1 2017-11-09 Jeff Law <law@redhat.com>
3 * tree-vrp.c (set_value_range): Do not reference vrp_equiv_obstack.
4 Get it from the existing bitmap instead.
5 (vrp_intersect_ranges_1): Likewise.
7 2017-11-09 Jakub Jelinek <jakub@redhat.com>
9 * gimple-ssa-store-merging.c (struct store_immediate_info): Add
11 (store_immediate_info::store_immediate_info): Add bitnotp argument,
13 (imm_store_chain_info::coalesce_immediate_stores): Break group
14 if bit_not_p is different.
15 (count_multiple_uses, split_group,
16 imm_store_chain_info::output_merged_store): Handle info->bit_not_p.
17 (handled_load): Avoid multiple chained BIT_NOT_EXPRs.
18 (pass_store_merging::process_store): Handle BIT_{AND,IOR,XOR}_EXPR
19 result inverted using BIT_NOT_EXPR, compute bit_not_p, pass it
20 to store_immediate_info ctor.
22 2017-11-09 Jim Wilson <jimw@sifive.com>
24 * collect2.c (OBJECT_FORMAT_COFF): Remove EXTENDED_COFF support.
25 (scan_prog_file): Likewise.
27 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
29 * bb-reorder.c (max_entry_frequency): Remove.
30 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
31 connect_traces, push_to_next_round_p): Remove prototypes.
32 (find_traces_1_round): Use counts only.
33 (push_to_next_round_p): Likewise.
34 (find_traces): Likewise.
35 (rotate_loop): Likewise.
36 (find_traces_1_round): Likewise.
37 (connect_traces): Likewise.
38 (edge_order): Likewise.
40 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
42 * config/arm/arm.c (output_return_instruction): Add comments to
43 indicate requirement for cmse_nonsecure_entry return to account
44 for the size of clearing instruction output here.
45 (thumb_exit): Likewise.
46 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
47 return in hardfloat mode.
49 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
51 * config/rs6000/rs6000.c (machine_function): Add a bool,
52 "toc_is_wrapped_separately".
53 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
54 if it wasn't explicitly set or unset, we are optimizing for speed, and
55 doing separate shrink-wrapping.
56 (rs6000_get_separate_components): Enable the TOC component if
57 saving the TOC register in the prologue.
58 (rs6000_components_for_bb): Handle the TOC component.
59 (rs6000_emit_prologue_components): Store the TOC register where needed.
60 (rs6000_set_handled_components): Mark TOC as handled, if handled.
61 (rs6000_emit_prologue): Don't save the TOC if that is already done.
63 2017-11-09 Martin Jambor <mjambor@suse.cz>
65 * ipa-param-manipulation.c: New file.
66 * ipa-param-manipulation.h: Likewise.
67 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
68 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
69 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
70 (ipa_parm_adjustment): Likewise.
71 (ipa_parm_adjustment_vec): Likewise.
72 (ipa_get_vector_of_formal_parms): Moved declaration to
73 ipa-param-manipulation.h.
74 (ipa_get_vector_of_formal_parm_types): Likewise.
75 (ipa_modify_formal_parameters): Likewise.
76 (ipa_modify_call_arguments): Likewise.
77 (ipa_combine_adjustments): Likewise.
78 (ipa_dump_param_adjustments): Likewise.
79 (ipa_modify_expr): Likewise.
80 (ipa_get_adjustment_candidate): Likewise.
81 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
82 ipa-param-manipulation.c.
83 (ipa_get_vector_of_formal_parm_types): Likewise.
84 (ipa_modify_formal_parameters): Likewise.
85 (ipa_modify_call_arguments): Likewise.
86 (ipa_modify_expr): Likewise.
87 (get_ssa_base_param): Likewise.
88 (ipa_get_adjustment_candidate): Likewise.
89 (index_in_adjustments_multiple_times_p): Likewise.
90 (ipa_combine_adjustments): Likewise.
91 (ipa_dump_param_adjustments): Likewise.
92 * tree-sra.c: Also include ipa-param-manipulation.h
93 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
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_masked_store): 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_align_stack_vars): 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_variable_length): 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_unaligned_possible): 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_element_align_preferred): Document.
