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20164 (gate_asan): Likewise.
20165 * asan.h (asan_no_sanitize_address_p): Remove the function.
20166 (sanitize_flags_p): New function.
20167 * builtins.def: Fix coding style.
20168 * common.opt: Use renamed enum value.
20169 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20170 * doc/extend.texi: Document no_sanitize attribute.
20171 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20172 to SANITIZE_UNDEFINED_NONDEFAULT.
20173 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20174 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20175 Use sanitize_flags_p.
20176 * gimplify.c (gimplify_function_tree): Likewise.
20177 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20178 * opts.c (parse_no_sanitize_attribute): New function.
20179 (common_handle_option): Use renamed enum value.
20180 * opts.h (parse_no_sanitize_attribute): Declare.
20181 * tree.c (sanitize_flags_p): New function.
20182 * tree.h: Declared here.
20183 * tsan.c: Use sanitize_flags_p.
20184 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20185 (instrument_mem_ref): Likewise.
20186 (instrument_bool_enum_load): Likewise.
20187 (do_ubsan_in_current_function): Remove the function.
20188 (pass_ubsan::execute): Use sanitize_flags_p.
20189 * ubsan.h: Remove do_ubsan_in_current_function
20190 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20191 (dump_function_to_file): Use it here.
20193 2017-06-13 Martin Jambor <mjambor@suse.cz>
20195 PR tree-optimization/80803
20196 PR tree-optimization/81063
20197 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20198 (propagate_subaccesses_across_link): Enqueue subtree whenever
20199 necessary instead of relying on the caller.
20201 2017-06-13 Martin Jambor <mjambor@suse.cz>
20203 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20204 that have a first_link.
20205 (sort_and_splice_var_accesses): Do not check first_link before
20207 (subtree_mark_written_and_enqueue): Likewise.
20208 (propagate_all_subaccesses): Likewise and do not stop at first
20209 parent with a first_link.
20211 2017-06-13 Martin Jambor <mjambor@suse.cz>
20213 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20216 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20218 * match.pd: New pattern.
20220 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20222 * tree-vrp.c (is_masked_range_test): New function.
20223 (register_edge_assert_for): Determine ranges for
20226 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20228 PR tree-optimization/67328
20229 * fold-const.c (maskable_range_p): New function.
20230 (build_range_check): Generate bittests if possible.
20232 2017-06-13 Martin Liska <mliska@suse.cz>
20234 * gimple-pretty-print.c (dump_probability): Add new argument.
20235 (dump_edge_probability): Dump both probability and count.
20236 (dump_gimple_label): Likewise.
20237 (dump_gimple_bb_header): Likewise.
20239 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20242 * config/avr/avr-devices.c: Fix indentation.
20243 * config/avr/gen-avr-mmcu-specs.c: Dito.
20245 2017-06-13 Richard Biener <rguenther@suse.de>
20247 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20248 instead get vector type from stmt_info.
20249 (vectorizable_reduction): Adjust. Remove dead code.
20251 2017-06-13 Richard Biener <rguenther@suse.de>
20253 PR middle-end/81065
20254 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20255 case of C * (x * C2 + C3).
20256 (fold_addr_of_array_ref_difference): Properly fold index difference.
20258 2017-06-12 David S. Miller <davem@davemloft.net>
20261 * config/sparc/sparc.md (return expander): Emit frame blockage if
20262 function uses alloca.
20264 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20266 * combine.c (make_field_assignment): Check len rather than the mode
20267 precision when calling force_to_mode.
20269 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20271 Support multilibs and devices that see flash in RAM address range.
20274 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20275 (avr_mcu_t) <flash_pm_offset>: New field.
20276 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20277 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20278 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20279 (AVR_TINY_PM_OFFSET): Remove macro.
20280 * config/avr/avr.opt (-mshort-calls): New option.
20281 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20282 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20283 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20284 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20285 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20286 instead of avr_arch->have_jmp_call.
20287 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20288 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20289 avr_arch->flash_pm_offset to define.
20290 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20291 new field flash_pm_offset. Add entry for avrxmega3.
20292 (avr_texinfo): Add entry for avrxmega3.
20293 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20294 attiny212, attiny214,
20295 attiny412, attiny414, attiny416, attiny417,
20296 attiny814, attiny816, attiny817,
20297 attiny1614, attiny1616, attiny1617,
20298 attiny3214, attiny3216, attiny3217.
20299 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20300 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20301 (avr_print_operand_address) [AVR_TINY]: Same.
20302 (avr_asm_init_sections) <readonly_data_section>: Only patch
20303 callback if avr_arch->flash_pm_offset = 0.
20304 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20305 for rodata if avr_arch->flash_pm_offset != 0.
20306 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20307 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20308 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20309 (m_options): Append opt_rcall.
20310 (m_dirnames): Append dir_rcall.
20311 * config/avr/t-multilib: Regenerate.
20313 * configure.ac [target=avr]: Check whether avrxmega3 default
20314 linker description file works as needed.
20315 * configure: Regenerate.
20316 * doc/avr-mmcu.texi: Regenerate.
20317 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20318 <__AVR_ARCH__>: Document avrxmega3 and 103.
20319 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20320 <__AVR_SHORT_CALLS__>: Document it.
20321 <__AVR_PM_BASE_ADDRESS__>: Document it.
20322 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20323 (AVR Variable Attributes) <progmem>: Document this is
20324 not needed for avrxmega3.
20325 (AVR Named Address Spaces) <__flash>: Dito.
20327 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20329 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20331 2017-06-12 Doug Rupp <rupp@adacore.com>
20333 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20334 Append vxworks-stdint.h to the tm_file list.
20335 * config/vxworks-stdint.h: New file.
20337 2017-06-12 Martin Liska <mliska@suse.cz>
20339 PR tree-optimization/81041
20340 * tree-profile.c (gimple_gen_ic_func_profiler):
20341 Create an extra BB in profile-generate
20342 (gimple_gen_time_profiler): Likewise.
20344 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20346 PR tree-optimization/81003
20347 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20348 (update_range_test): Use it instead of force_gimple_operand_gsi.
20350 2017-06-12 Richard Biener <rguenther@suse.de>
20352 PR tree-optimization/81053
20353 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20354 with backedge value not defined in loop. Simplify def stmt
20357 2017-06-11 Tom de Vries <tom@codesourcery.com>
20360 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20362 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20363 nvptx_cannot_force_const_mem.
20365 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20367 * opts.c (finish_options): Move test for flag_split_stack after
20368 it has been initialized.
20370 2017-06-11 Jason Merrill <jason@redhat.com>
20372 * tree.h (id_equal): New.
20373 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20374 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20375 instead of strcmp of IDENTIFIER_POINTER.
20377 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20379 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20380 (mark_all_inlined_calls_cdtor): Fix formating.
20381 (inline_transform): Rescale profile before inlining.
20383 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20385 * cgraph.h (cgraph_edge::clone): Update prototype.
20386 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20387 (cgraph_node::create_clone): Update.
20388 (cgraph_node::create_version_clone): Update.
20389 * tree-inline.c (copy_bb): Update.
20390 (expand_call_inline): Update.
20392 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20394 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20395 factored out from ...
20396 (rs6000_emit_prologue): ... here.
20398 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20400 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20401 factored out from ...
20402 (rs6000_emit_prologue): ... here.
20404 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20406 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20408 (handle_missing_profiles): Fix computation of tp_first_run.
20409 (counts_to_freqs): Do not touch freqs when count is 0.
20411 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20413 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20416 2017-06-10 Tom de Vries <tom@codesourcery.com>
20418 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20419 attributes): Document signal effective target.
20421 2017-06-10 Tom de Vries <tom@codesourcery.com>
20423 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20424 Document effective target stack_size.
20426 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20428 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20429 to the edit_context if they can be auto-applied.
20431 2017-06-9 Ian Lance Taylor <iant@golang.org>
20433 * opts.c (finish_options): If -fsplit-stack, disable implicit
20434 -forder-blocks-and-partition.
20435 * doc/invoke.texi (Optimize Options): Document that when using
20436 -fsplit-stack -forder-blocks-and-partition is not implicitly
20439 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20441 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20442 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20443 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20444 * builtins.def (abort, trap, unreachable): Declare cold.
20445 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20446 * tree-core.h (ECF_COLD): New.
20447 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20448 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20450 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20452 * predict.c (unlikely_executed_stmt_p): Cleanup.
20454 2017-06-09 Richard Biener <rguenther@suse.de>
20456 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20457 model if the ref is always written to.
20459 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20461 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20463 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20465 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20468 2017-06-09 Tom de Vries <tom@codesourcery.com>
20471 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20472 "target cannot support label values" when encountering LABEL_REF.
20474 2017-06-09 Martin Liska <mliska@suse.cz>
20476 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20477 (gimple_gen_ic_func_profiler): Emit direct comparison
20478 of __gcov_indirect_call_callee with NULL.
20479 (gimple_gen_time_profiler): Change probability from
20480 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20482 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20484 * profile.c (edge_gcov_counts): Turn to pointer.
20485 (compute_branch_probabilities, compute_branch_probabilities): Update.
20486 (branch_prob): Do not clear edge_gcov_count.
20487 * profile.h (edge_gcov_counts): Turn to pointer.
20488 (edge_gcov_count): Update.
20490 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20492 * gimple.h (gimple_check_failed): Mark cold.
20494 2017-06-09 Richard Biener <rguenther@suse.de>
20496 PR tree-optimization/66623
20497 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20498 refactor check_reduction into two parts, properly computing
20499 whether we have to check reduction validity for outer loop
20502 2017-06-09 Richard Biener <rguenther@suse.de>
20504 PR tree-optimization/79483
20505 * graphite-scop-detection.c (order): New global.
20506 (get_order): Compute bb to order mapping that satisfies code
20507 generation constraints.
20508 (cmp_pbbs): New helper.
20509 (build_scops): Start domwalk at entry block, sort generated
20512 2017-06-09 Richard Biener <rguenther@suse.de>
20514 PR middle-end/81007
20515 * ipa-polymorphic-call.c
20516 (ipa_polymorphic_call_context::restrict_to_inner_class):
20517 Skip FIELD_DECLs with error_mark_node type.
20518 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20521 2017-06-09 Martin Liska <mliska@suse.cz>
20523 * predict.c (struct branch_predictor): New struct.
20524 (test_prediction_value_range): New test.
20525 (predict_c_tests): New function.
20526 * selftest-run-tests.c (selftest::run_tests): Run the function.
20527 * selftest.h: Declare new tests.
20529 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20532 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20533 gen_add3_insn did not fail.
20534 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20535 r0, construct that number in a temporary reg and add that reg to r0.
20536 If asked to put the result in r0 as well, fail.
20538 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20540 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20541 for early expansion of vec_eqv.
20543 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20545 PR middle-end/81005
20546 * ubsan.c (instrument_null): Avoid pointless code temporary.
20547 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20549 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20553 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20556 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20559 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20561 * predict.c (unlikely_executed_edge_p): Move ahead.
20562 (probably_never_executed_edge_p): Use it.
20564 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20566 PR middle-end/79988
20567 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20568 gimple_call_builtin_p call.
20570 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20572 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20573 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20574 rtl_check_failed_type2, rtl_check_failed_code1,
20575 rtl_check_failed_code2, rtl_check_failed_code_mode,
20576 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20577 rtvec_check_failed_bounds, rtl_check_failed_flag,
20578 _fatal_insn_not_found, _fatal_insn): Likewise.
20579 * tree.h (tree_contains_struct_check_failed,
20580 tree_check_failed, tree_not_check_failed,
20581 tree_class_check_failed, tree_range_check_failed,
20582 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20583 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20584 tree_operand_check_failed, omp_clause_check_failed,
20585 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20588 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20590 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20591 flag_branch_probabilities.
20592 * ipa-inline.c (edge_badness): Likewise.
20593 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20594 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20595 * predict.c (maybe_hot_frequency_p): Likewise.
20596 (probably_never_executed): Likewise.
20597 * sched-ebb.c (schedule_ebbs): Likewise.
20598 * sched-rgn.c (find_single_block_region): Likewise.
20599 * tracer.c (tail_duplicate): Likewise.
20601 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20603 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20604 longer requires x_flag_profile_use.
20606 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20608 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20609 instead of flag_reorder_blocks_and_partition.
20610 * dbxout.c (dbxout_function_end): Likewise.
20611 * dwarf2out.c (gen_subprogram_die): Likewise.
20612 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20613 * hw-doloop.c (reorg_loops): Likewise.
20614 * varasm.c (assemble_start_function,
20615 assemble_end_function): Likewise.
20616 (decide_function_section): Do not check for
20617 flag_reorder_blocks_and_partition.
20619 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20621 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20623 (chkp_get_hard_register_fake_addr_expr): Ditto.
20624 (chkp_build_addr_expr): Add check for hard reg case.
20625 (chkp_parse_array_and_component_ref): Ditto.
20626 (chkp_find_bounds_1): Ditto.
20627 (chkp_process_stmt): Don't generate bounds store for
20630 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20632 * predict.c (maybe_hot_bb_p): Do not check profile status.
20633 (maybe_hot_edge_p): Likewise.
20634 (probably_never_executed): Check for zero counts even if profile
20636 (unlikely_executed_edge_p): New function.
20637 (unlikely_executed_stmt_p): New function.
20638 (unlikely_executed_bb_p): New function.
20639 (set_even_probabilities): Use unlikely predicates.
20640 (combine_predictions_for_bb): Likewise.
20641 (predict_paths_for_bb): Likewise.
20642 (predict_paths_leading_to_edge): Likewise.
20643 (determine_unlikely_bbs): New function.
20644 (estimate_bb_frequencies): Use it.
20645 (compute_function_frequency): Use zero counts even if profile is
20647 * profile-count.h: Fix typo.
20649 2017-08-08 Julia Koval <julia.koval@intel.com>
20651 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20652 _mm512_mask_cvtsepi16_storeu_epi8,
20653 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20654 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20655 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20656 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20657 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20658 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20659 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20660 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20661 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20662 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20663 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20664 __builtin_ia32_pmovuswb256mem_mask,
20665 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20666 __builtin_ia32_pmovwb512mem_mask): New builtins.
20668 2017-08-08 Julia Koval <julia.koval@intel.com>
20670 PR target/73350,80862
20671 * config/i386/subst.md (round): Fix round pattern.
20672 * config/i386/i386.c (ix86_erase_embedded_rounding):
20673 Fix erasing rounding for the fixed pattern.
20675 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20677 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20679 2017-06-08 Martin Liska <mliska@suse.cz>
20681 PR gcov-profile/80911
20682 * gcov.c (block_info::block_info): New constructor.
20684 2017-06-07 Carl Love <cel@us.ibm.com>
20686 * config/rs6000/rs6000-c: The return type of the following
20687 built-in functions was implemented as int not long long. Fix sign
20688 of return value for the unsigned version of vec_mulo and vec_mule.
20689 vector unsigned long long vec_bperm (vector unsigned long long,
20690 vector unsigned char)
20691 vector signed long long vec_mule (vector signed int,
20693 vector unsigned long long vec_mule (vector unsigned int,
20694 vector unsigned int)
20695 vector signed long long vec_mulo (vector signed int,
20697 vector unsigned long long vec_mulo (vector unsigned int,
20698 vector unsigned int)
20699 * doc/extend.texi: Fix the documentation for the built-in
20702 2017-06-07 Carl Love <cel@us.ibm.com>
20705 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20708 2017-06-07 Carl Love <cel@us.ibm.com>
20710 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20711 support, Generate doublehv for signed int/float for BE case only.
20713 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20715 * doc/invoke.texi (mcx16): Rewrite.
20717 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20719 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20720 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20721 *mov<mode>_softfloat, and an anonymous splitter): Use
20722 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20724 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20726 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20728 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20729 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20730 (rs6000_debug_reg_global): Adjust.
20731 (rs6000_init_hard_regno_mode_ok): Adjust.
20732 (rs6000_dbx_register_number): Adjust.
20733 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20734 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20735 Remove SPE_ACC and SPEFSCR.
20736 (REG_ALLOC_ORDER): Ditto.
20737 (FRAME_POINTER_REGNUM): Change to 111.
20738 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20739 (REG_CLASS_NAMES): Ditto.
20740 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20741 (REGISTER_NAMES): Ditto.
20742 (ADDITIONAL_REG_NAMES): Ditto.
20743 (rs6000_reg_names): Ditto.
20744 * config/rs6000/rs6000.md: Renumber some register number
20747 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20749 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20751 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20752 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20754 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20755 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20756 Delete the SPE high registers.
20757 (REG_ALLOC_ORDER): Ditto.
20758 (enum reg_class): Remove SPE_HIGH_REGS.
20759 (REG_CLASS_NAMES): Ditto.
20760 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20761 (REGISTER_NAMES): Ditto.
20762 (rs6000_reg_names): Ditto.
20763 * doc/tm.texi.in: Remove SPE as example.
20764 * doc/tm.texi: Regenerate.
20766 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20768 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20769 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20770 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20771 * config/rs6000/rs6000.md (type): Remove "brinc".
20773 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20775 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20776 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20777 * config/rs6000/linuxspe.h: Delete file.
20778 * config/rs6000/rs6000.md: Don't include spe.md.
20779 * config/rs6000/spe.h: Delete file.
20780 * config/rs6000/spe.md: Delete file.
20781 * config/rs6000/t-rs6000: Remove spe.md.
20783 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20785 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20786 (reg_or_none500mem_operand): Delete.
20787 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20788 instead of reg_or_none500mem_operand.
20790 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20792 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20793 handling of SPE flags.
20794 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20796 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20798 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20800 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20801 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20802 paired_divv2sf3): Similar.
20803 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20804 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20805 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20807 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20808 Rename the paired_* instruction patterns.
20809 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20811 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20812 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20813 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20814 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20815 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20816 PAIRED_VECTOR_MODE.
20817 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20818 (spe_func_has_64bit_regs_p): Delete.
20819 (spe_expand_predicate_builtin): Delete.
20820 (spe_expand_evsel_builtin): Delete.
20821 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20822 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20823 (invalid_e500_subreg): Delete.
20824 (rs6000_legitimize_address): Always force_reg op2 as well, for
20825 paired single memory accesses.
20826 (rs6000_member_type_forces_blk): Delete.
20827 (rs6000_spe_function_arg): Delete.
20828 (rs6000_expand_unop_builtin): Delete SPE handling.
20829 (rs6000_expand_binop_builtin): Ditto.
20830 (spe_expand_stv_builtin): Delete.
20831 (bdesc_2arg_spe): Delete.
20832 (spe_expand_builtin): Delete.
20833 (spe_expand_predicate_builtin): Delete.
20834 (spe_expand_evsel_builtin): Delete.
20835 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20836 (spe_init_builtins): Delete.
20837 (spe_func_has_64bit_regs_p): Delete.
20838 (savres_routine_name): Delete "info" parameter. Adjust callers.
20839 (rs6000_emit_stack_reset): Ditto.
20840 (rs6000_dwarf_register_span): Delete.
20841 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20842 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20843 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20845 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20847 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20848 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20849 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20850 mulv2sf3, divv2sf3): Delete expanders.
20852 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20854 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20856 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20858 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20859 * config/rs6000/rs6000.c: Ditto.
20861 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20863 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20864 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20865 comparison_operator.
20867 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20869 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20870 * config/rs6000/rs6000.opt: Ditto.
20871 * config/rs6000/t-rtems: Ditto.
20873 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20875 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20876 TARGET_E500_SINGLE by 0, simplify.
20877 * config/rs6000/rs6000.c: Ditto.
20878 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20879 (spe_build_register_parallel): Delete.
20880 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20881 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20882 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20883 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20884 (E500_CONVERT): Delete.
20885 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20887 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20889 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20890 * config/rs6000/dfp.md: Ditto.
20891 (negdd2, *negdd2_fpr): Merge.
20892 (absdd2, *absdd2_fpr): Merge.
20893 (negtd2, *negtd2_fpr): Merge.
20894 (abstd2, *abstd2_fpr): Merge.
20895 * config/rs6000/e500.h: Delete file.
20896 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20897 TARGET_FPRS by 1 and simplify.
20898 * config/rs6000/rs6000-c.c: Ditto.
20899 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20900 TARGET_DF_SPE by 0.
20901 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20903 * config/rs6000/rs6000.md: Ditto.
20904 (floatdidf2, *floatdidf2_fpr): Merge.
20905 (move_from_CR_gt_bit): Delete.
20906 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20907 (E500_CR_IOR_COMPARE): Delete.
20908 (All patterns that require !TARGET_FPRS): Delete.
20909 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20911 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20913 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20915 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20917 * graphds.c (add_edge): Intitialize edge's attached data.
20918 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20919 pointer parameter. Call pointed function on each edge during
20920 graph traversing. Skip traversing the edge when the function
20922 (graphds_dfs, graphds_scc): Ditto.
20923 (for_each_edge): New parameter. Pass the new parameter to callback
20925 * graphds.h (skip_edge_callback): New function pointer type.
20926 (graphds_dfs, graphds_scc): New function pointer parameter.
20927 (graphds_edge_callback, for_each_edge): New parameter.
20929 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20931 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20932 out code checking if runtime alias check is possible to below ...
20933 Call the new function.
20934 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20935 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20937 2017-06-07 Marek Polacek <polacek@redhat.com>
20940 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20941 TYPE_OVERFLOW_WRAPS check.
20943 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20945 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20946 if versioning is required.
20947 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20948 peeling with the check for versioning.
20950 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20952 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20953 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20954 Set true to new parameter if new ssa variable is defined.
20955 (vect_gen_vector_loop_niters): Refactor. Set range information
20956 for the new vector loop bound variable.
20957 (vect_do_peeling): Ditto.
20959 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20961 * tree-affine.c (ssa.h): Include header file.
20962 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20963 has wrapping overflow behavior.
20965 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20967 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20969 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20971 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20972 (tree_to_aff_combination): ... here.
20974 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20976 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20977 reg_pressure model function.
20978 (ivopts_global_cost_for_size): Delete.
20979 (determine_set_costs, iv_ca_recount_cost): Call new model function
20980 ivopts_estimate_reg_pressure.
20982 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20984 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20985 expensive than udiv. Remove floating point cases from mod.
20987 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20989 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20990 Increase idiv cost.
20992 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20994 * config/aarch64/aarch64.md
20995 (copysignsf3): Fix mask generation.
20997 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20999 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21001 (class dump_manager): Add register_dumps method.
21002 * dumpfile.c: Include langhooks.h.
21003 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21004 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21005 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21006 (dump_manager::dump_register): Start with 512 entries instead of 32.
21007 (dump_manager::register_dumps): New method.
21008 * toplev.c (general_init): Instead of invoking register_dumps
21009 langhook, invoke register_dumps method on the dump manager.
21010 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21013 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21015 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21016 Say that # requires an associated define_split to exist, and that
21017 the define_split must be suitable for use after register allocation.
21019 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21021 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21022 (compute_outgoing_frequencies): Also initialize zero counts.
21023 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21024 around loops; preserve more of profile when nothing changes.
21026 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21028 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21030 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21031 * config/arm/arm-cpu-cdata.h: Regenerate.
21032 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21033 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21034 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21035 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21036 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21038 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21040 * config/arm/t-rmprofile: Likewise.
21041 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21043 2017-06-06 David S. Miller <davem@davemloft.net>
21046 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21047 blockage if function uses alloca.
21049 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21051 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21052 New "uid" fields to hold pretty-print IDs of group and ref.
21053 Memory references are now identified as <group_id>:<ref_id>
21054 instead of using [random] addresses.
21055 (dump_mem_details): Simplify, no functional change.
21056 (dump_mem_ref): Simplify and make output more concise.
21057 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21058 (find_or_create_group): Initialize group uid.
21059 (record_ref): Initialize ref uid. Improve debug output.
21060 (prune_group_by_reuse, should_issue_prefetch_p,)
21061 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21062 (mark_nontemporal_store, determine_loop_nest_reuse):
21063 Improve debug output.
21065 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21067 * dbgcnt.def (prefetch): New debug counter.
21068 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21069 (schedule_prefetches): Stop issueing prefetches if debug counter
21072 2017-06-06 Tom de Vries <tom@codesourcery.com>
21074 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21075 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21077 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21079 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21080 Use aarch64_reg_or_zero predicate for operand 4.
21081 (aarch64_compare_and_swap<mode> define_insn_and_split):
21082 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21083 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21085 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21087 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21088 (arm_compute_save_core_reg_mask): This.
21089 (thumb1_compute_save_reg_mask): Rename into ...
21090 (thumb1_compute_save_core_reg_mask): This.
21091 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21092 (arm_compute_frame_layout): Likewise.
21094 2017-06-06 Richard Biener <rguenther@suse.de>
21096 PR tree-optimization/80974
21097 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21098 keep or clear leaders SSA info.
21100 2017-06-06 Tom de Vries <tom@codesourcery.com>
21102 * config/nvptx/nvptx.c (split_mode_p): New function.
21103 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21105 2017-06-06 Tom de Vries <tom@codesourcery.com>
21107 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21109 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21112 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21115 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21117 * shrink-wrap.c (handle_simple_exit): Update profile.
21118 (try_shrink_wrapping): Upate profile.
21120 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21122 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21123 (tree_guess_outgoing_edge_probabilities): New.
21124 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21125 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21127 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21129 * ipa-split.c (split_function): Initialize return bb profile.
21131 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21133 * profile.c (compute_branch_probabilities): Also initialize
21134 EXIT_BLOCK profile.
21136 2017-06-06 Richard Biener <rguenther@suse.de>
21138 PR tree-optimization/80928
21139 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21140 (vect_analyze_loop_operations): Properly guard analysis for
21142 (vect_transform_loop): Likewise.
21143 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21144 (vect_model_induction_cost): Do not cost for pure SLP.
21145 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21146 of induction in inner loop vectorization.
21147 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21148 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21149 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21151 (vect_analyze_slp_cost_1): Cost induction.
21152 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21153 (vect_get_slp_vect_defs): Likewise.
21154 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21155 (vect_transform_stmt): Handle SLP reductions.
21156 * tree-vectorizer.h (vectorizable_induction): Adjust.
21158 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21160 * config/rs6000/rs6000.c (make_resolver_func): Update
21161 init_lowered_empty_function call.
21163 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21165 * doc/invoke.texi: Document the -fprofile-abs-path option.
21166 * common.opt (fprofile-abs-path): New option.
21167 * gcov-io.h (gcov_write_filename): Declare.
21168 * gcov-io.c (gcov_write_filename): New function.
21169 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21170 * profile.c (output_location): Likewise.
21172 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21174 * shring-wrap.c: Revert accidental commit.
21176 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21178 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21180 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21182 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21184 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21185 profile in callgraph edge.
21186 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21187 (apply_scale): Likewise.
21188 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21189 Also scale profile when inlining function with zero profile.
21190 (initialize_cfun): Update exit block profile even when it is zero.
21191 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21192 when profile is read.
21194 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21196 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21197 (CLONE_*): New constants to define the processors we can generate
21198 code for with the target_clone attribute.
21199 (rs6000_clone_map): New array to identify which clone processors
21200 the current program is running on.
21201 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21202 target_clone attribute.
21203 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21204 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21205 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21206 (cpu_expand_builtin): Add support for target_clone attribute.
21207 (rs6000_valid_attribute_p): Allow "default" attribute.
21208 (get_decl_name): New debug function to simplify printing the
21209 current function name in debugging statements.
21210 (rs6000_clone_priority): New functions to support the target_clone
21211 attribute, and be able to generate code to switch between ISA 2.05
21212 through ISA 3.0 (power6 through power9).
21213 (rs6000_compare_version_priority): Likewise.
21214 (rs6000_get_function_versions_dispatcher): Likewise.
21215 (make_resolver_func): Likewise.
21216 (add_condition_to_bb): Likewise.
21217 (dispatch_function_versions): Likewise.
21218 (rs6000_generate_version_dispatcher_body): Likewise.
21219 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21220 (fusion_gpr_load_p): Fix a spacing issue.
21221 * doc/extend.texi (Common Function Attributes): Document that the
21222 PowerPC supports the target_clone attribute.
21224 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21226 * config/arm/arm.h: explain F symbol found in description of ARM
21227 register allocation in its legend.
21229 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21231 * config/mips/frame-header-opt.c: Include profile-count.h.
21232 * config/riscv/riscv.c: Include profile-count.h
21234 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21236 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21238 (sm_set_flag_if_changed): Add bbs field.
21239 (execute_sm_if_changed_flag_set): Pass BBS.
21240 (execute_sm): Update.
21242 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21244 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21247 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21249 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21250 (peephole2): New peephole2 to emit the above.
21251 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21253 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21255 * config/aarch64/aarch64.c (define_peephole2 above
21256 *sub_<shift>_<mode>): New peephole.
21258 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21260 * config/i386/i386.c (make_resolver_func): Update.
21261 * Makefile.in: Add profile-count.h and profile-count.o
21262 * auto-profile.c (afdo_indirect_call): Update to new API.
21263 (afdo_set_bb_count): Update.
21264 (afdo_propagate_edge): Update.
21265 (afdo_propagate_circuit): Update.
21266 (afdo_calculate_branch_prob): Update.
21267 (afdo_annotate_cfg): Update.
21268 * basic-block.h: Include profile-count.h
21269 (struct edge_def): Turn count to profile_count.
21270 (struct basic_block_def): Likewie.
21271 (REG_BR_PROB_BASE): Move to profile-count.h
21272 (RDIV): Move to profile-count.h
21273 * bb-reorder.c (max_entry_count): Turn to profile_count.
21274 (find_traces): Update.
21275 (rotate_loop):Update.
21276 (connect_traces):Update.
21277 (sanitize_hot_paths):Update.
21278 * bt-load.c (migrate_btr_defs): Update.
21279 * cfg.c (RDIV): Remove.
21280 (init_flow): Use alloc_block.
21281 (alloc_block): Uninitialize count.
21282 (unchecked_make_edge): Uninitialize count.
21283 (check_bb_profile): Update.
21284 (dump_edge_info): Update.
21285 (dump_bb_info): Update.
21286 (update_bb_profile_for_threading): Update.
21287 (scale_bbs_frequencies_int): Update.
21288 (scale_bbs_frequencies_gcov_type): Update.
21289 (scale_bbs_frequencies_profile_count): New.
21290 * cfg.h (update_bb_profile_for_threading): Update.
21291 (scale_bbs_frequencies_profile_count): Declare.
21292 * cfgbuild.c (compute_outgoing_frequencies): Update.
21293 (find_many_sub_basic_blocks): Update.
21294 * cfgcleanup.c (try_forward_edges): Update.
21295 (try_crossjump_to_edge): Update.
21296 * cfgexpand.c (expand_gimple_tailcall): Update.
21297 (construct_exit_block): Update.
21298 * cfghooks.c (verify_flow_info): Update.
21299 (dump_bb_for_graph): Update.
21300 (split_edge): Update.
21301 (make_forwarder_block): Update.
21302 (duplicate_block): Update.
21303 (account_profile_record): Update.
21304 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21305 (get_estimated_loop_iterations): Update.
21306 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21307 (single_likely_exit): Update.
21308 * cfgloopmanip.c (scale_loop_profile): Update.
21310 (set_zero_probability): Update.
21311 (lv_adjust_loop_entry_edge): Update.
21312 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21313 (purge_dead_edges): Update.
21314 (rtl_account_profile_record): Update.
21315 * cgraph.c (cgraph_node::create): Uninitialize count.
21316 (symbol_table::create_edge): Uninitialize count.
21317 (cgraph_update_edges_for_call_stmt_node): Update.
21318 (cgraph_edge::dump_edge_flags): Update.
21319 (cgraph_node::dump): Update.
21320 (cgraph_edge::maybe_hot_p): Update.
21321 * cgraph.h: Include profile-count.h
21322 (create_clone), create_edge, create_indirect_edge): Update.
21323 (cgraph_node): Turn count to profile_count.
21324 (cgraph_edge0: Likewise.
21325 (make_speculative, clone): Update.
21326 (create_edge): Update.
21327 (init_lowered_empty_function): Update.
21328 * cgraphclones.c (cgraph_edge::clone): Update.
21329 (duplicate_thunk_for_node): Update.
21330 (cgraph_node::create_clone): Update.
21331 * cgraphunit.c (cgraph_node::analyze): Update.
21332 (cgraph_node::expand_thunk): Update.
21333 * final.c (dump_basic_block_info): Update.
21334 * gimple-streamer-in.c (input_bb): Update.
21335 * gimple-streamer-out.c (output_bb): Update.
21336 * graphite.c (print_global_statistics): Update.
21337 (print_graphite_scop_statistics): Update.
21338 * hsa-brig.c: Include basic-block.h.
21339 * hsa-dump.c: Include basic-block.h.
21340 * hsa-gen.c (T sum_slice): Update.
21341 (convert_switch_statements):Update.
21342 * hsa-regalloc.c: Include basic-block.h.
21343 * ipa-chkp.c (chkp_produce_thunks): Update.
21344 * ipa-cp.c (struct caller_statistics): Update.
21345 (init_caller_stats): Update.
21346 (gather_caller_stats): Update.
21347 (ipcp_cloning_candidate_p): Update.
21348 (good_cloning_opportunity_p): Update.
21349 (get_info_about_necessary_edges): Update.
21350 (dump_profile_updates): Update.
21351 (update_profiling_info): Update.
21352 (update_specialized_profile): Update.
21353 (perhaps_add_new_callers): Update.
21354 (decide_about_value): Update.
21355 (ipa_cp_c_finalize): Update.
21356 * ipa-devirt.c (struct odr_type_warn_count): Update.
21357 (struct decl_warn_count): Update.
21358 (struct final_warning_record): Update.
21359 (possible_polymorphic_call_targets): Update.
21360 (ipa_devirt): Update.
21361 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21362 * ipa-icf.c (sem_function::merge): Update.
21363 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21364 * ipa-inline.c (compute_uninlined_call_time): Update.
21365 (compute_inlined_call_time): Update.
21366 (want_inline_small_function_p): Update.
21367 (want_inline_self_recursive_call_p): Update.
21368 (edge_badness): Update.
21369 (lookup_recursive_calls): Update.
21370 (recursive_inlining): Update.
21371 (inline_small_functions): Update.
21372 (dump_overall_stats): Update.
21373 (dump_inline_stats): Update.
21374 * ipa-profile.c (ipa_profile_generate_summary): Update.
21375 (ipa_propagate_frequency): Update.
21376 (ipa_profile): Update.
21377 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21378 * ipa-utils.c (ipa_merge_profiles): Update.
21379 * loop-doloop.c (doloop_modify): Update.
21380 * loop-unroll.c (report_unroll): Update.
21381 (unroll_loop_runtime_iterations): Update.
21382 * lto-cgraph.c (lto_output_edge): Update.
21383 (lto_output_node): Update.
21384 (input_node): Update.
21385 (input_edge): Update.
21386 (merge_profile_summaries): Update.
21387 * lto-streamer-in.c (input_cfg): Update.
21388 * lto-streamer-out.c (output_cfg): Update.
21389 * mcf.c (create_fixup_graph): Update.
21390 (adjust_cfg_counts): Update.
21391 (sum_edge_counts): Update.
21392 * modulo-sched.c (sms_schedule): Update.
21393 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21394 * predict.c (maybe_hot_count_p): Update.
21395 (probably_never_executed): Update.
21396 (dump_prediction): Update.
21397 (combine_predictions_for_bb): Update.
21398 (propagate_freq): Update.
21399 (handle_missing_profiles): Update.
21400 (counts_to_freqs): Update.
21401 (rebuild_frequencies): Update.
21402 (force_edge_cold): Update.
21403 * predict.h: Include profile-count.h
21404 (maybe_hot_count_p, counts_to_freqs): UPdate.
21405 * print-rtl-function.c: Do not include cfg.h
21406 * print-rtl.c: Include basic-block.h
21407 * profile-count.c: New file.
21408 * profile-count.h: New file.
21409 * profile.c (is_edge_inconsistent): Update.
21410 (correct_negative_edge_counts): Update.
21411 (is_inconsistent): Update.
21412 (set_bb_counts): Update.
21413 (read_profile_edge_counts): Update.
21414 (compute_frequency_overlap): Update.
21415 (compute_branch_probabilities): Update; Initialize and deinitialize
21417 (branch_prob): Update.
21418 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21419 (edge_gcov_count): New.
21420 (bb_gcov_count): New.
21421 * shrink-wrap.c (try_shrink_wrapping): Update.
21422 * tracer.c (better_p): Update.
21423 * trans-mem.c (expand_transaction): Update.
21424 (ipa_tm_insert_irr_call): Update.
21425 (ipa_tm_insert_gettmclone_call): Update.
21426 * tree-call-cdce.c: Update.
21427 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21428 (gimple_duplicate_sese_tail): Update.
21429 (gimple_account_profile_record): Update.
21430 (execute_fixup_cfg): Update.
21431 * tree-inline.c (copy_bb): Update.
21432 (copy_edges_for_bb): Update.
21433 (initialize_cfun): Update.
21434 (freqs_to_counts): Update.
21435 (copy_cfg_body): Update.
21436 (expand_call_inline): Update.
21437 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21438 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21439 (try_unroll_loop_completely): Update.
21440 (try_peel_loop): Update.
21441 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21442 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21443 * tree-ssa-loop-split.c (connect_loops): Update.
21444 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21445 * tree-ssa-reassoc.c (branch_fixup): Update.
21446 * tree-ssa-tail-merge.c (replace_block_by): Update.
21447 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21448 (compute_path_counts): Update.
21449 (update_profile): Update.
21450 (recompute_probabilities): Update.
21451 (update_joiner_offpath_counts): Update.
21452 (estimated_freqs_path): Update.
21453 (freqs_to_counts_path): Update.
21454 (clear_counts_path): Update.
21455 (ssa_fix_duplicate_block_edges): Update.
21456 (duplicate_thread_path): Update.
21457 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21458 (struct switch_conv_info): Update.
21459 * tree-tailcall.c (decrease_profile): Update.
21460 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21461 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21462 * value-prof.c (check_counter): Update.
21463 (gimple_divmod_fixed_value): Update.
21464 (gimple_mod_pow2): Update.
21465 (gimple_mod_subtract): Update.
21466 (gimple_ic_transform): Update.
21467 (gimple_stringop_fixed_value): Update.
21468 * value-prof.h (gimple_ic): Update.
21470 2017-06-02 Carl Love <cel@us.ibm.com>
21472 * config/rs6000/rs6000-c: Add support for built-in functions
21473 vector double vec_doublee (vector signed int);
21474 vector double vec_doublee (vector unsigned int);
21475 vector double vec_doublee (vector float);
21476 vector double vec_doubleh (vector signed int);
21477 vector double vec_doubleh (vector unsigned int);
21478 vector double vec_doubleh (vector float);
21479 vector double vec_doublel (vector signed int);
21480 vector double vec_doublel (vector unsigned int);
21481 vector double vec_doublel (vector float);
21482 vector double vec_doubleo (vector signed int);
21483 vector double vec_doubleo (vector unsigned int);
21484 vector double vec_doubleo (vector float);.
21485 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21486 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21488 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21489 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21490 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21492 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21493 vec_doublel, vec_doubleh.
21494 * doc/extend.texi: Update the built-in documentation file for the
21495 new built-in functions.
21497 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21500 * ipa-inline-analysis.c (free_growth_caches): Set
21501 edge_removal_hook_holder to NULL after removing it.
21503 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21505 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21506 comparision with zero.
21508 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21509 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21510 for early expansion of vec_min and vec_max builtins.
21511 (builtin_function_type): Add min/max unsigned variants to those
21512 identified as having unsigned arguments.
21514 2017-06-02 Olivier Hainque <hainque@adacore.com>
21516 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21518 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21520 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21521 Use VALL_F16 iterator rather than VALL.
21523 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21525 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21526 Emit CBNZ inside loop when doing a strong exchange and comparing
21527 against zero. Generate the CC flags after the loop.
21529 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21531 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21532 (dl_section_ref): New.
21533 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21534 On AIX, append an expression to subtract the size of the
21535 section length to dl_section_ref.
21537 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21539 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21540 for early expansion of vector absolute builtins.
21542 2017-06-02 Richard Biener <rguenther@suse.de>
21544 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21545 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21547 2017-06-02 Richard Biener <rguenther@suse.de>
21549 PR tree-optimization/80948
21550 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21551 stmt order as well.
21553 2017-06-02 Richard Biener <rguenther@suse.de>
21555 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21557 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21558 for inductions as relevant.
21560 2017-06-02 Richard Biener <rguenther@suse.de>
21562 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21563 (vectorizable_induction): ... this. Remove dead code.
21565 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21567 * builtins. (expand_builtin_alloca): Remove second parameter and
21568 infer its value from the first parameter instead.
21569 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21571 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21573 PR rtl-optimization/80903
21574 * loop-doloop.c (add_test): Unshare sequence.
21576 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21578 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21580 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21582 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21584 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21585 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21586 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21587 (xlogue_layout::get_stub_rtx): Make static.
21588 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21589 (xlogue_layout::compute_stub_managed_regs): Rename to...
21590 (xlogue_layout::count_stub_managed_regs): ...this.
21591 (xlogue_layout::is_stub_managed_reg): New function.
21592 (xlogue_layout::m_stub_names): Rename to...
21593 (xlogue_layout::s_stub_names): ...this, make static.
21594 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21595 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21596 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21597 xlogue_layout::s_stub_names): Instantiate statics.
21598 (stub_managed_regs): Remove.
21599 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21600 (disable_call_ms2sysv_xlogues): Rename to...
21601 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21602 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21604 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21605 change after reload_completed.
21606 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21608 (ix86_expand_prologue): Likewise.
21609 (ix86_expand_epilogue): Likewise.
21610 (ix86_expand_split_stack_prologue): Likewise.
21611 (ix86_compute_frame_layout): Remove frame parameter ...
21612 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21613 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21615 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21616 (ix86_frame): Move from here ...
21617 * config/i386/i386.h (ix86_frame): ... to here.
21618 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21619 complete ix86_frame data structure instead. Remove some_ld_name.
21621 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21623 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21624 symbols that hold a DECL_VALUE_EXPR.
21626 2017-06-01 Martin Jambor <mjambor@suse.cz>
21628 PR tree-optimization/80898
21629 * tree-sra.c (process_subtree_disqualification): Removed.
21630 (disqualify_candidate): Do not acll
21631 process_subtree_disqualification.
21632 (subtree_mark_written_and_enqueue): New function.
21633 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21634 RHS has been disqualified and re-queue LHS if necessary. Apart
21635 from that, ignore disqualified RHS.
21637 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21639 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21640 address fetch for z10 or later.
21642 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21644 * config/arc/arc.md (tst_movb): Add guard when splitting.
21646 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21648 * config/arc/arc.c (arc_can_eliminate): Test against
21649 arc_frame_pointer_needed.
21651 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21653 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21654 to prevent store reordering.
21655 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21656 (type): Add block type.
21657 (stack_tie): Define special instruction to be used in
21660 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21662 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21663 constraint. It is not valid for the pattern.
21664 (noncommutative_binary_comparison): Likewise.
21666 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21668 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21671 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21673 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21674 be used by the reg-alloc.
21676 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21678 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21679 reg-alloc when having mul64 or mul32x16 instructions.
21680 (mulsidi3): Likewise.
21681 (umulsidi3): Likewise.
21682 (mulsi32x16): New pattern.
21683 (mulsi64): Likewise.
21684 (mulsidi64): Likewise.
21685 (umulsidi64): Likewise.
21686 (MUL32x16_REG): Define.
21687 (mul64_600): Use MUL32x16_REG.
21688 (mac64_600): Likewise.
21689 (umul64_600): Likewise.
21690 (umac64_600): Likewise.
21692 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21694 * config/arc/arc.md (mulsi3_700): Make it commutative.
21696 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21698 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21700 (*sign_extendsidi2_insn): Likewise for movstosw.
21702 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21704 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21705 the type of the input discriminant value. Convert the
21706 discriminant value of signedness vary.
21708 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21710 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21711 Add to -Wall section.
21713 2017-06-01 Richard Biener <rguenther@suse.de>
21715 PR middle-end/66313
21716 * fold-const.c (fold_plusminus_mult_expr): If the factored
21717 factor may be zero use a wrapping type for the inner operation.
21718 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21719 and handle moved defs.
21720 (process_assignment): Properly guard the unary op case. Return a
21721 tri-state indicating that moving the stmt before the call may allow
21722 to continue. Pass through to_move.
21723 (find_tail_calls): Handle moving unrelated defs before
21726 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21729 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21730 splitter result in the canonical way.
21732 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21734 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21735 also for 32bit target. Update insn attributes.
21736 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21738 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21740 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21741 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21742 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21743 (_mm_maskz_min_ss): New intrinsics.
21745 2017-05-31 Martin Liska <mliska@suse.cz>
21747 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21748 Change comment style to one we normally use.
21749 (vectorizable_reduction): Likewise.
21750 (vectorizable_induction): Likewise.
21751 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21752 (vectorizable_call): Likewise.
21753 (vectorizable_simd_clone_call): Likewise.
21754 (vectorizable_conversion): Likewise.
21755 (vectorizable_assignment): Likewise.
21756 (vectorizable_shift): Likewise.
21757 (vectorizable_operation): Likewise.
21758 (vectorizable_store): Likewise.
21759 (vectorizable_load): Likewise.
21760 * tree-vectorizer.h: Likewise.
21762 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21764 * passes.c (emergency_dump_function): New.
21765 * tree-pass.h (emergency_dump_function): Declare.
21766 * plugin.c (plugins_internal_error_function): Remove.
21767 * plugin.h (plugins_internal_error_function): Remove declaration.
21768 * toplev.c (internal_error_function): New static function. Use it...
21769 (general_init): ...here.
21771 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21773 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21774 (arc_rtx_costs): Add costs for new patterns.
21775 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21776 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21778 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21780 * tree-ssa-strlen.c (get_next_strinfo): New function.
21781 (get_stridx_plus_constant): Use it.
21782 (zero_length_string): Likewise.
21783 (adjust_related_strinfos): Likewise.
21784 (adjust_last_stmt): Likewise.
21786 2017-05-31 Richard Biener <rguenther@suse.de>
21789 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21790 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21792 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21794 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21795 loop_vinfo argument and use of dependence distance vectors.
21796 Check instead whether the two references differ only in their
21797 initial value and assume that they have the same alignment if the
21798 difference is a multiple of the vector alignment.
21799 (vect_analyze_data_refs_alignment): Update call accordingly.
21801 2017-05-31 Martin Liska <mliska@suse.cz>
21804 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21806 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21808 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21809 (create_intersect_range_checks): Move from ...
21810 * tree-data-ref.c (create_intersect_range_checks_index)
21811 (create_intersect_range_checks): ... to here.