126 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
127 Alan Hayward <alan.hayward@arm.com>
128 David Sherwood <david.sherwood@arm.com>
130 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
131 previously undocumented selectors.
132 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
134 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
136 * doc/rtl.texi (const_vector): Say that elements can be
138 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
139 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
140 (gen_vec_duplicate): Use it instead of CONSTANT_P.
141 * optabs.c (expand_vector_broadcast): Likewise.
143 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
144 Alan Hayward <alan.hayward@arm.com>
145 David Sherwood <david.sherwood@arm.com>
147 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
148 scaled index even if the unscaled address was invalid.
149 Don't increase the complexity of using a scale in that case.
151 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
152 Alan Hayward <alan.hayward@arm.com>
153 David Sherwood <david.sherwood@arm.com>
155 * doc/rtl.texi: Rewrite the subreg rules so that they partition
156 the inner register into REGMODE_NATURAL_SIZE bytes rather than
157 UNITS_PER_WORD bytes.
158 * emit-rtl.c (validate_subreg): Divide subregs into blocks
159 based on REGMODE_NATURAL_SIZE of the inner mode.
160 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
161 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
162 * expmed.c (lowpart_bit_field_p): Divide the value up into
163 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
164 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
165 whether something is likely to occupy more than one register.
167 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
170 * ipa-inline-transform.c (update_noncloned_frequencies): Use
171 profile_count::adjust_for_ipa_scaling.
172 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
173 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
175 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
177 2017-11-09 Jakub Jelinek <jakub@redhat.com>
179 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
180 (split_group): Add total_orig and total_new arguments, estimate the
181 number of statements related to the store group without store merging
182 and with store merging.
183 (imm_store_chain_info::output_merged_store): Adjust split_group
184 callers, punt if estimated number of statements with store merging
185 is not smaller than estimated number of statements without it.
187 (handled_load): Remove has_single_use checks.
188 (pass_store_merging::process_store): Likewise.
190 2017-11-09 Richard Biener <rguenther@suse.de>
192 PR tree-optimization/82902
193 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
195 2017-11-09 Martin Liska <mliska@suse.cz>
198 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
201 2017-11-09 Martin Liska <mliska@suse.cz>
203 PR tree-optimization/82669
204 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
206 2017-11-09 Martin Liska <mliska@suse.cz>
208 PR gcov-profile/48463
209 * coverage.c (coverage_begin_function): Output also end locus
210 of a function and information whether the function is
212 * gcov-dump.c (tag_function): Parse and print the information.
213 * gcov.c (INCLUDE_MAP): Add include.
214 (INCLUDE_SET): Likewise.
215 (struct line_info): Move earlier in the source file because
216 of vector<line_info> in function_info structure.
217 (line_info::line_info): Likewise.
218 (line_info::has_block): Likewise.
219 (struct source_info): Add new member index.
220 (source_info::get_functions_at_location): New function.
221 (function_info::group_line_p): New function.
222 (output_intermediate_line): New function.
223 (output_intermediate_file): Use the mentioned function.
224 (struct function_start): New.
225 (struct function_start_pair_hash): Likewise.
226 (process_file): Add code that identifies group functions.
227 Assign lines either to global or function scope.
228 (generate_results): Skip artificial functions.
229 (find_source): Assign index for each source file.
230 (read_graph_file): Read new flag artificial and end_line.
231 (add_line_counts): Assign it either to global of function scope.
232 (accumulate_line_counts): Isolate core of the function to
233 accumulate_line_info and call it for both function and global
235 (accumulate_line_info): New function.
236 (output_line_beginning): Fix GNU coding style.
237 (print_source_line): New function.
238 (output_line_details): Likewise.
239 (output_function_details): Likewise.
240 (output_lines): Iterate both source (global) scope and function
242 (struct function_line_start_cmp): New class.
243 * doc/gcov.texi: Reflect changes in documentation.
245 2017-11-09 Jakub Jelinek <jakub@redhat.com>
248 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
249 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
250 and similarly for not instead of neg.
252 2017-11-08 Andi Kleen <ak@linux.intel.com>
254 * config/i386/i386.opt: Add -mforce-indirect-call.
255 * config/i386/predicates.md: Check for flag_force_indirect_call.
256 * doc/invoke.texi: Document -mforce-indirect-call
258 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
260 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
262 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
263 config/riscv/riscv.c (predict.h): New include.
264 (riscv_slow_unaligned_access_p): No longer static.
265 (riscv_block_move_straight): Add require.
266 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
268 2017-11-08 Jakub Jelinek <jakub@redhat.com>
271 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
272 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
273 nonimmediate_operand predicate for operand 1 instead of
276 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
278 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
280 * config/aarch64/constraints.md (Uml): New constraint.
281 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
284 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
286 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
287 of two vec_duplicates into a vec_concat.
289 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
291 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
292 Simplify vec_merge of vec_duplicate and vec_concat.
293 * config/aarch64/constraints.md (Utq): New constraint.
294 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
297 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
299 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
300 Simplify vec_merge of vec_duplicate and const_vector.
301 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
303 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
304 mode iterator. Update predicate on operand 1 to
305 handle non-const_vec constants. Delete constraints.
306 (*aarch64_combinez_be<mode>): Likewise for operand 2.
308 2017-11-08 Jakub Jelinek <jakub@redhat.com>
310 PR tree-optimization/78821
311 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
313 (store_operand_info::store_operand_info): Initialize it to false.
314 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
315 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
316 store in the group, and if chain_info is non-NULL, to ignore altogether
318 (compatible_load_p): Fail if bit_not_p does not match.
319 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
320 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
321 (pass_store_merging::process_store): Adjust
322 terminate_all_aliasing_chains calls to pass NULL in all current spots,
323 call terminate_all_aliasing_chains newly when adding a store into
324 a chain with non-NULL chain_info.
326 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
328 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
330 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
332 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
334 (aarch64_layout_frame): Initialise emit_frame_chain.
335 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
336 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
338 2017-11-08 Martin Liska <mliska@suse.cz>
340 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
341 of gimple_call_internal_p.
343 2017-11-07 Tom de Vries <tom@codesourcery.com>
345 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
347 2017-11-07 Tom de Vries <tom@codesourcery.com>
349 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
352 2017-11-08 Martin Liska <mliska@suse.cz>
355 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
357 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
359 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
361 2017-11-07 Jakub Jelinek <jakub@redhat.com>
364 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
365 (*cmp<mode>_ccz_1): New insn with $k alternative.
368 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
369 RTX_COMM_COMPARE as commutative as well.
370 (ix86_binary_operator_ok): Formatting fix.
371 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
372 *<code><mode>3<mask_name><round_saeonly_name>,
373 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
374 *<s>mul<mode>3_highpart<mask_name>,
375 *vec_widen_umult_even_v16si<mask_name>,
376 *vec_widen_umult_even_v8si<mask_name>,
377 *vec_widen_umult_even_v4si<mask_name>,
378 *vec_widen_smult_even_v16si<mask_name>,
379 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
380 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
381 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
382 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
383 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
384 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
385 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
386 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
387 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
388 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
389 ix86_binary_operator_ok. Formatting fixes.
390 (*<plusminus_insn><mode>3<mask_name><round_name>,
391 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
394 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
396 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
397 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
398 eq and ne if TARGET_ISEL.
399 (cmp): New code_iterator.
400 (UNS, UNSU_, UNSIK): New code_attrs.
401 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
402 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
404 ("eq<mode>3"): ... this.
405 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
407 ("ne<mode>3"): ... this.
409 2017-11-07 Julia Koval <julia.koval@intel.com>
412 * common/config/i386/i386-common.c
413 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
414 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
415 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
416 * config/i386/i386.opt: Ditto.
417 * config/i386/i386.c (ix86_target_string): Ditto.
418 (ix86_option_override_internal): Ditto.
419 (ix86_init_mpx_builtins): Move MPX to args2.
420 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
421 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
422 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
423 __builtin_ia32_vgf2p8affineinvqb_v32qi,
424 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
425 __builtin_ia32_vgf2p8affineinvqb_v16qi,
426 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
428 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
431 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
432 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
433 (zero-extendsidi peephole2): Remove peephole.