21812 (create_runtime_alias_checks): New function factored from ...
21813 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21814 here. Call above function.
21815 * tree-data-ref.h (create_runtime_alias_checks): New function.
21817 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21819 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21820 segment length for dr_b and compute it in wide_int.
21822 2017-05-31 Richard Biener <rguenther@suse.de>
21824 PR tree-optimization/80906
21825 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21826 and pass through iv_map.
21827 (copy_bb_and_scalar_dependences): Adjust.
21828 (translate_pending_phi_nodes): Likewise.
21829 (copy_loop_close_phi_args): Handle code-generating IVs instead
21832 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21834 * diagnostic-color.c (color_dict): Add "type-diff".
21835 (parse_gcc_colors): Update comment.
21836 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21837 -fdiagnostics-show-template-tree and -fno-elide-type.
21838 (GCC_COLORS): Add type-diff to example.
21840 (-fdiagnostics-show-template-tree): New.
21841 (-fno-elide-type): New.
21842 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21843 the pp_format_decoder callback. Call any m_format_postprocessor's
21845 (pretty_printer::pretty_printer): Initialize
21846 m_format_postprocessor.
21847 (pretty_printer::~pretty_printer): Delete any
21848 m_format_postprocessor.
21849 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21850 (class format_postprocessor): New class.
21851 (struct pretty_printer::format_decoder): Document the new parameters.
21852 (struct pretty_printer::m_format_postprocessor): New field.
21853 * tree-diagnostic.c (default_tree_printer): Update for new
21854 bool and const char ** params.
21855 * tree-diagnostic.h (default_tree_printer): Likewise.
21857 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21859 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21860 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21861 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21862 rs6000_gen_cell_microcode code.
21863 (rs6000_final_prescan_insn): Delete.
21864 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21865 "warn-cell-microcode" entries.
21866 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21867 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21868 throughout. Change cc_reg_not_micro_cr0_operand to
21869 cc_reg_not_cr0_operand throughout.
21870 (*extendhi<mode>2_noload): Delete.
21871 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21872 (mwarn-cell-microcode): Delete.
21873 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21874 -mgen-cell-microcode and -mwarn-cell-microcode.
21876 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21879 * config/i386/constraints.md (Yd): New constraint.
21881 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21882 and (?Yd, r) alternatives. Update insn attributes.
21883 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21884 and (?*Yd, r) alternatives. Update insn attributes.
21885 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21887 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21889 * gimplify.c (gimplify_modify_expr): Don't create a
21890 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21893 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21895 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21897 2017-05-30 Richard Biener <rguenther@suse.de>
21899 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21900 and reduc_def fields.
21901 (STMT_VINFO_REDUC_TYPE): New define.
21902 (STMT_VINFO_REDUC_DEF): Likewise.
21903 (vect_force_simple_reduction): Adjust prototype.
21904 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21905 (vect_is_simple_reduction): Remove check_reduction argument.
21906 (vect_force_simple_reduction): Adjust and set
21907 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21908 (vectorizable_reduction): Do not re-do reduction analysis
21909 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21910 * tree-parloops.c (gather_scalar_reductions): Adjust.
21912 2017-05-30 Richard Biener <rguenther@suse.de>
21914 PR middle-end/80901
21915 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21918 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21920 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21921 Introduce unknown_misalignment parameter and remove vf.
21922 (vect_peeling_hash_get_lowest_cost):
21923 Pass unknown_misalignment parameter.
21924 (vect_enhance_data_refs_alignment):
21925 Fix unsupportable data ref treatment.
21927 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21929 * tree-vect-data-refs.c (vect_get_data_access_cost):
21930 Workaround for SLP handling.
21931 (vect_enhance_data_refs_alignment):
21932 Compute costs for doing no peeling at all, compare to the best
21933 peeling costs so far and avoid peeling if cheaper.
21935 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21937 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21938 Return peeling info and set costs to zero for unlimited cost
21940 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21941 with unknown misalignment. Compute and costs for unknown
21942 misalignment, compare them to the costs for known misalignment
21943 and choose the cheapest for peeling.
21945 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21947 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21948 (vect_get_peeling_costs_all_drs): Create function.
21949 (vect_peeling_hash_get_lowest_cost):
21950 Use vect_get_peeling_costs_all_drs.
21951 (vect_peeling_supportable): Create function.
21952 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21954 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21956 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21957 DR_HAS_NEGATIVE_STEP.
21958 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21959 (vect_enhance_data_refs_alignment): Use.
21960 (vect_duplicate_ssa_name_ptr_info): Use.
21961 * tree-vectorizer.h (dr_misalignment): Use.
21962 (known_alignment_for_access_p): Use.
21964 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21967 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21969 (has_section_name): New function.
21971 2017-05-30 Martin Liska <mliska@suse.cz>
21974 * auto-profile.c (get_function_decl_from_block): Fix
21977 2017-05-30 Richard Biener <rguenther@suse.de>
21979 PR middle-end/80876
21980 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21982 2017-05-30 Martin Liska <mliska@suse.cz>
21984 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21985 * dumpfile.h (struct dump_file_info): Remove ctors.
21987 2017-05-30 Martin Liska <mliska@suse.cz>
21989 * predict.def: Fix GNU coding style.
21991 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21993 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21994 Mark 'to' argument with ATTRIBUTE_UNUSED.
21996 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21998 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21999 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22000 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22003 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22005 * doc/install.texi (Options specification): Restore entry of
22006 --enable-sjlj-exceptions.
22008 2017-05-27 Michael Eager <eager@eagercon.com>
22011 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22013 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22015 * config/microblaze/microblaze.h
22016 (FIXED_REGISTERS): Update in macro.
22017 (CALL_USED_REGISTERS): Update in macro.
22019 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22021 * doc/install.texi: Add links to macOS binary distributions.
22023 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22027 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22029 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22031 2017-05-26 Martin Liska <mliska@suse.cz>
22033 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22035 2017-05-26 Martin Liska <mliska@suse.cz>
22037 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22038 always leading ';; '.
22039 (dump_bb_info): Likewise.
22040 (brief_dump_cfg): Likewise.
22041 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22042 * dumpfile.c: Remove usage of TDF_VERBOSE.
22043 * dumpfile.h (enum dump_kind): Likewise.
22044 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22045 * print-tree.c (debug_verbose): Remove.
22046 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22047 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22048 * tree-diagnostic.c (default_tree_printer): Replace
22049 TDF_DIAGNOSTIC with TDF_SLIM.
22051 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22053 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22054 in parameter loop, rather than loop_vinfo.
22055 (create_intersect_range_checks): Ditto.
22056 (vect_create_cond_for_alias_checks): Update call to above functions.
22058 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22060 PR tree-optimization/80815
22061 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22062 for merging runtime alias checks. Handle negative DR_STEPs.
22064 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22066 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22068 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22069 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22070 out code pruning runtime alias checks.
22071 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22072 factored out from above.
22073 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22075 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22077 (prune_runtime_alias_test_list): New decalaration.
22079 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22081 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22082 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22083 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22084 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22085 (operator==, comp_dr_with_seg_len_pair): Ditto.
22086 (vect_prune_runtime_alias_test_list): Ditto.
22088 2017-05-26 Martin Liska <mliska@suse.cz>
22091 * params.def: Bound partial-inlining-entry-probability param.
22093 2017-05-26 Marek Polacek <polacek@redhat.com>
22096 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22099 2017-05-26 Richard Biener <rguenther@suse.de>
22101 PR tree-optimization/80842
22102 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22105 2017-05-26 Richard Biener <rguenther@suse.de>
22107 PR tree-optimization/80844
22108 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22110 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22112 * doc/md.texi (Machine Constraints): Update x86 family
22113 machine constraints section to match 'config/i386/constraints.md'.
22115 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22117 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22119 2017-05-25 Nathan Sidwell <nathan@acm.org>
22121 * doc/invoke.texi (--enable-languages): Update documentation.
22123 2017-05-25 Martin Liska <mliska@suse.cz>
22125 * dumpfile.c: Add TDF_FOLDING.
22126 * dumpfile.h (enum dump_kind): Likewise.
22127 * genmatch.c (dt_simplify::gen_1): Use it.
22129 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22131 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22133 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22135 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22136 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22138 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22140 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22141 * match.pd (X == C): Rewrite it here.
22142 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22143 with_certain_nonzero_bits2): New predicates.
22144 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22146 2017-05-24 Nathan Sidwell <nathan@acm.org>
22148 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22151 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22154 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22156 * config/powerpcspe: New port. Files are copied from the rs6000
22157 port, with "rs6000" in filenames replaced by "powerpcspe".
22159 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22161 PR rtl-optimization/80754
22162 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22164 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22166 * config/sparc/sparc.md (length): Return the correct value for -mflat
22167 sibcalls to match output_sibcall.
22169 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22173 * config/rs6000/rs6000.c (struct machine_function): Add new field
22175 (rs6000_get_separate_components): Init that field, use it.
22176 (rs6000_components_for_bb): Use the field.
22178 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22180 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22182 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22184 PR middle-end/80823
22185 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22187 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22190 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22191 address against address_operand predicate.
22192 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22194 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22196 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22197 parameters passed indirectly.
22199 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22201 * config/i386/i386.md (*movdi_internal): Remove SSE4
22202 alternative 18 (?r, *v). Update insn attributes.
22203 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22204 Update insn attributes.
22205 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22206 Update insn attributes.
22207 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22208 alternative 1 (r, v). Remove isa attribute.
22209 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22210 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22211 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22213 2017-05-23 Tom de Vries <tom@codesourcery.com>
22215 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22218 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22220 * cgraphunit.c (symbol_table::process_new_functions): Update.
22221 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22222 (inline_generate_summary): Rename to ...
22223 (ipa_fn_summary_generate): ... this one.
22224 (inline_read_summary): Rename to ...
22225 (ipa_fn_summary_read): ... this one.
22226 (inline_write_summary): Rename to ...
22227 (ipa_fn_summary_write): ... this one.
22228 (inline_free_summary): Rename to ...
22229 (ipa_free_fn_summary): ... this one.
22230 (pass_data_local_fn_summary, pass_local_fn_summary,
22231 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22232 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22233 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22234 make_pass_ipa_fn_summary): New.
22235 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22236 inline_write_summary, inline_free_summary): Remove.
22237 (ipa_free_fn_summary) : New.
22238 * ipa-inline.c (ipa_inline): Update.
22239 (pass_ipa_inline): Do not generate summaries.
22240 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22242 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22243 and add pass_ipa_fn_summary.
22244 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22246 (make_pass_inline_parameters): Remove.
22248 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22250 * omp-low.c (struct omp_context): Remove "default_kind" member.
22253 * omp-offload.c (execute_oacc_device_lower): Remove the
22254 parallelism dimensions function attributes for unparallelized
22255 OpenACC kernels constructs.
22257 2017-05-23 Martin Liska <mliska@suse.cz>
22259 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22261 (cgraph_edge::make_speculative): Likewise.
22262 (cgraph_edge::resolve_speculation): Likewise.
22263 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22264 (cgraph_node::dump): Likewise.
22265 * cgraph.h: Likewise.
22266 * cgraphunit.c (analyze_functions): Likewise.
22267 (symbol_table::compile): Likewise.
22268 * ipa-cp.c (print_all_lattices): Likewise.
22269 (determine_versionability): Likewise.
22270 (initialize_node_lattices): Likewise.
22271 (ipcp_verify_propagated_values): Likewise.
22272 (estimate_local_effects): Likewise.
22273 (update_profiling_info): Likewise.
22274 (create_specialized_node): Likewise.
22275 (perhaps_add_new_callers): Likewise.
22276 (decide_about_value): Likewise.
22277 (decide_whether_version_node): Likewise.
22278 (identify_dead_nodes): Likewise.
22279 (ipcp_store_bits_results): Likewise.
22280 * ipa-devirt.c (dump_targets): Likewise.
22281 (ipa_devirt): Likewise.
22282 * ipa-icf.c (sem_item::dump): Likewise.
22283 (sem_function::equals): Likewise.
22284 (sem_variable::equals): Likewise.
22285 (sem_item_optimizer::read_section): Likewise.
22286 (sem_item_optimizer::execute): Likewise.
22287 (congruence_class::dump): Likewise.
22288 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22289 (dump_inline_summary): Likewise.
22290 (estimate_node_size_and_time): Likewise.
22291 (inline_analyze_function): Likewise.
22292 * ipa-inline-transform.c (inline_call): Likewise.
22293 * ipa-inline.c (report_inline_failed_reason): Likewise.
22294 (want_early_inline_function_p): Likewise.
22295 (edge_badness): Likewise.
22296 (update_edge_key): Likewise.
22297 (inline_small_functions): Likewise.
22298 * ipa-profile.c (ipa_profile): Likewise.
22299 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22300 (ipa_make_edge_direct_to_target): Likewise.
22301 (remove_described_reference): Likewise.
22302 (ipa_impossible_devirt_target): Likewise.
22303 (propagate_controlled_uses): Likewise.
22304 (ipa_print_node_params): Likewise.
22305 (ipcp_transform_function): Likewise.
22306 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22307 (propagate_pure_const): Likewise.
22308 * ipa-reference.c (generate_summary): Likewise.
22309 (read_write_all_from_decl): Likewise.
22310 (propagate): Likewise.
22311 (ipa_reference_read_optimization_summary): Likewise.
22312 * ipa-utils.c (ipa_merge_profiles): Likewise.
22313 * ipa.c (walk_polymorphic_call_targets): Likewise.
22314 (symbol_table::remove_unreachable_nodes): Likewise.
22315 (ipa_single_use): Likewise.
22316 * passes.c (execute_todo): Likewise.
22317 * predict.c (drop_profile): Likewise.
22318 * symtab.c (symtab_node::get_dump_name): New function.
22319 (symtab_node::dump_name): Likewise.
22320 (symtab_node::dump_asm_name): Likewise.
22321 (symtab_node::dump_references): Likewise.
22322 (symtab_node::dump_referring): Likewise.
22323 (symtab_node::dump_base): Likewise.
22324 (symtab_node::debug_symtab): Likewise.
22325 * tree-sra.c (convert_callers_for_node): Likewise.
22326 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22327 * value-prof.c (init_node_map): Likewise.
22329 2017-05-23 Martin Liska <mliska@suse.cz>
22331 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22332 and symtab_node::debug_symtab to symbol_table::debug.
22333 * cgraphunit.c (analyze_functions): Use the renamed function.
22334 (symbol_table::compile): Likewise.
22335 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22336 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22337 * passes.c (execute_todo): Likewise.
22338 * symtab.c (symbol_table::dump): New function.
22339 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22341 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22343 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22344 that nonconst implies exec.
22346 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22348 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22349 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22350 (inline_edge_summary_vec): Turn into ...
22351 (ipa_call_summaries): ... this one.
22352 (redirect_to_unreachable, edge_set_predicate,
22353 evaluate_properties_for_edge, inline_summary_alloc,
22354 reset_ipa_call_summary, reset_inline_summary,
22355 inline_summary_t::duplicate): Update.
22356 (inline_edge_duplication_hook): Turn to ...
22357 (ipa_call_summary_t::duplicate): ... this one.
22358 (inline_edge_removal_hook): Turn to ...
22359 (ipa_call_summary_t::remove): ... this one.
22360 (dump_inline_edge_summary): Turn to ...
22361 (dump_ipa_call_summary): ... this one.
22362 (estimate_function_body_sizes): Update.
22363 (inline_update_callee_summaries): Update.
22364 (remap_edge_change_prob): Update.
22365 (remap_edge_summaries): Update.
22366 (inline_merge_summary): Update.
22367 (do_estimate_edge_time): Update.
22368 (inline_generate_summary): Update.
22369 (inline_read_section): Update.
22370 (inline_read_summary): Update.
22371 (inline_free_summary): Update.
22372 * ipa-inline.c (can_inline_edge_p): Update.
22373 (compute_inlined_call_time): Update.
22374 (want_inline_small_function_p): Update.
22375 (edge_badness): Update.
22376 (early_inliner): Update.
22377 * ipa-inline.h (inline_edge_summary): Turn to ...
22378 (ipa_call_summary): ... this one.
22379 (ipa_call_summary_t): New class.
22380 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22381 (ipa_call_summaries): New.
22382 (inline_edge_summary): Remove.
22383 (estimate_edge_growth): Update.
22384 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22385 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22386 * ipa-split.c (execute_split_functions): Update.
22387 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22389 2017-05-23 Tom de Vries <tom@codesourcery.com>
22391 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22392 attributes): Document rdrand effective target.
22394 2017-05-23 Tom de Vries <tom@codesourcery.com>
22396 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22397 attributes): Sort alphabetically.
22399 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22401 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22403 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22406 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22407 V2DF/V2DI splat into two separate patterns, one that handles
22408 registers, and the other that only handles memory. Drop support
22409 for splatting from a GPR on ISA 2.07 and then splitting the
22410 splat into direct move and splat.
22411 (vsx_splat_<mode>_reg): Likewise.
22412 (vsx_splat_<mode>_mem): Likewise.
22414 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22416 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22418 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22420 PR middle-end/80809
22421 * omp-low.c (finish_taskreg_remap): New function.
22422 (finish_taskreg_scan): If unit size of ctx->record_type
22423 is non-constant, unshare the size expression and replace
22424 decls in it with possible outer var refs.
22426 PR middle-end/80809
22427 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22428 GOVD_SHARED rather than GOVD_PRIVATE with it.
22429 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22430 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22432 PR middle-end/80853
22433 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22434 as last argument to build_outer_var_ref for pointer bases of array
22435 section reductions.
22437 2017-05-19 Martin Sebor <msebor@redhat.com>
22439 * print-tree.c (print_node): Print DECL_READ_P flag.
22441 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22443 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22444 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22445 * cgraph.c: Likewise.
22446 * cgraphunit.c: Likewise.
22447 * gengtype.c: Likewise.
22448 * ipa-cp.c: Likewise.
22449 * ipa-devirt.c: Likewise.
22450 * ipa-icf.c: Likewise.
22451 * ipa-predicate.c: Likewise.
22452 * ipa-profile.c: Likewise.
22453 * ipa-prop.c: Likewise.
22454 * ipa-split.c: Likewise.
22456 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22457 edge_predicate_pool, dump_inline_hints,
22458 inline_summary::account_size_time, redirect_to_unreachable,
22459 edge_set_predicate, set_hint_predicate,
22460 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22461 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22462 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22463 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22464 ipa_call_summary_t::remove, initialize_growth_caches,
22465 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22466 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22467 mark_modified, unmodified_parm_1, unmodified_parm,
22468 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22469 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22470 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22471 will_be_nonconstant_predicate, record_modified_bb_info,
22472 get_minimal_bb, record_modified, param_change_prob,
22473 phi_result_unknown_predicate, predicate_for_phi_result,
22474 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22475 estimate_function_body_sizes, compute_inline_parameters,
22476 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22477 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22478 inline_update_callee_summaries, remap_edge_change_prob,
22479 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22480 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22481 inline_analyze_function, inline_summary_t::insert,
22482 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22483 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22484 inline_free_summary): Move to ipa-fnsummary.h
22485 (predicate_t): Remove.
22486 * ipa-fnsummary.c: New file.
22487 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22488 (enum inline_hints_vals, inline_hints, agg_position_info,
22489 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22490 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22491 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22492 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22493 inline_read_summary, inline_write_summary, inline_free_summary,
22494 inline_analyze_function, initialize_inline_failed,
22495 inline_merge_summary, inline_update_overall_summary,
22496 compute_inline_parameters): Move to ipa-fnsummary.h
22497 * ipa-fnsummary.h: New file.
22498 * ipa-inline-transform.h: Include ipa-inline.h.
22499 * ipa-inline.c: LIkewise.
22501 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22503 * ipa-inline.c (edge_badness): Use inlined_time instead of
22504 inline_summaries->get.
22506 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22508 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22510 2017-05-22 Nathan Sidwell <nathan@acm.org>
22512 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22513 (fdump-lang): Document 'raw' option.
22514 * dumpfile.h (TDI_tu): Delete.
22515 * dumpfile.c (dump_files): Remove translation-unit.
22516 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22518 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22520 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22521 command option from $(AWK) call.
22522 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22523 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22524 [FORMAT]: Remove handling of variable.
22525 * config/avr/t-multilib: Regenerate.
22527 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22529 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22531 (dump_inline_summary): Do not print self_time.
22532 (estimate_function_body_sizes): Do not set self_time.
22533 (compute_inline_parameters): Likewise.
22534 (inline_read_section, inline_write_summary): Do not stream self_time.
22535 * ipa-inline.h (inline_summary): Drop self_time.
22537 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22539 * ipa-inline-analysis.c (account_size_time): Rename to ...
22540 (inline_summary::account_size_time): ... this one.
22541 (reset_ipa_call_summary): Turn to ...
22542 (ipa_call_summary::reset): ... this one.
22543 (reset_inline_summary): Turn to ...
22544 (inline_summary::reset): ... this one.
22545 (inline_summary_t::remove): Update.
22546 (inline_summary_t::duplicate): Update.
22547 (ipa_call_summary_t::remove): Update.
22548 (dump_inline_summary): Update.
22549 (estimate_function_body_sizes): Update.
22550 (compute_inline_parameters): Update.
22551 (estimate_node_size_and_time): Update.
22552 (inline_merge_summary): Update.
22553 (inline_update_overall_summary): Update.
22554 (inline_read_section): Update.
22555 (inline_write_summary): Update.
22556 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22557 add account_size_time and reset member functions.
22558 (ipa_call_summary): Add reset function.
22559 * ipa-predicate.h (predicate::operator &): Constify.
22561 2017-05-22 Richard Biener <rguenther@suse.de>
22563 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22565 2017-05-19 Jason Merrill <jason@redhat.com>
22567 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22569 2017-05-19 Marek Polacek <polacek@redhat.com>
22572 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22573 TYPE_OVERFLOW_WRAPS checks.
22575 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22577 * tree-core.h (enum omp_clause_default_kind): Add
22578 "OMP_CLAUSE_DEFAULT_PRESENT".
22579 * tree-pretty-print.c (dump_omp_clause): Handle it.
22580 * gimplify.c (enum gimplify_omp_var_data): Add
22581 "GOVD_MAP_FORCE_PRESENT".
22582 (gimplify_adjust_omp_clauses_1): Map it to
22583 "GOMP_MAP_FORCE_PRESENT".
22584 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22586 * gimplify.c (oacc_default_clause): Clarify.
22588 2017-05-19 Nathan Sidwell <nathan@acm.org>
22590 LANG_HOOK_REGISTER_DUMPS
22591 * toplev.c (general_init): Call register dump lang hook.
22592 * doc/invoke.texi: Document -fdump-lang option family.
22593 * dumpfile.c (dump_files): Remove class dump here.
22594 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22595 * dumpfile.h (tree_dump_index): Remove TDI_class.
22596 * langhooks-def.h (lhd_register_dumps): Declare.
22597 (LANG_HOOKS_REGISTER_DUMPS): Define.
22598 (LANG_HOOKS_INITIALIZER): Add it.
22599 * langhooks.c (lhd_register_dumps): Define.
22600 * langhooks.h (struct lang_hooks): Add register_dumps.
22602 2017-05-19 Nathan Sidwell <nathan@acm.org>
22604 * context.h (context::set_passes): New.
22605 * context.c (context::context): Do not create pass manager.
22606 * toplev.c (general_init): Create pass manager here.
22608 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22610 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22611 use this splitter if two add or or instructions would also work for
22612 the constant we want to generate.
22614 2017-05-19 Richard Biener <rguenther@suse.de>
22617 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22618 predicate evaluation.
22620 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22622 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22624 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22627 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22629 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22630 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22631 (inline_edge_summary_vec): Turn into ...
22632 (ipa_call_summaries): ... this one.
22633 (redirect_to_unreachable, edge_set_predicate,
22634 evaluate_properties_for_edge, inline_summary_alloc,
22635 reset_ipa_call_summary, reset_inline_summary,
22636 inline_summary_t::duplicate): Update.
22637 (inline_edge_duplication_hook): Turn to ...
22638 (ipa_call_summary_t::duplicate): ... this one.
22639 (inline_edge_removal_hook): Turn to ...
22640 (ipa_call_summary_t::remove): ... this one.
22641 (dump_inline_edge_summary): Turn to ...
22642 (dump_ipa_call_summary): ... this one.
22643 (estimate_function_body_sizes): Update.
22644 (inline_update_callee_summaries): Update.
22645 (remap_edge_change_prob): Update.
22646 (remap_edge_summaries): Update.
22647 (inline_merge_summary): Update.
22648 (do_estimate_edge_time): Update.
22649 (inline_generate_summary): Update.
22650 (inline_read_section): Update.
22651 (inline_read_summary): Update.
22652 (inline_free_summary): Update.
22653 * ipa-inline.c (can_inline_edge_p): Update.
22654 (compute_inlined_call_time): Update.
22655 (want_inline_small_function_p): Update.
22656 (edge_badness): Update.
22657 (early_inliner): Update.
22658 * ipa-inline.h (inline_edge_summary): Turn to ...
22659 (ipa_call_summary): ... this one.
22660 (ipa_call_summary_t): New class.
22661 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22662 (ipa_call_summaries): New.
22663 (inline_edge_summary): Remove.
22664 (estimate_edge_growth): Update.
22665 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22666 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22667 * ipa-split.c (execute_split_functions): Update.
22668 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22670 2017-05-19 Richard Biener <rguenther@suse.de>
22672 PR middle-end/80764
22673 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22675 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22677 * config/rs6000/rs6000.c (struct machine_function): Add field
22678 fpr_is_wrapped_separately.
22679 (rs6000_get_separate_components): Use 64 components. Handle the
22680 new FPR components.
22681 (rs6000_components_for_bb): Handle the FPR components.
22682 (rs6000_emit_prologue_components): Handle the FPR components.
22683 (rs6000_emit_epilogue_components): Handle the FPR components.
22684 (rs6000_set_handled_components): Handle the FPR components.
22685 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22686 that are already separately shrink-wrapped.
22687 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22688 that are already separately shrink-wrapped.
22690 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22693 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22696 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22697 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22698 cases where the register allocator uses a move and an offsettable
22699 memory operation to/from a FPR register on ISA 2.06/2.07.
22700 (define_peephole2 for Altivec d-form store): Likewise.
22702 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22705 * config/i386/mmx.md (*mov<mode>_internal): Enable
22706 alternatives 11, 12, 13 and 14 also for 32bit targets.
22707 Remove alternatives 15, 16, 17 and 18.
22708 * config/i386/sse.md (vec_concatv2di): Change
22709 alternative (!x, *y) to (x, ?!*Yn).
22711 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22713 * dumpfile.h (enum dump_kind): Remove stray comma.
22715 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22717 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22718 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22719 predicate::num_conditions
22720 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22721 (CHANGED): turn into predicate::changed.
22722 (agg_position_info): Move to ipa-predicate.h
22723 (add_condition, predicate::add_clause, predicate::operator &=,
22724 predicate::or_with, predicate::evaluate, predicate::probability,
22725 dump_condition, dump_clause, predicate::dump,
22726 predicate::remap_after_duplication, predicate::remap_after_inlining,
22727 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22728 (evaluate_conditions_for_known_args): Update.
22729 (set_cond_stmt_execution_predicate): Update.
22730 * ipa-inline.h: Include ipa-predicate.h
22731 (condition, inline_param_summary, conditions, agg_position_info,
22732 predicate): Move to ipa-predicate.h
22733 * ipa-predicate.c: New file.
22734 * ipa-predicate.h: New file.
22736 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22738 * final.c (leaf_function_p): Check we are not in a sequence.
22740 2017-05-18 Martin Liska <mliska@suse.cz>
22742 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22743 * dumpfile.c (dump_register): Use new enum dump_kind.
22744 (get_dump_file_name): Likewise.
22745 (dump_enable_all): Likewise.
22746 (dump_switch_p_1): Likewise.
22747 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22748 * dumpfile.h (enum dump_kind): New enum type.
22749 (struct dump_file_info): Create constructor and
22750 format fields and comments.
22751 * passes.c (pass_manager::register_one_dump_file):
22753 * statistics.c (statistics_early_init): Likewise.
22754 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22755 TDF_TREE with TDF_SLIM.
22756 (gather_memory_references_ref): Likewise.
22758 2017-05-18 Martin Liska <mliska@suse.cz>
22760 * vec.h (struct vnull): Use it.
22762 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22764 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22765 (true_predicate, false_predicate, true_predicate_p,
22766 false_predicate_p): Remove.
22767 (single_cond_predicate, not_inlined_predicate): Turn to member function
22769 (add_condition): Update.
22770 (add_clause): Turn to...
22771 (predicate::add_clause): ... this one; update; allow passing NULL
22773 (and_predicates): Turn to ...
22774 (predicate::operator &=): ... this one.
22775 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22776 (or_predicates): Turn to ...
22777 (predicate::or_with): ... this one.
22778 (evaluate_predicate): Turn to ...
22779 (predicate::evaluate): ... this one.
22780 (predicate_probability): Turn to ...
22781 (predicate::probability): ... this one.
22782 (dump_condition): Update.
22783 (dump_predicate): Turn to ...
22784 (predicate::dump): ... this one.
22785 (account_size_time): Update.
22786 (edge_set_predicate): Update.
22787 (set_hint_predicate): UPdate.
22788 (evaluate_conditions_for_known_args): Update.
22789 (evaluate_properties_for_edge): Update.
22790 (remap_predicate_after_duplication): Turn to...
22791 (predicate::remap_after_duplication): ... this one.
22792 (remap_hint_predicate_after_duplication): Update.
22793 (inline_summary_t::duplicate): UPdate.
22794 (dump_inline_edge_summary): Update.
22795 (dump_inline_summary): Update.
22796 (set_cond_stmt_execution_predicate): Update.
22797 (set_switch_stmt_execution_predicate): Update.
22798 (compute_bb_predicates): Update.
22799 (will_be_nonconstant_expr_predicate): Update.
22800 (will_be_nonconstant_predicate): Update.
22801 (phi_result_unknown_predicate): Update.
22802 (predicate_for_phi_result): Update.
22803 (array_index_predicate): Update.
22804 (estimate_function_body_sizes): Update.
22805 (estimate_node_size_and_time): Update.
22806 (estimate_ipcp_clone_size_and_time): Update.
22807 (remap_predicate): Rename to ...
22808 (predicate::remap_after_inlining): ... this one.
22809 (remap_hint_predicate): Update.
22810 (inline_merge_summary): Update.
22811 (inline_update_overall_summary): Update.
22812 (estimate_size_after_inlining): Update.
22813 (read_predicate): Rename to ...
22814 (predicate::stream_in): ... this one.
22815 (read_inline_edge_summary): Update.
22816 (write_predicate): Rename to ...
22817 (predicate::stream_out): ... this one.
22818 (write_inline_edge_summary): Update.
22819 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22820 (clause_t): Turn to uint32_t
22821 (predicate): Turn to class; implement constructor and operators
22823 (size_time_entry): Update.
22824 (inline_summary): Update.
22825 (inline_edge_summary): Update.
22827 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22829 * fold-const.c (fold_binary_loc): Move transformation...
22830 * match.pd (C - X CMP X): ... here.
22832 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22834 * config/sparc/sparc.c (sparc_option_override): Set function
22835 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22836 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22838 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22840 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22842 2017-05-18 Marek Polacek <polacek@redhat.com>
22845 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22846 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22847 gimple_assign_load_p.
22849 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22851 PR middle-end/80692
22852 * real.c (do_compare): Give decimal_do_compare preference over
22853 comparing just the signs.
22855 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22857 * doc/md.texi (Canonicalization of Instructions): Describe the
22858 canonical form of instructions that inherently set a condition
22861 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22863 PR middle-end/80775
22864 * tree-cfg.c: Move deletion of unreachable case statements to after
22865 the merging of consecutive case labels.
22867 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22869 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22870 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22871 restoring of callee-saved registers.
22873 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22875 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22876 * config/visium/visium.c (single_set_and_flags): Likewise.
22877 * config/visium/visium.md (Substitutions): Likewise.
22879 2017-05-17 Martin Liska <mliska@suse.cz>
22881 * cfg.c: Introduce dump_flags_t type and
22882 use it instead of int type.
22884 * cfghooks.c: Likewise.
22885 * cfghooks.h (struct cfg_hooks): Likewise.
22886 * cfgrtl.c: Likewise.
22887 * cfgrtl.h: Likewise.
22888 * cgraph.c (cgraph_node::get_body): Likewise.
22889 * coretypes.h: Likewise.
22890 * domwalk.c: Likewise.
22891 * domwalk.h: Likewise.
22892 * dumpfile.c (struct dump_option_value_info): Likewise.
22893 (dump_enable_all): Likewise.
22894 (dump_switch_p_1): Likewise.
22895 (opt_info_switch_p): Likewise.
22896 * dumpfile.h (enum tree_dump_index): Likewise.
22897 (struct dump_file_info): Likewise.
22898 * genemit.c: Likewise.
22899 * generic-match-head.c: Likewise.
22900 * gengtype.c (open_base_files): Likewise.
22901 * gimple-pretty-print.c: Likewise.
22902 * gimple-pretty-print.h: Likewise.
22903 * graph.c (print_graph_cfg): Likewise.
22904 * graphite-scop-detection.c (dot_all_sese): Likewise.
22905 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22906 * loop-unroll.c (report_unroll): Likewise.
22907 * passes.c (pass_manager::register_one_dump_file): Likewise.
22908 * print-tree.c: Likewise.
22909 * statistics.c: Likewise.
22910 * tree-cfg.c: Likewise.
22911 * tree-cfg.h: Likewise.
22912 * tree-dfa.c: Likewise.
22913 * tree-dfa.h: Likewise.
22914 * tree-dump.c (dump_function): Likewise.
22915 * tree-dump.h (struct dump_info): Likewise.
22916 * tree-pretty-print.c: Likewise.
22917 * tree-pretty-print.h: Likewise.
22918 * tree-ssa-live.c: Likewise.
22919 * tree-ssa-live.h: Likewise.
22920 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22921 * tree-vect-loop.c: Likewise.
22922 * tree-vect-slp.c: Likewise.
22924 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22925 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22927 PR tree-optimization/80457
22928 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22929 of all arguments to a statement as scalar_to_vec operations.
22930 (vectorizable_call): Adjust call to vect_model_simple_cost for
22932 (vectorizable_conversion): Likewise.
22933 (vectorizable_assignment): Likewise.
22934 (vectorizable_shift): Likewise.
22935 (vectorizable_operation): Likewise.
22936 (vectorizable_comparison): Likewise.
22937 (vect_is_simple_cond): Record the def types for operands.
22938 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22939 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22940 for statement argument count.
22942 2017-05-16 Carl Love <cel@us.ibm.com>
22944 * config/rs6000/rs6000-c: Add support for built-in functions
22945 vector unsigned long long vec_bperm (vector unsigned long long,
22946 vector unsigned char)
22947 vector signed long long vec_mule (vector signed int,
22949 vector unsigned long long vec_mule (vector unsigned int,
22950 vector unsigned int)
22951 vector signed long long vec_mulo (vector signed int,
22953 vector unsigned long long vec_mulo (vector unsigned int,
22954 vector unsigned int)
22955 vector signed char vec_sldw (vector signed char,
22956 vector signed char,
22958 vector unsigned char vec_sldw (vector unsigned char,
22959 vector unsigned char,
22961 vector signed short vec_sldw (vector signed short,
22962 vector signed short,
22964 vector unsigned short vec_sldw (vector unsigned short,
22965 vector unsigned short,
22967 vector signed int vec_sldw (vector signed int,
22970 vector unsigned int vec_sldw (vector unsigned int,
22971 vector unsigned int,
22973 vector signed long long vec_sldw (vector signed long long,
22974 vector signed long long,
22976 vector unsigned long long vec_sldw (vector unsigned long long,
22977 vector unsigned long long,
22979 * config/rs6000/rs6000-c: Add support for built-in functions
22980 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22981 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22982 * doc/extend.texi: Update the built-in documentation for the
22983 new built-in functions.
22985 2017-05-16 Marek Polacek <polacek@redhat.com>
22989 * tree.c (save_expr): Don't fold the expression.
22991 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22993 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22994 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22995 and (?*y,m). Update insn attributes.
22997 2017-05-16 Martin Liska <mliska@suse.cz>
22999 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23000 flags argument of print_gimple_stmt, print_gimple_expr,
23001 print_generic_stmt and print_generic_expr.
23002 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23003 * coretypes.h: Likewise.
23004 * except.c (dump_eh_tree): Likewise.
23005 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23006 * gimple-pretty-print.h: Likewise.
23007 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23008 (backprop::push_to_worklist): Likewise.
23009 (backprop::pop_from_worklist): Likewise.
23010 (backprop::process_use): Likewise.
23011 (backprop::intersect_uses): Likewise.
23012 (note_replacement): Likewise.
23013 * gimple-ssa-store-merging.c
23014 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23015 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23016 (pass_store_merging::execute): Likewise.
23017 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23018 (ssa_base_cand_dump_callback): Likewise.
23019 (dump_incr_vec): Likewise.
23020 (replace_refs): Likewise.
23021 (replace_mult_candidate): Likewise.
23022 (create_add_on_incoming_edge): Likewise.
23023 (create_phi_basis): Likewise.
23024 (insert_initializers): Likewise.
23025 (all_phi_incrs_profitable): Likewise.
23026 (introduce_cast_before_cand): Likewise.
23027 (replace_one_candidate): Likewise.
23028 * gimplify.c (gimplify_expr): Likewise.
23029 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23030 (set_rename): Likewise.
23031 (rename_uses): Likewise.
23032 (copy_loop_phi_nodes): Likewise.
23033 (add_close_phis_to_merge_points): Likewise.
23034 (copy_loop_close_phi_args): Likewise.
23035 (copy_cond_phi_args): Likewise.
23036 (graphite_copy_stmts_from_block): Likewise.
23037 (translate_pending_phi_nodes): Likewise.
23038 * graphite-poly.c (print_pdr): Likewise.
23039 (dump_gbb_cases): Likewise.
23040 (dump_gbb_conditions): Likewise.
23041 (print_scop_params): Likewise.
23042 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23043 (build_cross_bb_scalars_use): Likewise.
23044 (gather_bbs::before_dom_children): Likewise.
23045 * hsa-dump.c (dump_hsa_immed): Likewise.
23046 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23047 (get_replacement_map): Likewise.
23048 * ipa-inline-analysis.c (dump_condition): Likewise.
23049 (estimate_function_body_sizes): Likewise.
23050 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23051 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23052 * ipa-prop.c (ipa_dump_param): Likewise.
23053 (ipa_print_node_jump_functions_for_edge): Likewise.
23054 (ipa_modify_call_arguments): Likewise.
23055 (ipa_modify_expr): Likewise.
23056 (ipa_dump_param_adjustments): Likewise.
23057 (ipa_dump_agg_replacement_values): Likewise.
23058 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23059 * ipa-pure-const.c (check_stmt): Likewise.
23060 (pass_nothrow::execute): Likewise.
23061 * ipa-split.c (execute_split_functions): Likewise.
23062 * omp-offload.c (dump_oacc_loop_part): Likewise.
23063 (dump_oacc_loop): Likewise.
23064 * trans-mem.c (tm_log_emit): Likewise.
23065 (tm_memopt_accumulate_memops): Likewise.
23066 (dump_tm_memopt_set): Likewise.
23067 (dump_tm_memopt_transform): Likewise.
23068 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23069 (print_loop): Likewise.
23070 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23071 (chkp_gather_checks_info): Likewise.
23072 (chkp_get_check_result): Likewise.
23073 (chkp_remove_check_if_pass): Likewise.
23074 (chkp_use_outer_bounds_if_possible): Likewise.
23075 (chkp_reduce_bounds_lifetime): Likewise.
23076 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23077 (chkp_mark_completed_bounds): Likewise.
23078 (chkp_register_incomplete_bounds): Likewise.
23079 (chkp_mark_invalid_bounds): Likewise.
23080 (chkp_maybe_copy_and_register_bounds): Likewise.
23081 (chkp_build_returned_bound): Likewise.
23082 (chkp_get_bound_for_parm): Likewise.
23083 (chkp_build_bndldx): Likewise.
23084 (chkp_get_bounds_by_definition): Likewise.
23085 (chkp_generate_extern_var_bounds): Likewise.
23086 (chkp_get_bounds_for_decl_addr): Likewise.
23087 * tree-chrec.c (chrec_apply): Likewise.
23088 * tree-data-ref.c (dump_data_reference): Likewise.
23089 (dump_subscript): Likewise.
23090 (dump_data_dependence_relation): Likewise.
23091 (analyze_overlapping_iterations): Likewise.
23092 * tree-inline.c (expand_call_inline): Likewise.
23093 (tree_function_versioning): Likewise.
23094 * tree-into-ssa.c (dump_defs_stack): Likewise.
23095 (dump_currdefs): Likewise.
23096 (dump_names_replaced_by): Likewise.
23097 (dump_update_ssa): Likewise.
23098 (update_ssa): Likewise.
23099 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23100 * tree-parloops.c (build_new_reduction): Likewise.
23101 (try_create_reduction_list): Likewise.
23102 (ref_conflicts_with_region): Likewise.
23103 (oacc_entry_exit_ok_1): Likewise.
23104 (oacc_entry_exit_single_gang): Likewise.
23105 * tree-pretty-print.h: Likewise.
23106 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23107 (get_scalar_evolution): Likewise.
23108 (add_to_evolution): Likewise.
23109 (get_loop_exit_condition): Likewise.
23110 (analyze_evolution_in_loop): Likewise.
23111 (analyze_initial_condition): Likewise.
23112 (analyze_scalar_evolution): Likewise.
23113 (instantiate_scev): Likewise.
23114 (number_of_latch_executions): Likewise.
23115 (gather_chrec_stats): Likewise.
23116 (final_value_replacement_loop): Likewise.
23117 (scev_const_prop): Likewise.
23118 * tree-sra.c (dump_access): Likewise.
23119 (disqualify_candidate): Likewise.
23120 (create_access): Likewise.
23121 (reject): Likewise.
23122 (maybe_add_sra_candidate): Likewise.
23123 (create_access_replacement): Likewise.
23124 (analyze_access_subtree): Likewise.
23125 (analyze_all_variable_accesses): Likewise.
23126 (sra_modify_assign): Likewise.
23127 (initialize_constant_pool_replacements): Likewise.
23128 (find_param_candidates): Likewise.
23129 (decide_one_param_reduction): Likewise.
23130 (replace_removed_params_ssa_names): Likewise.
23131 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23132 * tree-ssa-copy.c (dump_copy_of): Likewise.
23133 (copy_prop_visit_cond_stmt): Likewise.
23134 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23135 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23136 (record_equivalences_from_stmt): Likewise.
23137 * tree-ssa-dse.c (compute_trims): Likewise.
23138 (delete_dead_call): Likewise.
23139 (delete_dead_assignment): Likewise.
23140 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23141 (forward_propagate_into_cond): Likewise.
23142 (pass_forwprop::execute): Likewise.
23143 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23144 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23146 (move_computations_worker): Likewise.
23147 (execute_sm): Likewise.
23148 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23149 (remove_exits_and_undefined_stmts): Likewise.
23150 (remove_redundant_iv_tests): Likewise.
23151 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23152 (adjust_iv_update_pos): Likewise.
23153 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23154 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23155 (value_replacement): Likewise.
23156 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23157 * tree-ssa-pre.c (print_pre_expr): Likewise.
23158 (get_representative_for): Likewise.
23159 (create_expression_by_pieces): Likewise.
23160 (insert_into_preds_of_block): Likewise.
23161 (eliminate_insert): Likewise.
23162 (eliminate_dom_walker::before_dom_children): Likewise.
23163 (eliminate): Likewise.
23164 (remove_dead_inserted_code): Likewise.
23165 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23166 * tree-ssa-reassoc.c (get_rank): Likewise.
23167 (eliminate_duplicate_pair): Likewise.
23168 (eliminate_plus_minus_pair): Likewise.
23169 (eliminate_not_pairs): Likewise.
23170 (undistribute_ops_list): Likewise.
23171 (eliminate_redundant_comparison): Likewise.
23172 (update_range_test): Likewise.
23173 (optimize_range_tests_var_bound): Likewise.
23174 (optimize_vec_cond_expr): Likewise.
23175 (rewrite_expr_tree): Likewise.
23176 (rewrite_expr_tree_parallel): Likewise.
23177 (linearize_expr): Likewise.
23178 (break_up_subtract): Likewise.
23179 (linearize_expr_tree): Likewise.
23180 (attempt_builtin_powi): Likewise.
23181 (attempt_builtin_copysign): Likewise.
23182 (transform_stmt_to_copy): Likewise.
23183 (transform_stmt_to_multiply): Likewise.
23184 (dump_ops_vector): Likewise.
23185 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23186 (print_scc): Likewise.
23187 (set_ssa_val_to): Likewise.
23188 (visit_reference_op_store): Likewise.
23189 (visit_use): Likewise.
23190 (sccvn_dom_walker::before_dom_children): Likewise.
23191 (run_scc_vn): Likewise.
23192 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23194 (expr_hash_elt::print): Likewise.
23195 (const_and_copies::pop_to_marker): Likewise.
23196 (const_and_copies::record_const_or_copy_raw): Likewise.
23197 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23198 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23199 (dump_predicates): Likewise.
23200 (find_uninit_use): Likewise.
23201 (warn_uninitialized_phi): Likewise.
23202 (pass_late_warn_uninitialized::execute): Likewise.
23203 * tree-ssa.c (verify_vssa): Likewise.
23204 (verify_ssa): Likewise.
23205 (maybe_optimize_var): Likewise.
23206 * tree-vrp.c (dump_value_range): Likewise.
23207 (dump_all_value_ranges): Likewise.
23208 (dump_asserts_for): Likewise.
23209 (register_edge_assert_for_2): Likewise.
23210 (vrp_visit_cond_stmt): Likewise.
23211 (vrp_visit_switch_stmt): Likewise.
23212 (vrp_visit_stmt): Likewise.
23213 (vrp_visit_phi_node): Likewise.
23214 (simplify_cond_using_ranges_1): Likewise.
23215 (fold_predicate_in): Likewise.
23216 (evrp_dom_walker::before_dom_children): Likewise.
23217 (evrp_dom_walker::push_value_range): Likewise.
23218 (evrp_dom_walker::pop_value_range): Likewise.
23219 (execute_early_vrp): Likewise.
23221 2017-05-16 Richard Biener <rguenther@suse.de>
23223 * dwarf2out.c (loc_list_from_tree_1): Do not create
23224 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23226 2017-05-16 Richard Biener <rguenther@suse.de>
23228 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23230 (note_variable_value_in_expr): If we resolved the decl ref
23231 do not push to the stack.