435 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
438 * stor-layout.c: Include attribs.h.
439 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
440 explicit lookup of "aligned" attribute.
442 2017-11-07 Andrew Waterman <andrew@sifive.com>
444 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
445 (riscv_expand_block_move): Likewise.
446 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
448 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
449 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
450 * config/riscv/riscv.c (riscv_block_move_straight): New function.
451 (riscv_adjust_block_mem): Likewise.
452 (riscv_block_move_loop): Likewise.
453 (riscv_expand_block_move): Likewise.
454 gcc/config/riscv/riscv.md (movmemsi): New pattern.
456 2017-11-07 Michael Clark <michaeljclark@mac.com>
458 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
459 (MUSL_DYNAMIC_LINKER): Likewise.
461 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
463 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
464 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
467 2017-11-07 Richard Biener <rguenther@suse.de>
469 * match.pd: Fix build.
471 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
472 Jackson Woodruff <jackson.woodruff@arm.com>
474 PR tree-optimization/71026
475 * match.pd: Canonicalize negate in division.
477 2017-11-07 Sudakshina Das <sudi.das@arm.com>
480 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
482 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
484 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
485 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
487 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
489 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
490 non-scalar integral types.
491 * match.pd (negate_expr_p): Handle MINUS_EXPR.
492 (-(A-B), -(~A)): New transformations.
494 2017-11-07 Tom de Vries <tom@codesourcery.com>
496 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
497 semicolon after "do {} while (0)".
498 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
499 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
500 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
501 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
502 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
504 2017-11-07 Tom de Vries <tom@codesourcery.com>
506 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
507 after "do {} while (0)".
508 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
509 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
510 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
511 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
512 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
514 2017-11-07 Tom de Vries <tom@codesourcery.com>
517 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
519 (arm_rtx_costs_internal): Add missing semicolon after
520 HANDLE_NARROW_SHIFT_ARITH call.
522 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
524 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
525 disable isel if it was not set explicitly.
527 2017-11-06 James Bowman <james.bowman@ftdichip.com>
529 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
531 (add_type_attribute) likewise.
533 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
535 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
536 of struct ix86_frame.
537 (ix86_initial_elimination_offset): Likewise.
538 (ix86_expand_split_stack_prologue): Likewise.
540 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
542 * tree-vrp.h (enum value_range_type): Update stale comment.
544 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
545 Alan Hayward <alan.hayward@arm.com>
546 David Sherwood <david.sherwood@arm.com>
548 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
549 (aarch64_expand_vec_perm_const): Take the number of units too.
550 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
551 (aarch64_expand_vec_perm_const): Likewise.
552 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
553 (vec_perm<mode>): Update accordingly.
555 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
556 Alan Hayward <alan.hayward@arm.com>
557 David Sherwood <david.sherwood@arm.com>
559 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
560 Take the number of units too.
561 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
562 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
563 but check for a vector mode before rather than after the call.
564 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
565 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
566 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
567 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
568 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
569 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
570 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
571 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
572 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
573 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
574 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
575 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
576 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
577 (aarch64_sqdmull2_n<mode>): Update accordingly.
579 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
580 Alan Hayward <alan.hayward@arm.com>
581 David Sherwood <david.sherwood@arm.com>
583 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
584 the number of units too.
585 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
586 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
587 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
588 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
589 (vec_store_lanesxi<mode>): Update accordingly.
591 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
592 Alan Hayward <alan.hayward@arm.com>
593 David Sherwood <david.sherwood@arm.com>
595 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
596 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
597 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
598 of units rather than the mode.
599 * config/aarch64/iterators.md (nunits): New mode attribute.
600 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
601 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
602 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
603 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
604 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
605 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
606 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
607 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
608 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
609 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
610 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
611 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
612 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
613 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
614 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
615 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
616 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
617 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
618 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
619 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
620 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
621 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
622 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
623 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
624 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
625 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
626 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
627 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
628 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
629 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
630 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
631 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
632 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
633 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
634 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
635 (aarch64_simd_vec_setv2di): Likewise.
637 2017-11-06 Carl Love <cel@us.ibm.com>
639 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
641 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
642 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
644 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
646 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
647 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
648 Add power 8 macro expansions.
649 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
650 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
651 power 8 instructions. (VSX_XXBR): Add iterator.
653 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
655 * config/arm/arm.md (predicable_short_it): Change default to "no",
656 improve documentation, remove uses that are identical to the default.
657 (enabled_for_depr_it): Rename to enabled_for_short_it.
658 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
659 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
660 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
661 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
662 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
664 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
667 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
668 float128 helper macros, which are no longer used after deleting
669 the old 'q' built-in functions, and moving the round to odd
670 built-in functions to being special built-in functions.
671 (BU_FLOAT128_2): Likewise.
672 (BU_FLOAT128_1_HW): Likewise.
673 (BU_FLOAT128_2_HW): Likewise.
674 (BU_FLOAT128_3_HW): Likewise.
675 (FABSQ): Delete old 'q' built-in functions.
676 (COPYSIGNQ): Likewise.
677 (SQRTF128_ODD): Move round to odd built-in functions to be
678 special built-in functions, so that we can handle
679 -mabi=ieeelongdouble.
680 (TRUNCF128_ODD): Likewise.
681 (ADDF128_ODD): Likewise.
682 (SUBF128_ODD): Likewise.
683 (MULF128_ODD): Likewise.
684 (DIVF128_ODD): Likewise.
685 (FMAF128_ODD): Likewise.
686 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
687 built-in names to 'f128'.
688 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
689 old 'q' built-in functions, as the machine independent code for
690 'f128' built-in functions handles this.
691 (rs6000_expand_builtin): Add expansion for float128 round to odd
692 functions, keying off on -mabi=ieeelongdouble of whether to use
693 the KFmode or TFmode variant.
694 (rs6000_init_builtins): Initialize the _Float128 round to odd
696 * doc/extend.texi (PowerPC Built-in Functions): Document the old
697 _Float128 'q' built-in functions are now mapped into the new
698 'f128' built-in functions.
700 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
702 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
703 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
705 2017-11-06 David Malcolm <dmalcolm@redhat.com>
708 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
709 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
710 * toplev.c: Include "ipa-fnsummary.h".
711 (toplev::finalize): Call ipa_fnsummary_c_finalize.
713 2017-11-06 Jakub Jelinek <jakub@redhat.com>
715 PR tree-optimization/82838
716 * gimple-ssa-store-merging.c
717 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
718 on a separate gimple_seq which is then appended to seq.
720 2017-11-06 Jeff Law <law@redhat.com>
723 * config/i386/i386.c (PROBE_INTERVAL): Remove.
724 (get_probe_interval): New functions.
725 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
726 (ix86_adjust_stack_and_probe): Likewise.
727 (output_adjust_stack_and_probe): Likewise.
728 (ix86_emit_probe_stack_range): Likewise.
729 (ix86_expand_prologue): Likewise.
731 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
733 PR tree-optimization/82816
734 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
735 if the modes of the two types are the same.
736 (convert_plusminus_to_widen): Likewise.
738 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
740 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
742 (usadv16qi): New define_expand.
743 (usadv8hi): New define_expand.
745 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
748 * ipa-inline-transform.c (update_noncloned_frequencies): Always
750 (inline_transform): Likewise.
751 * predict.c (counts_to_freqs): Remove useless conditional.
752 * profile-count.h (profile_count::apply_scale): Move sanity check.
753 * tree-inline.c (copy_bb): Always scale.
754 (copy_cfg_body): Likewise.
756 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
759 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
761 (*cmp_ite0): Add enabled_for_depr_it attribute.
762 (*cmp_ite1): Likewise.
764 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
766 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
769 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
771 * tree-vrp.c (vrp_int_const_binop): Return true on success and
772 return the value by pointer.
773 (extract_range_from_multiplicative_op_1): Update accordingly.
774 Return as soon as an operation fails.
776 2017-11-05 Tom de Vries <tom@codesourcery.com>
779 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
780 (DEF_SANITIZER_BUILTIN): ... here.