23233 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23235 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23236 operations in fast-math mode.
23237 (vaddq_f16): Likewise.
23238 (vmul_f16): Likewise.
23239 (vmulq_f16): Likewise.
23240 (vsub_f16): Likewise.
23241 (vsubq_f16): Likewise.
23242 * config/arm/neon.md (add<mode>3): New.
23244 (fma:<VH:mode>3): New. Also remove outdated comment.
23247 2017-05-16 Martin Liska <mliska@suse.cz>
23251 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23252 (odr_types_equivalent_p): Likewise.
23254 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23256 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23258 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23261 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23262 non-interunit SSE move alternatives with '?'.
23263 (zero-extendsidi peephole2): New peephole to skip intermediate
23264 general register in SSE zero-extend sequence.
23266 2017-05-15 Jeff Law <law@redhat.com>
23268 * reorg.c (relax_delay_slots): Create a new variable to hold
23269 the temporary target rather than clobbering TARGET_LABEL.
23271 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23272 missing argument to extract_bit_field call.
23273 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23275 2017-05-15 Martin Liska <mliska@suse.cz>
23278 * gcc.c (process_command): Do not allow empty argument of -o option.
23280 2017-05-15 Renlin Li <renlin.li@arm.com>
23282 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23283 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23284 * config/aarch64/constraints.md (Usf): Add long call check.
23285 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23286 (call_value): Likewise.
23287 (sibcall): Likewise.
23288 (sibcall_value): Likewise.
23290 (call_value_insn): New.
23291 (sibcall_insn): Update rtx pattern.
23292 (sibcall_value_insn): Likewise.
23293 (call_internal): Remove.
23294 (call_value_internal): Likewise.
23295 (sibcall_internal): Likewise.
23296 (sibcall_value_internal): Likewise.
23297 (call_reg): Likewise.
23298 (call_symbol): Likewise.
23299 (call_value_reg): Likewise.
23300 (call_value_symbol): Likewise.
23302 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23305 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23307 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23309 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23310 compatible with CCGOCmode and with CCZmode.
23312 2017-05-14 Martin Sebor <msebor@redhat.com>
23314 PR middle-end/77671
23315 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23316 (gimple_fold_builtin_snprintf): Same.
23317 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23318 (gimple_fold_builtin_snprintf): Same.
23319 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23320 of character types.
23321 (is_call_safe): New function.
23322 (try_substitute_return_value): Call it.
23323 (try_simplify_call): New function.
23324 (pass_sprintf_length::handle_gimple_call): Call it.
23326 2017-05-14 Martin Sebor <msebor@redhat.com>
23328 PR middle-end/80669
23329 * builtins.c (expand_builtin_stpncpy): Simplify.
23331 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23333 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23334 * config/i386/i386.h
23335 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23336 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23337 (struct machine_function): Add new members call_ms2sysv,
23338 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23339 (struct machine_frame_state): New fields sp_realigned and
23340 sp_realigned_offset.
23341 * config/i386/i386.c
23342 (enum xlogue_stub): New enum.
23343 (enum xlogue_stub_sets): New enum.
23344 (class xlogue_layout): New class.
23345 (struct ix86_frame): New fields stack_realign_allocate_offset,
23346 stack_realign_offset and outlined_save_offset. Modify comments to
23347 detail stack layout when using out-of-line stubs.
23348 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23349 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23350 -mcall-ms2sysv-xlogues.
23351 (stub_managed_regs): New static variable.
23352 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23353 registers managed by out-of-line stub.
23354 (disable_call_ms2sysv_xlogues): New function.
23355 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23356 m->call_ms2sysv when appropriate and compute frame layout for
23358 (sp_valid_at, fp_valid_at): New inline functions.
23359 (choose_basereg): New function.
23360 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23362 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23363 Use align parameter of choose_baseaddr to generated aligned SSE movs
23365 (pro_epilogue_adjust_stack): Modify to track
23366 machine_frame_state::sp_realigned.
23367 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23368 (ix86_nsaved_sseregs): Likewise.
23369 (ix86_emit_save_regs): Likewise.
23370 (ix86_emit_save_regs_using_mov): Likewise.
23371 (ix86_emit_save_sse_regs_using_mov): Likewise.
23372 (get_scratch_register_on_entry): Likewise.
23373 (gen_frame_set): New function.
23374 (gen_frame_load): Likewise.
23375 (gen_frame_store): Likewise.
23376 (emit_outlined_ms2sysv_save): Likewise.
23377 (emit_outlined_ms2sysv_restore): Likewise.
23378 (ix86_expand_prologue): Modify stack re-alignment code and call
23379 emit_outlined_ms2sysv_save when appropriate.
23380 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23381 parameter rtx_insn *insn, which allows the function to be used to only
23382 generate the notes.
23383 (ix86_expand_epilogue): Modify validity checks of frame and stack
23384 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23385 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23386 * config/i386/predicates.md
23387 (save_multiple): New predicate.
23388 (restore_multiple): Likewise.
23389 * config/i386/sse.md
23390 (save_multiple<mode>): New pattern.
23391 (save_multiple_realign<mode>): Likewise.
23392 (restore_multiple<mode>): Likewise.
23393 (restore_multiple_and_return<mode>): Likewise.
23394 (restore_multiple_leave_return<mode>): Likewise.
23395 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23397 2017-05-14 Julia Koval <julia.koval@intel.com>
23399 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23400 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23401 (__builtin_ia32_xsetbv): New builtins.
23402 * config/i386/i386.c (ix86_expand_special_args_builtin):
23404 (ix86_expand_builtin): Special expand for new intrinsics.
23405 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23406 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23407 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23409 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23411 * cfganal.c (inverted_post_order_compute): Change argument type
23413 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23414 * df-core.c (rest_of_handle_df_initialize): Adjust.
23415 (rest_of_handle_df_finish): Likewise.
23416 (df_analyze_1): Likewise.
23417 (df_analyze): Likewise.
23418 (loop_inverted_post_order_compute): Change argument to be a vec *.
23419 (df_analyze_loop): Adjust.
23420 (df_get_n_blocks): Likewise.
23421 (df_get_postorder): Likewise.
23422 * df.h (struct df_d): Change field to be a vec.
23423 * lcm.c (compute_laterin): Adjust.
23424 (compute_available): Likewise.
23425 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23426 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23427 * tree-ssa-pre.c (compute_antic): Likewise.
23429 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23431 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23432 (depth_first_search::depth_first_search): Change structure init
23433 function to this constructor.
23434 (depth_first_search::add_bb): Rename function to this member.
23435 (depth_first_search::execute): Likewise.
23436 (flow_dfs_compute_reverse_finish): Adjust.
23438 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23440 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23441 (longest_simple_path): Likewise.
23442 * shrink-wrap.c (spread_components): Likewise.
23443 (disqualify_problematic_components): Likewise.
23444 (emit_common_heads_for_components): Likewise.
23445 (emit_common_tails_for_components): Likewise.
23446 (insert_prologue_epilogue_for_components): Likewise.
23448 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23450 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23453 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23455 * df-core.c (df_set_blocks): Start using auto_bitmap.
23456 (df_compact_blocks): Likewise.
23457 * df-problems.c (df_rd_confluence_n): Likewise.
23458 * df-scan.c (df_insn_rescan_all): Likewise.
23459 (df_process_deferred_rescans): Likewise.
23460 (df_update_entry_block_defs): Likewise.
23461 (df_update_exit_block_uses): Likewise.
23462 (df_entry_block_bitmap_verify): Likewise.
23463 (df_exit_block_bitmap_verify): Likewise.
23464 (df_scan_verify): Likewise.
23465 * lra-constraints.c (lra_constraints): Likewise.
23466 (undo_optional_reloads): Likewise.
23467 (lra_undo_inheritance): Likewise.
23468 * lra-remat.c (calculate_gen_cands): Likewise.
23469 (do_remat): Likewise.
23470 * lra-spills.c (assign_spill_hard_regs): Likewise.
23471 (spill_pseudos): Likewise.
23472 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23473 (bitmap_set_subtract_values): Likewise.
23475 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23477 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23478 management with auto_bitmap.
23479 (fix_inter_tick): Likewise.
23480 (fix_recovery_deps): Likewise.
23481 * ira.c (add_store_equivs): Likewise.
23482 (find_moveable_pseudos): Likewise.
23483 (split_live_ranges_for_shrink_wrap): Likewise.
23484 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23485 (rtx_reuse_manager::seen_def_p): Likewise.
23486 (rtx_reuse_manager::set_seen_def): Likewise.
23487 * print-rtl.h (class rtx_reuse_manager): Likewise.
23489 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23491 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23493 (migrate_btr_def): Likewise.
23494 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23495 * df-core.c (loop_post_order_compute): Likewise.
23496 (loop_inverted_post_order_compute): Likewise.
23497 * hsa-common.h: Likewise.
23498 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23499 * init-regs.c (initialize_uninitialized_regs): Likewise.
23500 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23501 (inline_small_functions): Likewise.
23502 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23503 * ira.c (combine_and_move_insns): Likewise.
23504 (build_insn_chain): Likewise.
23505 * loop-invariant.c (find_invariants): Likewise.
23506 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23507 * predict.c (tree_predict_by_opcode): Likewise.
23508 (predict_paths_leading_to): Likewise.
23509 (predict_paths_leading_to_edge): Likewise.
23510 (estimate_loops_at_level): Likewise.
23511 (estimate_loops): Likewise.
23512 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23513 (spread_components): Likewise.
23514 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23515 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23516 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23517 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23518 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23519 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23520 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23521 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23522 (mark_threaded_blocks): Likewise.
23523 (thread_through_all_blocks): Likewise.
23524 * tree-ssa.c (verify_ssa): Likewise.
23525 (execute_update_addresses_taken): Likewise.
23526 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23528 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23530 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23532 (post_order_compute): Likewise.
23533 (inverted_post_order_compute): Likewise.
23534 (pre_and_rev_post_order_compute_fn): Likewise.
23536 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23538 * genrecog.c (int_set::int_set): Explicitly construct our
23539 auto_vec base class.
23540 * vec.h (auto_vec::auto_vec): New constructor.
23542 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23544 * bitmap.h (class auto_bitmap): New constructor taking
23545 bitmap_obstack * argument.
23547 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23549 * bitmap.h (class auto_bitmap): Change type of m_bits to
23550 bitmap_head, and adjust ctor / dtor and member operators.
23552 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23554 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23555 when returned register mode doesn't match original mode.
23557 2017-05-12 Jeff Law <law@redhat.com>
23558 Jakub Jelinek <jakub@redhat.com>
23560 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23561 we look for cc setter after the compare-elim changes.
23562 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23563 within the vector to match what compare-elim now expects.
23564 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23565 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23567 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23568 after the compare-elim changes.
23569 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23570 the vector to match what compare-elim now expects.
23571 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23572 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23573 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23574 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23575 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23577 * config/visium/visium.c (single_set_and_flags): Fix where
23578 we look for cc setter after the compare-elim changes.
23579 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23580 with the vector to match what compare-elim now expects.
23581 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23582 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23583 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23584 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23585 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23586 (neg<mode>2_insn_set_overflow): Likewise.
23588 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23590 PR middle-end/79794
23591 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23592 maybe_expand_insn call, set ops[0].target. If still set after call,
23593 set alt_rtl. Add extra arg to recursive calls.
23594 (extract_bit_field): Add alt_rtl argument. Pass to
23596 * expmed.h (extract_bit_field): Fix prototype.
23597 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23598 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23599 to extract_bit_field_calls.
23600 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23601 Pass alt_rtl to extract_bit_field calls.
23602 * calls.c (store_unaligned_arguments_into_psuedos)
23603 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23604 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23606 * optabs.h (struct expand_operand): Add target bitfield.
23608 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23610 * compare-elim.c (try_eliminate_compare): Canonicalize
23611 operation with embedded compare to
23612 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23613 (set (reg) (operation)].
23615 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23617 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23620 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23621 cost of adding a carry flag for ADC instruction.
23622 [case MINUS]: Ignore the cost of subtracting a carry flag
23623 for SBB instruction.
23625 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23627 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23629 * config/rs6000/bmiintrin.h: New file.
23630 * config/rs6000/bmi2intrin.h: New file.
23631 * config/rs6000/x86intrin.h: New file.
23633 2017-05-12 Jeff Law <law@redhat.com>
23635 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23638 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23640 PR middle-end/80707
23641 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23643 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23645 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23646 early expansion of vector divide builtins.
23647 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23648 builtins identified as having unsigned arguments.
23650 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23652 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23653 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23654 expansion of vector logical operations (and, andc, or, xor,
23657 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23659 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23660 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23662 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23664 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23665 early GIMPLE expansion of vector multiplies.
23667 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23669 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23670 TARGET_HAVE_MOVT conditional.
23671 (movt splitter): Likewise.
23673 2017-05-12 Richard Biener <rguenther@suse.de>
23675 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23676 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23677 Fold all stmts not inplace.
23679 2017-05-12 Richard Biener <rguenther@suse.de>
23681 PR tree-optimization/80713
23682 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23683 inserted_exprs bit for not removed stmts.
23685 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23687 PR middle-end/69921
23688 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23689 parallelized" attribute for parallelized OpenACC kernels.
23690 * omp-offload.c (execute_oacc_device_lower): Use it.
23692 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23693 Set "oacc kernels" attribute.
23694 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23695 parameter. Adjust all users.
23696 (oacc_fn_attrib_kernels_p): Remove function.
23697 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23698 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23699 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23700 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23701 assert "oacc kernels" attribute is set.
23703 2017-05-11 Carl Love <cel@us.ibm.com>
23705 * config/rs6000/rs6000-c: Add support for built-in functions
23706 vector unsigned char vec_popcnt (vector signed char)
23707 vector unsigned char vec_popcnt (vector unsigned char)
23708 vector unsigned short vec_popcnt (vector signed short)
23709 vector unsigned short vec_popcnt (vector unsigned short)
23710 vector unsigned int vec_popcnt (vector signed int)
23711 vector unsigned int vec_popcnt (vector unsigned int)
23712 vector unsigned long long vec_popcnt (vector signed long long)
23713 vector unsigned long long vec_popcnt (vector unsigned long long)
23714 vector signed long long vec_slo (vector signed long long,
23715 vector signed char)
23716 vector signed long long vec_slo (vector signed long long,
23717 vector unsigned char)
23718 vector unsigned long long vec_slo (vector unsigned long long,
23719 vector signed char)
23720 vector unsigned long long vec_slo (vector unsigned long long,
23721 vector unsigned char)
23722 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23723 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23724 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23725 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23726 * doc/extend.texi: Update the built-in documentation file for the
23727 new built-in functions.
23729 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23731 * attribs.h (sorted_attr_string): Move machine independent
23732 functions for target clone support from the i386 port to common
23733 code. Rename ix86_function_versions to common_function_versions.
23734 Rename make_name to make_unique_name.
23735 (common_function_versions): Likewise.
23736 (make_unique_name): Likewise.
23737 (make_dispatcher_decl): Likewise.
23738 (is_function_default_version): Likewise.
23739 * attribs.c (attr_strcmp): Likewise.
23740 (sorted_attr_string): Likewise.
23741 (common_function_versions): Likewise.
23742 (make_unique_name): Likewise.
23743 (make_dispatcher_decl): Likewise.
23744 (is_function_default_version): Likewise.
23745 * config/i386/i386.c (attr_strcmp): Likewise.
23746 (sorted_attr_string): Likewise.
23747 (ix86_function_versions): Likewise.
23748 (make_name): Likewise.
23749 (make_dispatcher_decl): Likewise.
23750 (is_function_default_version): Likewise.
23751 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23753 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23756 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23757 Account for direct move costs for vec_construct of integer
23760 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23763 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23764 (UNSPEC_STX_ATOMIC): Ditto.
23765 (loaddi_via_sse): New insn.
23766 (storedi_via_sse): Ditto.
23767 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23768 Update corresponding peephole2 patterns.
23769 (atomic_storedi_fpu): Ditto.
23771 2017-05-11 Julia Koval <julia.koval@intel.com>
23773 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23774 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23776 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23777 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23778 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23780 2017-05-11 Nathan Sidwell <nathan@acm.org>
23782 * graphite-poly.c: Include dumpfile.h.
23784 * dumpfle.h (dump_function): Declare here ...
23785 * tree-dump.h (dump_function): ... not here.
23786 * dumpfile.c: #include tree-cfg.h.
23787 (dump_function): Move here from ...
23788 * tree-dump.c (dump_function): ... here.
23789 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23790 * graphite-poly.c: Don't include tree-dump.h.
23791 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23792 * print-tree.c: Likewise.
23793 * stor-layout.c: Likewise.
23794 * tree-nested.c: Likewise.
23796 * dumpfile.c (dump_start): Use TDF_FLAGS.
23797 (dump_enable_all): Fix TDF_KIND check thinko.
23799 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23801 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23802 array entries to represent two legal parameterizations of the
23803 overloaded __builtin_cmpb function, as represented by the
23804 P6_OV_BUILTIN_CMPB constant.
23805 (altivec_resolve_overloaded_builtin): Add special case handling
23806 for the __builtin_cmpb function, as represented by the
23807 P6_OV_BUILTIN_CMPB constant.
23808 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23809 (BU_P6_64BIT_2): New macro.
23810 (BU_P6_OVERLOAD_2): New macro
23811 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23812 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23813 (CMPB): Add overload support to represent both 32-bit and 64-bit
23814 compare-bytes function.
23815 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23816 support for TARGET_CMPB.
23817 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23818 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23819 documentation of the __builtin_cmpb overloaded built-in function.
23821 2017-05-11 Richard Biener <rguenther@suse.de>
23823 PR tree-optimization/80705
23824 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23825 bases are not vectorizable.
23827 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23829 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23830 when counting register pressure.
23832 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23834 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23835 (iv_ca_more_deps): Renamed to ...
23836 (iv_ca_compare_deps): ... this.
23837 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23839 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23841 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23843 (determine_group_iv_costs): ... here.
23844 (find_inv_vars_cb): Record inv var if it's not recorded before.
23846 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23848 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23849 (get_shiftadd_cost): Ditto.
23851 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23853 * tree-ssa-address.c: Include header file.
23854 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23856 (add_to_parts): Refactor.
23857 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23858 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23863 PR tree-optimization/53090
23864 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23866 (extract_cond_operands): Detect condition with IV on both sides
23867 and return COMP_IV_EXPR_2.
23868 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23869 (rewrite_use_compare): Simplify by removing call to function
23870 extract_cond_operands.
23872 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23874 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23875 (extract_cond_operands): Detect condition comparing against non-
23876 invariant bound and return appropriate enum value.
23877 (find_interesting_uses_cond): Update use of extract_cond_operands.
23878 Handle its return value accordingly.
23879 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23881 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23883 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23884 nonlinear iv_use computation in loop invariant sensitive way.
23886 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23888 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23889 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23891 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23893 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23894 (dump_cand): Support iv_cand.inv_exprs.
23895 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23897 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23900 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23902 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23904 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23905 as local function. Include necessary header files.
23906 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23908 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23910 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23912 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23914 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23915 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23916 RSHIFT_EXPR and BIT_NOT_EXPR.
23918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23920 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23921 (adjust_setup_cost): New parameter supporting round up adjustment.
23922 (struct address_cost_data): Delete.
23923 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23924 (split_address_cost, ptr_difference_cost): Delete.
23925 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23926 (struct ainc_cost_data): New struct.
23927 (get_address_cost_ainc): New function.
23928 (get_address_cost, get_computation_cost): Reimplement.
23929 (determine_group_iv_cost_address): Record inv_expr for all uses of
23931 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23932 (iv_ca_has_deps): Reimplemented to ...
23933 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23935 (iv_ca_extend): Call iv_ca_more_deps.
23937 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23939 * tree-ssa-address.c (struct mem_address): Move to header file.
23940 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23941 * tree-ssa-address.h (struct mem_address): Move from C file.
23942 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23944 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23946 * tree-affine.h (aff_combination_type): New interface.
23947 (aff_combination_zero_p): Remove static.
23948 (aff_combination_const_p): New interface.
23949 (aff_combination_singleton_var_p): New interfaces.
23951 2017-05-11 Richard Biener <rguenther@suse.de>
23953 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23954 Skip unreachable blocks and destinations.
23955 (eliminate): Move stmt removal and fixup ...
23956 (fini_eliminate): ... here. Skip inserted exprs.
23957 (pass_pre::execute): Move fini_pre after fini_eliminate.
23958 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23959 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23960 PRE to get rid of dead code that has invalid SSA form and
23961 split critical edges again.
23963 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23965 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23967 2017-05-11 Richard Biener <rguenther@suse.de>
23969 * passes.c (execute_function_todo): Verify loops if they are
23970 said to be up-to-date.
23971 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23972 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23974 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23977 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23978 handle calling assemble_external ourself.
23981 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23982 modes with zero size. Enhance comment.
23984 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23986 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23987 built-ins for vec_xl and vec_xst with short and char pointer
23990 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23992 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23993 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23994 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23995 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23996 (_mm_maskz_min_round_ss): New intrinsics.
23997 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23998 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23999 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24000 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24001 (__builtin_ia32_minss_mask_round): New builtins.
24002 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24003 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24004 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24006 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24007 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24009 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24012 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24014 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24015 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24016 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24017 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24018 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24019 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24020 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24021 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24022 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24023 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24024 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24025 (__builtin_ia32_mulss_mask_round): New builtins.
24026 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24027 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24028 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24030 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24031 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24033 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24036 2017-05-10 Julia Koval <julia.koval@intel.com>
24038 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24039 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24040 (_mm256_setr_m128i): New intrinsics.
24042 2017-05-10 Julia Koval <julia.koval@intel.com>
24044 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24045 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24046 (_mm_maskz_rcp14_ss): New intrinsics.
24047 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24048 (__builtin_ia32_rcp14ss_mask): New builtins.
24049 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24051 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24053 PR tree-optimization/51513
24054 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24055 (assert_unreachable_fallthru_edge_p): Use it.
24056 (group_case_labels_stmt): Likewise.
24057 * tree-cfg.h: Prototype it.
24058 * stmt.c: Include cfghooks.h and tree-cfg.h.
24059 (emit_case_dispatch_table) <gap_label>: New local variable.
24060 Use it to fill dispatch table gaps.
24061 Test for default_label before updating probabilities.
24062 (expand_case) <default_label>: Remove unneeded initialization.
24063 Test for unreachable default case statement and remove its edge.
24064 Set default_label accordingly.
24065 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24067 2017-05-10 Carl Love <cel@us.ibm.com>
24069 * config/rs6000/rs6000-c: Add support for built-in functions
24070 vector signed char vec_neg (vector signed char)
24071 vector signed short int vec_neg (vector short int)
24072 vector signed int vec_neg (vector signed int)
24073 vector signed long long vec_neg (vector signed long long)
24074 vector float vec_neg (vector float)
24075 vector double vec_neg (vector double)
24076 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24078 * config/rs6000/altivec.h: Add define for vec_neg
24079 * doc/extend.texi: Update the built-in documentation for the
24080 new built-in functions.
24082 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24084 PR tree-optimization/77644
24085 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24087 2017-05-10 Nathan Sidwell <nathan@acm.org>
24089 * dumpfile.h (TDI_lang_all): New.
24090 (TDF_KIND): New. Renumber others
24091 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24093 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24095 (get_dump_file_name): Adjust suffix generation.
24096 (dump_enable_all): Use TDF_KIND.
24097 * doc/invoke.texi (-fdump-lang-all): Document.
24099 * dumpfile.h: Tabify.
24101 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24104 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24105 Move member access before delete.
24107 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24109 * tree-inline.c (expand_call_inline): Split block at stmt
24112 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24115 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24116 are now unused after splitting mov{sf,sd}_hardfloat.
24117 (f32_lr2): Likewise.
24118 (f32_lm): Likewise.
24119 (f32_lm2): Likewise.
24120 (f32_li): Likewise.
24121 (f32_li2): Likewise.
24122 (f32_lv): Likewise.
24123 (f32_sr): Likewise.
24124 (f32_sr2): Likewise.
24125 (f32_sm): Likewise.
24126 (f32_sm2): Likewise.
24127 (f32_si): Likewise.
24128 (f32_si2): Likewise.
24129 (f32_sv): Likewise.
24130 (f32_dm): Likewise.
24131 (f32_vsx): Likewise.
24132 (f32_av): Likewise.
24133 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24134 For movsf, order stores so the VSX stores occur before the GPR
24135 store which encourages the register allocator to use a traditional
24136 FPR instead of a GPR. For movsd, order the stores so that the GPR
24137 store comes before the VSX stores to allow the power6 to work.
24138 This is due to the power6 not having a 32-bit integer store
24139 instruction from a FPR.
24140 (movsf_hardfloat): Likewise.
24141 (movsd_hardfloat): Likewise.
24143 2017-05-09 Martin Sebor <msebor@redhat.com>
24145 PR translation/80280
24146 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24149 PR translation/80280
24150 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24151 data member added in r247778.
24152 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24154 2017-05-09 Nathan Sidwell <nathan@acm.org>
24156 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24158 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24161 2017-05-09 Marek Polacek <polacek@redhat.com>
24163 * doc/invoke.texi: Fix typo.
24165 2017-05-09 Richard Biener <rguenther@suse.de>
24167 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24168 (vrp_val_is_min): Likewise.
24169 (set_value_range_to_value): Likewise.
24170 (set_value_range_to_nonnegative): Likewise.
24171 (gimple_assign_nonzero_p): Likewise.
24172 (gimple_stmt_nonzero_p): Likewise.
24173 (vrp_int_const_binop): Likewise. Remove unreachable case.
24174 (adjust_range_with_scev): Adjust comments.
24175 (compare_range_with_value): Likewise.
24176 (extract_range_from_phi_node): Likewise.
24177 (test_for_singularity): Likewise.
24179 2017-05-09 Richard Biener <rguenther@suse.de>
24181 * tree-vrp.c (get_single_symbol): Add assert that we don't
24182 get overflowed constants as invariant part.
24183 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24184 checks. Use wi::cmp instead of recursing for integer constants.
24185 (compare_values): Just ignore whether we assumed undefined
24186 overflow instead of failing the compare.
24187 (extract_range_for_var_from_comparison_expr): Add comment before the
24188 TREE_NO_WARNING sets.
24189 (test_for_singularity): Likewise.
24190 (extract_range_from_comparison): Do not disable optimization
24191 when we assumed undefined overflow.
24192 (extract_range_basic): Remove init of unused var.
24194 2017-05-09 Richard Biener <rguenther@suse.de>
24196 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24197 (extract_range_from_multiplicative_op_1): Adjust.
24198 (extract_range_from_binary_expr_1): Use int_const_binop.
24200 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24203 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24204 rs6000_store_data_bypass_p in seven define_bypass directives and
24205 in several comments.
24206 * config/rs6000/rs6000-protos.h: Add prototype for
24207 rs6000_store_data_bypass_p function.
24208 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24209 function implements slightly different (rs6000-specific) semantics
24210 than store_data_bypass_p, returning false rather than aborting
24211 with assertion error when arguments do not satisfy the
24212 requirements of store data bypass.
24213 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24214 rs6000_store_data_bypass_p.
24216 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24218 * config/xtensa/xtensa-protos.h
24219 (xtensa_initial_elimination_offset): New declaration.
24220 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24221 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24222 macro definition, add case for FRAME_POINTER_REGNUM when
24223 FRAME_GROWS_DOWNWARD.
24224 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24225 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24226 xtensa_initial_elimination_offset.
24228 2017-05-08 Nathan Sidwell <nathan@acm.org>
24230 * doc/invoke.texi: Alphabetize -fdump options.
24232 2017-05-08 Martin Sebor <msebor@redhat.com>
24234 PR translation/80280
24235 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24237 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24239 * target.def (compute_frame_layout): New optional target hook.
24240 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24241 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24242 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24244 * reload1.c (verify_initial_elim_offsets): Likewise.
24245 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24246 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24247 (arm_get_frame_offsets): Split up into this ...
24248 (arm_compute_frame_layout): ... and this function.
24250 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24252 * config/aarch64/constraints.md (Usa): New constraint.
24253 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24255 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24257 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24258 with_multilib_list after it has been checked.
24260 2017-05-08 Richard Biener <rguenther@suse.de>
24262 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24263 (bitmap_set_subtract_values): Likewise.
24265 2017-05-08 Richard Biener <rguenther@suse.de>
24267 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24268 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24270 (gimple_stmt_nonzero_warnv_p): Rename to ...
24271 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24273 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24274 (extract_range_basic): Adjust, do not disable propagation on
24275 strict overflow sensitive simplification.
24276 (vrp_visit_cond_stmt): Likewise.
24278 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24280 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24281 body size unconditionally.
24283 2017-05-07 Jeff Law <law@redhat.com>
24286 2017-05-06 Jeff Law <law@redhat.com>
24287 PR tree-optimization/78496
24288 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24291 PR tree-optimization/78496
24292 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24293 (simplify_stmt_using_ranges): Call it.
24294 (vrp_dom_walker::before_dom_children): Extract equivalences
24295 from an ASSERT_EXPR with an equality comparison against a
24298 2017-05-06 Jeff Law <law@redhat.com>
24300 PR tree-optimization/78496
24301 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24304 PR tree-optimization/78496
24305 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24306 (simplify_stmt_using_ranges): Call it.
24307 (vrp_dom_walker::before_dom_children): Extract equivalences
24308 from an ASSERT_EXPR with an equality comparison against a
24311 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24313 * lra-constraints.c (lra_copy_reg_equiv): New function.
24314 (split_reg): Use it to copy equivalence information from the
24315 original register to the spill register.
24317 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24319 PR rtl-optimization/75964
24320 * simplify-rtx.c (simplify_const_relational_operation): Remove
24321 invalid handling of comparisons of integer ABS.
24323 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24325 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24326 initialize to zero.
24327 (init_regs): Remove declaration.
24328 (function_arg_advance_32): Initialize error_p as boolean variable.
24330 2017-05-05 Nathan Sidwell <nathan@acm.org>
24332 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24333 lines. Use for (;;).
24335 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24337 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24338 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24339 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24340 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24341 VF=2 that require versioning.
24343 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24345 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24348 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24350 * diagnostic.h (diagnostic_override_option_index): Convert from
24351 macro to inline function.
24353 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24355 * diagnostic.c (last_module_changed_p): New function.
24356 (set_last_module): New function.
24357 (diagnostic_report_current_module): Convert macro usage to
24358 the above functions.
24359 * diagnostic.h (diagnostic_context::last_module): Strengthen
24360 from const line_map * to const line_map_ordinary *.
24361 (diagnostic_last_module_changed): Delete macro.
24362 (diagnostic_set_last_module): Delete macro.
24364 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24366 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24367 with diagnostic_report_diagnostic.
24368 (diagnostic_n_impl_richloc): Likewise.
24369 * diagnostic.h (report_diagnostic): Delete macro.
24370 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24371 with diagnostic_report_diagnostic.
24372 * substring-locations.c (format_warning_va): Likewise.
24374 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24376 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24377 save/restor of format_spec. Move option-printing code to...
24378 (print_option_information): ...this new function, and
24379 reimplement by simply printing to the pretty_printer,
24380 rather than appending to the format string.
24382 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24384 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24385 handling logic into...
24386 (update_effective_level_from_pragmas): ...this new function.
24388 2017-05-04 Andrew Waterman <andrew@sifive.com>
24390 * config/riscv/riscv.opt (mstrict-align): New option.
24391 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24392 (SLOW_UNALIGNED_ACCESS): Define.
24393 (riscv_slow_unaligned_access): Declare.
24394 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24396 (riscv_slow_unaligned_access): New variable.
24397 (rocket_tune_info): Set slow_unaligned_access to true.
24398 (optimize_size_tune_info): Set slow_unaligned_access to false.
24399 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24400 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24401 (riscv_option_override): Set riscv_slow_unaligned_access.
24402 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24404 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24406 * config/riscv/riscv.md: Unify indentation.
24408 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24413 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24415 (extendsi<mode>2): Add support for doing sign extension via
24416 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24417 don't have ISA 3.0 instructions.
24418 (extendsi<mode>2 splitter): Likewise.
24419 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24420 generate the normal insns since SImode can now go in vector
24421 registers. Disallow the special UNSPECs needed for previous
24422 machines to hide SImode being used. Add new insns
24423 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24424 (fix_trunc<mode>si2_stfiwx): Likewise.
24425 (fix_trunc<mode>si2_internal): Likewise.
24426 (fixuns_trunc<mode>si2): Likewise.
24427 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24428 (fctiw<u>z_<mode>_smallint): Likewise.
24429 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24430 of floating point to 32-bit integer from doing a direct move to
24431 the GPR registers to do a store.
24432 (fctiwz_<mode>): Break long line.
24434 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24436 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24437 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24438 (addr_list, addr_offset_valid_p): New.
24439 (split_address_groups): Check offset validity with above function.
24440 (gt-tree-ssa-loop-ivopts.h): Include header file.
24442 2017-05-05 Nathan Sidwell <nathan@acm.org>
24444 * config.gcc (arm*-*-*): Add missing 'fi'.
24446 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24448 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24449 included in -fopt-info does not matter.
24450 * doc/optinfo.texi (-fopt-info): Fix description of default
24451 behavour. Explicitly say order of options included in -fopt-info
24454 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24456 * config.gcc: Allow combinations of aprofile and rmprofile values for
24457 --with-multilib-list.
24458 * config/arm/t-multilib: New file.
24459 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24460 variables. Remove setting of ISA and floating-point ABI in
24461 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24462 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24463 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24464 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24466 * config/arm/t-rmprofile: Likewise except for the matches changes.
24467 * doc/install.texi (--with-multilib-list): Document the combination of
24468 aprofile and rmprofile values and warn about pitfalls in doing that.
24470 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24472 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24473 (movdi_aarch64): Likewise.
24475 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24477 PR tree-optimization/80632
24478 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24480 (build_arrays): Initialize it for virtual phis.
24481 (fix_phi_nodes): Use it for virtual phis.
24483 PR tree-optimization/80558
24484 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24485 [x, y] op z into [x op, y op z] for op & or | if conditions
24488 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24489 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24492 * config/arm/arm.md (use_literal_pool): Remove.
24493 (64-bit immediate split): No longer takes cost into consideration
24494 if arm_disable_literal_pool is enabled.
24495 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24496 used when arm_disable_literal_pool is enabled.
24497 (arm_max_const_double_inline_cost): Remove use of
24498 arm_disable_literal_pool.
24499 (push_minipool_fix): Add assert.
24500 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24501 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24502 (no_literal_pool_sf_immediate): New.
24504 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24506 PR tree-optimization/80613
24507 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24508 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24510 2017-05-05 Richard Biener <rguenther@suse.de>
24512 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24514 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24516 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24517 of this flag from insn conditions due to removal from r247495.
24519 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24521 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24523 (arm_early_store_addr_dep_ptr): Likewise.
24524 * config/arm/aarch-common-protos.h
24525 (arm_early_load_addr_dep_ptr): Add prototype.
24526 (arm_early_store_addr_dep_ptr): Likewise.
24527 * config/arm/cortex-a53.md: Add new bypasses.
24529 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24531 * tree.c (next_type_uid): Change type to unsigned.
24532 (type_hash_canon): Decrement back next_type_uid if
24533 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24534 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24537 2017-05-04 Martin Sebor <msebor@redhat.com>
24539 * builtins.c: Fix a trivial typo in a comment.
24541 PR middle-end/79234
24542 * builtins.c (check_sizes): Adjust to handle reading past the end.
24543 Avoid printing excessive upper bound of ranges. Use %E to print
24544 tree nodes instead of converting them to %wu.
24545 (expand_builtin_memchr): New function.
24546 (compute_dest_size): Rename...
24547 (compute_objsize): ...to this.
24548 (expand_builtin_memcpy): Adjust.
24549 (expand_builtin_mempcpy): Adjust.
24550 (expand_builtin_strcat): Adjust.
24551 (expand_builtin_strcpy): Adjust.
24552 (check_strncat_sizes): Adjust.
24553 (expand_builtin_strncat): Adjust.
24554 (expand_builtin_strncpy): Adjust and simplify.
24555 (expand_builtin_memset): Adjust.
24556 (expand_builtin_bzero): Adjust.
24557 (expand_builtin_memcmp): Adjust.
24558 (expand_builtin): Handle memcmp.
24559 (maybe_emit_chk_warning): Check strncat just once.
24561 2017-05-04 Martin Sebor <msebor@redhat.com>
24563 PR preprocessor/79214
24564 PR middle-end/79222
24565 PR middle-end/79223
24566 * builtins.c (check_sizes): Add inlining context and issue
24567 warnings even when -Wno-system-headers is set.
24568 (check_strncat_sizes): Same.
24569 (expand_builtin_strncat): Same.
24570 (expand_builtin_memmove): New function.
24571 (expand_builtin_stpncpy): Same.
24572 (expand_builtin): Handle memmove and stpncpy.
24574 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24576 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24577 which is not used any more.
24579 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24581 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24583 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24585 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24586 (cortexa53_tunings): Likewise.
24587 (cortexa57_tunings): Likewise.
24588 (cortexa72_tunings): Likewise.
24589 (cortexa73_tunings): Likewise.
24591 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24593 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24594 Set loop alignment to 8.
24596 2017-05-04 Martin Sebor <msebor@redhat.com>
24598 PR translation/80280
24599 * builtins.c (expand_builtin_object_size): Add missing quoting to
24600 %D and like directives.
24601 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24602 (hsa_type_for_tree_type): Same.
24603 (verify_function_arguments): Same.
24604 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24605 * varasm.c (get_section): Same.
24608 2017-05-04 Martin Sebor <msebor@redhat.com>
24610 PR translation/80280
24611 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24613 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24615 * config/aarch64/aarch64.c (generic_addrcost_table):
24616 Change HI/TI mode setting.
24618 2017-05-04 Martin Jambor <mjambor@suse.cz>
24620 PR tree-optimization/80622
24621 * tree-sra.c (comes_initialized_p): New function.
24622 (build_accesses_from_assign): Only set write lazily when
24623 comes_initialized_p is false.
24624 (analyze_access_subtree): Use comes_initialized_p.
24625 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24626 instead of testing for PARM_DECL.
24628 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24630 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24631 constraint on operand 0 to allow more general addressing modes.
24632 Adjust output template.
24633 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24635 * config/aarch64/aarch64-protos.h
24636 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24637 * config/aarch64/constraints.md (Dp): New address constraint.
24638 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24641 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24643 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24644 update use of estimate_ipcp_clone_size_and_time.
24645 (estimate_local_effects): Update use of
24646 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24647 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24648 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24649 Return nonspecialized time.
24651 2017-05-04 Richard Biener <rguenther@suse.de>
24653 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24654 for the last VUSE which def dominates the PHI. Directly call
24656 (get_continuation_for_phi_1): Remove.
24658 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24660 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24661 to explain the use of truncating division. Cap the number of
24662 iterations to the maximum given by nb_iterations_upper_bound,
24665 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24667 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24668 * configure: Regenerate.
24669 * config.in: Regenerate.
24670 * config/i386/driver-mingw32.c: new file.
24671 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24672 * config.host: Link driver-mingw32.o on MinGW host.
24673 * doc/install.texi: Document new --enable-mingw-wildcard configure
24676 2017-05-04 Marek Polacek <polacek@redhat.com>
24678 PR tree-optimization/80612
24679 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24681 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24682 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24684 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24685 (movt splitter): Likewise.
24686 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24687 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24688 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24689 block for Thumb-1 with MOVT.
24690 (thumb2_legitimate_address_p): Move code block ...
24691 (can_avoid_literal_pool_for_label_p): ... into this new function.
24692 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24694 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24695 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24696 "M-profile targets with the MOVT instruction".
24698 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24700 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24701 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24702 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24704 2017-05-04 Martin Liska <mliska@suse.cz>
24706 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24707 variable cond_code.
24709 2017-05-04 Richard Biener <rguenther@suse.de>
24711 * tree.c (array_at_struct_end_p): Handle arrays at struct
24712 end with flexarrays more conservatively. Refactor and treat
24713 arrays of arrays or aggregates more strict. Fix
24714 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24715 * tree.c (array_at_struct_end_p): Adjust prototype.
24716 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24717 * gimple-fold.c (get_range_strlen): Likewise.
24718 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24720 2017-05-04 Richard Biener <rguenther@suse.de>
24722 PR tree-optimization/31130
24723 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24725 (supports_overflow_infinity): Likewise.
24726 (is_negative_overflow_infinity): Likewise.
24727 (is_positive_overflow_infinity): Likewise.
24728 (is_overflow_infinity): Likewise.
24729 (stmt_overflow_infinity): Likewise.
24730 (overflow_infinity_range_p): Likewise.
24731 (usable_range_p): Remove as always returning true.
24732 (make_overflow_infinity): Remove.
24733 (negative_overflow_infinity): Likewise.
24734 (positive_overflow_infinity): Likewise.
24735 (avoid_overflow_infinity): Likewise.
24736 (set_value_range): Adjust accordingly.
24737 (set_value_range_to_nonnegative): Likewise, remove now unused
24738 overflow_infinity arg.
24739 (vrp_operand_equal_p): Adjust.
24740 (update_value_range): Likewise.
24741 (range_int_cst_singleton_p): Likewise.
24742 (operand_less_p): Likewise.
24743 (compare_values_warnv): Likewise.
24744 (extract_range_for_var_from_comparison_expr): Likewise.
24745 (vrp_int_const_binop): Likewise.
24746 (zero_nonzero_bits_from_vr): Likewise.
24747 (extract_range_from_multiplicative_op_1): Likewise.
24748 (extract_range_from_binary_expr_1): Likewise.
24749 (extract_range_from_unary_expr): Likewise.
24750 (extract_range_from_comparison): Likewise.
24751 (extract_range_basic): Likewise.
24752 (adjust_range_with_scev): Likewise.
24753 (compare_ranges): Likewise.
24754 (compare_range_with_value): Likewise.
24755 (dump_value_range): Likewise.
24756 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24758 (simplify_cond_using_ranges): Adjust.
24760 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24762 * brig-builtins.def: Added a builtin for class_f64.
24763 * builtin-types.def: Added a builtin type needed by class_f64.
24765 2017-05-03 Jason Merrill <jason@redhat.com>
24767 * timevar.def: Add TV_CONSTEXPR.
24769 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24771 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24773 2017-05-03 Martin Jambor <mjambor@suse.cz>
24775 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24776 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24777 * ipa-cp.c (ipcp_propagate_stage): Do not call
24778 ipa_update_after_lto_read.
24779 * ipa-inline.c (ipa_inline): Likewise.
24781 2017-05-03 Martin Jambor <mjambor@suse.cz>
24783 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24784 tag. Added a default constructor and a destructor.
24785 (ipa_edge_args_sum_t): New class;
24786 (ipa_edge_args_sum): Declare.
24787 (ipa_edge_args_vector): Remove declaration.
24788 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24789 (ipa_free_edge_args_substructures): Remove declaration.
24790 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24791 (ipa_edge_args_info_available_for_edge_p): Likewise.
24792 * ipa-prop.c (ipa_edge_args_vector): Removed.
24793 (edge_removal_hook_holder): Likewise.
24794 (edge_duplication_hook_holder): Likewise.
24795 (ipa_edge_args_sum): New variable.
24796 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24797 ipa_edge_args_vector.
24798 (ipa_free_edge_args_substructures): Likewise.
24799 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24800 ipa_edge_args_vector.
24801 (ipa_edge_removal_hook): Turned into method
24802 ipa_edge_args_sum_t::remove.
24803 (ipa_edge_duplication_hook): Turned into method
24804 ipa_edge_args_sum_t::duplicate.
24805 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24806 registering edge hooks.
24807 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24808 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24809 ipa_edge_args_sum instead of ipa_edge_args_vector.
24810 * ipa-profile.c (ipa_profile): Likewise.
24812 2017-05-03 Martin Jambor <mjambor@suse.cz>
24814 * symbol-summary.h (function_summary): New method exists.
24815 (function_summary::symtab_removal): Deallocate through release.
24816 (call_summary): New class.
24817 (gt_ggc_mx): New overload.
24818 (gt_pch_nx): Likewise.
24819 (gt_pch_nx): Likewise.
24821 2017-05-03 Jeff Law <law@redhat.com>
24823 PR tree-optimization/78496
24824 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24825 from simplify_cond_using_ranges. Split off code to walk
24826 backwards through casts into ...
24827 (simplify_cond_using_ranges_2): New function.
24828 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24829 (execute_vrp): After identifying jump threads, call
24830 simplify_cond_using_ranges_2.
24832 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24835 * ipa-inline.h (inline_summary): Add ctor.
24836 (create_ggc): Do not use ggc_cleared_alloc.
24838 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24839 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24841 * gcc.c (handle_braces): Support escaping in switch matching
24843 * doc/invoke.texi (Spec Files): Document it.
24844 Remove superfluous @code markup in items.
24846 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24848 * diagnostic-show-locus.c (struct column_range): New struct.
24849 (get_affected_columns): New function.
24850 (get_printed_columns): New function.
24851 (struct correction): New struct.
24852 (correction::ensure_capacity): New function.
24853 (correction::ensure_terminated): New function.
24854 (struct line_corrections): New struct.
24855 (line_corrections::~line_corrections): New dtor.
24856 (line_corrections::add_hint): New function.
24857 (layout::print_trailing_fixits): Reimplement in terms of the new
24859 (selftest::test_overlapped_fixit_printing): New function.
24860 (selftest::diagnostic_show_locus_c_tests): Call it.
24862 2017-05-03 Nathan Sidwell <nathan@acm.org>
24864 Canonicalize canonical type hashing
24865 * tree.h (type_hash_canon_hash): Declare.
24866 * tree.c (type_hash_list, attribute_hash_list): Move into
24867 type_hash_canon_hash.
24868 (build_type_attribute_qual_variant): Break out hash code calc into
24869 type_hash_canon_hash.
24870 (type_hash_canon_hash): New. Generic type hash computation.
24871 (build_range_type_1, build_array_type_1, build_function_type,
24872 build_method_type_directly, build_offset_type, build_complex_type,
24873 make_vector_type): Call it.
24875 2017-05-03 Richard Biener <rguenther@suse.de>
24877 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24878 When all DRs have unknown misaligned do not always peel
24879 when there is a store but apply the same costing model as if
24880 there were only loads.
24882 2017-05-03 Richard Biener <rguenther@suse.de>
24885 PR tree-optimization/80492
24886 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24887 compare_base_decls returning dont-know properly.
24889 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24891 * config/arm/iterators.md (CCSI): New mode iterator.
24892 (arch): New mode attribute.
24893 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24894 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24895 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24896 code iterator for success result mode.
24897 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24898 the corresponding new insn generators.
24900 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24903 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24904 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24906 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24908 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24909 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24910 (DDR_A): Wrap DDR argument in brackets.
24911 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24912 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24913 (DDR_REVERSED_P): Likewise.
24915 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24917 PR tree-optimization/79472
24918 * tree-switch-conversion.c (struct switch_conv_info): Add
24919 contiguous_range and default_case_nonstandard fields.
24920 (collect_switch_conv_info): Compute contiguous_range and
24921 default_case_nonstandard fields, don't clear final_bb if
24922 contiguous_range and only the default case doesn't have the required
24924 (check_all_empty_except_final): Set default_case_nonstandard instead
24925 of failing if contiguous_range and the default case doesn't have empty
24927 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24928 and only the default case doesn't have the required constants. Skip
24930 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24931 if default_case_nonstandard.
24932 (build_constructors): Build constant 1 just once. Assert that default
24933 values aren't inserted in between cases if contiguous_range. Skip
24935 (build_arrays): Skip virtual phis.
24936 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24937 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24938 Handle virtual phis.
24939 (gen_inbound_check): Handle default_case_nonstandard case.
24940 (process_switch): Adjust check_final_bb caller. Call
24941 gather_default_values with the first non-default case instead of
24942 default case if default_case_nonstandard.
24944 2017-05-02 Nathan Sidwell <nathan@acm.org>
24946 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24947 check. Fix formatting.
24949 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24951 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24952 errors when comparing specialized and unspecialized times.
24954 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24956 * diagnostic-show-locus.c
24957 (layout::should_print_annotation_line_p): Make private.
24958 (layout::print_annotation_line): Make private.
24959 (layout::annotation_line_showed_range_p): Make private.
24960 (layout::show_ruler): Make private.
24961 (layout::print_source_line): Make private. Pass in line and
24962 line_width, rather than calling location_get_source_line. Drop
24964 (layout::print_leading_fixits): New method.
24965 (layout::print_any_fixits): Rename to...
24966 (layout::print_trailing_fixits): ...this, and make private.
24967 Don't print newline fixits.
24968 (diagnostic_show_locus): Move logic for printing one row into...
24969 (layout::print_line): ...this new function. Move the
24970 location_get_source_line call and error-handling from
24971 print_source_line to here. Call print_leading_fixits, and rename
24972 print_any_fixits to print_trailing_fixits.
24973 (selftest::test_fixit_insert_containing_newline): Update now that
24974 newlines are partially supported.
24975 (selftest::test_fixit_insert_containing_newline_2): New test.
24976 (selftest::test_fixit_replace_containing_newline): Update comments.
24977 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24978 * edit-context.c (class added_line): New class.
24979 (class edited_line): Describe newline handling in comment.
24980 (edited_line::actually_edited_p): New method.
24981 (edited_line::print_content): Delete redundant decl.
24982 (edited_line::m_predecessors): New field.
24983 (edited_file::print_content): Call edited_line::print_content.
24984 (edited_file::print_diff): Update to support newlines.
24985 (edited_file::print_diff_hunk): Likewise.
24986 (edited_file::print_run_of_changed_lines): New function.
24987 (edited_file::print_diff_line): Convert to...
24988 (print_diff_line): ...this.
24989 (edited_file::get_effective_line_count): New function.
24990 (edited_line::edited_line): Initialize new field m_predecessors.
24991 (edited_line::~edited_line): Clean up m_predecessors.
24992 (edited_line::apply_fixit): Handle newlines.
24993 (edited_line::get_effective_line_count): New function.
24994 (edited_line::print_content): New function.
24995 (edited_line::print_diff_lines): New function.
24996 (selftest::test_applying_fixits_insert_containing_newline): New
24998 (selftest::test_applying_fixits_replace_containing_newline): New
25000 (selftest::insert_line): New function.
25001 (selftest::test_applying_fixits_multiple_lines): Add example of
25003 (selftest::edit_context_c_tests): Call the new tests.
25005 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25007 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25008 parameter cand. Update dump information.
25009 (get_computation_cost): Update uses.
25011 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25013 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25014 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25015 (get_computation_at, rewrite_use_address): Update use of
25016 get_computation_aff.
25018 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25020 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25021 (get_computation): Delete.
25022 (get_computation_cost): Implement like get_computation_cost_at.
25023 Use get_computation_at.
25024 (get_computation_cost_at): Delete.
25025 (rewrite_use_nonlinear_expr): Use get_computation_at.
25026 (rewrite_use_compare, remove_unused_ivs): Ditto.
25028 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25030 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25032 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25034 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25035 (ivopts_global_cost_for_size): Rename parameter and update uses.
25036 (iv_ca_recount_cost): Update uses.
25037 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25038 candidates seperately in n_invs and n_cands.
25039 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25041 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25043 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25044 (find_inv_vars_cb): New.
25045 (find_depends): Renamed to ...
25046 (find_inv_vars): ... this.
25047 (add_candidate_1, force_var_cost): Call find_inv_vars.
25048 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25050 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25052 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25053 inv_vars. Add inv_exprs.
25054 (struct iv_cand): Rename depends_on to inv_vars.
25055 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25056 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25057 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25058 (dump_cand): Dump inv_vars.
25059 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25060 (record_invariant, find_depends, add_candidate_1): Ditto.
25061 (set_group_iv_cost, force_var_cost): Ditto.
25062 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25063 (get_computation_cost_at, get_computation_cost): Ditto.
25064 (determine_group_iv_cost_generic): Ditto.
25065 (determine_group_iv_cost_address): Ditto.
25066 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25067 (determine_group_iv_costs): Ditto.
25068 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25069 (iv_ca_set_remove_invariants): Renamed to ...
25070 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25071 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25072 (iv_ca_set_add_invariants): Renamed to ...
25073 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25074 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25075 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25076 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25077 (create_new_ivs): Remove useless dump.
25079 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25081 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25083 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25084 (iv_ca_set_no_cp, create_new_iv): Ditto.
25086 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25088 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25090 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25092 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25093 function tree_check2.
25095 2017-05-02 Martin Liska <mliska@suse.cz>
25097 * doc/gcov.texi: Add missing preposition.
25098 * gcov.c (function_info::function_info): Properly fill up
25099 all member variables.
25101 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25103 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25105 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25107 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25109 2017-05-02 Martin Liska <mliska@suse.cz>
25112 * lto-streamer-in.c (lto_read_tree_1): Remove
25113 LTO_STREAMER_DEBUG.
25114 * lto-streamer.c (struct tree_hash_entry): Likewise.
25115 (struct tree_entry_hasher): Likewise.
25116 (tree_entry_hasher::hash): Likewise.
25117 (tree_entry_hasher::equal): Likewise.
25118 (lto_streamer_init): Likewise.
25119 (lto_orig_address_map): Likewise.
25120 (lto_orig_address_get): Likewise.
25121 (lto_orig_address_remove): Likewise.
25122 * lto-streamer.h: Likewise.
25123 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25124 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25126 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25128 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25129 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25130 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25131 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25132 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25133 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25134 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25135 (mm_maskz_sub_ss): New intrinsics.
25136 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25137 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25138 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25139 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25140 (__builtin_ia32_subss_mask_round): New builtins.
25141 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25142 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25143 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25145 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25146 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25148 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25151 2017-05-02 Martin Jambor <mjambor@suse.cz>
25153 PR tree-optimization/78687
25154 * tree-sra.c (access): New field parent.
25155 (process_subtree_disqualification): New function.
25156 (disqualify_candidate): Call it.
25157 (build_accesses_from_assign): Reset write flag if creating an
25159 (build_access_subtree): Fill in parent field and also prpagate
25160 down grp_write flag.
25161 (create_artificial_child_access): New parameter set_grp_write, set
25162 grp_write to its value.
25163 (propagate_subaccesses_across_link): Also propagate grp_write flag
25165 (propagate_all_subaccesses): Push the closest parent back to work
25166 queue if add_access_to_work_queue returned true.
25168 2017-05-02 Richard Biener <rguenther@suse.de>
25170 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25171 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25172 -fstrict-overflow documentation.
25173 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25174 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25175 flag_strict_overflow.
25176 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25177 * lto-opts.c (lto_write_options): Do not stream it.
25178 * lto-wrapper.c (merge_and_complain): Do not handle it.
25179 * opts.c (default_options_table): Do not set -fstrict-overflow.
25180 (finish_options): Likewise do not clear it when sanitizing.
25181 * simplify-rtx.c (simplify_const_relational_operation): Do not
25182 test flag_strict_overflow.
25184 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25186 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25187 using enabled attribute.
25188 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25189 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25190 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25191 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25192 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25193 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25194 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25195 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25196 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25197 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25199 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25201 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25203 2017-05-02 Richard Biener <rguenther@suse.de>
25205 PR tree-optimization/80591
25207 2017-04-10 Richard Biener <rguenther@suse.de>
25209 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25212 2017-05-02 Richard Biener <rguenther@suse.de>
25214 PR tree-optimization/80549
25215 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25216 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25217 headers if they do not have a preheader.
25219 2017-05-02 Martin Liska <mliska@suse.cz>
25222 * common.opt: Fix typo.
25223 * doc/invoke.texi: Likewise.
25225 2017-05-01 Jan Beulich <jbeulich@suse.com>
25227 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25228 swapping, add (x,x,m,x,n) alternative.
25230 2017-05-01 Nathan Sidwell <nathan@acm.org>
25232 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25233 unnecessary unadjusted_alignment check.
25235 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25238 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25240 * gimplify.c (gimplify_cilk_detach): New function.
25241 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25242 * tree-core.h: Document EXPR_CILK_SPAWN.
25243 * tree.h (EXPR_CILK_SPAWN): Define.
25245 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25247 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25248 to use new fixit_hint representation, using the "replace" logic.
25249 (get_line_span_for_fixit_hint): Likewise.
25250 (layout::print_any_fixits): Likewise.
25251 (selftest::test_one_liner_many_fixits): Rename to...
25252 (selftest::test_one_liner_many_fixits_1): ...this, and update
25253 comment and expected output to reflect that the multiple fix-it
25254 hints are now consolidated into one insertion.
25255 (selftest::test_one_liner_many_fixits_2): New test.
25256 (selftest::test_diagnostic_show_locus_one_liner): Update for
25258 (selftest::test_fixit_consolidation): Update for fix-it API
25260 * diagnostic.c (print_parseable_fixits): Likewise.
25261 * edit-context.c (edited_line::m_line_events): Convert from
25262 auto_vec <line_event *> to auto_vec <line_event>.
25263 (class line_event): Convert from abstract base class to a concrete
25264 class, taking over the role of replace_event.
25265 (class insert_event): Delete.
25266 (class replace_event): Rename to class line_event. Convert to
25268 (edit_context::add_fixits): Reimplement.
25269 (edit_context::apply_insert): Delete.
25270 (edit_context::apply_replace): Rename to...
25271 (edit_context::apply_fixit): ...this. Convert to half-open range.
25272 (edited_file::apply_insert): Delete.
25273 (edited_file::apply_replace): Rename to...
25274 (edited_file::apply_fixit): ...this.
25275 (edited_line::~edited_line): Drop deletion of events.
25276 (edited_line::apply_insert): Delete.
25277 (edited_line::apply_replace): Rename to...
25278 (edited_line::apply_fixit): ...this. Convert to half-open range.
25279 Update for change to type of m_line_events.
25280 * edit-context.h (edit_context::apply_insert): Delete.
25281 (edit_context::apply_replace): Rename to...
25282 (edit_context::apply_fixit): ...this.
25284 2017-05-01 Martin Sebor <msebor@redhat.com>
25286 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25289 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25292 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25293 __get_cpuid_max returns 0.
25294 (__get_cpuid_count): Ditto.
25296 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25298 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25299 replacement expression is another instance of one of its arguments.
25301 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25304 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25305 check for stack push/pop autoinc.
25306 * config/i386/i386.c (ix86_agi_dependent): Return false
25307 if the only reason why modified_in_p returned true is that
25308 addr is SP based and set_insn is a push or pop.
25310 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25312 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25315 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25318 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25319 (account_size_time): Use two predicates - exec_pred and
25321 (evaluate_conditions_for_known_args): Compute both clause and
25323 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25324 (inline_summary_t::duplicate): Update.
25325 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25327 (compute_inline_parameters): Likewise.
25328 (estimate_edge_size_and_time): Update caluclation of time.
25329 (estimate_node_size_and_time): Compute both time and nonspecialized
25331 (estimate_ipcp_clone_size_and_time): Update.
25332 (inline_merge_summary): Update.
25333 (do_estimate_edge_time): Update.
25334 (do_estimate_edge_size): Update.
25335 (do_estimate_edge_hints): Update.
25336 (inline_read_section, inline_write_summary): Stream both new predicates.
25337 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25339 (compute_inlined_call_time): Cleanup.
25340 (big_speedup_p): Update.
25341 (edge_badness): Update.
25342 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25343 (size_time_entry): Replace predicate by exec_predicate and
25344 nonconst_predicate.
25345 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25346 (estimate_edge_time): Return also nonspec_time.
25347 (reset_edge_growth_cache): Update.
25349 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25351 PR rtl-optimization/80491
25352 * ifcvt.c (noce_process_if_block): When looking for x setter
25353 with missing else_bb, don't check only the insn right before
25354 cond_earliest, but look for the last insn that x is modified in
25355 within the same bb.
25357 PR rtl-optimization/80491
25358 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25360 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25362 PR tree-optimization/80487
25363 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25365 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25367 PR tree-optimization/79697
25368 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25369 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25370 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25372 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25373 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25375 2017-04-28 Martin Sebor <msebor@redhat.com>
25377 PR tree-optimization/80523
25378 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25379 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25381 (maybe_warn, format_directive, parse_directive): Use new functions.
25382 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25384 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25386 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25388 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25390 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25391 target_header_dir): Set correctly.
25392 * configure: Regenerated.
25393 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25394 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25395 instead of SYSTEM_HEADER_DIR.
25397 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25399 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25400 (estimate_local_effects): Likewise.
25401 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25402 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25403 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25404 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25405 do_estimate_edge_time, estimate_edge_time): Likewise.
25406 * ipa-inline-analysis.c (estimate_node_size_and_time,
25407 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25408 (estimate_time_after_inlining): Remove.
25410 2017-04-28 Martin Liska <mliska@suse.cz>
25412 * doc/gcov.texi: Enhance documentation of gcov.
25414 2017-04-28 Martin Liska <mliska@suse.cz>
25416 * doc/gcov.texi: Sort options in alphabetic order.
25417 * doc/gcov-dump.texi: Likewise.
25418 * doc/gcov-tool.texi: Likewise.
25419 * gcov.c (print_usage): Likewise.
25420 * gcov-dump.c (print_usage): Likewise.
25421 * gcov-tool.c (print_merge_usage_message): Likewise.
25422 (print_rewrite_usage_message): Likewise.
25423 (print_overlap_usage_message): Likewise.
25425 2017-04-28 Martin Liska <mliska@suse.cz>
25427 PR gcov-profile/53915
25428 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25430 2017-04-28 Martin Liska <mliska@suse.cz>
25432 PR gcov-profile/79891
25433 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25434 is marked by compiler as living on a line.
25435 (get_cycles_count): Remove usage of the union.
25436 (output_intermediate_file): Likewise.
25437 (find_source): Fix GNU coding style.
25438 (accumulate_line_counts): Remove old non-all block mode.
25439 (output_lines): Remove usage of the union.
25440 * profile.c (output_location): Include all BBs, even if
25441 belonging to a same line (and file) as a previous BB.
25443 2017-04-28 Martin Liska <mliska@suse.cz>
25445 * gcov.c (process_args): Handle new argument 'w'.
25446 (read_graph_file): Assign ID to BBs.
25447 (output_branch_count): Display BB # if verbose flag is set.
25448 (output_lines): Likewise for arcs.
25449 (print_usage): Add '--verbose' option help.
25450 * doc/gcov.texi: Document --verbose (-w) option.
25452 2017-04-28 Martin Liska <mliska@suse.cz>
25454 * gcov.c (struct block_location_info): New struct.
25455 (process_file): Fill up the new structure.
25456 (read_graph_file): Replace usage of encoding by the newly added
25458 (add_line_counts): Likewise.
25459 (accumulate_line_counts): Remove usage of the union.
25460 (function_info::function_info): New function.
25461 (function_info::~function_info): Likewise.
25462 (process_file): Call delete instead of release_function.
25463 (release_function): Release the function.
25464 (release_structures): Call delete instead of release_function.
25465 (solve_flow_graph): Replace usage of num_blocks.
25466 (find_exception_blocks): Likewise.
25467 (output_lines): Fix GNU coding style.
25469 2017-04-28 Martin Liska <mliska@suse.cz>
25472 * coverage.c (coverage_remove_note_file): New function.
25473 * coverage.h: Declare the function.
25474 * toplev.c (finalize): Clean if an error has been seen.
25476 2017-04-28 Martin Liska <mliska@suse.cz>
25478 PR gcov-profile/80031
25479 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25480 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25481 * gcov.c (read_graph_file): Read just number of blocks.
25482 * profile.c (branch_prob): Do not stream 0 flags per a basic
25485 2017-04-28 Martin Liska <mliska@suse.cz>
25487 * gcov-dump.c (tag_*): Add new argument to declarations.
25488 (dump_gcov_file): Likewise.
25489 (tag_blocks): Add and use new argument depth.
25490 (tag_arcs): Likewise.
25491 (tag_lines): Likewise.
25492 (tag_counters): Likewise.
25493 (tag_summary): Likewise.
25494 (dump_working_sets): Use depth to do a proper indentation.
25496 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25499 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25500 * symtab.c (symtab_node::debug_symtab): Move definition here.
25502 2017-04-28 Richard Biener <rguenther@suse.de>
25504 * lto-streamer.h (LTO_major_version): Bump to 7.
25506 2017-04-28 Richard Biener <rguenther@suse.de>
25508 * tree-vrp.c (assert_info): New struct.
25509 (add_assert_info): New helper.
25510 (register_edge_assert_for_2): Refactor to add asserts to a vector
25512 (register_edge_assert_for_1): Likewise.
25513 (register_edge_assert_for): Likewise.
25514 (finish_register_edge_assert_for): New helper actually registering
25515 asserts where live on edge.
25516 (find_conditional_asserts): Adjust.
25517 (find_switch_asserts): Likewise.
25518 (evrp_dom_walker::try_find_new_range): Generalize.
25519 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25521 2017-04-27 Marek Polacek <polacek@redhat.com>
25524 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25525 arg10 and arg11 to itype.
25527 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25529 * doc/extend.texi (Object Size Checking): Improve grammar.
25531 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25534 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25535 that the logic for permitting reciprocal estimates matches that
25538 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25541 * tree.c (type_cache_hasher::equal): Only compare
25542 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25543 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25544 non-aggregate element types.
25545 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25546 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25548 2017-04-27 Richard Biener <rguenther@suse.de>
25550 PR middle-end/80533
25551 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25552 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25553 keeping a reference to a trailing array.
25555 2017-04-27 Richard Biener <rguenther@suse.de>
25557 PR middle-end/80539
25558 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25559 being in loop-closed SSA form conservatively.
25560 (chrec_fold_multiply_poly_poly): Likewise.
25562 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25564 PR middle-end/79665
25565 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25566 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25568 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25571 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25572 (aarch64_function_arg_alignment): Return unsigned int again, but still
25573 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25574 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25575 Don't emit -Wpsabi note.
25576 (aarch64_function_arg_boundary): Likewise.
25577 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25580 2017-04-26 Nathan Sidwell <nathan@acm.org>
25582 * tree.h (crc32_unsigned_n): Declare.
25583 (crc32_unsigned, crc32_unsigned): Make inline.
25584 * tree.c (crc32_unsigned_bits): Replace with ...
25585 (crc32_unsigned_n): ... this.
25586 (crc32_unsigned, crc32_byte): Remove.
25587 (crc32_string): Remove unnecessary braces.
25589 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25591 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25592 * ipa-inline-analysis.c (MAX_TIME): Remove.
25593 (account_size_time): Use sreal for time.
25594 (dump_inline_summary): Update.
25595 (estimate_function_body_sizes): Update.
25596 (estimate_edge_size_and_time): Update.
25597 (estimate_calls_size_and_time): Update.
25598 (estimate_node_size_and_time): Update.
25599 (inline_merge_summary): Update.
25600 (inline_update_overall_summary): Update.
25601 (estimate_time_after_inlining): Update.
25602 (inline_read_section): Update.
25603 (inline_write_summary): Update.
25604 * ipa-inline.c (compute_uninlined_call_time): Update.
25605 (compute_inlined_call_time): Update.
25606 (recursive_inlining): Update.
25607 (inline_small_functions): Update.
25608 (dump_overall_stats): Update.
25609 * ipa-inline.h: Include sreal.h.
25610 (size_time_entry): Turn time to sreal.
25611 (inline_summary): Turn self_time nad time to sreal.
25613 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25615 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25617 (sreal::stream_out, sreal::stream_in): New.
25618 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25620 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25622 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25625 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25628 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25630 (dimode_scalar_chain::compute_convert_gain): Ditto.
25631 (dimode_scalar_chain::make_vector_copies): Ditto.
25632 (dimode_scalar_chain::convert_reg): Ditto.
25633 (dimode_scalar_chain::convert_insn): Ditto.
25634 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25635 (VI248_AVX512BW_1): New mode iterator.
25636 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25637 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25640 2017-04-25 Martin Sebor <msebor@redhat.com>
25642 PR tree-optimization/80497
25643 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25644 constants are representable in HOST_WIDE_INT.
25645 (parse_directive): Ditto.
25647 2017-04-25 Martin Sebor <msebor@redhat.com>
25650 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25651 (new_zero_array): Adjust signature.
25652 (dom_info::dom_init): Used unsigned rather that size_t.
25653 (dom_info::dom_info): Same.
25655 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25656 Jakub Jelinek <jakub@redhat.com>
25659 * config/arm/arm.c: Include gimple.h.
25660 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25661 returns negative, increment ncrn only if it returned positive.
25662 (arm_needs_doubleword_align): Return int instead of bool,
25663 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25664 members, but if there is any such non-FIELD_DECL
25665 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25666 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25667 returns negative, increment nregs only if it returned positive.
25668 (arm_setup_incoming_varargs): Likewise.
25669 (arm_function_arg_boundary): Emit -Wpsabi note if
25670 arm_needs_doubleword_align returns negative, return
25671 DOUBLEWORD_ALIGNMENT only if it returned positive.
25673 2017-04-25 Marek Polacek <polacek@redhat.com>
25676 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25677 first argument to type.
25679 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25682 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25683 type checks to test for compatibility instead of equality.
25685 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25686 Jakub Jelinek <jakub@redhat.com>
25689 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25691 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25692 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25693 the alignment computation, but return their maximum in warn_alignment.
25694 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25695 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25697 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25698 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25701 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25703 * config/arc/simdext.md (dmpyh): Fix typo.
25705 2017-04-25 Richard Biener <rguenther@suse.de>
25707 PR tree-optimization/80492
25708 * alias.c (compare_base_decls): Handle registers with asm
25709 specification conservatively.
25710 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25711 compare_base_decls returning dont-know properly.
25713 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25715 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25716 (legitimate_offset_address_p): New function.
25717 (arc_legitimate_address_p): Use above function.
25719 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25721 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25723 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25725 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25726 ACCH registers whenever they are available.
25728 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25730 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25731 double regs fix when not used.
25733 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25735 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25737 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25738 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25740 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25742 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25743 when emitting short ADD instructions.
25745 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25747 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25749 (cmpsi_cc_c_insn): Likewise.
25750 (cbranchsi4_scratch): Compute proper instruction length using
25751 compact_hreg_operand.
25752 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25754 2017-04-25 Richard Biener <rguenther@suse.de>
25756 PR middle-end/80509
25757 * passes.c (pass_manager::pass_manager): Initialize
25758 m_name_to_pass_map.
25760 2017-04-25 Richard Biener <rguenther@suse.de>
25762 PR tree-optimization/79201
25763 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25765 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25768 * config/s390/vector.md: Split MEM->GPR vector moves for
25769 non-s_operand addresses.
25771 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25774 * config/s390/predicates.md (reload_const_wide_int_operand): New
25776 * config/s390/s390.md ("movti"): Remove d/P alternative.
25777 ("movti_bigconst"): New pattern definition.
25779 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25782 * s390-protos.h (s390_expand_cs_hqi): Removed.
25783 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25784 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25785 modes as well as CCZ1mode and CCZmode.
25786 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25787 signature of s390_emit_compare_and_swap.
25788 (s390_expand_cs_hqi): Likewise, make static.
25789 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25790 compare-and-swap, in some cases.
25791 (s390_expand_cs): Wrapper function.
25792 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25794 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25795 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25796 patterns for small and large integers. Forbid symref memory operands.
25797 Move expander to s390.c. Require cc register.
25798 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25799 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25800 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25801 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25802 symref memory operands. Remove CC mode and call s390_match_ccmode
25804 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25806 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25808 * config/s390/s390.md (define_peephole2): New peephole to help
25809 combining the load-and-test pattern with volatile memory.
25811 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25813 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25814 with CCZmode for TARGET_Z196.
25816 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25818 PR rtl-optimization/80501
25819 * combine.c (make_compound_operation_int): Set subreg_code to SET
25820 even for AND with mask of the sign bit of mode.
25822 PR rtl-optimization/80500
25823 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25824 sum's initial value.
25826 2017-04-25 Julian Brown <julian@codesourcery.com>
25827 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25829 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25831 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25833 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25835 2017-04-25 Julian Brown <julian@codesourcery.com>
25836 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25838 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25839 (thunderx2t99_sha): New Reservation.
25841 2017-04-25 Julian Brown <julian@codesourcery.com>
25842 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25844 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25845 type for 1-element load.
25847 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25849 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25851 2017-04-24 Martin Jambor <mjambor@suse.cz>
25853 PR tree-optimization/80293
25854 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25855 char arrays not totally scalarizable if it is false.
25856 (analyze_all_variable_accesses): Pass correct value in the new
25857 parameter. Add a statistics counter.
25859 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25861 PR middle-end/79931
25862 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25864 2017-04-24 Richard Biener <rguenther@suse.de>
25866 PR tree-optimization/80494
25867 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25868 out for complex types.
25870 2017-04-24 Richard Biener <rguenther@suse.de>
25872 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25873 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25874 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25875 (DFS): Adjust and never fail.
25876 (sccvn_dom_walker::fail): Remove.
25877 (sccvn_dom_walker::before_dom_children): Adjust.
25878 (run_scc_vn): Likewise and never fail.
25879 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25880 (pass_fre::execute): Likewise.
25882 2017-04-24 Richard Biener <rguenther@suse.de>
25884 PR tree-optimization/79725
25885 * tree-ssa-sink.c (statement_sink_location): Return whether
25886 failure reason was zero uses. Move that check later.
25887 (sink_code_in_bb): Deal with zero uses by removing the stmt
25890 2017-04-24 Richard Biener <rguenther@suse.de>
25893 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25894 pointer-based references.
25896 2017-04-24 Richard Biener <rguenther@suse.de>
25899 * pass_manager.h (pass_manager::operator new): Remove.
25900 (pass_manager::operator delete): Likewise.
25901 * passes.c (pass_manager::operator new): Remove.
25902 (pass_manager::operator delete): Likewise.
25903 (pass_manager::pass_manager): Zero individual pass members.
25905 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25908 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25909 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25910 Check "XEXP (src, 1)" operand here.
25911 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25912 Check "XEXP (src, 1)" operand here.
25913 (dimode_scalar_chain::make_vector_copies): Detect count register
25914 of a shift instruction. Zero extend count register from QImode
25915 to DImode to satisfy vector shift pattern count operand predicate.
25916 Substitute vector shift count operand with a DImode copy.
25917 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25920 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25922 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25923 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25924 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25925 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25926 (UNSPEC_NOREX_MEM): Remove definition.
25928 2017-04-21 Richard Biener <rguenther@suse.de>
25930 PR tree-optimization/79547
25931 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25932 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25933 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25934 without any constraints.
25936 2017-04-21 Richard Biener <rguenther@suse.de>
25938 PR tree-optimization/78847
25939 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25941 2017-04-21 Richard Biener <rguenther@suse.de>
25943 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25944 (build_distinct_type_copy): Likewise.
25945 (build_variant_type_copy): Likewise.
25946 * tree.c (build_qualified_type): Pass down mem-stat info.
25947 (build_distinct_type_copy): Likewise.
25948 (build_variant_type_copy): Likewise.
25950 2017-04-21 Richard Biener <rguenther@suse.de>
25952 PR tree-optimization/80237
25953 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25955 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25956 for a simplified result.
25958 2016-04-21 Richard Biener <rguenther@suse.de>
25960 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25961 sth as strict as a simple_iv but a chrec without symbols and an
25962 operand defined in the loop we are peeling (and not some subloop).
25963 (propagate_constants_for_unrolling): Propagate all constants.
25965 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25968 * config/i386/i386.c (print_reg): Remove assert for disalowed
25969 regno values, call output_operand_lossage instead.
25971 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25974 * config/i386/constraints.md (Yc): New register constraint.
25975 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25976 Use Yc constraint for alternative 2 of operand 0. Remove
25977 preferred_for_speed attribute.
25979 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25981 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25982 lastprivate clauses in SIMT case.
25984 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25986 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25988 2017-04-20 Richard Biener <rguenther@suse.de>
25990 PR tree-optimization/57796
25991 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25992 as N scalar stores.
25993 (vect_model_load_cost): Cost gathers as N scalar loads.
25995 2017-04-20 Richard Biener <rguenther@suse.de>
25997 * ggc-page.c (ggc_allocated_p): Rename to ...
25998 (safe_lookup_page_table_entry): ... this and return the lookup
26000 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26002 2017-04-20 Richard Biener <rguenther@suse.de>
26004 PR tree-optimization/80453
26005 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26006 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26007 from the conditions.
26008 (vn_phi_eq): Pass them down.
26009 (vn_phi_lookup): Record them.
26010 (vn_phi_insert): Likewise.
26012 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26014 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26015 uninitialized variable warning to avoid buffer overrun.
26017 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26020 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26021 is suppressed for '{ 0 }' in C.
26023 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26025 * BASE-VER: Set to 8.0.0.
26027 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26029 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26030 priority .init_array and .fini_array section with SECTION_NOTYPE
26033 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26035 PR middle-end/80423
26036 * tree.h (build_array_type): Add typeless_storage default argument.
26037 * tree.c (type_cache_hasher::equal): Also compare
26038 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26039 (build_array_type): Add typeless_storage argument, set
26040 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26042 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26043 (build_array_type): Likewise. Add typeless_storage argument.
26045 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26046 Jakub Jelinek <jakub@redhat.com>
26048 PR tree-optimization/80426
26049 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26050 operation on symbolic operands, also compute the overflow for the
26051 invariant part when the operation degenerates into a negation.
26053 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26056 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26057 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26060 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26062 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26065 * config/avr/avr.c (tree.h): Include it.
26066 (cgraph.h): Include it.
26067 (avr_encode_section_info): Don't warn for uninitialized progmem
26068 variable if it's just an alias.
26070 2017-04-19 Richard Biener <rguenther@suse.de>
26073 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26074 when needed by AutoPGO.
26076 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26079 * doc/lto.texi: Remove an extra 'that'.
26081 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26083 PR rtl-optimization/80429
26084 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26085 are only used in debug insns.
26087 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26088 Vladimir Makarov <vmakarov@redhat.com>
26090 * config/sparc/predicates.md (input_operand): Add comment. Return
26091 true for any memory operand when LRA is in progress.
26092 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26094 2017-04-18 Jeff Law <law@redhat.com>
26097 * mips.md ({return,simple_return}_internal): Do not overwrite
26100 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26102 PR tree-optimization/80443
26103 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26104 instead of adding 1, subtract -1 and similarly instead of subtracting
26107 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26109 PR rtl-optimization/80357
26110 * haifa-sched.c (tmp_bitmap): New variable.
26111 (model_recompute): Handle duplicate use records.
26112 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26113 (free_global_sched_pressure_data): Free it.
26115 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26118 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26119 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26120 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26121 instead of SYSTEM_HEADER_DIR.
26123 2017-04-18 Jeff Law <law@redhat.com>
26125 PR middle-end/80422
26126 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26127 predecessors after walking up the insn chain.
26129 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26132 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26133 sizetype type into debug info.
26135 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26138 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26139 unneeded test for TARGET_UPPER_REGS_SF.
26140 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26142 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26145 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26146 instead of gsi_after_labels.
26148 2017-04-18 Jeff Law <law@redhat.com>
26150 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26154 2017-04-13 Jeff Law <law@redhat.com>
26155 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26156 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26158 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26161 * config/avr/avr.c (intl.h): Include it.
26162 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26164 2017-04-18 Martin Liska <mliska@suse.cz>
26166 PR gcov-profile/78783
26167 * gcov-tool.c (gcov_output_files): Validate that destination
26168 file is either removed by the tool or by a user.
26170 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26171 Guy Benyei <guybe@mellanox.com>
26173 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26174 block, and do not negate it, the stored id is already negative.
26176 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26178 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26180 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26183 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26184 masks of options that should be turned off if the VSX vector
26185 options are turned off.
26186 (OTHER_P8_VECTOR_MASKS): Likewise.
26187 (OTHER_VSX_VECTOR_MASKS): Likewise.
26188 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26189 rs6000_disable_incompatible_switches to validate no type switches
26191 (rs6000_incompatible_switch): New function to disallow turning on
26192 other vector options if -mno-vsx, -mno-power8-vector, or
26193 -mno-power9-vector are specified.
26195 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26197 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26199 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26201 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26202 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26203 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26204 (ARG_POINTER_CFA_OFFSET): Likewise.
26206 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26208 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26209 conditions to take advantage of various optimizations.
26211 2017-04-13 Jeff Law <law@redhat.com>
26213 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26214 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26215 (zero_extendsidi2_dext): Likewise.
26217 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26220 * fold-const.c (fold_ternary_loc): Revert
26221 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26224 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26226 PR rtl-optimization/80343
26227 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26228 new scratch pseudo.
26230 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26233 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26234 to ubsan_encode_value.
26236 2017-04-13 Jeff Law <law@redhat.com>
26238 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26239 appearing in DEBUG_INSNs.
26241 2017-04-13 Martin Liska <mliska@suse.cz>
26243 PR gcov-profile/80413
26244 * gcov-io.c (gcov_write_string): Copy to buffer just when
26245 allocated size is greater than zero.
26247 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26250 * dwarf2out.c (decls_for_scope): Ignore declarations of
26251 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26253 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26256 * ipa-visibility.c (non_local_p): Fix typos.
26257 (localize_node): When localizing symbol in same comdat group,
26258 dissolve the group only when we know external symbols are going
26260 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26262 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26264 PR tree-optimization/79390
26265 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26266 order does not result in usable sequence, retry with reversed operand
26272 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26273 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26274 op0 instead of fold_convert_loc (loc, type, arg0).
26276 2017-04-12 Jeff Law <law@redhat.com>
26278 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26279 has a delay slot in the generated code.
26281 * config/cris/cris.md (cris_preferred_reload_class): Return
26282 GENNONACR_REGS rather than GENERAL_REGS.
26284 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26287 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26288 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26289 signedness of the result type.
26291 2017-04-12 Richard Biener <rguenther@suse.de>
26292 Jeff Law <law@redhat.com>
26294 PR tree-optimization/80359
26295 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26296 trim stores to TARGET_MEM_REFs.
26298 2017-04-12 Richard Biener <rguenther@suse.de>
26300 PR tree-optimization/79390
26301 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26302 threading case even more.
26304 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26307 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26308 for quad_address_p for TImode, instead of just not indexed_address.
26310 2017-04-12 Richard Biener <rguenther@suse.de>
26311 Bernd Edlinger <bernd.edlinger@hotmail.de>
26313 PR middle-end/79671
26314 * alias.c (component_uses_parent_alias_set_from): Handle
26315 TYPE_TYPELESS_STORAGE.
26316 (get_alias_set): Likewise.
26317 * tree-core.h (tree_type_common): Add typeless_storage flag.
26318 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26319 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26320 for types containing members with TYPE_TYPELESS_STORAGE.
26321 (place_field): Likewise.
26322 (layout_type): Likewise for ARRAY_TYPE.
26323 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26324 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26325 TYPE_TYPELESS_STORAGE.
26326 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26328 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26331 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26332 first argument to type.
26334 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26338 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26339 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26340 (rs6000_expand_binop_builtin): Likewise.
26341 (rs6000_expand_ternop_builtin): Likewise; also add missing
26342 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26343 vshasigma built-ins.
26344 * doc/extend.texi: Document that vec_xxpermdi's third argument
26345 must be a constant.
26347 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26349 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26350 Use shift_const cost parameter when calculating gain of STV shifts.
26352 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26354 PR rtl-optimization/70478
26355 * lra-constraints.c (process_alt_operands): Check memory for
26356 disfavoring memory insn operand.
26358 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26360 PR middle-end/80100
26361 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26362 left shift in unsigned HOST_WIDE_INT type.
26364 PR rtl-optimization/80385
26365 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26366 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26369 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26370 if they have any depend clauses.
26372 2017-04-11 Martin Liska <mliska@suse.cz>
26375 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26376 * ipa-split.c (split_function): Create a local comdat symbol
26377 if caller is in a comdat group.
26379 2017-04-11 Martin Liska <mliska@suse.cz>
26382 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26385 2017-04-11 Martin Sebor <msebor@redhat.com>
26387 PR middle-end/80364
26388 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26389 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26391 (directive::set_width, directive::set_precision, format_character):
26393 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26396 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26399 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26400 conflict, set target->arch_name instead of target->cpu_name.
26402 2017-04-11 Richard Biener <rguenther@suse.de>
26404 PR tree-optimization/80374
26405 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26406 build_zero_cst, remove fold_convertible_p check again.
26408 2017-04-11 Martin Liska <mliska@suse.cz>
26411 * ubsan.c (instrument_object_size): Do not instrument register
26414 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26417 * config/i386/i386-builtin-types.def
26418 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26419 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26420 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26421 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26422 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26423 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26424 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26425 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26426 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26427 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26428 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26429 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26430 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26431 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26432 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26433 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26434 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26435 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26436 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26437 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26438 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26439 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26440 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26441 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26442 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26443 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26444 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26445 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26446 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26447 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26448 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26449 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26450 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26451 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26452 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26453 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26454 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26455 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26456 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26457 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26458 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26459 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26460 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26461 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26462 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26463 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26465 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26466 flag to second_arg_count, handle 4 argument function type _COUNT
26467 aliases, handle second_arg_count on second argument rather than last.
26469 2017-04-10 Jeff Law <law@redhat.com>
26471 PR tree-optimization/80374
26472 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26473 record anything if we can not convert integer_zero_node to the
26476 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26479 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26480 Enhance special handling given to the TARGET_P9_MINMAX option in
26481 relation to certain other options.
26483 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26485 PR tree-optimization/80153
26486 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26487 remove POINTER_PLUS_EXPR's base part directly, rather than through
26490 2017-04-10 Richard Biener <rguenther@suse.de>
26491 Bin Cheng <bin.cheng@arm.com>
26493 PR tree-optimization/80153
26494 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26495 the first element of pointer type aff_tree. Build result expr in
26497 (add_elt_to_tree): Convert to type unconditionally. Remove other
26498 fold_convert calls.
26499 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26500 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26502 2017-04-10 Richard Biener <rguenther@suse.de>
26504 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26507 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26509 PR rtl-optimization/70478
26510 * lra-constraints.c (curr_small_class_check): New.
26511 (update_and_check_small_class_inputs): New.
26512 (process_alt_operands): Update curr_small_class_check. Disfavor
26513 alternative insn memory operands. Check available regs for small
26516 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26519 * config/mips/mips.opt (-mvirt): Update description.
26520 * doc/invoke.texi (-mvirt): Likewise.
26522 2017-04-10 Richard Biener <rguenther@suse.de>
26524 PR middle-end/80362
26525 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26526 looking for NEGATE_EXPR in -A / -B to A / B folding.
26528 2017-04-10 Martin Liska <mliska@suse.cz>
26530 PR gcov-profile/80224
26531 * gcov.c (print_usage): Fix usage string.
26532 (get_gcov_intermediate_filename): Remove.
26533 (output_gcov_file): Use both for normal and intermediate format.
26534 (generate_results): Do not initialize special file for
26535 intermediate format.
26537 2017-04-10 Richard Biener <rguenther@suse.de>
26539 PR tree-optimization/80304
26540 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26543 2017-04-10 Nathan Sidwell <nathan@acm.org>
26546 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26547 (rs6000_init_builtins): Use it.
26549 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26551 * config/arm/arm.md (<mrc>): Add mode to SET source.
26552 (<mrrc>): Likewise.
26554 2017-04-10 Richard Biener <rguenther@suse.de>
26556 PR middle-end/80344
26557 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26559 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26562 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26563 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26564 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26565 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26566 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26567 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26568 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26569 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26570 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26571 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26572 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26573 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26574 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26575 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26576 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26577 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26578 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26579 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26580 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26581 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26582 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26583 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26584 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26586 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26588 PR rtl-optimization/70478
26589 * lra-constraints.c: Reverse the last patch.
26591 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26593 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26594 Add comment for WCHAR_T.
26596 2017-04-08 Martin Liska <mliska@suse.cz>
26599 2017-04-07 Martin Liska <mliska@suse.cz>
26602 * ipa-split.c (split_function): Add function part to a same comdat
26605 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26608 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26610 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26612 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26613 (vec_pairing): ...to this.
26614 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26615 (rs6000_sched_init): Adjust for name change.
26616 (struct rs6000_sched_context): Likewise.
26617 (rs6000_init_sched_context): Likewise.
26618 (rs6000_set_sched_context): Likewise.
26620 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26626 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26628 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26629 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26630 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26632 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26634 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26636 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26638 PR rtl-optimization/70703
26639 * ira-color.c (update_conflict_hard_regno_costs): Use
26640 int64_t instead of HOST_WIDE_INT.
26642 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26644 PR rtl-optimization/70478
26645 * lra-constraints.c (process_alt_operands): Disfavor alternative
26646 insn memory operands.
26648 2017-04-07 Jeff Law <law@redhat.com>
26650 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26651 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26653 2017-04-07 Martin Liska <mliska@suse.cz>
26656 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26657 Show error message instead of an ICE.
26659 2017-04-07 Martin Liska <mliska@suse.cz>
26662 * ipa-split.c (split_function): Add function part to a same comdat
26665 2017-04-07 Richard Biener <rguenther@suse.de>
26667 PR middle-end/80341
26668 * tree.c (get_unwidened): Also handle ! for_type case for
26670 * convert.c (do_narrow): Split out from ...
26671 (convert_to_integer_1): ... here. Do not pass final truncation
26672 type to get_unwidened for TRUNC_DIV_EXPR.
26674 2017-04-07 Richard Biener <rguenther@suse.de>
26676 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26678 (aff_combination_const): Adjust.
26679 (aff_combination_scale): Likewise.
26680 (aff_combination_add_elt): Likewise.
26681 (aff_combination_add_cst): Likewise.
26682 (aff_combination_convert): Likewise.
26683 (add_elt_to_tree): Likewise. Remove unused argument.
26684 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26686 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26688 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26690 * config/arm/arm.c (arm_default_short_enums): Use
26691 ARM_DEFAULT_SHORT_ENUMS.
26692 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26694 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26697 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26698 members with redundant out-of-class redeclaration.
26700 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26703 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26704 * config/i386/i386.md (*zero_extendsidi2):
26705 Add (?*x,*x) and (?*v,*v) alternatives.
26707 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26710 * config/i386/i386.c (ix86_expand_builtin)
26711 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26712 mode from insn data. Convert operands to insn operand mode.
26713 Copy operands that don't satisfy insn predicate to a register.
26715 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26717 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26720 2017-04-06 Richard Biener <rguenther@suse.de>
26722 PR tree-optimization/80334
26723 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26724 preserve alignment of accesses.
26726 2017-04-06 Richard Biener <rguenther@suse.de>
26728 PR tree-optimization/80262
26729 * tree-sra.c (build_ref_for_offset): Preserve address-space
26731 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26732 Drop useless address-space information on MEM_REF offsets.
26734 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26736 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26738 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26740 PR rtl-optimization/70703
26741 * ira-color.c (update_conflict_hard_regno_costs): Use
26742 HOST_WIDE_INT instead of long.
26744 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26747 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26748 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26750 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26751 for x86_64 target. Handle -m3dnowa option.
26753 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26755 PR rtl-optimization/70703
26756 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26757 (update_conflict_hard_regno_costs): Use long instead of unsigned
26758 arithmetic for cost calculation.
26760 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26761 Bernd Edlinger <bernd.edlinger@hotmail.de>
26764 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26767 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26770 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26771 ptr_mode with Pmode throughout.
26772 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26773 into probe_stack_range and use DImode.
26775 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26778 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26779 call_eh_return is true.
26781 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26783 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26784 Initialize last_match_fntype_index.
26786 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26789 * tree-nvr.c: Include internal-fn.h.
26790 (pass_return_slot::execute): Ignore internal calls without
26793 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26794 Richard Biener <rguenther@suse.de>
26797 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26798 captures used multiple times, except for the last use.
26799 * generic-match-head.c: Include gimplify.h.
26801 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26803 PR tree-optimization/79390
26804 * target.h (struct noce_if_info): Declare.
26805 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26806 * target.def (noce_conversion_profitable_p): New target hook.
26807 * ifcvt.h (struct noce_if_info): New type, moved from ...
26808 * ifcvt.c (struct noce_if_info): ... here.
26809 (noce_conversion_profitable_p): Renamed to ...
26810 (default_noce_conversion_profitable_p): ... this. No longer
26812 (noce_try_store_flag_constants, noce_try_addcc,
26813 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26814 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26815 instead of noce_conversion_profitable_p.
26816 * config/i386/i386.c: Include ifcvt.h.
26817 (ix86_option_override_internal): Don't override
26818 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26819 (ix86_noce_conversion_profitable_p): New function.
26820 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26821 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26822 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26823 * doc/tm.texi: Regenerated.
26825 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26827 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26830 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26833 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26834 instructions for small multiply cores.
26836 2017-04-04 Jeff Law <law@redhat.com>
26838 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26840 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26841 that are not set by the loop over the elements.
26843 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26846 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26847 int mode, convert_modes it to mode as unsigned, otherwise use
26848 lowpart_subreg to mode rather than SImode.
26849 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26850 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26851 Use DImode instead of SImode for the shift count operand.