781 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
782 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
784 2017-11-05 Tom de Vries <tom@codesourcery.com>
787 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
789 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
790 ASM_OUTPUT_BEFORE_CASE_LABEL call.
791 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
793 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
794 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
795 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
797 2017-11-05 Tom de Vries <tom@codesourcery.com>
800 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
803 2017-11-04 Michael Clark <michaeljclark@mac.com>
805 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
806 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
808 (*addsi3_extended): Likewise.
809 (*addsi3_extended2): Likewise.
810 (<optab>si3): Likewise.
811 (<optab>di3): Likewise.
812 (<optab><mode>3): Likewise.
813 (<*optabe>si3_internal): Likewise.
814 (zero_extendqi<SUPERQI:mode>2): Likewise.
815 (*add<mode>hi3): Likewise.
816 (*xor<mode>hi3): Likewise.
817 (<optab>di3): Likewise.
818 (*<optab>si3_extend): Likewise.
819 (*sge<u>_<X:mode><GPR:mode>): Likewise.
820 (*slt<u>_<X:mode><GPR:mode>): Likewise.
821 (*sle<u>_<X:mode><GPR:mode>): Likewise.
823 2017-11-04 Andrew Waterman <andrew@sifive.com>
825 * config/riscv/riscv.c (riscv_option_override): Conditionally set
826 TARGET_STRICT_ALIGN based upon -mtune argument.
828 2017-11-04 Andrew Waterman <andrew@sifive.com>
830 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
832 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
834 config/i386/i386.c (choose_basereg): Use optional scratch
835 register and add assertion.
836 (x86_emit_outlined_ms2sysv_save): Use scratch register when
837 needed, and don't allocate stack.
838 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
839 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
840 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
842 2017-11-03 Jeff Law <law@redhat.com>
844 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
845 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
846 to probe at the start of a noreturn function.
848 2017-11-03 Jakub Jelinek <jakub@redhat.com>
850 PR tree-optimization/78821
851 * gimple-ssa-store-merging.c: Update the file comment.
852 (MAX_STORE_ALIAS_CHECKS): Define.
853 (struct store_operand_info): New type.
854 (store_operand_info::store_operand_info): New constructor.
855 (struct store_immediate_info): Add rhs_code and ops data members.
856 (store_immediate_info::store_immediate_info): Add rhscode, op0r
857 and op1r arguments to the ctor, initialize corresponding data members.
858 (struct merged_store_group): Add load_align_base and load_align
860 (merged_store_group::merged_store_group): Initialize them.
861 (merged_store_group::do_merge): Update them.
862 (merged_store_group::apply_stores): Pick the constant for
863 encode_tree_to_bitpos from one of the two operands, or skip
864 encode_tree_to_bitpos if neither operand is a constant.
865 (class pass_store_merging): Add process_store method decl. Remove
866 bool argument from terminate_all_aliasing_chains method decl.
867 (pass_store_merging::terminate_all_aliasing_chains): Remove
868 var_offset_p argument and corresponding handling.
869 (stmts_may_clobber_ref_p): New function.
870 (compatible_load_p): New function.
871 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
872 if there is overlap and rhs_code is not INTEGER_CST. For
873 non-overlapping stores terminate group if rhs is not mergeable.
874 (get_alias_type_for_stmts): Change first argument from
875 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
876 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
877 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
879 (get_location_for_stmts): Change first argument from
880 auto_vec<gimple *> & to vec<gimple *> &.
881 (struct split_store): Remove orig_stmts data member, add orig_stores.
882 (split_store::split_store): Create orig_stores rather than orig_stmts.
883 (find_constituent_stmts): Renamed to ...
884 (find_constituent_stores): ... this. Change second argument from
885 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
886 to info structures rather than the statements.
887 (split_group): Rename ALLOW_UNALIGNED argument to
888 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
889 it. Adjust find_constituent_stores caller.
890 (imm_store_chain_info::output_merged_store): Handle rhs_code other
891 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
892 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
893 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
894 (mem_valid_for_store_merging): New function.
895 (handled_load): New function.
896 (pass_store_merging::process_store): New method.
897 (pass_store_merging::execute): Use process_store method. Adjust
898 terminate_all_aliasing_chains caller.
900 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
902 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
903 Return true for more constants, symbols and label references.
904 (aarch64_valid_floating_const): Remove unused function.