26852 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26855 2017-04-04 Richard Biener <rguenther@suse.de>
26857 PR middle-end/80281
26858 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26859 arithmetic done for the negate or the plus. Simplify.
26860 (A - (-B) -> A + B): Likewise.
26861 * fold-const.c (split_tree): Make sure to not negate pointers.
26863 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26865 PR rtl-optimization/60818
26866 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26867 a compare of comparisons with the thing compared if this results
26868 in a different machine mode.
26870 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26872 * alias.c (base_alias_check): Fix typo in comment.
26873 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26874 * cgraphunit.c (symbol_table::compile): Likewise.
26875 * collect2.c (maybe_run_lto_and_relink): Likewise.
26876 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26877 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26878 * config/avr/avr.c (avr_map_op_t): Likewise.
26879 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26880 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26881 * config/epiphany/epiphany.md (movcc): Likewise.
26882 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26883 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26885 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26886 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26887 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26888 * config/sh/sh.c (sh_rtx_costs): Likewise.
26889 * fold-const.c (fold_truth_andor): Likewise.
26890 * genautomata.c (collapse_flag): Likewise.
26891 * gengtype.h (struct type::u::s): Likewise.
26892 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26893 * input.c (FORMAT_AMOUNT): Likewise.
26894 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26895 (known_aggs_to_agg_replacement_list): Likewise.
26896 * ipa-inline-analysis.c: Likewise.
26897 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26898 * ipa-polymorphic-call.c
26899 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26900 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26901 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26903 * modulo-sched.c (apply_reg_moves): Likewise.
26904 * omp-expand.c (build_omp_regions_1): Likewise.
26905 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26906 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26907 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26908 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26909 * value-prof.c: Likewise.
26910 * var-tracking.c (val_reset): Likewise.
26912 2017-04-03 Richard Biener <rguenther@suse.de>
26914 PR tree-optimization/80275
26915 * fold-const.c (split_address_to_core_and_offset): Handle
26918 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26920 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26921 descriptors is at least equal to that of functions.
26923 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26925 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26927 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26930 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26931 (mov<IMOD4:mode>): New expander.
26932 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26934 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26936 PR rtl-optimization/79405
26937 * fwprop.c (propagations_left): New variable.
26938 (forward_propagate_into): Decrement it.
26939 (fwprop_init): Initialize it.
26940 (fw_prop): If the variable has reached zero, stop propagating.
26941 (fwprop_addr): Ditto.
26943 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26946 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26947 a FUNCTION_DECL, pass it as decl instead of origin to
26950 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26952 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26955 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26958 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26959 TARGET_VSX_SMALL_INTEGER.
26961 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26963 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26964 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26966 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26968 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26969 extraction from odd-numbered MSA register.
26971 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26973 PR middle-end/80173
26974 * expmed.c (store_bit_field_1): Don't attempt to create
26975 a word subreg out of hard registers wider than word if they
26976 have HARD_REGNO_NREGS of 1 for their mode.
26978 PR middle-end/80163
26979 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26980 conversions to integer types wider than word and pointer.
26983 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26984 (rtx_equal_for_cselib_p): Pass 0 to it.
26985 * cselib.c (cselib_hasher::equal): Likewise.
26986 (rtx_equal_for_cselib_1): Add depth argument. If depth
26987 is 128, don't look up VALUE locs and punt. Increment
26988 depth in recursive calls when walking VALUE locs.
26990 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26992 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26993 (make_gcov_file_name): Use the canonical path name for generating
26995 (read_line): Fix handling of files with ascii null bytes.
26997 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26999 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27000 to initialise a vector register instead
27001 of using a const_int.
27003 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27005 PR translation/80189
27006 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27007 diagnostic messages.
27009 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27012 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27013 (dfp_diex_<mode>): Update mode of operand 1.
27014 * doc/extend.texi (dxex, dxexq): Document change to return type.
27015 (diex, diexq): Document change to argument type.
27017 2017-03-30 Martin Jambor <mjambor@suse.cz>
27020 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27021 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27022 it reflects the signature changes performed at the callee side.
27023 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27024 to cgraph_build_function_type_skip_args.
27025 (build_function_decl_skip_args): Adjust call to the above function.
27027 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27030 * config/i386/sse.md
27031 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27032 register as dest whenever it is a MEM not rtx_equal_p to the
27033 corresponding dup operand, and when forcing into reg move the
27034 reg into the memory afterwards.
27035 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27036 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27037 for the force_reg mode.
27038 (avx512vl_vextractf128<mode>): Use register as dest either
27039 always when a MEM, or when it is a MEM not rtx_equal_p to the
27040 corresponding dup operand, or even not when it is a CONST_VECTOR
27041 depending on the mode and lo vs. hi.
27042 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27044 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27045 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27046 Likewise. Require that operands[2] is even.
27047 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27048 Remove extraneous parens. Require that operands[2] is a multiple
27050 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27051 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27052 disallow memory then.
27054 2017-03-30 Richard Biener <rguenther@suse.de>
27056 PR tree-optimization/77498
27057 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27058 to non-constants over backedges.
27060 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27062 PR rtl-optimization/80233
27063 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27064 as last_combined_insn. Do not test for BARRIER_P separately.
27066 2017-03-29 Andreas Schwab <schwab@suse.de>
27069 * calls.c (prepare_call_address): Convert funexp to Pmode before
27070 copying to temp reg.
27072 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27074 PR tree-optimization/80158
27075 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27076 Handle possible future case of more than one alternate
27078 (replace_rhs_if_not_dup): Likewise.
27079 (replace_one_candidate): Likewise.
27081 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27083 PR rtl-optimization/80193
27084 * ira.c (ira): Do not check allocation for LRA.
27086 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27088 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27089 (nvptx_output_simt_exit): Declare.
27090 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27091 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27092 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27093 (nvptx_declare_function_name): ...here. Emit declaration of local
27094 memory space buffer for omp_simt_enter insn.
27095 (nvptx_output_unisimt_switch): New.
27096 (nvptx_output_softstack_switch): New.
27097 (nvptx_output_simt_enter): New.
27098 (nvptx_output_simt_exit): New.
27099 * config/nvptx/nvptx.h (struct machine_function): New fields
27100 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27101 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27102 (UNSPECV_SIMT_EXIT): Ditto.
27103 (omp_simt_enter_insn): New insn.
27104 (omp_simt_enter): New expansion.
27105 (omp_simt_exit): New insn.
27106 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27108 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27109 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27110 (expand_GOMP_SIMT_EXIT): New.
27111 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27112 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27113 (GOMP_SIMT_EXIT): Ditto.
27114 * target-insns.def (omp_simt_enter): New insn.
27115 (omp_simt_exit): Ditto.
27116 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27118 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27119 (lower_rec_input_clauses): Likewise.
27120 (lower_lastprivate_clauses): Handle SIMT privatization.
27122 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27123 (ompdevlow_adjust_simt_enter): New.
27124 (find_simtpriv_var_op): New.
27125 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27126 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27128 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27129 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27130 (copy_decl_for_dup_finish): Ditto.
27132 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27134 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27137 * config/i386/i386.c (ix86_option_override_internal): Always
27138 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27140 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27142 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27144 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27146 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27147 mark new edge's irreducible flag accordign to it.
27148 (vect_do_peeling): Check loop preheader edge's irreducible flag
27149 and pass it to function slpeel_add_loop_guard.
27151 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27153 PR tree-optimization/80218
27154 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27155 Update block frequencies and counts.
27157 2017-03-28 Richard Biener <rguenther@suse.de>
27159 PR tree-optimization/78644
27160 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27161 of a simplification result we may not use it at all.
27163 2017-03-28 Richard Biener <rguenther@suse.de>
27166 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27167 without arguments, generate default definition of a SSA name.
27169 2017-03-28 Richard Biener <rguenther@suse.de>
27171 PR middle-end/80222
27172 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27173 TYPE_REF_CAN_ALIAS_ALL references.
27174 * fold-const.c (fold_indirect_ref_1): Likewise.
27176 2017-03-28 Martin Liska <mliska@suse.cz>
27179 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27180 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27182 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27183 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27185 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27186 (EXTRA_SPECS): Define.
27187 (SUBTARGET_EXTRA_SPECS): Likewise.
27188 (SUBTARGET_CPP_SPEC): Likewise.
27189 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27190 SUBTARGET_EXTRA_SPECS.
27191 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27193 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27195 * config/arc/simdext.md (vst64_insn): Update pattern.
27196 (vld32wh_insn): Likewise.
27197 (vld32wl_insn): Likewise.
27198 (vld64_insn): Likewise.
27199 (vld32_insn): Likewise.
27201 2017-03-28 Marek Polacek <polacek@redhat.com>
27204 * fold-const.c (fold_comparison): Use protected_set_expr_location
27205 instead of SET_EXPR_LOCATION.
27207 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27209 * tree.c (add_expr): Avoid name lookup warning.
27211 2017-03-27 Jeff Law <law@redhat.com>
27213 PR tree-optimization/80216
27214 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27215 function name. Limit recursion depth.
27216 (record_temporary_equivalences): Corresponding changes.
27218 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27220 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27223 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27226 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27228 * cfgcleanup.c (reg_note_cfa_p): New array.
27229 (insns_have_identical_cfa_notes): New function.
27230 (old_insns_match_p): Don't cross-jump in between /f
27231 and non-/f instructions. If both i1 and i2 are frame related,
27232 verify all CFA notes, their order and content.
27234 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27237 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27238 HImode and SImode with zero extend to DImode to one insn.
27239 (bswap<mode>2_extenddi): Likewise.
27240 (bswapsi2_extenddi): Likewise.
27241 (bswaphi2_extendsi): Likewise.
27242 (bswaphi2): Combine bswap HImode and SImode into one insn.
27243 Separate memory insns from swapping register.
27244 (bswapsi2): Likewise.
27245 (bswap<mode>2): Likewise.
27246 (bswaphi2_internal): Delete, no longer used.
27247 (bswapsi2_internal): Likewise.
27248 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27249 store, and gpr<-gpr swap insns.
27250 (bswap<mode>2_store): Likewise.
27251 (bswaphi2_reg): Register only splitter, combine with the splitter.
27252 (bswaphi2 splitter): Likewise.
27253 (bswapsi2_reg): Likewise.
27254 (bswapsi2 splitter): Likewise.
27255 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27256 the insns into load, store, and register/register insns.
27257 (bswapdi2_ldbrx): Likewise.
27258 (bswapdi2_load): Likewise.
27259 (bswapdi2_store): Likewise.
27260 (bswapdi2_reg): Likewise.
27262 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27264 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27265 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27267 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27270 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27272 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27273 special handling for target option conflicts between dform
27274 options (-mpower9-dform, -mpower9-dform-vector,
27275 -mpower9-dform-scalar) and -mno-direct-move.
27277 2017-03-27 Richard Biener <rguenther@suse.de>
27279 PR tree-optimization/80181
27280 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27282 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27284 * config/arc/predicates.md (move_double_src_operand): Replace the
27285 call to move_double_src_operand with a call to address_operand.
27287 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27289 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27290 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27291 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27293 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27295 * config/arc/predicates.md (long_immediate_loadstore_operand):
27296 Consider scaled addresses cases.
27298 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27300 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27301 restored when in interrupt.
27302 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27303 doesn't have delay slot.
27305 2017-03-27 Richard Biener <rguenther@suse.de>
27308 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27309 inlined thunk clones.
27311 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27314 * asan.c (instrument_derefs): Copy over last operand from
27315 original COMPONENT_REF to the new COMPONENT_REF with
27316 DECL_BIT_FIELD_REPRESENTATIVE.
27317 * ubsan.c (instrument_object_size): Likewise.
27319 2017-03-27 Richard Biener <rguenther@suse.de>
27321 PR tree-optimization/80170
27322 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27323 sure DR/SCEV didnt fold in constants we do not see when looking
27324 at the reference base alignment.
27326 2017-03-27 Richard Biener <rguenther@suse.de>
27328 PR middle-end/80171
27329 * gimple-fold.c (fold_ctor_reference): Properly guard against
27330 NULL return value from canonicalize_constructor_val.
27332 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27335 * config/i386/i386.c (ix86_expand_builtin)
27336 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27337 flags reg setting and flags reg using instructions.
27338 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27339 clobbering instructions to zero extend op2.
27341 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27343 * doc/install.texi (Configuration) <--with-aix-soname>:
27344 Update link to AIX ld.
27346 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27348 PR rtl-optimization/80160
27349 PR rtl-optimization/80159
27350 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27351 reg_alternate_class into account.
27353 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27356 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27357 to consider in curr_insn_transform.
27359 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27361 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27362 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27363 and emit_mode_inner.
27365 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27367 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27368 argument to the overloaded builtin variants. Use the new flag to
27369 deprecate certain builtin variants.
27370 * config/s390/s390-builtin-types.def: Add new builtin types.
27371 * config/s390/s390-builtins.h: Support new flags field for
27372 overloaded builtins.
27373 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27374 (s390_macro_to_expand): Enable vector float data type.
27375 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27376 builtins by incrementing the __VEC__ version number.
27377 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27379 (s390_resolve_overloaded_builtin): Emit error messages depending
27380 on the builtin flags.
27381 * config/s390/s390.c (s390_expand_builtin): Support additional
27382 flags argument. Change error message to match the messages
27383 emitted in s390-c.c.
27384 * config/s390/s390.md: New UNSPEC_* constants.
27385 (op_type): Add new instruction types.
27386 * config/s390/vecintrin.h: Add new builtins and test data class
27388 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27389 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27390 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27391 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27392 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27393 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27395 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27396 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27397 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27398 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27400 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27401 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27402 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27403 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27405 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27406 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27407 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27408 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27409 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27410 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27411 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27413 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27414 ("vec_scatter_element<V_HW_4:mode>_DI")
27415 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27416 ("vec_fpint<mode>", "vflls")
27417 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27418 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27419 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27420 ("*vec_cmphe<mode>_cc"): ... these.
27422 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27423 mode constant instead of magic value.
27425 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27427 * config/s390/s390.c (s390_expand_vec_compare): Support other
27428 vector floating point modes than just V2DF.
27429 (s390_expand_vcond): Likewise.
27430 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27431 (s390_cannot_change_mode_class): Prevent mode changes between TF
27432 and V1TF in vector registers.
27433 * config/s390/s390.md (DF, SF): New mode attributes.
27434 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27435 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27436 SFmode support for VRs.
27437 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27439 (VFT, VF_HW): New mode iterators.
27440 (vw, sdx): New mode attributes.
27441 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27442 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27443 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27444 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27445 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27446 also the new vector floating point modes. Renaming to ...
27448 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27449 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27450 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27451 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27452 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27453 ("vec_unordered<mode>"): ... these.
27455 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27456 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27457 ("*vec_extendv2df"): New insn definitions.
27459 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27461 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27462 ("mulditi3_2", "*muldi3_sign"): New patterns.
27463 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27464 rename the pattern definition.
27466 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27468 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27470 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27472 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27474 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27475 instruction if possible.
27476 * config/s390/vector.md (vec_halfnumelts): New mode
27478 ("*vec_vllezlf<mode>"): New pattern.
27480 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27482 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27483 ("popcountv4si2", "popcountv2di2"): Rename to ...
27484 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27485 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27487 ("popcount<mode>2_vxe"): New pattern.
27489 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27491 * common/config/s390/s390-common.c (processor_flags_table): Add
27493 * config.gcc: Add arch12.
27494 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27495 Default to arch12 for unknown CPU model numbers.
27496 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27497 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27498 PROCESSOR_max sanity check.
27499 * config/s390/s390-opts.h (enum processor_type): Add
27501 * config/s390/s390.c (processor_table): Add arch12.
27502 (s390_expand_builtin): Add check for B_VXE flag.
27503 (s390_issue_rate): Add PROCESSOR_ARCH12.
27504 (s390_get_sched_attrmask): Likewise.
27505 (s390_get_unit_mask): Likewise.
27506 (s390_sched_score): Enable z13 scheduling for arch12.
27507 (s390_sched_reorder): Likewise.
27508 (s390_sched_variable_issue): Likewise.
27509 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27511 (s390_tune_attr): Use z13 scheduling also for arch12.
27512 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27513 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27514 (TARGET_VXE_P): New macros.
27515 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27516 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27517 * config/s390/s390.opt: Add arch12 as processor_type.
27519 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27521 * config/s390/s390.md
27522 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27523 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27524 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27526 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27527 Rename expanders to ...
27529 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27530 ("fixuns_truncdddi2_emu"): ... these.
27532 ("fixuns_trunc<mode>si2_emu"): New expander.
27534 ("*fixuns_truncdfdi2_z13"): Rename to ...
27535 ("*fixuns_truncdfdi2_vx"): ... this.
27537 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27539 * config/s390/2964.md: Remove the single element vector compare
27540 instructions which are no longer used.
27541 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27543 (s390_canonicalize_comparison): Remove handling of DFmode
27545 (s390_expand_vec_compare_scalar): Remove function.
27546 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27547 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27549 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27551 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27553 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27555 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27556 will anyway by matched by mov<mode>_64dfp.
27558 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27560 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27561 vlef/vstef. Add missing operand to vleif.
27563 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27565 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27566 pair for all vector types with 64 bit elements.
27567 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27568 * config/s390/vector.md (V_HW_64): ... here.
27569 (V_128_NOSINGLE): New mode iterator.
27570 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27571 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27572 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27573 ("*vec_load_pairv2di"): Change to ...
27574 ("*vec_load_pair<mode>"): ... this one.
27576 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27578 * config/s390/constraints.md: Add comments.
27579 (jKK): Reject element sizes > 8 bytes.
27580 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27582 * config/s390/s390.md: Add the s_operand checks formerly in
27583 s390_split_ok_p to various splitters where they are still
27585 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27586 for 128 bit vectors. Plus two splitters.
27588 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27590 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27593 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27596 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27597 error if the boundary argument is not constant.
27599 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27601 PR rtl-optimization/80112
27602 * loop-doloop.c (doloop_condition_get): Don't check condition
27603 if cmp isn't SET with IF_THEN_ELSE src.
27605 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27607 PR tree-optimization/80158
27608 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27609 replacing a candidate statement, also replace it for the
27610 candidate's alternate interpretation.
27611 (replace_rhs_if_not_dup): Likewise.
27612 (replace_one_candidate): Likewise.
27614 2017-03-24 Richard Biener <rguenther@suse.de>
27616 PR tree-optimization/80167
27617 * graphite-isl-ast-to-gimple.c
27618 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27620 (translate_isl_ast_to_gimple::get_rename): Likewise.
27622 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27624 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27625 handling of certain combinations of target options, including the
27626 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27627 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27629 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27632 * config/arm/arm.md (*load_multiple): Add reload_completed to
27633 matching condition.
27635 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27636 Richard Biener <rguenth@suse.de>
27638 PR tree-optimization/79908
27639 PR tree-optimization/80136
27640 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27641 been cast away, gimplify_and_add suffices.
27643 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27645 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27647 2017-03-23 Richard Biener <rguenther@suse.de>
27649 PR tree-optimization/80032
27650 * gimplify.c (gimple_push_cleanup): Forced unconditional
27651 cleanups still have to go to the conditional_cleanups
27654 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27656 PR tree-optimization/80072
27657 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27659 (next_operand_entry_id): Change type to unsigned int.
27660 (sort_by_operand_rank): Make sure to return the right return value
27661 even if unsigned fields are bigger than INT_MAX.
27662 (struct oecount): Change cnt and id type to unsigned int.
27663 (oecount_hasher::equal): Formatting fix.
27664 (oecount_cmp): Make sure to return the right return value
27665 even if unsigned fields are bigger than INT_MAX.
27666 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27669 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27670 TREE_READONLY on result if writing it more than once.
27673 * doc/invoke.texi (-fsanitize=thread): Document that with
27674 -fnon-call-exceptions atomics are not able to throw
27678 * tsan.c: Include tree-eh.h.
27679 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27680 maybe_clean_or_replace_eh_stmt where needed.
27681 (instrument_memory_accesses): Add cfg_changed argument.
27682 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27683 if it returned true.
27684 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27686 PR rtl-optimization/63191
27687 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27688 wrapper function, moved the whole old content into ...
27689 (ix86_delegitimize_address_1): ... this. New inline function.
27690 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27691 true as last argument instead of ix86_delegitimize_address.
27693 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27695 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27696 cortexa57_branch_cost.
27698 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27700 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27702 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27705 * doc/md.texi (Constraints): Document wA constraint.
27706 * config/rs6000/constraints.md (wA): New.
27707 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27708 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27709 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27710 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27712 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27715 * gimplify.c (is_oacc_declared): New function.
27716 (oacc_default_clause): Use it to set default flags for acc declared
27717 variables inside parallel regions.
27718 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27719 declared variables.
27720 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27721 declare attribute to any decl as necessary.
27723 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27726 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27727 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27728 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27729 (arm_arch_lpae): This.
27730 * config/arm/arm.c (arm_arch7ve): Rename into ...
27731 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27732 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27735 2017-03-22 Martin Liska <mliska@suse.cz>
27738 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27739 error message instead of an ICE.
27741 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27743 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27745 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27747 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27749 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27752 2017-03-21 Martin Sebor <msebor@redhat.com>
27754 * doc/extend.texi: Use "cannot" instead of "can't."
27755 * doc/hostconfig.texi: Same.
27756 * doc/install.texi: Same.
27757 * doc/invoke.texi: Same.
27758 * doc/loop.texi: Same.
27759 * doc/md.texi: Same.
27760 * doc/objc.texi: Same.
27761 * doc/rtl.texi: Same.
27762 * doc/tm.texi: Same.
27763 * doc/tm.texi.in: Same.
27764 * doc/trouble.texi: Same.
27766 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27769 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27770 (collect_checksum_attributes): Set it.
27771 (die_checksum_ordered): Use it.
27773 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27775 PR tree-optimization/79908
27776 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27777 change: For a VA_ARG whose LHS has been cast away, use
27778 force_gimple_operand to construct the side effects.
27780 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27782 PR translation/80001
27783 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27784 more amenable to translation.
27785 (oacc_loop_auto_partitions): Likewise.
27787 2017-03-21 Marek Polacek <polacek@redhat.com>
27788 Martin Sebor <msebor@redhat.com>
27790 PR tree-optimization/80109
27791 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27792 on INTEGRAL_TYPE_P.
27794 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27795 Segher Boessenkool <segher@kernel.crashing.org>
27798 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27799 check reg_used_between_p between insn and one of succ or succ2
27800 depending on if succ is artificial insn not inserted into insn
27803 2017-03-21 Martin Liska <mliska@suse.cz>
27805 PR gcov-profile/80081
27806 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27807 * doc/gcc.texi: Include gcov-dump stuff.
27808 * doc/gcov-dump.texi: New file.
27810 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27812 PR rtl-optimization/79150
27813 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27814 conditional jump, if the jump is the last insn of the loop.
27816 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27817 Richard Biener <rguenth@suse.de>
27819 PR tree-optimization/79908
27820 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27821 been cast away, use force_gimple_operand to construct the side
27824 2017-03-21 Martin Liska <mliska@suse.cz>
27826 PR libfortran/79956
27827 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27830 2017-03-21 Brad Spengler <spender@grsecurity.net>
27833 * plugin.c (htab_hash_plugin): New function.
27834 (add_new_plugin): Use it and adjust.
27835 (parse_plugin_arg_opt): Adjust.
27836 (init_one_plugin): Likewise.
27838 2017-03-21 Richard Biener <rguenther@suse.de>
27840 PR tree-optimization/80032
27841 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27842 if set force the cleanup to happen unconditionally.
27843 (gimplify_target_expr): Push inserted clobbers with force_uncond
27844 to avoid them being removed by control-dependent DCE.
27846 2017-03-21 Richard Biener <rguenther@suse.de>
27848 PR tree-optimization/80122
27849 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27850 va_arg_pack_len when the inlined call stmt requires pack
27852 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27854 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27857 * tsan.c (instrument_builtin_call): If the memory model argument
27858 is not a constant, assume it is valid.
27861 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27864 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27866 PR rtl-optimization/79910
27867 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27868 if its dest is used by an insn before I2 (other than the combined
27869 insns themselves, which are properly handled already).
27871 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27874 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27876 * combine.c (record_used_regs): New static function.
27877 (try_combine): Handle situations where there is an additional
27878 instruction between I2 and I3 which needs to have a LOG_LINK
27882 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27884 * combine.c (try_combine): Delete redundant i1 test. Call
27885 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27887 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27890 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27891 alternatives 13/14.
27893 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27895 PR tree-optimization/80054
27896 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27897 the optimization if a PHI or any of its arguments is not dominated
27898 by the candidate's basis. Use gphi* rather than gimple* as
27900 (replace_profitable_candidates): Clean up a gimple* variable that
27901 should be a gphi* variable.
27903 2017-03-20 Martin Sebor <msebor@redhat.com>
27906 * doc/extend.texi (attribute constructor): Document present limitation.
27908 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27911 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27912 __POWER9_VECTOR__ #ifdef control, change template definition to
27913 use Power9-specific built-in function.
27914 (vec_any_eq): Likewise.
27915 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27916 to control outcomes from this test.
27917 (vector_ae_<mode>p): For VEC_F modes, likewise.
27919 2017-03-20 Ian Lance Taylor <iant@google.com>
27921 * config/i386/i386.c (ix86_function_regparm): Save an extra
27922 register for -fsplit-stack with DECL_STATIC_CHAIN.
27924 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27927 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27928 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27930 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27932 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27934 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27937 2017-03-20 Marek Polacek <polacek@redhat.com>
27940 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27942 2017-03-20 Richard Biener <rguenther@suse.de>
27944 PR tree-optimization/80113
27945 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27946 allocate extra SSA name for PHI def.
27947 (add_close_phis_to_outer_loops): Likewise.
27948 (add_close_phis_to_merge_points): Likewise.
27949 (copy_loop_close_phi_args): Likewise.
27950 (copy_cond_phi_nodes): Likewise.
27952 2017-03-20 Martin Liska <mliska@suse.cz>
27954 PR middle-end/79753
27955 * tree-chkp.c (chkp_build_returned_bound): Do not build
27956 returned bounds for a LHS that's not a BOUNDED_P type.
27958 2017-03-20 Martin Liska <mliska@suse.cz>
27962 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27963 COMPLEX_CST and VECTOR_CST.
27965 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27968 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27969 target operand. A new splitter adds the clobber statement in case
27970 the target operand is dead anyway.
27972 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27974 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27975 to age-old versions of binutils and glibc.
27977 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27979 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27981 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27983 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27985 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27987 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27988 requirement for binutils 2.13.
27990 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27992 * combine.c (try_combine): Delete redundant i1 test. Call
27993 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27995 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27997 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27998 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28000 <riscv64-*-elf>: Re-arrange section
28001 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28002 <riscv32-*-linux>: Likewise.
28003 <riscv64-*-elf>: Likewise
28004 <riscv64-*-linux>: Likewise.
28006 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28009 * aarch64.opt(verbose-cost-dump): Fix typo.
28011 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28014 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28015 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28017 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28019 * reload.c (find_reloads): When reloading a nonoffsettable address,
28020 use RELOAD_OTHER for it and its address reloads.
28022 PR rtl-optimization/79910
28023 * combine.c (record_used_regs): New static function.
28024 (try_combine): Handle situations where there is an additional
28025 instruction between I2 and I3 which needs to have a LOG_LINK
28028 2017-03-17 Jeff Law <law@redhat.com>
28030 PR tree-optimization/71437
28031 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28032 conditional in the hash table first.
28033 (vrp_dom_walker::before_dom_children): Extract condition from
28034 ASSERT_EXPR. Record condition, its inverion and any implied
28035 conditions as well.
28037 2017-03-17 Marek Polacek <polacek@redhat.com>
28038 Markus Trippelsdorf <markus@trippelsdorf.de>
28040 PR tree-optimization/80079
28041 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28044 2017-03-17 Richard Biener <rguenther@suse.de>
28046 PR middle-end/80075
28047 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28048 Properly verify the LHS before the RHS possibly claims to be
28050 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28053 2017-03-17 Martin Jambor <mjambor@suse.cz>
28055 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28056 (List of -O2 options): Likewise.
28057 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28060 2017-03-17 Tom de Vries <tom@codesourcery.com>
28062 * gcov-dump.c (print_usage): Print bug_report_url.
28064 2017-03-17 Richard Biener <rguenther@suse.de>
28066 PR middle-end/80050
28067 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28068 (parser::peek): Likewise.
28070 2017-03-17 Richard Biener <rguenther@suse.de>
28072 PR tree-optimization/80048
28073 * sese.c (free_sese_info): Properly release rename_map and
28074 copied_bb_map elements.
28076 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28078 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28079 Add linked-list forward and backlinks. Insert on
28080 construction, remove on destruction.
28081 (class pass_store_merging): Add m_stores_head field.
28082 (pass_store_merging::terminate_and_process_all_chains):
28083 Iterate over m_stores_head list.
28084 (pass_store_merging::terminate_all_aliasing_chains):
28086 (pass_store_merging::execute): Check for debug stmts first.
28087 Push new chains onto the m_stores_head stack.
28089 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28092 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28093 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28094 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28096 2017-03-16 Jeff Law <law@redhat.com>
28098 PR tree-optimization/71437
28099 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28100 member function. Implementation moved into after_dom_children
28101 member function and into the threader's thread_outgoing_edges
28103 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28104 some code into new thread_outgoing_edges.
28105 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28106 definition. Simplify marker handling (do it here). Assume we always
28107 have the available expression and the const/copies tables.
28108 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28110 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28111 * tree-vrp.c (equiv_stack): No longer file scoped.
28112 (vrp_dom_walker): New class.
28113 (vrp_dom_walker::before_dom_children): New member function.
28114 (vrp_dom_walker::after_dom_children): Likewise.
28115 (identify_jump_threads): Setup domwalker. Use it rather than
28116 walking edges in a random order by hand. Simplify setup/finalization.
28117 (finalize_jump_threads): Remove.
28118 (vrp_finalize): Do not call identify_jump_threads here.
28119 (execute_vrp): Do it here instead and call thread_through_all_blocks
28122 PR tree-optimization/71437
28123 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28125 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28127 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28128 (dom_opt_dom_walker::thread_across_edge): Remove
28129 handle_dominating_asserts argument. All callers changed.
28130 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28131 changes. Remove calls to lhs_of_dominating_assert. Other
28132 uses of handle_dominating_asserts turn into unconditional code
28133 (simplify_control_stmt_condition_1): Likewise.
28134 (simplify_control_stmt_condition): Likewise.
28135 (thread_through_normal_block, thread_across_edge): Likewise.
28136 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28137 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28138 object if it is not an SSA_NAME.
28139 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28140 before calling into the VRP specific simplifiers.
28141 (identify_jump_threads): Remove handle_dominating_asserts
28144 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28147 * tree-diagnostic.c (default_tree_printer): No longer static.
28148 * tree-diagnostic.h (default_tree_printer): New prototype.
28150 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28152 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28153 Change ins into fmov.
28155 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28157 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28158 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28159 Use h_con constraint for operand 1.
28160 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28161 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28163 2017-03-15 Jeff Law <law@redhat.com>
28165 PR tree-optimization/71437
28166 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28167 (record_temporary_equivalences): Use it.
28169 PR tree-optimization/71437
28170 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28171 tree-ssa-scopedtables.
28172 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28173 (record_conditions, record_cond, vuse_eq): Likewise.
28174 (record_edge_info): Adjust to API tweak of record_conditions.
28175 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28176 (record_temporary_equivalences, optimize_stmt): Likewise.
28177 (eliminate_redundant_computations): Likewise.
28178 (record_equivalences_from_stmt): Likewise.
28179 * tree-ssa-scopedtables.c: Include options.h and params.h.
28180 (vuse_eq): New function, moved from tree-ssa-dom.c
28181 (build_and_record_new_cond): Likewise.
28182 (record_conditions): Likewise. Accept vector of conditions rather
28183 than edge_equivalence structure for first argument.
28184 for the first argument.
28185 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28186 from tree-ssa-dom.c.
28187 (avail_exprs_stack::record_cond): Likewise.
28188 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28189 from tree-ssa-dom.c.
28190 (avail_exprs_stack): Add new member functions lookup_avail_expr
28192 (record_conditions): Declare.
28194 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28197 * lra-constraints.c (process_alt_operands): Increase reject for
28198 reloading an input/output operand.
28200 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28203 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28204 insns to convert from signed/unsigned char/short to IEEE 128-bit
28206 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28208 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28211 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28212 subreg of inner mode for values already in registers.
28214 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28216 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28217 iteration reg is used after the loop.
28219 2017-03-14 Martin Sebor <msebor@redhat.com>
28221 PR tree-optimization/79800
28222 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28223 precision in negative-positive range.
28224 (format_floating): Call non-const overload with adjusted precision.
28226 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28229 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28232 2017-03-14 Martin Sebor <msebor@redhat.com>
28234 PR middle-end/80020
28235 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28236 * builtins.def (aligned_alloc): Use it.
28239 * Makefile.in (GTFILES): Add calls.c.
28240 * calls.c: Include "gt-calls.h".
28242 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28244 PR rtl-optimization/79728
28245 * regs.h (struct target_regs): New field
28246 x_contains_allocatable_regs_of_mode.
28247 (contains_allocatable_regs_of_mode): New macro.
28248 * reginfo.c (init_reg_sets_1): Initialize it, and change
28249 contains_reg_of_mode so it includes global regs as well.
28250 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28251 rather than contains_regs_of_mode.
28253 2017-03-14 Martin Liska <mliska@suse.cz>
28255 * doc/invoke.texi: Document options that can't be combined with
28256 -fcheck-pointer-bounds.
28258 2017-03-14 Martin Liska <mliska@suse.cz>
28260 PR middle-end/79831
28261 * doc/invoke.texi (-Wchkp): Document the option.
28263 2017-03-14 Martin Liska <mliska@suse.cz>
28265 * Makefile.in: Install gcov-dump.
28267 2017-03-14 Martin Liska <mliska@suse.cz>
28269 * multiple_target.c (expand_target_clones): Bail out for
28270 an invalid attribute.
28272 2017-03-14 Richard Biener <rguenther@suse.de>
28274 * alias.c (struct alias_set_entry): Pack properly.
28275 * cfgloop.h (struct loop): Likewise.
28276 * cse.c (struct set): Likewise.
28277 * ipa-utils.c (struct searchc_env): Likewise.
28278 * loop-invariant.c (struct invariant): Likewise.
28279 * lra-remat.c (struct cand): Likewise.
28280 * recog.c (struct change_t): Likewise.
28281 * rtl.h (struct address_info): Likewise.
28282 * symbol-summary.h (function_summary): Likewise.
28283 * tree-loop-distribution.c (struct partition): Likewise.
28284 * tree-object-size.c (struct object_size_info): Likewise.
28285 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28286 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28287 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28288 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28289 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28290 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28291 (struct _stmt_vec_info): Likewise.
28293 2017-03-14 Martin Liska <mliska@suse.cz>
28296 * multiple_target.c (create_dispatcher_calls): Check that
28297 a target can create a function dispatcher.
28299 2017-03-14 Martin Liska <mliska@suse.cz>
28302 * multiple_target.c (expand_target_clones): Drop local.local
28303 flag for default implementation.
28305 2017-03-14 Richard Biener <rguenther@suse.de>
28307 PR tree-optimization/80030
28308 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28310 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28312 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28313 gcc_fallthrough() instead of __attribute__((fallthrough));
28315 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28317 * doc/gcc.texi: Remove "up" link to (DIR).
28318 * doc/gccint.texi: Ditto.
28320 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28322 * doc/install.texi (Specific) <avr>: Remove reference to
28325 2017-03-13 Jeff Law <law@redhat.com>
28327 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28328 attribute rather than comments.
28330 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28331 match_scratch operand is highest.
28333 2017-03-13 Martin Liska <mliska@suse.cz>
28335 PR middle-end/78339
28336 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28337 is a CHKP clone, use original declaration.
28339 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28341 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28342 (arc_conditional_register_usage): Use a different allocation order
28343 when optimizing for size.
28344 * common/config/arc/arc-common.c (arc_option_optimization_table):
28345 Section anchors default on when optimizing for size.
28347 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28349 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28351 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28353 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28354 * config/arc/arc.md (cpu_facility): Add cd variant.
28355 (*movqi_insn): Add code density variant.
28356 (*movhi_insn): Likewise.
28357 (*movqi_insn): Likewise.
28358 (*addsi3_mixed): Likewise.
28359 (subsi3_insn): Likewise.
28361 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28363 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28365 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28367 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28368 expressions with MINUS and UNARY ops.
28370 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28373 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28375 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28376 between vec_select and vector argument.
28377 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28378 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28379 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28380 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28381 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28382 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28384 2017-03-13 Richard Biener <rguenther@suse.de>
28387 * params.def (vect-max-peeling-for-alignment): Fix typo.
28389 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28391 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28392 issue that only occurred with binutils below 2.18.
28394 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28396 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28397 refer to binutils 2.11/2.12 minimum.
28399 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28401 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28402 ftp.kernel.org and simplify binutils requirement.
28404 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28406 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28408 (Optimize Options): Ditto. Also remove redundancy.
28410 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28412 PR translation/79848
28413 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28415 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28416 to G_ to avoid double translation.
28418 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28420 PR translation/79923
28421 * auto-profile.c (get_combined_location): Convert leading
28422 character of diagnostics to lower case and remove trailing period.
28423 (read_profile): Likewise for various diagnostics.
28424 * config/arm/arm.c (arm_option_override): Remove trailing period
28425 from various diagnostics.
28426 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28427 (msp430_expand_delay_cycles): Likewise.
28429 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28432 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28433 full command-line argument, rather than just "str".
28434 (aarch64_validate_march): Likewise.
28435 (aarch64_validate_mtune): Likewise.
28437 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28439 PR rtl-optimization/78911
28440 * lra-assigns.c (must_not_spill_p): New function.
28441 (spill_for): Use it.
28443 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28445 PR tree-optimization/79981
28446 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28447 ATOMIC_COMPARE_EXCHANGE ifn result.
28448 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28449 IFN_ATOMIC_COMPARE_EXCHANGE.
28451 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28454 * opts.c (parse_sanitizer_options): Add missing question mark to
28455 "did you mean" message.
28457 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28459 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28461 (VMULEUH_UNS): Likewise.
28462 (VMULOUB_UNS): Likewise.
28463 (VMULOUH_UNS): Likewise.
28464 * config/rs6000/rs6000.c (builtin_function_type): Remove
28465 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28467 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28470 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28471 x with result of extra_parsing_for_operand_code_0.
28472 (function_reader::extra_parsing_for_operand_code_0): Convert
28473 return type from void to rtx, returning x. When reading
28474 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28475 larger size containing struct block_symbol.
28477 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28479 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28480 -mfloat128-hardware without -m64.
28482 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28485 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28486 entries to the case statement that marks unsigned arguments to
28487 overloaded functions.
28489 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28491 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28492 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28494 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28497 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28498 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28500 2017-03-10 Martin Liska <mliska@suse.cz>
28504 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28505 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28508 2017-03-10 Olivier Hainque <hainque@adacore.com>
28510 * tree-switch-conversion (array_value_type): Start by resetting
28511 candidate type to it's main variant.
28513 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28515 PR rtl-optimization/79909
28516 * combine.c (try_combine): Use simplify_replace_rtx on individual
28517 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28518 of the whole CALL_INSN_FUNCTION_USAGE.
28520 PR tree-optimization/79972
28521 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28522 get_range_info on SSA_NAMEs. Formatting fixes.
28524 2017-03-10 Richard Biener <rguenther@suse.de>
28525 Jakub Jelinek <jakub@redhat.com>
28527 PR tree-optimization/77975
28528 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28529 edge to be constant.
28530 (get_val_for): For constant x return it. Formatting fix.
28531 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28532 would use the same bases as the current one.
28534 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28536 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28537 instead of vec_select for V1TImode.
28538 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28540 (VSX_LE_128): Add V1TI to this mode iterator.
28541 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28542 (*vsx_le_perm_store_<mode>): Likewise.
28543 (pre-reload splitter for VSX stores): Likewise.
28544 (post-reload splitter for VSX stores): Likewise.
28545 (*vsx_xxpermdi2_le_<mode>): Likewise.
28546 (*vsx_lxvd2x2_le_<mode>): Likewise.
28547 (*vsx_stxvd2x2_le_<mode>): Likewise.
28549 2017-03-09 Michael Eager <eager@eagercon.com>
28551 Correct failures with --enable-checking=yes,rtl.
28553 * config/microblaze/microblaze.c (microblaze_expand_shift):
28554 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28555 test for const0_rtx.
28556 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28557 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28559 2017-03-09 Richard Biener <rguenther@suse.de>
28561 PR tree-optimization/79977
28562 * graphite-scop-detection.c (scop_detection::merge_sese):
28563 Handle the case of extra exits to blocks dominating the entry.
28565 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28567 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28570 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28572 PR rtl-optimization/79949
28573 * lra-constraints.c (process_alt_operands): Check memory when
28574 trying to predict a cycle. Print about the overall increase.
28576 2017-03-09 Richard Biener <rguenther@suse.de>
28578 PR middle-end/79971
28579 * gimple-expr.c (useless_type_conversion_p): Preserve
28580 TYPE_SATURATING for fixed-point types.
28582 2017-03-09 Richard Biener <rguenther@suse.de>
28585 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28586 alignment of BLKmode params.
28588 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28591 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28592 (VALL_NO_V2Q): Likewise.
28594 * config/aarch64/aarch64-simd.md
28595 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28597 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28598 VALL_NO_V2Q mode iterator.
28599 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28601 2017-03-09 Martin Liska <mliska@suse.cz>
28603 PR tree-optimization/79631
28604 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28605 tree_int_cst_sign_bit just for INTEGER constants.
28607 2017-03-09 Martin Liska <mliska@suse.cz>
28611 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28614 2017-03-09 Marek Polacek <polacek@redhat.com>
28617 * tree.c (inchash::add_expr): Handle TREE_VEC.
28619 2017-03-09 Martin Liska <mliska@suse.cz>
28622 (chkp_narrow_size_and_offset): New function.
28623 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28624 (void chkp_parse_bit_field_ref): New function.
28625 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28626 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28628 2017-03-09 Martin Liska <mliska@suse.cz>
28631 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28632 (chkp_find_bounds_1): Remove gcc_unreachable.
28634 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28637 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28638 BUILT_IN_SYNC*, determine the access type from the size suffix and
28639 always build a MEM_REF with that type. Handle forgotten
28640 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28643 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28644 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28645 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28646 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28647 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28648 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28649 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28650 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28651 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28652 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28653 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28654 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28655 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28656 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28657 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28658 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28659 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28660 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28661 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28662 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28663 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28664 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28665 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28666 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28667 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28668 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28669 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28670 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28671 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28672 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28673 definitions outside of __OPTIMIZE__ guarded section.
28676 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28677 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28678 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28679 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28682 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28684 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28685 ("vfenez<mode>"): Add missing constraints.
28687 2017-03-08 Martin Sebor <msebor@redhat.com>
28690 * config/nds32/nds32.c (nds32_option_override):
28691 Fix misspelled diagnostic.
28693 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28696 * gimplify.c (gimplify_omp_for): Replace index var in outer
28697 taskloop statement with an artificial variable and add
28698 OMP_CLAUSE_PRIVATE clause for it.
28700 2017-03-08 Richard Biener <rguenther@suse.de>
28702 PR tree-optimization/79955
28703 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28704 for accesses that are completely outside of the variable.
28706 2017-03-08 Andrew Haley <aph@redhat.com>
28708 PR tree-optimization/79943
28709 * tree-ssa-loop-split.c (compute_new_first_bound): When
28710 calculating the new upper bound, (END-BEG) should be added, not
28713 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28715 * config/avr/avr.md (setmemhi): Make sure match_dup
28716 operand number comes before match_scratch.
28718 2017-03-08 Richard Biener <rguenther@suse.de>
28720 PR tree-optimization/79920
28721 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28722 with ncopies == 1 to ...
28723 (vect_transform_slp_perm_load): ... here. Properly compute
28724 all element loads by iterating VF times over the group. Do
28725 not handle ncopies (computed in a broken way) in
28726 vect_create_mask_and_perm.
28728 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28731 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28732 is a uniform vector, use uniform_vector_p return value instead of
28733 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28735 2017-03-07 Marek Polacek <polacek@redhat.com>
28737 PR middle-end/79809
28738 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28739 (alloca_call_type): Likewise.
28741 2017-03-07 Martin Liska <mliska@suse.cz>
28743 * gcov.c (process_args): Put comment to correct location.
28745 2017-03-07 Martin Liska <mliska@suse.cz>
28747 PR middle-end/68270
28748 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28749 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28750 (chkp_narrow_bounds_for_field): Likewise.
28751 (chkp_parse_array_and_component_ref): Pass one more argument to
28754 2017-03-07 Richard Biener <rguenther@suse.de>
28756 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28759 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28761 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28762 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28764 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28767 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28768 to end of description.
28769 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28771 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28773 PR rtl-optimization/79901
28774 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28776 (*avx512f_<code><mode>3<mask_name>): ... this.
28777 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28778 iterator instead of VI8_AVX2_AVX512BW.
28780 PR rtl-optimization/79901
28781 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28782 min/max expander, expand it using expand_vec_cond_expr.
28785 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28788 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28791 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28792 to void * for PCH reasons.
28793 * dwarf2out.c (output_loc_operands, output_die): Cast
28794 v.val_vec.array to unsigned char *.
28796 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28799 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28802 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28804 PR rtl-optimization/79571
28805 * lra-constraints.c (process_alt_operands): Calculate static
28806 reject and subtract it from overall when only addresses will be
28809 2017-03-06 Julia Koval <julia.koval@intel.com>
28812 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28813 incoming stack boundary to 128 for 64-bit targets.
28815 2017-03-06 Richard Biener <rguenther@suse.de>
28817 PR tree-optimization/79894
28818 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28819 to NULL after folding it.
28821 2017-03-06 Richard Biener <rguenther@suse.de>
28823 PR tree-optimization/79824
28824 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28825 check disabling peeling for gaps.
28827 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28829 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28830 attributes): Document gettimeofday.
28832 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28834 * config/s390/s390.c (s390_option_override_internal): Set
28835 PARAM_MIN_VECT_LOOP_BOUND
28837 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28839 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28840 * config/s390/s390.md: Likewise.
28842 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28845 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28846 (<avx2_avx512>_perm<mode>): Rename to ...
28847 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28849 (<avx512>_perm<mode>_mask): Rename to ...
28850 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28852 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28853 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28854 instead of VI8F_256_512.
28855 (avx512f_perm<mode>): New define_expand.
28856 (avx512f_perm<mode>_mask): Likewise.
28857 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28858 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28860 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28862 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28863 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28865 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28867 2017-03-06 Martin Liska <mliska@suse.cz>
28870 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28871 when having a SSA NAME w/o VAR_DECL assigned to it.
28873 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28875 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28876 msa_dpsub_<su>_d): Fix MODE for vec_select.
28878 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28880 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28882 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28884 2017-03-06 Richard Biener <rguenther@suse.de>
28886 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28887 * plugin.c (register_plugin_info): Likewise.
28888 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28890 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28892 * config/i386/sse.md (sse_storehps, sse_storelps,
28893 avx_<castmode><avxsizesuffix>_<castmode>,
28894 avx512f_<castmode><avxsizesuffix>_<castmode>,
28895 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28896 in condition that at least one operand is not a MEM.
28898 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28900 PR middle-end/79805
28901 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28902 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28904 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28905 gimple_call_nothrow_p flag based on whether original builtin can throw.
28906 If it can, emit following stmts on the fallthrough edge.
28907 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28908 don't create new bb if inserting just debug stmts on the edge, try to
28909 insert them on the fallthru bb or just reset debug stmts.
28911 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28914 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28915 restore recog_data (including the operand rtxes inside it) around
28916 the call to get_insn_template.
28918 2017-03-03 Martin Sebor <msebor@redhat.com>
28920 PR tree-optimization/79699
28921 * context.c (context::~context): Free MPFR caches to avoid
28922 a memory leak on program exit.
28924 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28926 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28927 Use wide_int::ulow () instead of .elt (0).
28929 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28931 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28932 (*pushxf): Limit oF constraint to 32bit targets and add oC
28933 constraint for 64bit targets.
28934 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28935 (*pushdf): Change rmF constraint to rmC.
28937 2017-03-03 Martin Liska <mliska@suse.cz>
28939 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28940 Remove unused variable.
28942 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28945 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28946 is a memory operand, increase num_memory.
28947 (ix86_expand_args_builtin): Likewise.
28949 2017-03-03 Jan Hubicka <jh@suse.cz>
28952 * ipa-devirt.c (maybe_record_node): Properly handle
28953 __cxa_pure_virtual visibility.
28955 2017-03-03 Martin Liska <mliska@suse.cz>
28957 PR tree-optimization/79803
28958 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28960 (pass_loop_prefetch::execute): Disabled optimization if an
28961 assumption about L1 cache size is not met.
28963 2017-03-03 Martin Liska <mliska@suse.cz>
28965 PR rtl-optimization/79574
28966 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28967 (hash_scan_set): Likewise.
28968 (dump_hash_table): Likewise.
28969 (hoist_code): Likewise.
28971 2017-03-03 Richard Biener <rguenther@suse.de>
28973 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28974 in place of uhigh ().
28975 (fixed_convert_from_real): Likewise.
28977 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28980 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28982 2017-03-03 Richard Biener <rguenther@suse.de>
28984 PR middle-end/79818
28985 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28986 TYPE_OVERFLOW_UNDEFINED check.
28988 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28990 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28992 (vector_ae_<mode>_p): Likewise.
28993 (vector_nez_<mode>_p): Likewise.
28994 (vector_ne_v2di_p): Likewise.
28995 (vector_ae_v2di_p): Likewise.
28996 (vector_ne_<mode>_p): Likewise.
28997 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28999 (vsx_tsqrt<mode>2_fe): Likewise.
29001 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29004 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29006 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29008 PR rtl-optimization/79780
29009 * cprop.c (one_cprop_pass): When second and further conditional trap
29010 in a single basic block is turned into an unconditional trap, turn it
29011 into a deleted note to avoid RTL verification failures.
29013 2017-03-02 Richard Biener <rguenther@suse.de>
29015 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29017 2017-03-02 Richard Biener <rguenther@suse.de>
29019 PR tree-optimization/79345
29021 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29022 param and abort the walk, returning -1 if it is hit.
29023 (walk_aliased_vdefs): Take a limit param and pass it on.
29024 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29025 defaulting to 0 and return a signed int.
29026 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29027 (check_defs): New helper.
29028 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29029 about uninitialized memory.
29030 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29031 bogus uninitialized warning.
29032 (fixed_convert_from_real): Likewise.
29034 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29036 PR tree-optimization/66768
29037 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29038 iv_use if base object can't be determined.
29040 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29042 PR tree-optimization/79345
29043 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29044 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29045 (get_pattern_stats): Initialize it.
29046 * genemit.c (gen_expand): Verify match_scratch numbers come after
29047 match_operand/match_dup numbers.
29048 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29049 match_scratch numbers.
29050 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29052 * config/s390/s390.md (trunctdsd2): Likewise.
29054 2017-03-02 Richard Biener <rguenther@suse.de>
29056 * wide-int.h (wide_int_storage::operator=): Implement in terms
29059 2017-03-02 Richard Biener <rguenther@suse.de>
29061 PR tree-optimization/79777
29062 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29063 the to insert expression to sth existing.
29065 2017-03-01 Martin Sebor <msebor@redhat.com>
29067 PR middle-end/79692
29068 * gimple-ssa-sprintf.c
29069 (directive::known_width_and_precision): New function.
29070 (format_integer): Use it.
29071 (get_mpfr_format_length): Consider the full range of precision
29072 when computing %g output with the # flag. Set the likely byte
29073 count to 3 rather than 1 when precision is indeterminate.
29074 (format_floating): Correct the lower bound of precision.
29076 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29078 * doc/invoke.texi: Document default code model for 64-bit Linux.
29080 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29083 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29084 udiv rather than div since input pattern is unsigned.
29086 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29088 * config/i386/i386.c (print_reg): Warn for values of
29089 unsupported size in integer register.
29091 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29094 * config/rs6000/predicates.md (current_file_function_operand): Do
29095 not allow self calls to be local if the function is replaceable.
29097 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29100 * config/rs6000/altivec.h (vec_ctz and others): Change the
29101 preprocessor macro that controls conditional compilation from
29102 _ARCH_PWR9 to __POWER9_VECTOR__.
29103 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29104 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29105 control (instead of _ARCH_PWR9 control) so that template
29106 definition uses power9-specific function.
29107 (vec_any_eq): Likewise.
29108 (vec_all_ne): Change macro definition to use a power9-specific
29109 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29110 _ARCH_PWR9 control).
29111 (vec_any_eq) Likewise.
29112 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29113 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29114 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29115 support for xvcmpnedp instruction.
29116 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29117 macro expansion so that Power9 implementation of vec_all_ne does
29118 not use the AltiVec predicate framework.
29119 (VCMPNEH_P): Likewise.
29120 (VCMPNEW_P): Likewise.
29121 (VCMPNED_P): Likewise.
29122 (VCMPNEFP_P): Likewise.
29123 (VCMPNEDP_P): Likewise.
29124 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29125 implementation of vec_any_eq to not use AltiVec predicate
29127 (VCMPAEH_P): Likewise.
29128 (VCMPAEW_P): Likewise.
29129 (VCMPAED_P): Likewise.
29130 (VCMPAEFP_P): Likewise.
29131 (VCMPAEDP_P): Likewise.
29132 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29133 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29134 not use the AltiVec predicate framework.
29135 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29136 of vec_any_eq to not use AltiVec predicate framework.
29137 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29138 support for predefined __POWER9_VECTOR__ macro to indicate that
29139 Power9 instruction selection is enabled.
29140 (altivec_overloaded_builtins): Remove extraneous
29141 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29142 function argument types RS6000_BTI_bool_V16QI and
29143 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29144 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29145 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29146 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29147 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29148 Power9 for implementations of vec_cmpne. Change the signature for
29149 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29150 (representing vec_all_ne) to remove the previously described first
29151 argument of type RS6000_BTI_INTSI, as this was an artifact of
29152 reliance on the AltiVec predicate framework, which is no longer
29153 used in the implementation of these functions. Add
29154 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29155 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29156 since, unlike the AltiVec predicate framework implementation, we
29157 do not share function descriptors between vec_alle and vec_anyeq.
29158 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29159 set of modes that receive special treatment even when
29160 TARGET_P9_VECTOR is true. The special treatment emits code that
29161 does not depend on Power9 instructions.
29162 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29163 define_expand to not rely on AltiVec predicate framework.
29164 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29166 (vector_ne_v2di_p): Change this define_expand to not rely on
29167 AltiVec predicate framework.
29168 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29170 (vector_ne_<mode>_p): Change this define_expand to not rely on
29171 AltiVec predicate framework.
29172 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29174 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29175 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29176 define_insn pattern.
29177 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29178 define_insn pattern because the xvcmpne<VSs>. instruction is not
29180 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29181 instruction is not supported.
29183 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29185 * config/nvptx/nvptx.c: Include intl.h.
29187 2017-03-01 Martin Jambor <mjambor@suse.cz>
29190 * ipa-prop.h (ipa_bits): Removed field known.
29191 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29192 to pointers. Adjusted their comments to warn about their sharing.
29193 (ipcp_transformation_summary): Change bits to a vector of pointers.
29194 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29195 (ipa_get_ipa_bits_for_value): Declare.
29196 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29197 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29198 (ipa_bits_hash_table): Likewise.
29199 (ipa_vr_ggc_hash_traits): Likewise.
29200 (ipa_vr_hash_table): Likewise.
29201 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29202 being pointers and vr_known being removed.
29203 (ipa_set_jf_unknown): Likewise.
29204 (ipa_get_ipa_bits_for_value): New function.
29205 (ipa_set_jfunc_bits): Likewise.
29206 (ipa_get_value_range): New overloaded functions.
29207 (ipa_set_jfunc_vr): Likewise.
29208 (ipa_compute_jump_functions_for_edge): Use the above functions to
29209 construct bits and vr parts of jump functions.
29210 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29211 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29213 (ipcp_grow_transformations_if_necessary): Also allocate
29214 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29216 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29217 them. Fix too long lines.
29218 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29219 vr_known being removed.
29220 (ipa_read_jump_function): Use new setter functions to construct bits
29221 and vr parts of jump functions or set them to NULL.
29222 (write_ipcp_transformation_info): Adjust for bits being pointers.
29223 (read_ipcp_transformation_info): Likewise.
29224 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29226 Include gt-ipa-prop.h.
29227 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29229 (ipcp_store_bits_results): Likewise.
29230 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29231 Do not write to existing jump functions but use a temporary instead.
29233 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29236 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29237 attempt to use its first operand as BIT_FIELD_REF base.
29239 2017-03-01 Richard Biener <rguenther@suse.de>
29241 PR middle-end/79721
29242 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29243 interpolating formula in wrapping arithmetic.
29244 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29246 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29248 PR tree-optimization/79734
29249 * tree-vect-generic.c (expand_vector_condition): Optimize
29250 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29251 Handle VEC_COND_EXPR where comparison has different inner width from
29252 type's inner width.
29254 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29256 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29257 markup, and similar issues. Remove @opindex entries for things
29258 that aren't options. Add missing -mmpy-option entries.
29260 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29262 PR tree-optimization/79737
29263 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29264 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29265 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29266 instead of byte_size. Formatting fix.
29267 (shift_bytes_in_array_right): Formatting fix.
29269 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29272 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29273 condition on optimize for the leaf function test.
29275 2017-02-28 Martin Liska <mliska@suse.cz>
29278 * read-rtl-function.c (function_reader::handle_unknown_directive):
29279 Bail out when one uses -flto.
29281 2017-02-28 Martin Liska <mliska@suse.cz>
29283 * common.opt: Replace space with tabular for options of <number>
29285 * config/i386/i386.opt: Show <number> value for
29286 -mlarge-data-threshold.
29287 * opts.c (print_filtered_help): Do not display number in hexadecimal
29290 2017-02-28 Martin Liska <mliska@suse.cz>
29292 * common.opt: Fix --help=option -Q for options which are of
29295 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29297 * config/i386/i386.c (print_reg): Error out for values
29298 of 8-bit size in invalid integer register.
29300 2017-02-28 Martin Sebor <msebor@redhat.com>
29302 PR tree-optimization/79691
29303 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29305 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29308 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29309 gcc_unreachable with output_operand_lossage.
29311 2017-02-28 Richard Biener <rguenther@suse.de>
29313 PR tree-optimization/79740
29314 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29316 (visit_nary_op): Insert the nary into the hashtable if we
29317 pattern-matched sth.
29318 * tree-ssa-pre.c (eliminate_insert): Robustify.
29320 2017-02-28 Richard Biener <rguenther@suse.de>
29322 PR middle-end/79731
29323 * fold-const.c (decode_field_reference): Reject out-of-bound
29326 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29328 * config/i386/i386.c: Include intl.h.
29329 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29330 instead of just cond ? "..." : "...".
29331 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29332 * coverage.c (read_counts_file): Likewise.
29333 * omp-offload.c: Include intl.h.
29334 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29335 of just cond ? "..." : "...".
29336 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29337 of just cond ? "..." : "...".
29339 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29342 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29344 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29345 'tune for' CPU name.
29346 * config/arm/arm-cpu-data.h: Regenerated.
29348 2017-02-28 Richard Biener <rguenther@suse.de>
29350 PR tree-optimization/79732
29351 * tree-inline.c (expand_call_inline): Do not shadow var.
29353 2017-02-28 Richard Biener <rguenther@suse.de>
29355 PR tree-optimization/79723
29356 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29357 address-space properly.
29359 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29361 * doc/optinfo.texi (Optimization groups): Fix option used for
29363 * doc/invoke.texi (-fopt-info): Document "omp".
29364 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29365 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29366 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29367 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29368 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29370 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29372 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29373 OPTGROUP_OMP with "omp".
29375 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29378 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29379 for arithmetic shift of unsigned V2DI.
29381 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29383 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29384 arc/linux.h headers.
29385 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29386 (LINK_SPEC): Likewise.
29387 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29388 (EXTRA_SPECS): Likewise.
29389 (STARTFILE_SPEC): Likewise.
29390 (ENDFILE_SPEC): Likewise.
29391 (LIB_SPEC): Likewise.
29392 (TARGET_SDATA_DEFAULT): Likewise.
29393 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29394 (MULTILIB_DEFAULTS): Likewise.
29395 (DWARF2_UNWIND_INFO): Likewise.
29396 * config/arc/big.h: New file.
29397 * config/arc/elf.h: Likewise.
29398 * config/arc/linux.h: Likewise.
29399 * config/arc/t-uClibc: Remove.
29401 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29403 PR tree-optimization/77536
29404 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29405 (tree_transform_and_unroll_loop): Use above function to compute the
29406 estimated niter of unrolled loop and use it when scaling profile.
29407 Also use count info rather than frequency if it's non-zero.
29408 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29409 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29410 (vect_transform_loop): Call above function.
29412 2017-02-27 Richard Biener <rguenther@suse.de>
29414 PR tree-optimization/45397
29415 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29416 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29417 (visit_nary_op): Add pattern matching for CSEing sign-changed
29418 or truncated operations with wider ones.
29420 2017-02-27 Richard Biener <rguenther@suse.de>
29422 PR tree-optimization/79690
29423 * tree-vect-stmts.c (vectorizable_store): Use vector type
29424 built from the DR with address-space.
29426 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29428 * doc/invoke.texi (Optimize Options): Refine the description
29429 of asan-use-after-return.
29431 2017-02-25 Alan Modra <amodra@gmail.com>
29433 PR rtl-optimization/79584
29434 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29435 base, not ad->base_term, the reg within base. Remove assertion
29436 that ad->base == ad->base_term. Replace gen_int_mode using
29437 bogus mode with const0_rtx.
29439 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29441 PR middle-end/79396
29442 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29443 FMA_EXPR like tcc_binary or tcc_unary.
29445 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29448 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29450 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29451 (output_loc_operands): Handle DW_OP_call_ref and
29452 DW_OP_GNU_variable_value.
29453 (struct variable_value_struct): New type.
29454 (struct variable_value_hasher): Likewise.
29455 (variable_value_hash): New variable.
29456 (string_types): Remove.
29457 (copy_loc_descr): New function.
29458 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29459 (prepend_loc_descr_to_each): New function.
29460 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29461 instead of add_loc_descr_to_each if the first argument is single
29462 location list and the second has multiple.
29463 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29464 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29465 when looking for variable value which doesn't have other location info.
29466 (loc_list_from_tree): Formatting fix.
29467 (gen_array_type_die): Simplify DW_AT_string_length handling.
29468 (adjust_string_types): Remove.
29469 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29470 string_types. Call resolve_variable_values.
29471 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29472 (resolve_addr_in_expr): Likewise. Add A argument.
29473 (copy_deref_exprloc): Remove deref argument. Adjust for the
29474 original expression being DW_OP_GNU_variable_value with optionally
29475 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29476 optionally after it.
29477 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29478 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29479 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29480 (variable_value_hasher::hash, variable_value_hasher::equal): New
29482 (resolve_variable_value_in_expr, resolve_variable_value,
29483 resolve_variable_values, note_variable_value_in_expr,
29484 note_variable_value): New functions.
29485 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29488 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29491 * opts.h (handle_generated_option): Add GENERATED_P argument.
29492 * opts-common.c (handle_option): Adjust function comment.
29493 (handle_generated_option): Add GENERATED_P argument, pass it to
29495 (control_warning_option): Pass false to handle_generated_option
29497 * opts.c (maybe_default_option): Pass true to handle_generated_option
29499 * optc-gen.awk: Likewise.
29501 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29503 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29504 a REG, look at the REG it is a SUBREG of.
29505 (splitter for cmpeqsi_t): Ditto.
29507 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29509 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29510 the special USEs with the pattern of the insn, not the insn itself.
29512 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29515 * doc/invoke.texi: Document -mload-store-pairs.
29517 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29518 Sandra Loosemore <sandra@codesourcery.com>
29520 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29521 argument isn't a CONST_INT.
29522 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29523 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29524 (nios2_validate_compare): Bypass alternate compare logic if *op2
29525 is not a CONST_INT.
29526 (ldstwm_operation_p): Return false if first_base is not a REG or
29527 if first_offset is not a CONST_INT.
29529 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29531 * config/cris/cris.md: Use correct operand in a define_peephole2.
29533 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29535 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29537 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29539 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29540 this_insn if it is an INSN or JUMP_INSN.
29541 (force_offsettable): Look at base, not at addr.
29542 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29543 on things that aren't necessarily CONST_INTs.
29545 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29547 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29548 -mfpmath=sse is the default also for x86-32 targets with SSE2
29549 instruction set when @option{-ffast-math} is enabled
29551 2017-02-24 Jeff Law <law@redhat.com>
29553 PR rtl-optimizatoin/79286
29554 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29557 2017-02-24 Richard Biener <rguenther@suse.de>
29559 PR tree-optimization/79389
29560 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29563 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29565 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29566 function comment to reflect reality.
29567 (loop_exits_before_overflow): Fix typo in function description.
29569 2017-02-24 Richard Biener <rguenther@suse.de>
29571 PR tree-optimization/79389
29572 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29573 properly that a threading opportunity exists. Detect conditional
29574 copy/constant propagation opportunities.
29576 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29578 * config/visium/visium.md (type): Add trap.
29579 (b): New mode attribute.
29580 (*btst): Rename into...
29581 (*btst<mode>): ...this and adjust.
29582 (*cbranchsi4_btst_insn): Rename into...
29583 (*cbranch<mode>4_btst_insn): ...this and adjust.
29584 (trap): New define_insn.
29586 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29588 PR tree-optimization/79389
29589 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29590 (noce_reversed_cond_code): New function.
29591 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29592 reversed_comparison_code. Formatting fix.
29593 (noce_try_store_flag): Test rev_cond != NULL in addition to
29594 reversed_comparison_code.
29595 (noce_try_store_flag_constants): Likewise.
29596 (noce_try_store_flag_mask): Likewise.
29597 (noce_try_addcc): Use rev_cond if non-NULL instead of
29598 reversed_comparison_code.
29599 (noce_try_cmove_arith): Likewise. Formatting fixes.
29600 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29601 (noce_find_if_block): Initialize rev_cond.
29602 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29603 instead of false as last argument never attempt to reverse it
29606 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29608 PR tree-optimization/79663
29609 * tree-predcom.c (combine_chains): Process refs in reverse order
29610 only for ZERO length chains, and add explaining comment.
29612 2017-02-23 Jeff Law <law@redhat.com>
29614 PR tree-optimization/79578
29615 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29616 in call to operand_equal_p.
29618 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29621 * config/i386/cpuid.h: Fix another undefined behavior.
29623 2017-02-23 Richard Biener <rguenther@suse.de>
29625 PR tree-optimization/79683
29626 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29627 vector types for data-refs.
29629 2017-02-23 Martin Liska <mliska@suse.cz>
29631 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29633 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29635 PR middle-end/79665
29636 * internal-fn.c (get_range_pos_neg): Moved to ...
29637 * tree.c (get_range_pos_neg): ... here. No longer static.
29638 * tree.h (get_range_pos_neg): New prototype.
29639 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29640 are known to be in between 0 and signed maximum inclusive, try to
29641 expand both unsigned and signed divmod and use the cheaper one from
29644 2017-02-22 Jeff Law <law@redhat.com>
29646 PR tree-optimization/79578
29647 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29648 to compare base operands.
29650 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29653 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29654 gpc_reg_operand instead of fpr_reg_operand.
29656 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29658 * config/mips/mips.c (mips_return_in_memory): Force FP
29659 vector types to be returned in memory for o32 ABI.
29661 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29663 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29664 instead of DW_TAG_member for static data member declarations and don't
29665 set no_linkage_name for static inline data members.
29666 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29669 2017-02-22 Martin Liska <mliska@suse.cz>
29671 * doc/invoke.texi: Replace inequality signs with square brackets
29674 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29677 * lra-constraints.c (simplify_operand_subreg): Handle
29678 WORD_REGISTER_OPERATIONS targets.
29680 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29683 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29684 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29685 elimination by swapping fld*.
29687 2017-02-22 Richard Biener <rguenther@suse.de>
29689 PR tree-optimization/79673
29690 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29691 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29692 irrelevant address-space qualifiers and avoiding a
29693 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29695 2017-02-22 Richard Biener <rguenther@suse.de>
29697 PR tree-optimization/79666
29698 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29699 to not symbolically negate if that may introduce undefined
29702 2017-02-22 Martin Liska <mliska@suse.cz>
29705 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29706 * data-streamer-out.c (streamer_write_gcov_count_stream):
29708 * value-prof.c (stream_out_histogram_value): Make assert more
29709 precise based on type of counter.
29711 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29714 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29715 Use nonimmediate_operand instead of memory_operand for operand 1.
29716 (float-extend standard_x87sse_constant_load splitter): Ditto.
29718 2017-02-21 Jeff Law <law@redhat.com>
29720 PR tree-optimization/79621
29721 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29722 blocks with edges to themselves.
29724 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29727 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29728 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29729 Use gimple_call_builtin_p.
29732 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29733 on temporarily removed DEBUG_INSNs.
29735 PR tree-optimization/79649
29736 * tree-loop-distribution.c (classify_partition): Give up on
29737 non-generic address space loads/stores.
29739 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29741 * doc/loop.texi (Loop manipulation): Remove nonexistent
29742 tree_ssa_loop_version from the documentation.
29743 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29745 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29748 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29749 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29750 * config/rs6000/rs6000.c: Include except.h.
29751 (rs6000_expand_split_stack_prologue): Call
29752 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29754 2017-02-21 Martin Jambor <mjambor@suse.cz>
29757 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29758 have been analyzed.
29760 2017-02-21 Martin Jambor <mjambor@suse.cz>
29762 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29763 for backward compatibility only.
29764 * doc/invoke.texi (Option Summary): Remove all references to
29765 -fipa-cp-alignment.
29767 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29771 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29773 * lra-constraints.c (curr_insn_transform): Handle
29774 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29776 2017-02-21 Martin Liska <mliska@suse.cz>
29778 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29780 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29782 PR translation/79638
29783 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29785 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29788 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29789 (arm_function_ok_for_sibcall): Return false for an indirect call by
29790 descriptor if all the argument registers are used.
29791 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29792 alignment of the function.
29794 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29796 PR tree-optimization/61441
29797 * simplify-rtx.c (simplify_const_unary_operation): For
29798 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29799 the sNaN unmodified.
29801 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29803 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29804 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29805 instead of SYSTEM_HEADER_DIR.
29807 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29808 Martin Liška <mliska@suse.cz>
29810 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29811 Fix typos and grammar, use active voice, and clarify.
29813 2017-02-20 Marek Polacek <polacek@redhat.com>
29815 PR middle-end/79537
29816 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29819 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29821 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29824 * config/i386/i386.c (ix86_expand_builtin): Handle
29825 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29826 ix86_builtins_isa[fcode].isa as a requirement of those
29827 flags and any other flag in the bitmask.
29828 (ix86_init_mmx_sse_builtins): Use 0 instead of
29829 ~OPTION_MASK_ISA_64BIT as mask.
29830 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29831 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29832 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29833 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29835 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29838 * lra-constraints.c (split_reg): Check requested split mode
29839 is supported by the register.
29841 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29843 * lra-constraints.c (simplify_operand_subreg): Remove early
29846 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29849 * lra-constraints.c (curr_insn_transform): Tighten condition
29850 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29852 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29855 * lra-constraints.c (curr_insn_transform): Handle
29856 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29858 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29861 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29863 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29865 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29868 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29871 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29873 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29876 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29878 * final.c (last_columnnum, override_columnnum): New variables.
29879 (final_start_function): Set last_columnnum, pass it to begin_prologue
29880 hook and pass 0 to dwarf2out_begin_prologue.
29881 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29882 to source_line debug hook.
29883 (notice_source_line): Compute last_columnnum and for debug_column_info
29884 return true on column changes.
29885 * debug.h (struct gcc_debug_hooks): Add column argument to
29886 source_line and begin_prologue hooks.
29887 (debug_nothing_int_charstar_int_bool): Remove prototype.
29888 (debug_nothing_int_int_charstar,
29889 debug_nothing_int_int_charstar_int_bool): New prototypes.
29890 (dwarf2out_begin_prologue): Add column argument.
29891 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29892 begin_prologue hooks.
29893 (debug_nothing_int_charstar_int_bool): Remove.
29894 (debug_nothing_int_int_charstar,
29895 debug_nothing_int_int_charstar_int_bool): New functions.
29896 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29897 through to dwarf2out_source_line.
29898 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29899 (dwarf2out_source_line): Add column argument, emit it if requested.
29900 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29902 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29903 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29904 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29905 through to dwarf2out_begin_prologue.
29906 (vmsdbgout_source_line): Add column argument, pass it through to
29907 dwarf2out_source_line.
29908 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29909 dbxout_source_line caller.
29910 (dbxout_source_line): Add column argument.
29912 * common.opt (gno-column-info, gcolumn-info): New options.
29913 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29914 (check_die): Also test for multiple DW_AT_decl_column attributes.
29915 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29916 DW_AT_decl_column if requested.
29917 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29919 (gen_variable_die): Likewise.
29920 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29921 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29924 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29925 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29926 (ix86_handle_option): Handle OPT_m3dnowa.
29927 * doc/invoke.texi (-m3dnowa): Document.
29928 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29929 -m3dnowa instead of -m3dnow -march=athlon.
29932 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29933 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29935 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29938 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29939 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29940 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29941 generator for vsx_xxpermdi_<mode>_be.
29942 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29943 force big-endian semantics.
29944 (vsx_xxpermdi_<mode>_be): New define_expand with same
29945 implementation as previous version of vsx_xxpermdi_<mode>.
29947 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29949 PR tree-optimization/79327
29950 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29951 variable, its initialization and use.
29953 2017-02-17 Julia Koval <julia.koval@intel.com>
29955 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29956 (OPTION_MASK_ISA_PKU_UNSET): New.
29957 (ix86_handle_option): Handle -mrdpid.
29958 * config/i386/cpuid.h (bit_RDPID): New.
29959 * config/i386/driver-i386.c (host_detect_local_cpu):
29960 Detect RDPID feature.
29961 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29962 * config/i386/i386-c.c (ix86_target_macros_internal):
29964 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29965 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29966 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29967 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29968 * config/i386/i386.md (define_insn "rdpid"): New.
29969 * config/i386/i386.opt Add -mrdpid.
29970 * config/i386/immintrin.h (_rdpid_u32): New.
29972 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29974 PR rtl-optimization/79541
29975 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29976 instead of transforming it into USE.
29978 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29980 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29981 If HONOR_SNANS (SFmode) force the input to a register.
29982 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29983 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29984 an frsp or similar insn.
29986 2017-02-17 Martin Liska <mliska@suse.cz>
29988 PR rtl-optimization/79577
29989 * params.def (selsched-max-sched-times): Increase minimum to 1.
29991 2017-02-17 Martin Liska <mliska@suse.cz>
29993 PR rtl-optimization/79574
29994 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29996 2017-02-17 Martin Liska <mliska@suse.cz>
29998 PR tree-optimization/79529
29999 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30000 ssa_defined_default_def_p to handle cases which are implicitly
30002 * tree-ssa.c (ssa_defined_default_def_p): New function.
30003 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30004 which are implicitly defined.
30005 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30007 2017-02-17 Richard Biener <rguenther@suse.de>
30009 PR middle-end/79576
30010 * params.def (max-ssa-name-query-depth): Limit to 10.
30012 2017-02-17 Richard Biener <rguenther@suse.de>
30014 PR tree-optimization/79552
30015 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30018 2017-02-17 Richard Biener <rguenther@suse.de>
30021 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30023 2017-02-17 Marek Polacek <polacek@redhat.com>
30025 PR middle-end/79536
30026 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30027 (fold_negate_expr): New wrapper.
30029 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30031 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30032 Correct terminology and de-emphasize pre-standard behavior.
30034 2017-02-16 Alan Modra <amodra@gmail.com>
30036 PR rtl-optimization/79286
30037 * ira.c (def_dominates_uses): New function.
30038 (update_equiv_regs): Don't create an equivalence for insns that
30039 may trap where the register def does not dominate the use.
30041 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30043 PR rtl-optimization/78127
30044 * lra.c (lra): Call lra_eliminate before finish the loop after
30047 2017-02-16 Richard Biener <rguenther@suse.de>
30049 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30051 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30052 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30053 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30054 (isl_val_int_from_wi): New function.
30055 (extract_affine_gmp): Rename to ...
30056 (extract_affine_wi): ... this, take a widest_int.
30057 (extract_affine_int): Just wrap extract_affine_wi.
30058 (add_param_constraints): Use isl_val_int_from_wi.
30059 (add_loop_constraints): Likewise, and extract_affine_wi.
30061 2017-02-15 Jeff Law <law@redhat.com>
30063 PR middle-end/79521
30064 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30065 ira_init_register_move_cost_if_necessary.
30067 2017-02-15 Martin Sebor <msebor@redhat.com>
30069 PR middle-end/32003
30070 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30071 removed in a prior commit.
30073 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30075 PR tree-optimization/79347
30076 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30077 counters during peeling.
30079 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30081 * Makefile.in (site.exp): Remove "set ISLVER".
30083 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30086 * real.c (real_from_integer): Call real_convert even for decimal.
30088 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30091 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30093 2017-02-14 Andrew Pinski <apinski@cavium.com>
30095 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30096 cores and change the partno/implementer to be correct.
30097 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30098 the 'B" as the implementer.
30099 * config/aarch64/aarch64-tune.md: Regenerate.
30101 2017-02-14 Carl Love <cel@us.ibm.com>
30103 * config/rs6000/rs6000.c: Add case statement entry to make the
30104 xvcvuxdsp built-in argument unsigned.
30105 * config/rs6000/vsx.md: Fix the source and return operand types so they
30106 match the instruction definitions from the ISA document. Fix typo
30107 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30110 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30113 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30114 member early_clobber_alts.
30115 * lra-lives.c (reg_early_clobber_p): New.
30116 (process_bb_lives): Use it.
30117 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30118 (debug_operand_data): Initialize early_clobber_alts.
30119 (setup_operand_alternative): Set up early_clobber_alts.
30120 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30121 alternatives to new_insn_reg.
30122 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30124 (lra_update_insn_regno_info): Pass the new arg.
30126 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30128 PR middle-end/79505
30129 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30130 (new_oacc_loop_raw): Don't clear already cleared fields.
30133 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30134 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30135 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30137 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30140 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30142 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30145 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30146 the extra instruction to the right place to store 128-bit constant
30149 2017-02-14 Martin Sebor <msebor@redhat.com>
30151 PR middle-end/79448
30152 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30153 warning for strings of unknown length.
30155 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30157 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30159 2017-02-14 Jeff Law <law@redhat.com>
30162 * ira-costs.c (scan_one_insn): Initialize register move costs
30163 for pseudos seen in USE/CLOBBER insns.
30165 PR tree-optimization/79095
30166 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30167 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30168 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30169 if the operands are known to be not equal, then the resulting range
30171 (intersect_ranges): If the new range is ~[0,0] and the old range is
30172 wide, then prefer ~[0,0].
30173 * tree-vrp.c (overflow_comparison_p_1): New function.
30174 (overflow_comparison_p): New function.
30175 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30176 if NAME is used in an overflow test.
30177 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30178 overflow check that can be expressed as an equality test, then adjust
30179 ops to be that equality test.
30181 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30183 * config/s390/s390-builtin-types.def: Remove flags argument.
30184 * config/s390/s390.c (s390_init_builtins): Likewise.
30186 2017-02-14 Martin Liska <mliska@suse.cz>
30188 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30189 vector. Fix trailing white spaces.
30191 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30193 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30196 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30198 PR rtl-optimization/68664
30199 * config/arm/arm.c (arm_sched_can_speculate_insn):
30200 New function. Declare prototype.
30201 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30203 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30205 PR rtl-optimization/68664
30206 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30208 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30210 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30212 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30213 max skip bytes for function, loop and jump.
30215 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30217 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30218 ABS_EXPR for gimple dump.
30220 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30223 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30225 PR tree-optimization/79408
30226 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30227 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30228 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30229 also if rhs1 is INTEGER_CST.
30231 2017-02-14 Richard Biener <rguenther@suse.de>
30233 PR middle-end/79432
30234 * tree-into-ssa.c (insert_phi_nodes): When the function can
30235 have abnormal edges rewrite SSA names with broken use-def
30236 dominance out of SSA and register them for PHI insertion.
30238 2017-02-13 Martin Sebor <msebor@redhat.com>
30240 PR middle-end/79496
30241 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30242 clearing info.nowrite flag when snprintf size argument is a range.
30244 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30246 * cprop.c (cprop_jump): Add missing space in string literal.
30247 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30248 (get_constraint_for_component_ref): Likewise.
30249 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30250 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30251 * lra-constraints.c (process_alt_operands): Likewise.
30252 * ipa-inline.c (inline_small_functions): Likewise.
30253 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30254 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30255 * trans-mem.c (diagnose_tm_1_op): Likewise.
30256 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30257 (grid_parallel_clauses_gridifiable): Likewise.
30259 * config/nvptx/mkoffload.c (process): Add space in between
30262 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30263 "MOD4_SSE_REGS" and "ALL_REGS".
30265 * spellcheck.c (test_data): Add , in between "foo" and "food".
30267 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30270 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30271 boundary crossing check and subsequent code generation agree.
30273 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30275 * config/aarch64/aarch64.c (has_memory_op): Delete.
30276 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30279 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30281 PR rtl-optimization/79388
30282 PR rtl-optimization/79450
30283 * combine.c (distribute_notes): When removing TEM_INSN for which
30284 corresponding dest has last value recorded, invalidate that last
30287 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30289 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30290 of explicit '@'. Add missing assembly comment marker on branch costs
30293 2017-02-13 Nathan Sidwell <nathan@acm.org>
30295 * gengtype-lex.l (<in_struct>): Add '/'.
30297 2017-02-13 Martin Liska <mliska@suse.cz>
30300 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30302 2017-02-13 Richard Biener <rguenther@suse.de>
30304 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30306 * configure: Re-generate.
30307 * config.in: Likewise.
30308 * graphite-dependences.c: Simplify as if
30309 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30310 * graphite-isl-ast-to-gimple.c: Likewise.
30311 * graphite-optimize-isl.c: Likewise.
30312 * graphite-poly.c: Likewise.
30313 * graphite-sese-to-poly.c: Likewise.
30314 * graphite.h: Likewise.
30315 * toplev.c: Include isl/version.h and use isl_version () for
30316 printing the ISL version.
30317 * doc/install.texi: Update ISL requirement.
30319 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30321 * doc/standards.texi (Standards): Update reference to
30324 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30326 * doc/extend.texi (Named Address Spaces): sourceware.org now
30328 * doc/install.texi (Binaries): Ditto.
30331 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30333 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30334 standard terminology "stringize"/"stringizing" throughout.
30335 * doc/cppinternals.texi: Likewise.
30337 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30339 * doc/extend.texi: Fix some spelling mistakes and typos.
30340 * doc/invoke.texi: Likewise.
30342 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30345 * params.def (inline-min-speedup) Change from 10 to 8.
30347 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30349 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30352 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30355 * ipa-inline-analysis.c (get_minimal_bb): New function.
30356 (record_modified): Use it.
30357 (remap_edge_change_prob): Handle also ancestor functions.
30359 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30361 * doc/contrib.texi (Contributors): Remove broken link into
30362 the Mauve CVS repository.
30364 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30366 PR middle-end/79454
30367 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30368 result computation whenever lhs doesn't have vector mode, not
30369 just when it has BLKmode.
30371 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30373 * doc/makefile.texi (profiledbootstrap): Refer to the
30374 installation instructions only in textual form.
30376 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30379 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30381 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30383 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30384 (Specific): Update mingw-w64 reference.
30386 (Specific): Remove broken link to Renesas RX processor.
30388 2017-02-10 Richard Biener <rguenther@suse.de>
30390 * toplev.c (process_options): Do not mention obsolete graphite
30391 options when printing sorry message about missing graphite support.
30392 Mention -floop-nest-optimize.
30394 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30396 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30397 (vtst_p16): Likewise.
30398 (vtstq_p8): Likewise.
30399 (vtstq_p16): Likewise.
30401 (vtstq_p64): Likewise.
30402 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30403 (vset_lane_p64): New.
30404 (vsetq_lane_p64): New.
30406 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30408 PR tree-optimization/79411
30409 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30410 stmt operands are SSA_NAMEs used in abnormal phis.
30411 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30414 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30417 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30420 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30422 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30424 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30426 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30430 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30431 not arbitrary TREE_CONSTANT.
30434 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30435 "omp declare target link" attribute unless is_global_var.
30436 * omp-offload.c (find_link_var_op): Likewise.
30438 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30439 Chung-Lin Tang <cltang@codesourcery.com>
30441 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30443 (gimplify_adjust_omp_clauses): Don't delete TILE.
30444 (gimplify_omp_for): Deal with TILE.
30445 * internal-fn.c (expand_GOACC_TILE): New function.
30446 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30448 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30449 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30451 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30452 avoid DIV for outermost collapse var.
30453 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30454 Remove out of date comments, fix whitespace.
30455 * omp-general.c (omp_extract_for_data): Deal with tiling.
30456 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30457 adjust OLF_DIM_BASE value.
30458 (struct omp_for_data): Add tiling field.
30459 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30460 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30461 for auto loops. Remove default auto determining, moved to
30462 oacc_loop_fixed_partitions.
30463 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30464 stmts, add e_mask field.
30465 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30466 (oacc_thread_numbers): Use oacc_dim_call.
30467 (oacc_xform_tile): New.
30468 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30469 (finish_oacc_loop): Adjust for ifns vector.
30470 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30471 add case for GOACC_TILE fns.
30472 (oacc_loop_xform_loop): Delete.
30473 (oacc_loop_process): Iterate over call list directly, and add
30474 handling for GOACC_TILE fns.
30475 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30477 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30478 vector partitioning to outer loops. Assign 2 partitions to loops
30479 when available. Add TILE handling.
30480 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30481 (execite_oacc_device_lower): Process GOACC_TILE fns,
30482 ignore unknown specs.
30483 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30484 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30485 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30487 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30489 * configure.ac (ACX_BUGURL): Update.
30490 * configure: Regenerate.
30492 2017-02-09 Richard Biener <rguenther@suse.de>
30494 PR tree-optimization/69823
30495 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30496 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30498 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30500 * config/arc/arc-c.def: Add __NPS400__ definition.
30501 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30502 (TARGET_NPS400): Define.
30504 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30506 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30508 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30509 pointer, arch_info.
30510 (arc_cpu_types): Fill the arch_info field with a pointer into the
30511 arc_arch_types table.
30512 (arc_selected_cpu): Declare.
30513 * config/arc/arc.c (arc_selected_cpu): Make global.
30514 (arc_selected_arch): Delete.
30515 (arc_base_cpu): Delete.
30516 (arc_override_options): Remove references to deleted variables,
30517 update access to arch information.
30518 (ARC_OPT): Update access to arch information.
30519 (ARC_OPTX): Likewise.
30520 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30521 (TARGET_ARC600): Update access to arch information.
30522 (TARGET_ARC601): Likewise.
30523 (TARGET_ARC700): Likewise.
30524 (TARGET_EM): Likewise.
30525 (TARGET_HS): Likewise.
30526 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30529 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30532 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30533 condition/operands for integer GE/LE/GEU/LEU operations.
30535 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30537 PR translation/79397
30538 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30541 2017-02-08 Martin Jambor <mjambor@suse.cz>
30544 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30545 whether allocation happened.
30546 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30547 nothing was allocated.
30549 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30551 PR tree-optimization/79408
30552 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30553 constant, but SSA_NAME with a known integer range, use the minimum
30554 of that range instead of op1 to determine if modulo can be replaced
30555 with its first operand.
30557 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30559 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30561 2017-02-08 Richard Biener <rguenther@suse.de>
30563 PR tree-optimization/71824
30564 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30565 Check all loops contained in the merged region.
30567 2017-02-07 Andrew Pinski <apinski@cavium.com>
30569 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30571 2017-02-07 Andrew Pinski <apinski@cavium.com>
30573 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30574 (thunderxt88): Likewise.
30575 (thunderxt81): Disable LSE and change v8.1 to v8.
30576 (thunderxt83): Likewise.
30578 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30579 Richard Biener <rguenther@suse.de>
30581 PR middle-end/79399
30582 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30583 type from int to size_t.
30584 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30586 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30588 PR rtl-optimization/79386
30589 * cprop.c (bypass_conditional_jumps): Initialize
30590 bypass_last_basic_block already before splitting bbs after
30591 unconditional traps...
30592 (bypass_conditional_jumps): ... rather than here.
30595 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30596 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30597 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30598 fix -masm=intel patterns.
30600 2017-02-07 Richard Biener <rguenther@suse.de>
30602 PR tree-optimization/79256
30603 PR middle-end/79278
30604 * builtins.c (get_object_alignment_2): Use min_align_of_type
30605 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30606 and ADJUST_FIELD_ALIGN.
30608 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30610 * doc/tm.texi: Regenerate.
30611 * stor-layout.c (layout_decl): Adjust.
30612 (update_alignment_for_field): Likewise.
30613 (place_field): Likewise.
30614 (min_align_of_type): Likewise.
30615 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30616 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30617 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30618 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30619 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30620 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30621 * config/i386/i386.c (x86_field_alignment): Likewise.
30622 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30623 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30624 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30625 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30626 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30627 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30631 2017-01-30 Richard Biener <rguenther@suse.de>
30633 PR tree-optimization/79256
30634 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30635 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30638 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30640 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30641 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30642 builtins to SImode and emit a zero-extend, if necessary.
30644 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30646 * docs/invoke.texi (RISC-V Options): Alphabetize.
30648 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30650 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30653 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30655 * config/riscv/riscv.c: New file.
30656 * common/config/riscv/riscv-common.c: Likewise.
30657 * config.gcc: Likewise.
30658 * config/riscv/constraints.md: Likewise.
30659 * config/riscv/elf.h: Likewise.
30660 * config/riscv/generic.md: Likewise.
30661 * config/riscv/linux.h: Likewise.
30662 * config/riscv/multilib-generator: Likewise.
30663 * config/riscv/peephole.md: Likewise.
30664 * config/riscv/pic.md: Likewise.
30665 * config/riscv/predicates.md: Likewise.
30666 * config/riscv/riscv-builtins.c: Likewise.
30667 * config/riscv/riscv-c.c: Likewise.
30668 * config/riscv/riscv-ftypes.def: Likewise.
30669 * config/riscv/riscv-modes.def: Likewise.
30670 * config/riscv/riscv-opts.h: Likewise.
30671 * config/riscv/riscv-protos.h: Likewise.
30672 * config/riscv/riscv.h: Likewise.
30673 * config/riscv/riscv.md: Likewise.
30674 * config/riscv/riscv.opt: Likewise.
30675 * config/riscv/sync.md: Likewise.
30676 * config/riscv/t-elf-multilib: Likewise.
30677 * config/riscv/t-linux: Likewise.
30678 * config/riscv/t-linux-multilib: Likewise.
30679 * config/riscv/t-riscv: Likewise.
30680 * configure.ac: Likewise.
30681 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30682 Waterman as RISC-V maintainers.
30683 * doc/install.texi: Add RISC-V entries.
30684 * doc/invoke.texi: Add RISC-V options section.
30685 * doc/md.texi: Add RISC-V constraints section.
30686 * configure: Regenerated.
30688 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30691 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30692 false values to be constant vectors with all 0 or all 1 bits set.
30693 (vcondu<mode><mode>): Likewise.
30694 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30696 (fpmask_comparison_operator): Update comment.
30697 (vecint_comparison_operator): New predicate.
30698 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30699 vector conditionals when the true and false values are constant
30700 vectors with all 0 bits or all 1 bits set.
30702 2017-02-06 Martin Sebor <msebor@redhat.com>
30704 PR tree-optimization/79376
30705 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30707 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30709 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30710 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30711 to simplify split condition.
30713 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30715 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30716 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30719 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30721 PR rtl-optimization/68664
30722 * target.def (can_speculate_insn): New hook.
30723 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30724 * doc/tm.texi: Regenerate.
30725 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30726 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30727 (rs6000_sched_can_speculate_insn): New function.
30729 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30731 PR tree-optimization/79284
30732 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30733 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30734 vectorizable_mask_load_store, vectorizable_operation,
30735 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30736 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30737 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30738 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30739 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30740 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30741 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30742 is_gimple_assign (stmt). Replace another such test with
30743 is_gimple_assign (stmt).
30745 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30748 * config/avr/avr.c (rtl-iter.h): Include it.
30749 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30750 (avr_legitimate_combined_insn): ...and implementation.
30752 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30754 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30755 * config/s390/s390.c (s390_const_operand_ok)
30756 (s390_canonicalize_comparison, s390_extract_part)
30757 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30758 (s390_contiguous_bitmask_p, s390_rtx_costs)
30759 (legitimize_pic_address): Likewise.
30760 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30761 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30762 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30763 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30764 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30766 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30768 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30769 REGNO($0) == REGNO($1).
30771 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30773 * config/s390/linux.h(SIZE_TYPE): Add comment.
30775 2017-02-06 Julian Brown <julian@codesourcery.com>
30776 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30777 Virendra Pathak <virendra.pathak@broadcom.com>
30779 * config/aarch64/aarch64-cores.def: Change the scheduler
30781 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30782 * config/aarch64/thunderx2t99.md: New file.
30784 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30786 * doc/standards.texi (Go Language): Update link to language
30789 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30791 * tree-eh.c (lower_resx): Sanitize profile.
30792 (cleanup_empty_eh_move_lp): Likewise.
30794 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30797 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30799 * cfgloopmanip.h (loop_version): Update prototype.
30800 * modulo-sched.c (sms_schedule): Update call of loop_version.
30801 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30802 * tree-parloops.c (gen_parallel_loop): Likewise.
30803 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30804 * tree-ssa-loop-split.c (split_loop): Likewise.
30805 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30806 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30808 2017-02-05 Martin Liska <mliska@suse.cz>
30811 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30813 (print_operand_address): Initialize a struct to zero.
30815 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30817 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30818 garbage collector only in textual form.
30820 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30822 * doc/extend.texi (x86 specific memory model extensions for
30823 transactional memory): Simplify a phrase.
30825 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30828 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30829 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30830 (atomic_storedi_1): Likewise.
30832 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30834 PR tree-optimization/79338
30835 * tree-parloops.c (gather_scalar_reductions): Don't call
30836 vect_analyze_loop_form for loop->inner before destroying loop's
30839 2017-02-03 Martin Sebor <msebor@redhat.com>
30841 PR tree-optimization/79327
30842 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30843 when precision has resulted in leading zeros.
30844 (format_integer): Adjust the likely counter to assume an unknown
30845 argument that may be zero is non-zero.
30847 2017-02-03 Jason Merrill <jason@redhat.com>
30850 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30851 avoid copying non-taken branch.
30853 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30855 PR tree-optimization/79340
30856 * tree-vect-loop.c (vectorizable_reduction): Release
30857 vec_defs elements after safe_splicing them into other vectors.
30860 PR tree-optimization/79327
30861 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30862 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30864 (format_integer): Use wide_int_to_tree instead of build_int_cst
30865 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30866 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30867 of shortest and longest sequence.
30869 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30871 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30872 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30874 2017-02-03 Walter Lee <walt@tilera.com>
30877 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30878 after initial stackframe link reg save.
30879 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30881 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30884 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30885 wu for stxssp alternative.
30887 2017-02-03 Martin Sebor <msebor@redhat.com>
30889 PR tree-optimization/79352
30890 * gimple-fold.c (get_range_strlen): Add argument.
30891 (get_range_strlen): Change return type to bool.
30892 (get_maxval_strlen): Pass in a dummy argument.
30893 * gimple-fold.h (get_range_strlen): Change return type to bool.
30894 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30895 * tree.h (array_at_struct_end_p): Add argument.
30896 * tree.c (array_at_struct_end_p): Handle it.
30898 2017-02-03 Martin Liska <mliska@suse.cz>
30901 * multiple_target.c (create_dispatcher_calls): Redirect edge
30902 from a caller of a dispatcher.
30903 (expand_target_clones): Make the clones local.
30904 (ipa_target_clone): Do both target clones and resolvers.
30905 (ipa_dispatcher_calls): Remove the pass.
30906 (pass_dispatcher_calls::gate): Likewise.
30907 (make_pass_dispatcher_calls): Likewise.
30908 * passes.def (pass_target_clone): Put as very first IPA early
30911 2017-02-03 Martin Liska <mliska@suse.cz>
30913 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30914 in case of a function with ifunc attribute.
30916 2017-02-03 Martin Liska <mliska@suse.cz>
30918 * cgraph.c (cgraph_node::dump): Dump function version info.
30919 * symtab.c (symtab_node::dump_base): Add missing new line.
30921 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30923 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30924 (ifcombine_ifandif): Use it.
30926 2017-02-03 Martin Liska <mliska@suse.cz>
30928 * doc/invoke.texi: Document default value for
30929 use-after-scope-direct-emission-threshold.
30931 2017-02-03 Martin Liska <mliska@suse.cz>
30933 PR tree-optimization/79339
30934 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30935 (format_floating): Likewise.
30937 2017-02-03 Martin Liska <mliska@suse.cz>
30940 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30942 (ipa_node_params_t::remove): Likewise.
30943 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30944 initialization from removed ipa_node_params_t::insert.
30945 (ipa_node_params::~ipa_node_params): Move from removed
30946 ipa_node_params_t::release.
30947 * symbol-summary.h (symbol_summary::m_released): New member.
30948 Do not release a summary twice. Do not allow to call finalizer
30949 for types of a summary that live in GGC memory.
30951 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30953 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30956 2017-02-02 Martin Sebor <msebor@redhat.com>
30958 PR middle-end/79275
30959 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30960 (format_string): Tighten up the range of output for non-constant
30961 strings and correct the expected range for wide non-constant strings.
30963 2017-02-02 Martin Sebor <msebor@redhat.com>
30965 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30967 PR middle-end/32003
30968 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30970 (-fdump-tree-@var): Add to index and document how to come up
30971 with pass-specific option and dump file names.
30972 (-fdump-passes): Clarify where to look for output.
30974 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30976 PR middle-end/77445
30977 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30978 statistics of the analyzed path; allow threading for speed when
30979 any of BBs along the path are optimized for speed.
30981 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30983 PR middle-end/78468
30984 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30985 settings of the virtual registers.
30988 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30990 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30991 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30994 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30996 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30997 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30999 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31001 * config/s390/s390.md: Add missing comments with the expanded
31003 * config/s390/vector.md: Likewise.
31004 * config/s390/vx-builtins.md: Likewise.
31006 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31009 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31010 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31011 conditions on a single line.
31013 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31015 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31016 __S390_VX__ to __VX__.
31018 2017-02-01 Andrew Pinski <apinski@cavium.com>
31020 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31021 stmt_info to record_stmt_cost.
31022 (vect_get_known_peeling_cost): Pass stmt_info if known to
31024 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31025 cpu_vector_cost field into
31026 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31027 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31028 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31029 splitting of scalar_stmt_cost and vec_stmt_cost.
31030 (thunderx_vector_cost): Likewise.
31031 (cortexa57_vector_cost): LIkewise.
31032 (exynosm1_vector_cost): Likewise.
31033 (xgene1_vector_cost): Likewise.
31034 (thunderx2t99_vector_cost): Improve after the splitting of the two
31036 (aarch64_builtin_vectorization_cost): Update for the splitting of
31037 scalar_stmt_cost and vec_stmt_cost.
31039 2017-02-01 Torvald Riegel <triegel@redhat.com>
31040 Richard Henderson <rth@redhat.com>
31042 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31043 conditional on existance of a fast atomic load.
31044 * optabs-query.c (can_atomic_load_p): New function.
31045 * optabs-query.h (can_atomic_load_p): Declare it.
31046 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31047 no fast atomic load is available for the particular size of access.
31048 (expand_atomic_compare_and_swap): Likewise.
31049 (expand_atomic_load): Likewise.
31050 (expand_atomic_store): Likewise.
31051 (expand_atomic_fetch_op): Likewise.
31052 * testsuite/lib/target-supports.exp
31053 (check_effective_target_sync_int_128): Remove x86 because it provides
31054 no fast atomic load.
31055 (check_effective_target_sync_int_128_runtime): Likewise.
31057 2017-02-01 Richard Biener <rguenther@suse.de>
31059 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31060 (graphite_transform_loops): Provide opt-info for optimized nests.
31061 * tree-parloop.c (parallelize_loops): Provide opt-info for
31062 parallelized loops.
31064 2017-02-01 Richard Biener <rguenther@suse.de>
31066 PR middle-end/79315
31067 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31068 was not set before.
31070 2017-02-01 Richard Biener <rguenther@suse.de>
31072 PR tree-optimization/71824
31073 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31074 Verify the loops are valid in the merged SESE region.
31075 (scop_detection::can_represent_loop_1): Check analyzing the
31076 evolution of the number of iterations in the region succeeds.
31078 2017-01-31 Ian Lance Taylor <iant@golang.org>
31080 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31081 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31083 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31085 PR preprocessor/79210
31086 * input.c (get_substring_ranges_for_loc): Replace line_width
31087 assertion with error-handling.
31089 2017-01-31 Richard Biener <rguenther@suse.de>
31091 PR tree-optimization/77318
31092 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31093 (create_pw_aff_from_tree): Take loop parameter.
31094 (add_condition_to_pbb): Pass loop of the condition to
31095 create_pw_aff_from_tree.
31097 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31099 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31100 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31102 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31106 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31108 (convert_int_to_float128): Likewise.
31109 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31110 (convert_int_to_float128): Likewise.
31111 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31112 (UNSPEC_IEEE128_CONVERT): Likewise.
31113 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31114 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31115 Use local variables for IBM extended format.
31116 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31117 (fix_trunc<mode>si2_fprs): Likewise.
31118 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31119 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31120 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31121 to know that we can now have integers of all sizes in vector
31123 (fix<uns>_<mode>di2_hw): Likewise.
31124 (float<uns>_<mode>si2_hw): Likewise.
31125 (fix_<mode>si2_hw): Likewise.
31126 (fixuns_<mode>si2_hw): Likewise.
31127 (float<uns>_<mode>di2_hw): Likewise.
31128 (float_<mode>di2_hw): Likewise.
31129 (float_<mode>si2_hw): Likewise.
31130 (floatuns_<mode>di2_hw): Likewise.
31131 (floatuns_<mode>si2_hw): Likewise.
31132 (xscvqp<su>wz_<mode>): Delete, no longer used.
31133 (xscvqp<su>dz_<mode>): Likewise.
31134 (xscv<su>dqp_<mode>): Likewise.
31135 (ieee128_mfvsrd_64bit): Likewise.
31136 (ieee128_mfvsrd_32bit): Likewise.
31137 (ieee128_mfvsrwz): Likewise.
31138 (ieee128_mtvsrw): Likewise.
31139 (ieee128_mtvsrd_64bit): Likewise.
31140 (ieee128_mtvsrd_32bit): Likewise.
31142 2017-01-31 Martin Liska <mliska@suse.cz>
31145 * ipa-prop.c (ipa_free_all_node_params): Call release method
31146 instead of ~sumbol_summary to not to trigger double times
31149 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31151 PR tree-optimization/71691
31152 * bitmap.h (class auto_bitmap): New.
31153 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31154 is_maybe_undefined instead of ssa_undefined_value_p.
31156 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31158 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31159 __S390_ARCH_LEVEL__ to __ARCH__.
31161 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31163 PR tree-optimization/79267
31164 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31165 if should_remove_lhs_p is true.
31167 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31170 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31171 (add_alignment_attribute): New.
31172 (base_type_die): Add alignment attribute.
31173 (subrange_type_die): Likewise.
31174 (modified_type_die): Likewise.
31175 (gen_array_type_die): Likewise.
31176 (gen_descr_array_type_die: Likewise.
31177 (gen_enumeration_type_die): Likewise.
31178 (gen_subprogram_die): Likewise.
31179 (gen_variable_die): Likewise.
31180 (gen_field_die): Likewise.
31181 (gen_ptr_to_mbr_type_die): Likewise.
31182 (gen_struct_or_union_type_die): Likewise.
31183 (gen_subroutine_type_die): Likewise.
31184 (gen_typedef_die): Likewise.
31185 (base_type_cmp): Compare alignment attribute.
31187 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31190 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31191 (setb_unsigned) New pattern for setb with CCUNS.
31192 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31193 subfc./subfe sequence to avoid overflow problems. Generate a
31194 shorter sequence with cmpld/setb for power9.
31195 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31196 for generating subfc. instruction.
31197 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31198 now uses this instruction.
31200 2017-01-30 Ian Lance Taylor <iant@google.com>
31203 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31204 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31206 2017-01-30 Martin Sebor <msebor@redhat.com>
31208 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31209 Move constant to the right of a relational operator.
31210 (get_mpfr_format_length, format_character, format_string): Ditto.
31211 (should_warn_p, maybe_warn): Same.
31213 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31215 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31218 * asan.c (get_translation_unit_decl): Remove function.
31219 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31221 2017-01-30 Martin Liska <mliska@suse.cz>
31223 PR gcov-profile/79259
31224 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31225 -fprofile-generate.
31227 2017-01-30 Martin Liska <mliska@suse.cz>
31230 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31231 Initialize variables with NULL value.
31233 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31236 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31238 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31240 2017-01-30 Richard Biener <rguenther@suse.de>
31242 PR tree-optimization/79276
31243 * tree-vrp.c (process_assert_insertions): Properly adjust common
31244 when removing a duplicate.
31246 2017-01-30 Richard Biener <rguenther@suse.de>
31248 PR tree-optimization/79256
31249 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31250 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31252 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31254 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31257 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31258 ("*r<noxa>sbg_<mode>_sll_bitmask")
31259 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31260 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31261 Use contiguous_bitmask_nowrap_operand.
31263 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31266 * config/rs6000/altivec.h (vec_xl): Revise #define.
31267 (vec_xst): Likewise.
31269 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31271 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31273 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31275 PR rtl-optimization/79194
31276 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31277 traps before call to bypass_conditional_jumps.
31279 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31281 PR tree-optimization/71374
31282 * lra-constraints.c (check_conflict_input_operands): New.
31283 (match_reload): Use it.
31285 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31288 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31289 account to calculate conflict_set.
31291 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31293 PR rtl-optimization/78559
31294 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31295 other_insn in combine.
31297 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31299 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31300 uint16_type_node for BT_UINT16.
31302 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31304 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31305 "RTL Tests" to menu.
31306 (GIMPLE Tests): New node.
31307 (RTL Tests): New node.
31309 2017-01-27 Richard Biener <rguenther@suse.de>
31311 PR tree-optimization/79245
31312 * tree-loop-distribution.c (distribute_loop): Apply cost
31313 modeling also to detected patterns.
31315 2017-01-27 Richard Biener <rguenther@suse.de>
31317 PR tree-optimization/71433
31318 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31319 (compare_assert_loc): New function.
31320 (process_assert_insertions): Sort and optimize assert locations
31321 to remove duplicates and push down identical assertions on
31322 edges to their destination block.
31324 2017-01-27 Richard Biener <rguenther@suse.de>
31326 PR tree-optimization/79244
31327 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31328 out SSA names even if abnormal.
31330 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31332 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31333 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31334 instead of MPFR_RNDN.
31336 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31339 * arm.c (arm_option_override): Don't call build_target_option_node
31340 until after doing all option overrides.
31341 (arm_valid_target_attribute_tree): Likewise.
31343 2017-01-27 Martin Liska <mliska@suse.cz>
31345 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31346 for {cd}tors and C++ {cd}tors.
31348 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31350 * config/s390/s390.md ("*setmem_long_and")
31351 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31353 2017-01-26 Martin Sebor <msebor@redhat.com>
31355 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31358 2017-01-26 Martin Sebor <msebor@redhat.com>
31360 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31361 HAVE_DFmode before using XFmode or DFmode.
31362 (parse_directive): Avoid using the z length modifier to avoid
31363 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31365 PR middle-end/78703
31366 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31367 to accept adjustment as an array.
31368 (get_int_range): New function.
31369 (struct directive): Make width and prec arrays.
31370 (directive::set_width, directive::set_precision): Call get_int_range.
31371 (format_integer, format_floating): Handle width and precision ranges.
31372 (format_string, parse_directive): Same.
31374 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31377 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31378 comdat_type_p set, just clone them, but keep the children in the
31382 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31383 which have direct callers with -fvar-tracking-assignments enabled
31385 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31386 inside of type units.
31388 2017-01-26 Martin Sebor <msebor@redhat.com>
31390 PR middle-end/78703
31391 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31393 (struct format_result): Replace number_chars, number_chars_min,
31394 and number_chars_max with a single member of struct result_range.
31396 (format_result::operator+=): Adjust.
31397 (struct fmtresult): Remove bounded. Handle likely and unlikely
31399 (fmtresult::adjust_for_width_or_precision): New function.
31400 (fmtresult:type_max_digits): New function.
31401 (bytes_remaining): Handle likely and unlikely counters.
31402 (min_bytes_remaining): Remove.
31403 (format_percent): Simplify.
31404 (format_integer, format_floating): Set likely and unlikely counters.
31405 (get_string_length, format_character, format_string): Same.
31406 (format_plain, should_warn_p): New function.
31407 (maybe_warn): Call should_warn_p. Update diagnostic messages
31408 and handle those for all directives, including plain strings.
31409 (format_directive): Handle likely and unlikely counters.
31410 Remove unnecessary quoting from diagnostics. Add an informational
31412 (add_bytes): Remove.
31413 (pass_sprintf_length::compute_format_length): Simplify.
31414 (try_substitute_return_value): Handle likely and unlikely counters.
31416 2017-01-26 Carl Love <cel@us.ibm.com>
31418 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31419 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31421 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31424 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31425 endianess for subregs into account.
31426 * lra-constraints.c (lra_constraints): Do risky transformations
31427 always on the first iteration.
31428 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31429 last_call_used_reg_set.
31430 (process_bb_lives): Define and use last_call_used_reg_set.
31431 * lra.c (lra): Always continue after lra_constraints on the first
31434 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31436 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31438 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31440 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31442 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31443 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31444 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31445 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31446 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31447 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31448 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31449 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31450 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31452 2017-01-26 Marek Polacek <polacek@redhat.com>
31455 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31456 for the third operand.
31458 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31460 PR middle-end/79236
31461 * omp-low.c (struct omp_context): Add simt_stmt field.
31462 (scan_omp_for): Return omp_context *.
31463 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31464 context to the _simt_ SIMD stmt.
31465 (lower_omp_for): For combined SIMD with sibling _simt_
31466 SIMD, make sure to use the same decls in _looptemp_
31467 clauses as in the sibling.
31469 2017-01-26 David Sherwood <david.sherwood@arm.com>
31471 PR middle-end/79212
31472 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31475 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31478 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31479 emit fld b; fld a; if possible.
31481 * brig-builtins.def: Update copyright years.
31482 * config/arm/arm_acle_builtins.def: Update copyright years.
31484 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31487 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31488 constraint instead of o for the stxsd instruction.
31490 2017-01-25 Carl Love <cel@us.ibm.com>
31492 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31493 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31495 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31497 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31499 2017-01-25 Richard Biener <rguenther@suse.de>
31501 PR tree-optimization/69264
31502 * target.def (vector_alignment_reachable): Improve documentation.
31503 * doc/tm.texi: Regenerate.
31504 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31506 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31507 earlier changes with respect to TYPE_USER_ALIGN.
31508 (vector_alignment_reachable_p): Likewise. Improve dumping.
31510 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31513 * config/arm/arm.md (xordi3): Force constant operand into a register
31516 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31518 * doc/invoke.texi (-fstore-merging): Correct default optimization
31519 levels at which it is enabled.
31520 (-O): Move -fstore-merging from list to...
31523 2017-01-25 Richard Biener <rguenther@suse.de>
31526 * omp-expand.c: Include debug.h.
31527 (expand_omp_taskreg): Make sure to generate early debug before
31528 outlining anything from a function.
31529 (expand_omp_target): Likewise.
31530 (grid_expand_target_grid_body): Likewise.
31532 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31535 * asan.c (get_translation_unit_decl): New function.
31536 (asan_add_global): Extract modules file name from globals
31537 TRANSLATION_UNIT_DECL name.
31539 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31542 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31543 for long calls with APCS frame and VFP.
31545 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31547 * cfg.c (original_copy_tables_initialized_p): New function.
31548 * cfg.h (original_copy_tables_initialized_p): New decl.
31549 * cfgrtl.c (relink_block_chain): Guard the call to
31550 free_original_copy_tables with a call to
31551 original_copy_tables_initialized_p.
31552 * cgraph.h (symtab_node::native_rtl_p): New decl.
31553 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31554 (symtab_node::needed_p): Don't assert for early assembly output
31555 for __RTL functions.
31556 (cgraph_node::finalize_function): Set "force_output" for __RTL
31558 (cgraph_node::analyze): Bail out early for __RTL functions.
31559 (analyze_functions): Update assertion to support __RTL functions.
31560 (cgraph_node::expand): Bail out early for __RTL functions.
31561 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31563 * function.h (struct function): Update comment for field
31565 * gimple-expr.c: Include "tree-pass.h".
31566 (gimple_has_body_p): Return false for __RTL functions.
31567 * Makefile.in (OBJS): Add run-rtl-passes.o.
31568 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31570 (gcc::pass_manager::get_clean_slate): New accessor.
31571 * passes.c: Include "insn-addr.h".
31572 (should_skip_pass_p): Add logging. Update logic for running
31573 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31574 property-provider override so it is only done for gimple passes.
31576 (skip_pass): New function.
31577 (execute_one_pass): Call skip_pass when skipping passes.
31578 * read-md.c (md_reader::read_char): Support filtering
31579 the input to a subset of line numbers.
31580 (md_reader::md_reader): Initialize fields
31581 m_first_line and m_last_line.
31582 (md_reader::read_file_fragment): New function.
31583 * read-md.h (md_reader::read_file_fragment): New decl.
31584 (md_reader::m_first_line): New field.
31585 (md_reader::m_last_line): New field.
31586 * read-rtl-function.c (function_reader::create_function): Only
31587 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31588 curr_properties. Set DECL_INITIAL to a dummy block.
31589 (read_rtl_function_body_from_file_range): New function.
31590 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31592 * run-rtl-passes.c: New file.
31593 * run-rtl-passes.h: New file.
31595 2017-01-24 Jeff Law <law@redhat.com>
31597 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31600 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31602 PR tree-optimization/79159
31603 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31604 (record_nonwrapping_iv): Improve boundary using above function if no
31605 value range information.
31607 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31608 Martin Jambor <mjambor@suse.cz>
31610 * brig-builtins.def: New file.
31611 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31612 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31613 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31614 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31615 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31616 * builtin-types.def (BT_INT8): New.
31617 (BT_INT16): Likewise.
31618 (BT_UINT8): Likewise.
31619 (BT_UINT16): Likewise.
31620 (BT_FN_ULONG): Likewise.
31621 (BT_FN_UINT_INT): Likewise.
31622 (BT_FN_UINT_ULONG): Likewise.
31623 (BT_FN_UINT_LONG): Likewise.
31624 (BT_FN_UINT_PTR): Likewise.
31625 (BT_FN_ULONG_PTR): Likewise.
31626 (BT_FN_INT8_FLOAT): Likewise.
31627 (BT_FN_INT16_FLOAT): Likewise.
31628 (BT_FN_UINT32_FLOAT): Likewise.
31629 (BT_FN_UINT16_FLOAT): Likewise.
31630 (BT_FN_UINT8_FLOAT): Likewise.
31631 (BT_FN_UINT64_FLOAT): Likewise.
31632 (BT_FN_UINT16_UINT32): Likewise.
31633 (BT_FN_UINT32_UINT16): Likewise.
31634 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31635 (BT_FN_INT_PTR_INT): Likewise.
31636 (BT_FN_UINT_PTR_UINT): Likewise.
31637 (BT_FN_LONG_PTR_LONG): Likewise.
31638 (BT_FN_ULONG_PTR_ULONG): Likewise.
31639 (BT_FN_VOID_UINT64_UINT64): Likewise.
31640 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31641 (BT_FN_INT8_INT8_INT8): Likewise.
31642 (BT_FN_INT16_INT16_INT16): Likewise.
31643 (BT_FN_INT_INT_INT): Likewise.
31644 (BT_FN_UINT_FLOAT_UINT): Likewise.
31645 (BT_FN_FLOAT_UINT_UINT): Likewise.
31646 (BT_FN_ULONG_UINT_UINT): Likewise.
31647 (BT_FN_ULONG_UINT_PTR): Likewise.
31648 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31649 (BT_FN_UINT_UINT_UINT): Likewise.
31650 (BT_FN_VOID_UINT_PTR): Likewise.
31651 (BT_FN_UINT_UINT_PTR: Likewise.
31652 (BT_FN_UINT32_UINT64_PTR): Likewise.
31653 (BT_FN_INT_INT_UINT_UINT): Likewise.
31654 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31655 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31656 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31657 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31658 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31659 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31660 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31661 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31662 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31663 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31664 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31665 * doc/frontends.texi: List BRIG FE.
31666 * doc/install.texi (Testing): Add BRIG tesring requirements.
31667 * doc/invoke.texi (Overall Options): Mention BRIG.
31668 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31670 2017-01-24 Richard Biener <rguenther@suse.de>
31672 PR translation/79208
31673 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31675 2017-01-24 Martin Jambor <mjambor@suse.cz>
31678 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31679 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31680 and known_contexts.
31682 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31684 PR middle-end/79123
31685 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31686 casts from signed to unsigned really don't have a range.
31688 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31690 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31691 GMP_RNDx for compatiblity.
31693 2017-01-24 Martin Liska <mliska@suse.cz>
31696 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31697 that would prevent us to call alloca with -1 as argument.
31699 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31701 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31702 Avoid -Wformat-security warning.
31704 2017-01-23 Andrew Pinski <apinski@cavium.com>
31706 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31709 2017-01-23 Martin Sebor <msebor@redhat.com>
31711 PR middle-end/78703
31712 * gimple-ssa-sprintf.c (warn_level): New global.
31713 (format_integer): Use it here and throughout the rest of the file.
31714 Use the same switch to compute sign as base.
31715 (maybe_warn): New function.
31716 (format_directive): Factor out warnings into maybe_warn.
31717 Add debugging output. Use warn_level.
31718 (add_bytes): Use warn_level.
31719 (pass_sprintf_length::compute_format_length): Add debugging output.
31720 (try_substitute_return_value): Same.
31721 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31723 PR middle-end/78703
31724 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31725 (struct fmtresult, format_integer, format_floating): Adjust.
31726 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31727 (get_string_length, format_string,format_directive): Same.
31728 (pass_sprintf_length::compute_format_length): Same.
31729 (try_substitute_return_value): Simplify slightly.
31731 PR middle-end/78703
31732 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31733 (fmtresult::operator+=): Outlined.
31734 (struct fmtresult): Add ctors.
31735 (struct conversion_spec): Rename...
31736 (struct directive): ...to this. Add and remove data members.
31737 (directive::set_width, directive::set_precision): New functions.
31738 (format_percent): Use fmtresult ctor.
31739 (get_width_and_precision): Remove.
31740 (format_integer): Make naming changes. Avoid computing width and
31742 (format_floating): Same. Adjust indentation.
31743 (format_character, format_none): New functions.
31744 (format_string): Moved character handling to format_character.
31745 (format_directive): Remove arguments, change return type.
31746 (parse_directive): New function.
31747 (pass_sprintf_length::compute_format_length): Move directive
31748 parsing to parse_directive.
31750 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31752 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31753 (assign_assembler_name_if_needed): ... this.
31754 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31755 (assign_assembler_name_if_needed): ... this.
31756 (free_lang_data_in_cgraph): Adjust callers.
31757 * cgraphunit.c (cgraph_node::analyze): Likewise.
31758 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31761 2017-01-23 Richard Biener <rguenther@suse.de>
31763 PR tree-optimization/79088
31764 PR tree-optimization/79188
31765 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31766 resetting loop bounds after last path deletion. Reset loop
31767 bounds of the target loop, make code match the comments.
31768 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31769 Make sure loops need no fixups.
31771 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31773 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31774 exponent support with double type for first argument.
31775 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31776 type returned by __builtin_vec_extract_sig,
31777 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31778 functions from "vector int" to "vector unsigned int" or from
31779 "vector long long int" to "vector unsigned long long int".
31780 Changed type returned by __builtin_vec_extract_exp,
31781 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31782 functions from "vector int" to "vector unsigned int" or from
31783 "vector long long int" to "vector unsigned long long int".
31784 Changed return type of __builtin_vec_test_data_class,
31785 __builtin_vec_test_data_class_sp, and
31786 __builtin_vec_test_data_class_dp from "vector int" to
31787 "vector bool int" or from "vector long long int" to "vector bool
31788 long long int" and changed second argument type from "unsigned
31789 int" to "int". Added new overloaded function forms "vector float
31790 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31791 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31792 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31793 double, vector unsigned long long int)" and "vector double
31794 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31795 long int)". Changed return type of
31796 __builtin_scalar_test_data_class and
31797 __builtin_scalar_test_data_class_sp and
31798 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31799 int" and changed second argument from "unsigned int" to "int".
31800 Changed type returned by __builtin_scalar_test_neg,
31801 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31802 from "int" to "bool int". Added new overloaded function form
31803 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31804 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31805 exponent double-precision with floating point first argument.
31806 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31807 documentation of scalar_test_data_class, scalar_test_neg,
31808 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31809 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31810 vec_test_data_class built-in functions to reflect refinements in
31811 their type signatures.
31813 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31815 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31817 (aarch64_elf_asm_destructor): Likewise.
31819 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31821 PR rtl-optimization/78634
31822 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31823 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31824 * ifcvt.c (noce_try_cmove): Add missing cost check.
31826 PR rtl-optimization/71724
31827 * combine.c (if_then_else_cond): Look for situations where it is
31828 beneficial to undo the work of one of the recursive calls.
31830 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31832 PR tree-optimization/70754
31833 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31834 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31835 combined stmt before it if not NULL.
31836 (combine_chains): Process refs reversely and compute dominance point
31839 2017-01-23 Martin Liska <mliska@suse.cz>
31841 PR tree-optimization/79196
31842 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31843 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31845 (strlen_optimize_stmt): Call the renamed function.
31847 2017-01-23 Michael Matz <matz@suse.de>
31849 PR tree-optimization/78384
31850 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31852 2017-01-23 Richard Biener <rguenther@suse.de>
31854 PR tree-optimization/79186
31855 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31856 both incoming edges before moving an assert.
31858 2017-01-23 Martin Jambor <mjambor@suse.cz>
31860 * ipa-prop.c (load_from_param_1): Removed.
31861 (load_from_unmodified_param): Bits from load_from_param_1 put back
31863 (load_from_param): Removed.
31864 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31865 with stmt. Reverted back to use of load_from_unmodified_param.
31867 2017-01-23 Martin Jambor <mjambor@suse.cz>
31870 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31871 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31872 field a pointer to garbage collected vector, mark lattices and
31873 ipcp_orig_node with GTY((skip)).
31874 (ipa_get_param_count): Adjust to descriptors being a pointer.
31875 (ipa_get_param): Likewise.
31876 (ipa_get_type): Likewise.
31877 (ipa_get_param_move_cost): Likewise.
31878 (ipa_set_param_used): Likewise.
31879 (ipa_get_controlled_uses): Likewise.
31880 (ipa_set_controlled_uses): Likewise.
31881 (ipa_is_param_used): Likewise.
31882 (ipa_node_params_t): Move into garbage collector. New methods insert
31884 (ipa_node_params_sum): Annotate wth GTY(()).
31885 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31887 (ipa_load_from_parm_agg): Adjust declaration.
31888 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31889 * ipa-profile.c (ipa_profile): Likewise.
31890 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31891 (ipa_populate_param_decls): Make descriptors parameter garbage
31893 (ipa_dump_param): Adjust to descriptors being a pointer.
31894 (ipa_alloc_node_params): Likewise.
31895 (ipa_initialize_node_params): Likewise.
31896 (load_from_param_1): Make descriptors parameter garbage collected.
31897 (load_from_unmodified_param): Likewise.
31898 (load_from_param): Likewise.
31899 (ipa_load_from_parm_agg): Likewise.
31900 (ipa_node_params::~ipa_node_params): Removed.
31901 (ipa_free_all_node_params): Remove call to delete operator.
31902 (ipa_node_params_t::insert): New.
31903 (ipa_node_params_t::remove): Likewise.
31904 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31905 copy known_csts and known_contexts vectors.
31906 (ipa_read_node_info): Adjust to descriptors being a pointer.
31907 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31909 (ipcp_transform_function): Make descriptors variable garbage
31912 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31914 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31915 * config/i386/avx512dqintrin.h: Ditto.
31916 * config/i386/avx512fintrin.h: Ditto.
31917 * config/i386/i386.c: Handle new builtins.
31918 * config/i386/i386-builtin.def: Add new builtins.
31919 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31920 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31922 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31923 Martin Liska <mliska@suse.cz>
31925 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31926 * asan.c (asan_expand_poison_ifn): Support stores and use
31927 appropriate ASAN report function.
31928 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31929 * internal-fn.def (ASAN_POISON_USE): Declare.
31930 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31931 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31932 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31933 ASAN_POISON calls w/o LHS.
31934 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31935 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31936 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31937 * gimplify.c (asan_poison_variables): Add attribute
31938 use_after_scope_memory to variables that really needs to live
31940 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31941 having the attribute.
31943 2017-01-23 Martin Liska <mliska@suse.cz>
31945 * asan.c (create_asan_shadow_var): New function.
31946 (asan_expand_poison_ifn): Likewise.
31947 * asan.h (asan_expand_poison_ifn): New declaration.
31948 * internal-fn.c (expand_ASAN_POISON): Likewise.
31949 * internal-fn.def (ASAN_POISON): New builtin.
31950 * sanopt.c (pass_sanopt::execute): Expand
31951 asan_expand_poison_ifn.
31952 * tree-inline.c (copy_decl_for_dup_finish): Make function
31954 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31955 * tree-ssa.c (is_asan_mark_p): New function.
31956 (execute_update_addresses_taken): Rewrite local variables
31957 (identified just by use-after-scope as addressable) into SSA.
31959 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31961 * doc/install.texi (Specific): opensource.apple.com uses https
31962 now. Remove trailing slash.
31964 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31966 * README.Portability: Remove note on an Irix compatibility issue.
31968 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31970 * gcov.c (INCLUDE_ALGORITHM): Define.
31971 (INCLUDE_VECTOR): Define.
31972 No longer include <vector> and <algorithm> directly.
31974 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31976 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31978 * doc/invoke.texi (Code Gen Options): Ditto.
31980 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31983 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31985 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31987 rtl-optimization/79125
31988 * cprop.c (local_cprop_pass): Handle cases where we make an
31989 unconditional trap.
31991 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31995 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31996 read from, for big endian.
31998 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32000 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32001 register pauth builtins for LP64 only.
32003 2017-01-20 Marek Polacek <polacek@redhat.com>
32006 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32009 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32011 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32013 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32014 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32015 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32017 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32020 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32021 in big-endian mode when they are not a single duplicated value.
32023 2017-01-20 Richard Biener <rguenther@suse.de>
32025 * BASE-VER: Bump to 7.0.1.
32027 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32029 * omp-low.c (omplow_simd_context): New struct. Use it...
32030 (lower_rec_simd_input_clauses): ...here and...
32031 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32032 references to idx, lane, max_vf, is_simt.
32034 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32036 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32039 2017-01-20 Martin Jambor <mjambor@suse.cz>
32041 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32042 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32044 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32045 (GTFILES): Rename hsa.c to hsa-common.c.
32046 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32047 * hsa-dump.c: Likewise.
32048 * hsa-gen.c: Likewise.
32049 * hsa-regalloc.c: Likewise.
32050 * ipa-hsa.c: Likewise.
32051 * omp-expand.c: Likewise.
32052 * omp-low.c: Likewise.
32053 * toplev.c: Likewise.
32055 2017-01-20 Marek Polacek <polacek@redhat.com>
32058 * doc/invoke.texi: Document -Wduplicated-branches.
32059 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32060 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32061 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32062 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32063 return 0 only when not OEP_LEXICOGRAPHIC.
32064 (fold_build_cleanup_point_expr): Use the expression
32065 location when building CLEANUP_POINT_EXPR.
32066 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32067 * tree.c (add_expr): Handle error_mark_node.
32069 2017-01-20 Martin Liska <mliska@suse.cz>
32072 * tree-profile.c (init_ic_make_global_vars): Do not call
32074 (gimple_init_gcov_profiler): Likewise.
32076 2017-01-20 Martin Liska <mliska@suse.cz>
32079 * cgraph.h (maybe_create_reference): Remove argument and
32081 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32083 * ipa-cp.c (create_specialized_node): Likewise.
32084 * symtab.c (symtab_node::maybe_create_reference): Handle
32085 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32087 2017-01-20 Martin Liska <mliska@suse.cz>
32089 * read-rtl-function.c (function_reader::create_function): Use
32090 build_decl instread of build_decl_stat.
32092 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32094 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32095 * config/i386/avx512dqintrin.h: Ditto.
32096 * config/i386/avx512fintrin.h: Ditto.
32097 * config/i386/i386-builtin-types.def: Add new types.
32098 * config/i386/i386.c: Handle new types.
32099 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32100 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32101 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32102 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32103 (__builtin_ia32_kshiftridi): New.
32104 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32106 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32110 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32111 define to rs6000_init_stack_protect_guard.
32112 (rs6000_init_stack_protect_guard): New function.
32114 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32115 Yunqiang Su <yunqiang.su@imgtec.com>
32117 * config.gcc (supported_defaults): Add madd4.
32118 (with_madd4): Add validation.
32119 (all_defaults): Add madd4.
32120 * config/mips/mips.opt (mmadd4): New option.
32121 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32123 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32125 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32126 (ISA_HAS_FUSED_MADD4): Likewise.
32127 * doc/invoke.texi (-mmadd4): Document the new option.
32128 * doc/install.texi (--with-madd4): Document the new option.
32130 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32132 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32133 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32134 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32135 (aarch64_init_pauth_hint_builtins): New.
32136 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32137 (aarch64_expand_builtin): Expand new builtins.
32139 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32141 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32142 * combine-stack-adj.c (no_unhandled_cfa): Handle
32143 REG_CFA_TOGGLE_RA_MANGLE.
32144 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32145 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32146 info for return address signing.
32147 (aarch64_expand_epilogue): Likewise.
32149 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32151 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32152 * config/aarch64/aarch64-protos.h
32153 (aarch64_return_address_signing_enabled): New declaration.
32154 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32156 (aarch64_expand_prologue): Sign return address before it's pushed onto
32158 (aarch64_expand_epilogue): Authenticate return address fetched from
32160 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32161 (aarch64_attributes): New function attributes for "sign-return-address".
32162 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32163 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32164 ("*do_return"): Generate combined instructions according to key index.
32165 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32166 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32168 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32169 * config/aarch64/aarch64.opt (msign-return-address=): New.
32170 * doc/extend.texi (AArch64 Function Attributes): Documents
32171 "sign-return-address=".
32172 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32174 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32176 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32177 overall option summary.
32179 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32181 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32182 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32183 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32184 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32186 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32188 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32189 -mpower9-minmax by default for -mcpu=power9.
32190 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32191 128-bit floating point.
32193 2017-01-20 Alan Modra <amodra@gmail.com>
32195 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32196 optimizing for size.
32198 2017-01-20 Alan Modra <amodra@gmail.com>
32201 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32202 for strcmp and strncmp from corresponding builtin decl.
32204 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32206 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32207 instead of i386/rtems-64.h.
32208 * config/i386/rtems-64.h: Remove.
32210 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32214 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32216 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32218 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32220 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32221 Change int to HOST_WIDE_INT.
32222 * config/aarch64/aarch64-protos.h
32223 (aarch64_simd_gen_const_vector_dup): Likewise.
32224 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32226 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32228 * langhooks-def.h (lhd_type_for_size): New decl.
32229 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32230 * langhooks.c (lhd_type_for_size): New function, taken from
32233 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32235 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32236 define_bypass for CR latency.
32237 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32238 (power9-alu2): Add define_bypass for CR latency.
32240 (power9-mul): Update insn latency.
32241 (power9-mul-compare): Update insn latency, bypass latency and remove
32244 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32246 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32248 * config/aarch64/aarch64.md
32249 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32250 aarch64_nopcrelative_literal_loads.
32251 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32253 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32255 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32256 TARGET_LOONGSON_3A.
32257 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32259 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32262 * config.gcc (supported_defaults): Add lxc1-sxc1.
32263 (with_lxc1_sxc1): Add validation.
32264 (all_defaults): Add lxc1-sxc1.
32265 * config/mips/mips.opt (mlxc1-sxc1): New option.
32266 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32268 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32269 __mips_no_lxc1_sxc1.
32270 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32271 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32272 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32274 2017-01-19 Richard Biener <rguenther@suse.de>
32276 PR tree-optimization/72488
32277 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32278 sure to restore SSA info.
32279 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32281 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32283 PR rtl-optimization/79121
32284 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32285 of the inner type when shifting an extended value.
32287 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32290 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32291 interposable aliases.
32293 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32296 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32297 Use the evmergelohi instruction.
32298 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32299 (mov_sitf_e500_subreg8_2_be): Likewise.
32300 (mov_sitf_e500_subreg12_2_le): Likewise.
32301 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32302 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32303 (mov_sitf_e500_subreg8_2_le): Likewise.
32304 (mov_sitf_e500_subreg12_2_be): Likewise.
32306 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32308 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32309 attribute from vecsimple to vecperm.
32310 (altivec_vbpermq2): Likewise.
32312 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32315 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32317 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32318 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32319 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32320 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32321 case where N arg is SIZE_MAX.
32322 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32323 (cmpstrsi): Add pattern.
32325 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32327 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32328 __builtin_vec_revb builtins.
32329 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32330 built-in functions to support generation of the ISA 3.0 XXBR<x>
32331 vector byte reverse instructions.
32332 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32333 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32334 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32335 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32336 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32337 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32338 (P9V_BUILTIN_VEC_REVB): Likewise.
32339 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32340 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32341 (p9_xxbrq_v16qi): Likewise.
32342 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32343 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32344 (p9_xxbrh_v8hi): Likewise.
32345 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32346 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32347 vec_revb built-in functions.
32349 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32351 PR rtl-optimization/78952
32352 * config/i386/i386.md (any_extract): New code iterator.
32353 (*insvqi_2): Use any_extract for source operand.
32354 (*insvqi_3): Use any_shiftrt for source operand.
32356 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32358 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32360 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32362 2017-01-18 Matthias Klose <doko@ubuntu.com>
32364 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32366 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32368 * config/rs6000/altivec.h (vec_bperm): Change #define.
32369 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32370 (altivec_vbpermq2): New define_insn.
32371 (altivec_vbpermd): Likewise.
32372 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32373 function interface.
32374 (VBPERMD): Likewise.
32375 (VBPERM): New polymorphic function interface.
32376 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32377 Add entries for P9V_BUILTIN_VEC_VBPERM.
32378 * doc/extend.texi: Add interfaces for vec_bperm.
32380 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32382 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32383 first letter of error messages.
32384 (s390_resolve_overloaded_builtin): Likewise.
32385 * config/s390/s390.c (s390_expand_builtin): Likewise.
32386 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32387 (s390_valid_target_attribute_inner_p): Likewise.
32388 * config/s390/s390.md ("tabort"): Likewise.
32390 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32392 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32393 (ISA_AVOID_DIV_HILO): New macro.
32394 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32395 (ISA_HAS_DDIV): Likewise.
32397 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32399 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32401 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32403 * doc/invoke.texi (fabi-version): Spelling fix.
32405 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32408 * doc/invoke.texi (fabi-version): Mention mangling fix for
32411 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32414 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32416 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32419 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32420 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32422 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32424 (stack_protect_test): Ditto.
32425 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32426 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32428 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32429 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32430 -mstack-protector-guard-offset=.
32431 (RS/6000 and PowerPC Options): Ditto.
32433 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32435 * config/i386/i386.h (MASK_CLASS_P): New define.
32436 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32437 there are no registers from different register sets also when
32438 mask registers are used. Update function comment.
32439 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32440 to (*k/*r) and (*k/*km) alternatives.
32442 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32444 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32445 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32446 (EH_RETURN_HANDLER_RTX): New define.
32447 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32448 Force frame pointer in EH return functions.
32449 (aarch64_expand_epilogue): Add barrier for eh_return.
32450 (aarch64_final_eh_return_addr): Remove.
32451 (aarch64_eh_return_handler_rtx): New function.
32452 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32454 (aarch64_eh_return_handler_rtx): New prototype.
32456 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32458 * config/rs6000/altivec.h (vec_rlmi): New #define.
32459 (vec_vrlnm): Likewise.
32460 (vec_rlnm): Likewise.
32461 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32462 (UNSPEC_VRLNM): Likewise.
32463 (VIlong): New mode iterator.
32464 (altivec_vrl<VI_char>mi): New define_insn.
32465 (altivec_vrl<VI_char>nm): Likewise.
32466 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32468 (VRLDNM): Likewise.
32469 (RLNM): New polymorphic function entry.
32470 (VRLWMI): New monomorphic function entry.
32471 (VRLDMI): Likewise.
32472 (RLMI): New polymorphic function entry.
32473 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32474 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32475 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32478 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32481 * dwarf2out.c (field_byte_offset): Restore the
32482 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32483 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32484 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32487 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32490 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32492 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32495 * dwarf2out.c (add_data_member_location_attribute): For constant
32496 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32497 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32498 DW_AT_byte_size attributes.
32500 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32502 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32503 after forcing to constant memory when the code model is medium.
32505 2017-01-17 Julia Koval <julia.koval@intel.com>
32508 * config/i386/avx512fintrin.h
32509 (_mm512_i32gather_ps): Change __addr type to void const*.
32510 (_mm512_mask_i32gather_ps): Ditto.
32511 (_mm512_i32gather_pd): Ditto.
32512 (_mm512_mask_i32gather_pd): Ditto.
32513 (_mm512_i64gather_ps): Ditto.
32514 (_mm512_mask_i64gather_ps): Ditto.
32515 (_mm512_i64gather_pd): Ditto.
32516 (_mm512_mask_i64gather_pd): Ditto.
32517 (_mm512_i32gather_epi32): Ditto.
32518 (_mm512_mask_i32gather_epi32): Ditto.
32519 (_mm512_i32gather_epi64): Ditto.
32520 (_mm512_mask_i32gather_epi64): Ditto.
32521 (_mm512_i64gather_epi32): Ditto.
32522 (_mm512_mask_i64gather_epi32): Ditto.
32523 (_mm512_i64gather_epi64): Ditto.
32524 (_mm512_mask_i64gather_epi64): Ditto.
32525 (_mm512_i32scatter_ps): Change __addr type to void*.
32526 (_mm512_mask_i32scatter_ps): Ditto.
32527 (_mm512_i32scatter_pd): Ditto.
32528 (_mm512_mask_i32scatter_pd): Ditto.
32529 (_mm512_i64scatter_ps): Ditto.
32530 (_mm512_mask_i64scatter_ps): Ditto.
32531 (_mm512_i64scatter_pd): Ditto.
32532 (_mm512_mask_i64scatter_pd): Ditto.
32533 (_mm512_i32scatter_epi32): Ditto.
32534 (_mm512_mask_i32scatter_epi32): Ditto.
32535 (_mm512_i32scatter_epi64): Ditto.
32536 (_mm512_mask_i32scatter_epi64): Ditto.
32537 (_mm512_i64scatter_epi32): Ditto.
32538 (_mm512_mask_i64scatter_epi32): Ditto.
32539 (_mm512_i64scatter_epi64): Ditto.
32540 (_mm512_mask_i64scatter_epi64): Ditto.
32541 * config/i386/avx512pfintrin.h
32542 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32543 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32544 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32545 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32546 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32547 (_mm512_prefetch_i32scatter_ps): Ditto.
32548 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32549 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32550 (_mm512_prefetch_i64scatter_pd): Ditto.
32551 (_mm512_prefetch_i64scatter_ps): Ditto.
32552 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32553 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32554 * config/i386/avx512vlintrin.h
32555 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32556 (_mm_mmask_i32gather_ps): Ditto.
32557 (_mm256_mmask_i32gather_pd): Ditto.
32558 (_mm_mmask_i32gather_pd): Ditto.
32559 (_mm256_mmask_i64gather_ps): Ditto.
32560 (_mm_mmask_i64gather_ps): Ditto.
32561 (_mm256_mmask_i64gather_pd): Ditto.
32562 (_mm_mmask_i64gather_pd): Ditto.
32563 (_mm256_mmask_i32gather_epi32): Ditto.
32564 (_mm_mmask_i32gather_epi32): Ditto.
32565 (_mm256_mmask_i32gather_epi64): Ditto.
32566 (_mm_mmask_i32gather_epi64): Ditto.
32567 (_mm256_mmask_i64gather_epi32): Ditto.
32568 (_mm_mmask_i64gather_epi32): Ditto.
32569 (_mm256_mmask_i64gather_epi64): Ditto.
32570 (_mm_mmask_i64gather_epi64): Ditto.
32571 (_mm256_i32scatter_ps): Change __addr type to void*.
32572 (_mm256_mask_i32scatter_ps): Ditto.
32573 (_mm_i32scatter_ps): Ditto.
32574 (_mm_mask_i32scatter_ps): Ditto.
32575 (_mm256_i32scatter_pd): Ditto.
32576 (_mm256_mask_i32scatter_pd): Ditto.
32577 (_mm_i32scatter_pd): Ditto.
32578 (_mm_mask_i32scatter_pd): Ditto.
32579 (_mm256_i64scatter_ps): Ditto.
32580 (_mm256_mask_i64scatter_ps): Ditto.
32581 (_mm_i64scatter_ps): Ditto.
32582 (_mm_mask_i64scatter_ps): Ditto.
32583 (_mm256_i64scatter_pd): Ditto.
32584 (_mm256_mask_i64scatter_pd): Ditto.
32585 (_mm_i64scatter_pd): Ditto.
32586 (_mm_mask_i64scatter_pd): Ditto.
32587 (_mm256_i32scatter_epi32): Ditto.
32588 (_mm256_mask_i32scatter_epi32): Ditto.
32589 (_mm_i32scatter_epi32): Ditto.
32590 (_mm_mask_i32scatter_epi32): Ditto.
32591 (_mm256_i32scatter_epi64): Ditto.
32592 (_mm256_mask_i32scatter_epi64): Ditto.
32593 (_mm_i32scatter_epi64): Ditto.
32594 (_mm_mask_i32scatter_epi64): Ditto.
32595 (_mm256_i64scatter_epi32): Ditto.
32596 (_mm256_mask_i64scatter_epi32): Ditto.
32597 (_mm_i64scatter_epi32): Ditto.
32598 (_mm_mask_i64scatter_epi32): Ditto.
32599 (_mm256_i64scatter_epi64): Ditto.
32600 (_mm256_mask_i64scatter_epi64): Ditto.
32601 (_mm_i64scatter_epi64): Ditto.
32602 (_mm_mask_i64scatter_epi64): Ditto.
32603 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32604 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32605 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32606 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32607 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32608 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32609 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32610 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32611 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32612 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32613 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32614 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32615 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32616 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32617 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32618 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32619 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32620 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32621 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32622 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32623 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32624 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32625 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32626 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32627 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32628 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32629 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32630 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32631 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32632 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32633 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32634 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32635 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32636 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32637 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32638 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32639 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32640 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32641 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32642 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32643 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32644 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32645 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32646 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32647 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32648 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32649 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32650 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32651 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32652 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32653 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32654 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32655 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32656 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32657 definitions accordingly.
32659 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32660 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32663 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32666 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32669 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32670 hard regs for inner regno.
32672 2017-01-17 Martin Liska <mliska@suse.cz>
32675 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32676 assumption and add comment.
32678 2017-01-17 Nathan Sidwell <nathan@acm.org>
32680 * ipa-visibility.c (localize_node): New function, broken out of ...
32681 (function_and_variable_visibility): ... here. Call it.
32683 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32685 PR middle-end/77445
32686 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32687 correctly set frequency of oudgoing edge.
32688 (duplicate_thread_path): Fix profile updating.
32690 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32693 * configure.ac: Add GCC_BASE_VER.
32694 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32695 version from BASE-VER file.
32696 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32697 (gcc.o): Depend on $(BASEVER).
32698 * common.opt (dumpfullversion): New option.
32699 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32700 * doc/invoke.texi: Document -dumpfullversion.
32701 * doc/install.texi: Document --with-gcc-major-version-only.
32702 * configure: Regenerated.
32704 2017-01-17 Richard Biener <rguenther@suse.de>
32706 PR tree-optimization/71433
32707 * tree-vrp.c (register_new_assert_for): Merge same asserts
32708 on all incoming edges.
32709 (process_assert_insertions_for): Handle insertions at the
32712 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32714 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32715 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32717 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32720 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32723 2017-01-17 Alan Modra <amodra@gmail.com>
32726 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32727 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32728 symbolic stack limit when pic.
32730 2017-01-16 Martin Sebor <msebor@redhat.com>
32732 PR tree-optimization/78608
32733 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32735 2017-01-16 Jeff Law <law@redhat.com>
32738 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32739 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32740 for several include directories that may be relative to sysroot.
32741 * config/i386/x-mingw32 (gplus_includedir): Define.
32742 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32743 (native_system_includedir): Likewise.
32744 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32745 override if TARGET_SYSTEM_ROOT is defined.
32746 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32748 PR tree-optimization/79090
32749 PR tree-optimization/33562
32750 PR tree-optimization/61912
32751 PR tree-optimization/77485
32752 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32753 and computed trims into the dump file.
32755 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32757 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32759 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32762 * gimplify.c (gimplify_init_constructor): If want_value and
32763 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32767 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32768 sequence. Formatting fixes.
32769 (doloop_optimize): Formatting fixes.
32772 * gcc.c (debug_level_greater_than_spec_func): New function.
32773 (static_spec_functions): Add debug-level-gt spec function.
32774 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32776 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32777 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32778 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32779 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32780 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32781 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32783 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32785 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32786 QImode fixups to general and mask registers only.
32788 2017-01-16 Carl Love <cel@us.ibm.com>
32790 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32791 for built-in functions
32792 vector signed char vec_nabs (vector signed char)
32793 vector signed short vec_nabs (vector signed short)
32794 vector signed int vec_nabs (vector signed int)
32795 vector signed long long vec_nabs (vector signed long long)
32796 vector float vec_nabs (vector float)
32797 vector double vec_nabs (vector double)
32798 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32800 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32801 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32802 * doc/extend.texi: Update the documentation file for the new built-in
32805 2017-01-16 Martin Sebor <msebor@redhat.com>
32807 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32810 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32812 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32813 UNSPEC_VSX__XXSPLTD to require special splat handling.
32815 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32818 * system.h: Poison strndup.
32820 2017-01-16 Alan Modra <amodra@gmail.com>
32823 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32826 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32828 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32830 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32832 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32833 call recog here. Assert that INSN_CODE (insn) is non-negative.
32835 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32838 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32840 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32841 in the currently scheduled RTL fragment.
32843 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32845 PR rtl-optimization/78751
32846 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32849 2017-01-14 Jeff Law <law@redhat.com>
32851 PR tree-optimization/79090
32852 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32853 variable length stores.
32854 (compute_trims): Delete dead assignment to *trim_tail.
32855 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32858 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32860 PR rtl-optimization/78626
32861 PR rtl-optimization/78727
32862 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32863 of a block, and split such blocks after everything else is finished.
32865 2017-01-14 Alan Modra <amodra@gmail.com>
32868 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32869 target legitimate_combined_insn.
32870 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32871 (rs6000_legitimate_combined_insn): New function.
32872 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32874 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32875 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32876 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32878 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32880 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32882 2017-01-13 Jeff Law <law@redhat.com>
32884 PR tree-optimization/33562
32885 PR tree-optimization/61912
32886 PR tree-optimization/77485
32887 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32889 (delete_dead_assignment): Likewise.
32890 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32891 statement to delete_dead_call and delete_dead_assignment.
32893 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32896 * substring-locations.c (format_warning_va): Strengthen case 1 so
32897 that both endpoints of the substring must be within the format
32898 range for just the substring to be printed.
32900 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32902 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32903 * config/i386/i386.c (ix86_target_string): Add missing options
32904 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32905 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32906 flags_other and ix86_target_other to flags2_other. Display unknown
32908 (ix86_valid_target_attribute_inner_p): Add missing options and
32909 reorder options by implied ISAs, as in ix86_target_string.
32911 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32913 * hash-table.h (hash_table::too_empty_p): New function.
32914 (hash_table::expand): Use it.
32915 (hash_table::traverse): Likewise.
32916 (hash_table::empty_slot): Use sizeof (value_type) instead of
32917 sizeof (PTR) to convert bytes to elements. Shrink the table
32918 if the current size is excessive for the current number of
32921 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32923 * ira-costs.c (record_reg_classes): Break from the inner loop
32924 early once alt_fail is known to be true. Update outer loop
32925 handling accordingly.
32927 2017-01-13 Jeff Law <law@redhat.com>
32929 * tree-ssa-dse.c (decrement_count): New function.
32930 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32931 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32932 when we know the partially dead statement is a mem* function.
32934 PR tree-optimization/61912
32935 PR tree-optimization/77485
32936 * tree-ssa-dse.c: Include expr.h.
32937 (maybe_trim_constructor_store): New function.
32938 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32940 PR tree-optimization/33562
32941 PR tree-optimization/61912
32942 PR tree-optimization/77485
32943 * doc/invoke.texi: Document new dse-max-object-size param.
32944 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32945 * tree-ssa-dse.c: Include params.h.
32946 (dse_store_status): New enum.
32947 (initialize_ao_ref_for_dse): New, partially extracted from
32949 (valid_ao_ref_for_dse, normalize_ref): New.
32950 (setup_live_bytes_from_ref, compute_trims): Likewise.
32951 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32952 (maybe_trim_partially_dead_store): Likewise.
32953 (maybe_trim_complex_store): Likewise.
32954 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32955 Track what bytes live from the original store. Return tri-state
32956 for dead, partially dead or live.
32957 (dse_dom_walker): Add constructor, destructor and new private members.
32958 (delete_dead_call, delete_dead_assignment): New extracted from
32960 (dse_optimize_stmt): Make a member of dse_dom_walker.
32961 Use initialize_ao_ref_for_dse.
32963 PR tree-optimization/33562
32964 PR tree-optimization/61912
32965 PR tree-optimization/77485
32966 * sbitmap.h (bitmap_count_bits): Prototype.
32967 (bitmap_clear_range, bitmap_set_range): Likewise.
32968 * sbitmap.c (bitmap_clear_range): New function.
32969 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32971 2017-01-13 Martin Liska <mliska@suse.cz>
32974 * function.c (set_cfun): Add new argument force.
32975 * function.h (set_cfun): Likewise.
32976 * ipa-inline-transform.c (inline_call): Use the function when
32977 strict alising from is dropped for function we inline to.
32979 2017-01-13 Richard Biener <rguenther@suse.de>
32981 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32982 for dumping GIMPLE INTEGER_CSTs.
32984 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32986 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32987 to 201112L since C++17.
32989 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32992 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32993 if -fsanitize=kernel-address is present.
32995 2017-01-13 Richard Biener <rguenther@suse.de>
32997 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32998 as _Literal ( type ) number in case usual suffixes do not
32999 preserve all information.
33001 2017-01-13 Richard Biener <rguenther@suse.de>
33003 PR tree-optimization/77283
33004 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33005 and ssa-iterators.h.
33006 (is_feasible_trace): Implement a cost model based on joiner
33009 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33012 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33013 char or short to __float128/_Float128 directly.
33015 2017-01-12 Martin Sebor <msebor@redhat.com>
33017 to -Wformat-overflow.
33018 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33019 (min_bytes_remaining): Same.
33020 (get_string_length): Same.
33021 (format_string): Same.
33022 (format_directive): Same.
33024 (pass_sprintf_length::handle_gimple_call): Same.
33026 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33028 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33029 info.nowrite calls with no lhs that can't throw. Return bool
33030 whether gsi_remove has been called or not.
33031 (pass_sprintf_length::handle_gimple_call): Return bool whether
33032 try_substitute_return_value called gsi_remove. Formatting fix.
33033 (pass_sprintf_length::execute): Don't use gsi_remove if
33034 handle_gimple_call returned true.
33037 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33038 be removed due to side-effects, don't remove following barrier nor
33039 turn the successor edge into fallthru edge.
33041 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33044 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33045 element-reversing loads and stores as not swappable.
33047 2017-01-12 Nathan Sidwell <nathan@acm.org>
33048 Nicolai Stange <nicstange@gmail.com>
33050 * combine.c (try_combine): Don't ignore result of overlap checking
33051 loop. Combine overlap & asm check into single loop.
33053 2017-01-12 Richard Biener <rguenther@suse.de>
33055 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33056 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33058 2017-01-12 Richard Biener <rguenther@suse.de>
33060 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33061 and TS_TARGET_OPTION directly derive from TS_BASE.
33062 * tree-core.h (tree_optimization_option): Derive from tree_base.
33063 (tree_target_option): Likewise.
33065 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33067 * config/i386/i386.c (memory_address_length): Increase len
33068 only when rip_relative_addr_p returns false.
33070 2017-01-11 Julia Koval <julia.koval@intel.com>
33072 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33073 (OPTION_MASK_ISA_SGX_SET): New.
33074 (ix86_handle_option): Handle OPT_msgx.
33075 * config.gcc: Added sgxintrin.h.
33076 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33077 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33078 * config/i386/i386.c (ix86_target_string): Add -msgx.
33080 (ix86_option_override_internal): Handle new options.
33081 (ix86_valid_target_attribute_inner_p): Add sgx.
33082 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33083 * config/i386/i386.opt: Add msgx.
33084 * config/i386/sgxintrin.h: New file.
33085 * config/i386/x86intrin.h: Add sgxintrin.h.
33087 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33090 * fold-const.c (maybe_nonzero_address): Return 1 for function
33092 (tree_single_nonzero_warnv_p): Don't handle function local objects
33096 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33099 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33102 * opts.c: Include "spellcheck.h"
33103 (struct string_fragment): New struct.
33104 (struct edit_distance_traits<const string_fragment &>): New
33106 (get_closest_sanitizer_option): New function.
33107 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33109 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33111 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33113 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33114 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33115 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33116 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33117 for initial die_offset if dwarf_split_debug_info.
33118 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33119 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33121 (output_skeleton_debug_sections): Formatting fix. Use
33122 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33123 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33125 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33127 * config/arm/cortex-a53.md: Add bypasses for
33128 cortex_a53_r2f_cvt.
33129 (cortex_a53_r2f): Only use for transfers.
33130 (cortex_a53_f2r): Likewise.
33131 (cortex_a53_r2f_cvt): Add reservation for conversions.
33132 (cortex_a53_f2r_cvt): Likewise.
33134 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33136 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33137 to all inlined functions, change static to extern.
33139 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33142 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33144 (arm_assemble_integer): Likewise.
33146 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33148 * config.gcc: Use new awk script to check CPU, FPU and architecture
33149 parameters for --with-... options.
33150 * config/arm/parsecpu.awk: New file
33151 * config/arm/arm-cpus.in: New file.
33152 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33154 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33156 * config/arm/t-arm: Update dependency rules.
33157 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33158 of processing .def files.
33159 * config/arm/genopt.sh: Deleted.
33160 * config/arm/gentune.sh: Deleted.
33161 * config/arm/arm-cores.def: Deleted.
33162 * config/arm/arm-arches.def: Deleted.
33163 * config/arm/arm-fpus.def: Deleted.
33164 * config/arm/arm-tune.md: Regenerated.
33165 * config/arm/arm-tables.opt: Regenerated.
33166 * config/arm/arm-cpu.h: New generated file.
33167 * config/arm/arm-cpu-data.h: New generated file.
33168 * config/arm/arm-cpu-cdata.h: New generated file.
33170 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33173 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33175 (input_varpool_node): Unpack dynamically_initialized bit.
33177 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33179 PR rtl-optimization/79032
33180 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33181 the alignment of the adjusted memory reference against that of MODE,
33182 instead of the alignment of the original memory reference.
33184 2017-01-11 Martin Jambor <mjambor@suse.cz>
33186 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33188 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33189 decorated functions.
33191 2017-01-11 Richard Biener <rguenther@suse.de>
33193 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33194 set range/nonnull info for PHI results. Do not set it on
33195 stmts marked for removal.
33197 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33199 * expr.c (store_field): In the bitfield case, fetch the return value
33200 from the registers before applying a single big-endian adjustment.
33201 Always do a final load for a BLKmode value not larger than a word.
33203 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33206 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33207 that we correctly handle column numbers greater than
33208 LINE_MAP_MAX_COLUMN_NUMBER.
33210 2017-01-10 Martin Sebor <msebor@redhat.com>
33212 PR middle-end/78245
33213 * gimple-ssa-sprintf.c (get_destination_size): Call
33214 {init,fini}object_sizes.
33215 * tree-object-size.c (addr_object_size): Adjust.
33216 (pass_through_call): Adjust.
33217 (pass_object_sizes::execute): Adjust.
33218 * tree-object-size.h (fini_object_sizes): Declare.
33220 2017-01-10 Martin Sebor <msebor@redhat.com>
33222 PR tree-optimization/78775
33223 * builtins.c (get_size_range): Move...
33224 * calls.c: ...to here.
33225 (alloc_max_size): Accept zero argument.
33226 (operand_signed_p): Remove.
33227 (maybe_warn_alloc_args_overflow): Call get_size_range.
33228 * calls.h (get_size_range): Declare.
33230 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33232 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33233 from TI's devices.csv file as of September 2016.
33234 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33236 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33238 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33239 * doc/invoke.texi: Likewise.
33240 * doc/md.texi: Likewise.
33241 * doc/objc.texi: Likewise.
33243 2017-01-10 Joshua Conner <joshconner@google.com>
33245 * config/arm/fuchsia-elf.h: New file.
33246 * config/fuchsia.h: New file.
33247 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33248 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33250 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33252 2016-01-10 Richard Biener <rguenther@suse.de>
33254 PR tree-optimization/79034
33255 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33256 Propagate out degenerate PHIs in the joiner.
33258 2017-01-10 Martin Liska <mliska@suse.cz>
33260 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33261 (sort_congruence_classes_by_decl_uid): Likewise.
33262 (sort_congruence_class_groups_by_decl_uid): Likewise.
33263 (sem_item_optimizer::merge_classes): Sort class, groups in these
33264 classes and members in the groups by DECL_UID of declarations.
33265 This would make merge operations stable.
33267 2017-01-10 Martin Liska <mliska@suse.cz>
33269 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33270 usage of m_classes_vec.
33271 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33272 (sem_item_optimizer::get_group_by_hash): Likewise.
33273 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33274 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33275 (sem_item_optimizer::verify_classes): Likewise.
33276 (sem_item_optimizer::process_cong_reduction): Likewise.
33277 (sem_item_optimizer::dump_cong_classes): Likewise.
33278 (sem_item_optimizer::merge_classes): Likewise.
33279 * ipa-icf.h (congruence_class_hash): Rename from
33280 congruence_class_group_hash. Remove declaration of m_classes_vec.
33282 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33284 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33285 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33286 * config.gcc: Add avx512vpopcntdqintrin.h.
33287 * config/i386/avx512vpopcntdqintrin.h: New.
33288 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33289 * config/i386/i386-builtin-types.def: Add new types.
33290 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33291 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33292 __builtin_ia32_vpopcountq_v8di_mask): New.
33293 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33294 __AVX512VPOPCNTDQ__.
33295 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33296 (PTA_AVX512VPOPCNTDQ): Define.
33297 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33298 TARGET_AVX512VPOPCNTDQ_P): Define.
33299 * config/i386/i386.opt: Add mavx512vpopcntdq.
33300 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33301 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33303 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33305 PR middle-end/77484
33306 * predict.def (PRED_CALL): Set to 67.
33308 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33310 * expr.c (store_field): In the bitfield case, if the value comes from
33311 a function call and is of an aggregate type returned in registers, do
33312 not modify the field mode; extract the value in all cases if the mode
33313 is BLKmode and the size is not larger than a word.
33315 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33318 * config/i386/cpuid.h: Fix undefined behavior.
33320 2017-01-04 Jeff Law <law@redhat.com>
33322 PR tree-optimization/79007
33323 PR tree-optimization/67955
33324 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33325 conservative for pt.null when flag_non_call_exceptions is on.
33327 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33329 PR translation/79019
33330 PR translation/79020
33331 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33332 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33333 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33335 * config/avr/avr.opt (maccumulate-args): Likewise.
33336 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33337 * common.opt (freport-bug): Likewise.
33338 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33339 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33340 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33341 translatable string.
33342 * config/i386/i386.c (function_value_32): Likewise.
33343 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33344 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33346 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33347 * common/config/msp430/msp430-common.c (msp430_handle_option):
33349 * symtab.c (symtab_node::verify_base): Likewise.
33350 * opts.c (set_debug_level): Likewise.
33351 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33352 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33353 missing whitespace to translatable strings.
33354 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33355 * config/sh/superh.h: Likewise.
33356 * config/i386/xopintrin.h: Likewise.
33357 * config/i386/znver1.md: Likewise.
33358 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33359 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33360 * double-int.h (struct double_int): Likewise.
33361 * double-int.c (div_and_round_double): Likewise.
33362 * wide-int.cc: Likewise.
33363 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33364 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33365 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33366 (crossjumps_occurred): ... this.
33367 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33370 PR tree-optimization/78899
33371 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33372 returning bool return struct loop *, NULL for failure and the new
33374 (versionable_outer_loop_p): Don't version outer loop if it has
33375 dont_vectorized bit set.
33376 (tree_if_conversion): When versioning outer loop, ensure
33377 tree_if_conversion is performed also on the inner loop of the
33378 non-vectorizable outer loop copy.
33379 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33380 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33381 prevent vectorization of it.
33382 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33383 the outer loop vectorization of the non-scalar version is attempted
33384 before vectorization of the inner loop in scalar version. If
33385 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33386 vectorization of its inner loop.
33387 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33388 has 2 inner loops, rename also on edges from bb whose single pred
33389 is outer_loop->header. Fix typo in function comment.
33391 2017-01-09 Martin Sebor <msebor@redhat.com>
33394 * asan.c (asan_emit_stack_protection): Increase local buffer size
33395 to avoid snprintf truncation warning.
33397 2017-01-09 Andrew Pinski <apinski@cavium.com>
33399 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33400 to reference thunderx2t99 for the tuning structure
33401 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33403 (thunderx2t99_extra_costs): This.
33404 * config/aarch64/aarch64-tune.md: Regenerate.
33405 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33406 (vulcan_addrcost_table): This.
33407 (vulcan_regmove_cost): Rename to ...
33408 (thunderx2t99_regmove_cost): This.
33409 (vulcan_vector_cost): Rename to ...
33410 (thunderx2t99_vector_cost): this.
33411 (vulcan_branch_cost): Rename to ...
33412 (thunderx2t99_branch_cost): This.
33413 (vulcan_tunings): Rename to ...
33414 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33415 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33417 2017-01-09 Martin Jambor <mjambor@suse.cz>
33421 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33422 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33423 (propagate_vr_accross_jump_function): Use the above function for all
33424 value range computations for pass-through jump functions and type
33425 converasion from explicit value range values.
33426 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33427 parameters from TYPE_ARG_TYPES.
33428 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33429 (ipa_write_node_info): Stream type of the actual argument.
33430 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33432 2017-01-09 Martin Liska <mliska@suse.cz>
33435 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33436 (lookup_compiler): Do not show error message with have_E.
33438 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33440 PR tree-optimization/78938
33441 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33442 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33443 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33444 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33447 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33449 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33452 2017-01-09 Richard Biener <rguenther@suse.de>
33454 PR tree-optimization/78997
33455 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33456 name condition properly.
33458 2017-01-09 Richard Biener <rguenther@suse.de>
33461 * dwarf2out.c (is_cxx): New overload with context.
33462 (is_naming_typedef_decl): Use it.
33464 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33466 * invoke.texi (Option Summary): Correct spacing in option lists
33467 and add line breaks to fix over-long lines.
33469 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33471 PR middle-end/17660
33473 * extend.texi (Common Variable Attributes): Add xref to GCC
33474 Internals manual to explain mode attribute keywords.
33476 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33479 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33480 and Preprocessor Options.
33481 (Options for Linking): Document -pthread here....
33482 (RS/6000 and PowerPC Options): ...not here.
33483 (Solaris 2 Options): ...or here.
33484 * doc/cppopts.texi: Document -pthread.
33486 2017-01-08 Martin Sebor <msebor@redhat.com>
33488 PR middle-end/77708
33489 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33490 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33491 New member functions.
33492 (format_directive): Used them.
33494 (pass_sprintf_length::handle_gimple_call): Same.
33495 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33496 to avoid truncation for any argument.
33497 (extract_affine_mul): Same.
33498 * tree.c (get_file_function_name): Same.
33500 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33502 PR middle-end/77484
33503 * predict.def (PRED_INDIR_CALL): Set to 86.
33505 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33507 PR preprocessor/54124
33508 * doc/cppopts.texi: Reformat -d subtable to list the full name
33509 of the options. Add cross-reference to the docs for the general
33510 compiler -d options.
33511 * doc/invoke.texi (Developer Options): Add cross-reference to the
33512 preprocessor-specific -d option documentation.
33514 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33516 PR preprocessor/13498
33517 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33518 redudant material, and reflect new command-line options.
33519 (System Headers): Likewise.
33521 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33523 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33524 -isystem, and -idirafter. Copy-edit.
33525 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33526 default for -ftrack-macro-expansion. Delete obsolete and
33527 badly-formatted implementation details about -fdebug-cpp output.
33528 * doc/cppwarnopts.texi: Copy-edit.
33530 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33533 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33534 that the transition from a max line width >= 1<<10 to narrower
33535 lines works correctly.
33537 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33539 * doc/options.texi (PerFunction): New.
33540 * opt-functions.awk (switch_flags): Map both Optimization and
33541 PerFunction to CL_OPTIMIZATION.
33542 * opth-gen.awk: Test for PerFunction flag along with
33544 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33545 it only when the latter is present. Skip those that don't in
33546 the hash function generator.
33547 * common.opt (fvar-tracking): Mark as PerFunction instead of
33549 (fvar-tracking-assignments): Likewise.
33550 (fvar-tracking-assignments-toggle): Likewise.
33551 (fvar-tracking-uninit): Likewise.
33553 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33555 PR translation/79018
33556 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33559 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33562 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33563 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33564 TARGET_LONG_JUMP_TABLE_OFFSETS.
33565 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33566 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33567 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33568 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33569 * config/m68k/m68k.md (tablejump expander): Likewise.
33570 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33571 TARGET_LONG_JUMP_TABLE_OFFSETS.
33572 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33573 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33575 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33576 David Holsgrove <david.holsgrove@xilinx.com>
33578 * common/config/microblaze/microblaze-common.c
33579 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33580 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33582 * config/microblaze/microblaze.c (microblaze_must_save_register)
33583 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33585 (microblaze_eh_return): New function.
33586 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33587 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33588 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33589 * config/microblaze/microblaze.md (eh_return): New pattern.
33591 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33593 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33594 GCC_DIAGNOSTIC_STRINGIFY): Define.
33596 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33598 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33600 * config/arm/arm.md (<mcrr>): New.
33602 * config/arm/arm.c (arm_arch5te): New.
33603 (arm_option_override): Set arm_arch5te.
33604 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33606 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33607 (arm_mcrr_qualifiers): ... this. New.
33608 (MRRC_QUALIFIERS): Define to...
33609 (arm_mrrc_qualifiers): ... this. New.
33610 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33612 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33613 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33614 (MRRCI, mrrc, MRRC): New.
33615 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33616 VUNSPEC_MRRC2): New.
33618 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33620 * config/arm/arm.md (<mcr>): New.
33622 * config/arm/arm.c (arm_coproc_builtin_available): Add
33623 support for mcr, mrc, mcr2 and mrc2.
33624 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33625 (arm_mcr_qualifiers): ... this. New.
33626 (MRC_QUALIFIERS): Define to ...
33627 (arm_mrc_qualifiers): ... this. New.
33628 (MCR_QUALIFIERS): Define to ...
33629 (arm_mcr_qualifiers): ... this. New.
33630 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33632 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33633 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33634 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33635 VUNSPEC_MRC2): New.
33637 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33639 * config/arm/arm.md (*ldc): New.
33643 * config/arm/arm.c (arm_coproc_builtin_available): Add
33644 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33645 (arm_coproc_ldc_stc_legitimate_address): New.
33646 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33647 'qualifier_const_pointer'.
33648 (LDC_QUALIFIERS): Define to...
33649 (arm_ldc_qualifiers): ... this. New.
33650 (STC_QUALIFIERS): Define to...
33651 (arm_stc_qualifiers): ... this. New.
33652 * config/arm/arm-protos.h
33653 (arm_coproc_ldc_stc_legitimate_address): New.
33654 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33655 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33656 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33657 stc2, stcl, stc2l): New.
33658 * config/arm/constraints.md (Uz): New.
33659 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33660 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33661 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33662 VUNSPEC_STC2L): New.
33664 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33666 * config/arm/arm.md (<cdp>): New.
33667 * config/arm/arm.c (neon_const_bounds): Rename this ...
33668 (arm_const_bounds): ... this.
33669 (arm_coproc_builtin_available): New.
33670 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33671 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33672 (CDP_QUALIFIERS): Define to...
33673 (arm_cdp_qualifiers): ... this. New.
33675 (arm_expand_builtin_args): Add case for 6 arguments.
33676 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33677 (arm_const_bounds): ... this.
33678 (arm_coproc_builtin_available): New.
33679 * config/arm/arm_acle.h (__arm_cdp): New.
33681 * config/arm/arm_acle_builtins.def (cdp): New.
33683 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33684 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33685 'arm_const_bounds'.
33686 * config/arm/types.md (coproc): New.
33687 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33688 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33689 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33690 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33692 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33694 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33695 (UBINOP_QUALIFIERS): New.
33697 (acle_builtin_data): New. Change comment.
33698 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33699 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33700 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33701 arm_acle_builtins.def.
33702 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33703 (arm_init_acle_builtins): New.
33704 (CRC32_BUILTIN): Remove.
33705 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33706 crc32cb, crc32ch and crc32cw.
33707 (arm_init_crc32_builtins): Remove.
33708 (arm_init_builtins): Use arm_init_acle_builtins rather
33709 than arm_init_crc32_builtins.
33710 (arm_expand_acle_builtin): New.
33711 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33712 * config/arm/arm_acle_builtins.def: New.
33714 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33716 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33717 (arm_builtin_datum): ... this.
33718 (arm_init_neon_builtin): Rename to ...
33719 (arm_init_builtin): ... this. Add a new parameters PREFIX
33720 and USE_SIG_IN_NAME.
33721 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33722 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33723 'arm_builtin_datum'.
33724 (arm_init_vfp_builtins): Likewise.
33725 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33726 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33727 (arm_expand_neon_args): Rename to ...
33728 (arm_expand_builtin_args): ... this. Rename builtin_arg
33729 enum values and differentiate between ARG_BUILTIN_MEMORY
33730 and ARG_BUILTIN_NEON_MEMORY.
33731 (arm_expand_neon_builtin_1): Rename to ...
33732 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33733 values, arm_expand_builtin_args and add bool parameter NEON.
33734 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33735 (arm_expand_vfp_builtin): Likewise.
33736 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33738 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33740 PR middle-end/77484
33741 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33742 * predict.c (tree_estimate_probability_bb): Reverse direction of
33743 polymorphic call predictor.
33745 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33747 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33748 (determine_pass_name_match): ...this new function and...
33749 (should_skip_pass_p): ...this new function.
33751 2017-01-06 Nathan Sidwell <nathan@acm.org>
33753 * ipa-visibility.c (function_and_variable_visibility): Reformat
33754 comments and long lines. Remove extrneous if.
33755 * symtab.c (symtab_node::make_decl_local): Fix code format.
33756 (symtab_node::set_section_for_node): Fix comment typo.
33758 2017-01-06 Martin Liska <mliska@suse.cz>
33761 * lra-constraints.c: Rename invariant to lra_invariant.
33762 * predict.c (set_even_probabilities): Initialize e to NULL.
33764 2017-01-05 Martin Sebor <msebor@redhat.com>
33766 PR tree-optimization/78910
33767 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33768 (format_integer): Correct off-by-one error in the handling
33769 of precision with negative numbers in signed conversions..
33771 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33773 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33775 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33777 PR tree-optimization/71016
33778 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33779 factor_out_conditional_conversion. Formatting fix.
33780 (factor_out_conditional_conversion): Add cond_stmt argument.
33781 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33782 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33785 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33787 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33788 read-rtl-function.o, and selftest-rtl.o.
33789 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33790 (selftest::aarch64_test_loading_full_dump): New function.
33791 (selftest::aarch64_run_selftests): New function.
33792 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33793 selftest::aarch64_run_selftests.
33794 * config/i386/i386.c
33795 (selftest::ix86_test_loading_dump_fragment_1): New function.
33796 (selftest::ix86_test_loading_call_insn): New function.
33797 (selftest::ix86_test_loading_full_dump): New function.
33798 (selftest::ix86_test_loading_unspec): New function.
33799 (selftest::ix86_run_selftests): Call the new functions.
33800 * emit-rtl.c (maybe_set_max_label_num): New function.
33801 * emit-rtl.h (maybe_set_max_label_num): New decl.
33802 * function.c (instantiate_decls): Guard call to
33803 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33804 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33806 * gensupport.c (gen_reader::gen_reader): Pass "false"
33807 for new "compact" param of rtx_reader.
33808 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33809 rather than an empty string for NULL strings.
33810 * read-md.c: Potentially include config.h rather than bconfig.h.
33811 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33812 (have_error): New global, copied from errors.c.
33813 (md_reader::read_name): Rename to...
33814 (md_reader::read_name_1): ...this, adding "out_loc" param,
33815 and converting "missing name or number" to returning false, rather
33817 (md_reader::read_name): Reimplement in terms of read_name_1.
33818 (md_reader::read_name_or_nil): New function.
33819 (md_reader::read_string): Handle "(nil)" by returning NULL.
33820 (md_reader::md_reader): Add new param "compact".
33821 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33822 (md_reader::read_file): New method.
33823 * read-md.h (md_reader::md_reader): Add new param "compact".
33824 (md_reader::read_file): New method.
33825 (md_reader::is_compact): New accessor.
33826 (md_reader::read_name): Convert return type from void to file_location.
33827 (md_reader::read_name_or_nil): New decl.
33828 (md_reader::read_name_1): New decl.
33829 (md_reader::m_compact): New field.
33830 (noop_reader::noop_reader): Pass "false" for new "compact" param
33832 (rtx_reader::rtx_reader): Add new "compact" param.
33833 (rtx_reader::read_rtx_operand): Make virtual and convert return
33834 type from void to rtx.
33835 (rtx_reader::read_until): New decl.
33836 (rtx_reader::handle_any_trailing_information): New virtual function.
33837 (rtx_reader::postprocess): New virtual function.
33838 (rtx_reader::finalize_string): New virtual function.
33839 (rtx_reader::m_in_call_function_usage): New field.
33840 (rtx_reader::m_reuse_rtx_by_id): New field.
33841 * read-rtl-function.c: New file.
33842 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33843 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33844 (selftest::verify_three_block_rtl_cfg): New decl.
33845 * read-rtl-function.h: New file.
33846 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33847 For host, include function.h, memmodel.h, and emit-rtl.h.
33848 (one_time_initialization): New function.
33849 (struct compact_insn_name): New struct.
33850 (compact_insn_names): New array.
33851 (find_code): Handle insn codes in compact dumps.
33852 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33853 (bind_subst_iter_and_attr): Likewise.
33854 (add_condition_to_string): Likewise.
33855 (add_condition_to_rtx): Likewise.
33856 (apply_attribute_uses): Likewise.
33857 (add_current_iterators): Likewise.
33858 (apply_iterators): Likewise.
33859 (initialize_iterators): Guard usage of apply_subst_iterator with
33860 #ifdef GENERATOR_FILE.
33861 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33862 (md_reader::read_mapping): Likewise.
33863 (add_define_attr_for_define_subst): Likewise.
33864 (add_define_subst_attr): Likewise.
33865 (read_subst_mapping): Likewise.
33866 (check_code_iterator): Likewise.
33867 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33869 (one_time_initialization): New function.
33870 (rtx_reader::read_until): New method.
33871 (read_flags): New function.
33872 (parse_reg_note_name): New function.
33873 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33874 Handle reuse_rtx ids.
33875 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33876 Add parsing support for RTL dumps, mirroring the special-cases in
33877 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33878 values, and calling handle_any_trailing_information.
33879 (rtx_reader::read_rtx_operand): Convert return type from void
33880 to rtx, returning return_rtx. Handle case 'e'. Call
33881 finalize_string on XSTR and XTMPL fields.
33882 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33883 "(nil)" values were omitted. Call the postprocess vfunc on the
33885 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33886 class ctor. Initialize m_in_call_function_usage. Call
33887 one_time_initialization.
33888 * rtl-tests.c (selftest::test_uncond_jump): Call
33889 set_new_first_and_last_insn.
33890 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33891 * selftest-rtl.c: New file.
33892 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33893 (selftest::get_insn_by_uid): New decl.
33894 * selftest-run-tests.c (selftest::run_tests): Call
33895 read_rtl_function_c_tests.
33896 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33897 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33900 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33902 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33903 operands in a special way. Assert that pos+len <= mode precision.
33905 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33907 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33908 3 argument Alias with unlimited for the negative form.
33909 (fno-vect-cost-model): Removed.
33911 2017-01-05 Martin Liska <mliska@suse.cz>
33913 * hsa-gen.c (gen_hsa_divmod): New function.
33914 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33916 2017-01-05 Martin Liska <mliska@suse.cz>
33919 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33922 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33924 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33925 small constant length operands.
33927 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33929 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33930 between loop iterations.
33932 2017-01-05 Martin Liska <mliska@suse.cz>
33935 * gimplify.c (gimplify_decl_expr): Compare to
33936 asan_poisoned_variables instread of checking flags.
33937 (gimplify_target_expr): Likewise.
33938 (gimplify_expr): Likewise.
33939 (gimplify_function_tree): Conditionally initialize
33940 asan_poisoned_variables.
33942 2017-01-04 Jeff Law <law@redhat.com>
33944 PR tree-optimizatin/78812
33945 * rtl.h (contains_mem_rtx_p): Prototype.
33946 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33947 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33948 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33949 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33950 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33952 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33954 * input.c (assert_char_at_range): Default-initialize actual_range.
33956 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33958 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33959 to match the caller.
33961 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33963 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33964 insns after final jump in test to emit dummy move.
33966 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33968 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33969 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33971 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33973 * multiple_target.c (create_dispatcher_calls): Init e_next.
33974 * tree-ssa-loop-split.c (split_loop): Init border.
33975 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33978 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33983 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33984 (altivec_register_operand): Do not return true if the operand
33985 contains a SUBREG mixing SImode and SFmode.
33986 (vsx_register_operand): Likewise.
33987 (vsx_reg_sfsubreg_ok): New predicate.
33988 (vfloat_operand): Do not return true if the operand contains a
33989 SUBREG mixing SImode and SFmode.
33990 (vint_operand): Likewise.
33991 (vlogical_operand): Likewise.
33992 (gpc_reg_operand): Likewise.
33993 (int_reg_operand): Likewise.
33994 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33995 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33996 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33998 (rs6000_emit_move_si_sf_subreg): New helper function.
33999 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34000 fixup SUBREGs involving SImode and SFmode.
34001 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34002 numbers for the new peephole2 optimization.
34003 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34004 the GLIBC math library that do AND/IOR/XOR operations on single
34005 precision floating point.
34006 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34007 target macros to say whether we need to avoid SUBREGs mixing
34009 (TARGET_ALLOW_SF_SUBREG): Likewise.
34010 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34011 (UNSPEC_SI_FROM_SF): Likewise.
34012 (iorxor): Change spacing.
34013 (and_ior_xor): New iterator for AND, IOR, and XOR.
34014 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34015 (movdi_from_sf_zero_ext): Likewise.
34016 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34017 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34018 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34019 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34020 (fms<mode>4): Likewise.
34021 (fnma<mode>4): Likewise.
34022 (fnms<mode>4): Likewise.
34023 (nfma<mode>4): Likewise.
34024 (nfms<mode>4): Likewise.
34026 2017-01-04 Marek Polacek <polacek@redhat.com>
34029 * doc/invoke.texi: Document -Wpointer-compare.
34031 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34033 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34036 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34037 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34040 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34043 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34044 documentation of the powerpc_popcntb_ok attribute.
34045 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34046 code to issue warning messages if a requested CPU configuration is
34047 not supported by the binary (assembler and loader) toolchain.
34048 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34049 made to define a built-in function that has been disabled.
34050 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34051 made to define a built-in function that has been disabled.
34052 (altivec_init_builtins): Add comment explaining why definition
34053 of the DST built-in functions is not preceded by an assertion
34054 check. Add assertions to prevent ICE if attempts are made to
34055 define an altivec predicate or an abs* built-in function that has
34057 (htm_init_builtins): Add comment explaining why definition of the
34058 htm built-in functions is not preceded by an assertion check.
34060 2017-01-04 Jeff Law <law@redhat.com>
34062 PR tree-optimizatin/67955
34063 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34064 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34065 the points-to solution does not include pt_null. Use DECL_PT_UID
34068 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34070 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34071 Use gen_int_mode instead of gen_lopwart for const_int operands.
34073 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34075 PR tree-optimization/71563
34076 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34077 out of range value - has low bits known to be zero.
34079 2017-01-04 Alan Modra <amodra@gmail.com>
34081 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34082 * configure: Regenerate.
34083 * config.in: Regenerate.
34085 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34088 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34089 a substring of the message, but strcmp with the whole message. Ifdef
34090 ENABLE_NLS, translate the message first using dgettext.
34092 2017-01-03 Jeff Law <law@redhat.com>
34094 PR tree-optimizatin/78856
34095 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34096 (mark_threaded_blocks): Remove code to truncate thread paths that
34097 cross multiple loop headers. Instead invalidate the cached loop
34098 iteration information and handle case of a thread path walking
34099 into an irreducible region.
34101 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34104 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34105 assertions. Add support for doing the signbit if the IEEE 128-bit
34106 floating point value is in a GPR.
34107 * config/rs6000/rs6000.md (Fsignbit): Delete.
34108 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34109 Update the length attribute if the value is in a GPR.
34110 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34111 the sign or zero extension instruction, since the value is always 0/1.
34112 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34115 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34116 extracting SImode to a GPR register so that we can generate a
34117 store, limit the vector to be in a traditional Altivec register
34118 for the vextuwrx instruction.
34120 2017-01-03 Ian Lance Taylor <iant@google.com>
34122 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34124 2017-01-03 Martin Sebor <msebor@redhat.com>
34126 PR tree-optimization/78696
34127 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34128 precision. Use MPFR for %f for greater fidelity. Correct handling
34130 (pass_sprintf_length::compute_format_length): Set width and precision
34131 specified by asrerisk to void_node for vararg functions.
34132 (try_substitute_return_value): Adjust dump output.
34134 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34136 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34138 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34140 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34141 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34142 -mlra/-mno-lra was passed to the compiler.
34144 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34146 PR rtl-optimization/65618
34147 * emit-rtl.c (try_split): Move initialization of "before" and
34148 "after" to just before the call to emit_insn_after_setloc.
34150 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34152 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34154 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34156 * dwarf2out.c (gen_enumeration_type_die): When
34157 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34159 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34161 PR tree-optimization/78965
34162 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34163 Change first argument from const call_info & to call_info &. For %n
34164 set info.nowrite to false.
34166 PR middle-end/78901
34167 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34168 possibly throwing calls.
34170 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34171 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34172 and fns handling, rather than in a separate case SSA_NAME.
34174 2017-01-02 Jeff Law <law@redhat.com>
34176 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34177 fixes for first_period and second_period variables.
34179 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34182 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34183 (*insvqi_1): New insn pattern.
34184 (*insvqi_1_mem_rex64): Ditto.
34185 (*insvqi_2): Ditto.
34186 (*insvqi_3): Rename from *insvqi.
34188 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34190 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34192 * doc/cfg.texi (Edges): Remove reference to Java.
34193 (Maintaining the CFG): Ditto.
34195 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34197 PR middle-end/77674
34198 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34199 transparent aliases.
34201 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34203 PR middle-end/77484
34204 * predict.def (PRED_CALL): Update hitrate.
34205 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34206 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34207 into direct/indirect/polymorphic variants.
34209 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34211 Update copyright years.
34213 * gcc.c (process_command): Update copyright notice dates.
34214 * gcov-dump.c (print_version): Ditto.
34215 * gcov.c (print_version): Ditto.
34216 * gcov-tool.c (print_version): Ditto.
34217 * gengtype.c (create_file): Ditto.
34218 * doc/cpp.texi: Bump @copying's copyright year.
34219 * doc/cppinternals.texi: Ditto.
34220 * doc/gcc.texi: Ditto.
34221 * doc/gccint.texi: Ditto.
34222 * doc/gcov.texi: Ditto.
34223 * doc/install.texi: Ditto.
34224 * doc/invoke.texi: Ditto.
34226 Copyright (C) 2017 Free Software Foundation, Inc.
34228 Copying and distribution of this file, with or without modification,
34229 are permitted in any medium without royalty provided the copyright
34230 notice and this notice are preserved.