906 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
909 * config/aarch64/aarch64.c (aarch64_layout_frame):
910 Undo forcing of LR at bottom of frame.
912 2017-11-03 Jeff Law <law@redhat.com>
915 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
916 for int_registers_saved.
918 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
919 extracted from tree-ssa-dom.c.
920 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
921 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
922 (record_equivalences_from_incoming_edge): Add additional argument
923 to single_pred_edge_ignoring_loop_edges call.
924 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
925 (uncprop_dom_walker::before_dom_children): Add additional argument
926 to single_pred_edge_ignoring_loop_edges call.
927 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
928 single_pred_edge_ignoring_loop_edges rather than open coding.
929 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
931 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
933 * match.pd (-(-A)): Rewrite.
935 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
937 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
938 (rs6000_emit_int_cmove): New declaration.
939 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
940 (rs6000_emit_sISEL): Delete.
941 (rs6000_emit_int_cmove): Make non-static.
942 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
943 instead of rs6000_emit_sISEL.
945 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
947 * asan.c (create_cond_insert_point): Maintain profile.
948 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
950 * basic-block.h (struct basic_block_def): Remove frequency.
951 (EDGE_FREQUENCY): Use to_frequency
952 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
954 (find_traces): Update to use to_frequency.
955 (find_traces_1_round): Likewise; use only IPA counts.
956 (bb_to_key): Likewise.
957 (connect_traces): Use IPA counts only.
958 (copy_bb_p): Update to use to_frequency.
959 (fix_up_crossing_landing_pad): Likewise.
960 (sanitize_hot_paths): Likewise.
961 * bt-load.c (basic_block_freq): Likewise.
962 * cfg.c (init_flow): Set count_max to uninitialized.
963 (check_bb_profile): Remove frequencies; check counts.
964 (dump_bb_info): Do not dump frequencies.
965 (update_bb_profile_for_threading): Update counts only.
966 (scale_bbs_frequencies_int): Likewise.
967 (MAX_SAFE_MULTIPLIER): Remove.
968 (scale_bbs_frequencies_gcov_type): Update counts only.
969 (scale_bbs_frequencies_profile_count): Update counts only.
970 (scale_bbs_frequencies): Update counts only.
971 * cfg.h (struct control_flow_graph): Add count-max.
972 (update_bb_profile_for_threading): Update prototype.
973 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
974 (find_many_sub_basic_blocks): Likewise.
975 * cfgcleanup.c (try_forward_edges): Likewise.
976 (try_crossjump_to_edge): Likewise.
977 * cfgexpand.c (expand_gimple_cond): Likewise.
978 (expand_gimple_tailcall): Likewise.
979 (construct_init_block): Likewise.
980 (construct_exit_block): Likewise.
981 * cfghooks.c (verify_flow_info): Check consistency of counts.
982 (dump_bb_for_graph): Do not dump frequencies.
983 (split_block_1): Do not update frequencies.
984 (split_edge): Do not update frequencies.
985 (make_forwarder_block): Do not update frequencies.
986 (duplicate_block): Do not update frequencies.
987 (account_profile_record): Do not update frequencies.
988 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
989 for global heuristics.
990 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
991 (expected_loop_iterations_unbounded): Use counts only.
992 * cfgloopmanip.c (scale_loop_profile): Simplify.
993 (create_empty_loop_on_edge): Simplify
995 (duplicate_loop_to_header_edge): Simplify
996 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
997 (update_br_prob_note): Take care of removing note when profile
999 (relink_block_chain): Do not dump frequency.
1000 (rtl_account_profile_record): Use to_frequency.
1001 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
1002 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
1003 (cgraph_update_edges_for_call_stmt_node): Likewise.
1004 (cgraph_edge::verify_count_and_frequency): Update.
1005 (cgraph_node::verify_node): Temporarily disable frequency verification.
1006 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
1007 to_cgraph_frequency.
1008 (cgraph_edge::rebuild_edges): Convert to ipa counts.
1009 * cgraphunit.c (init_lowered_empty_function): Do not initialize
1011 (cgraph_node::expand_thunk): Update profile.
1012 * except.c (dw2_build_landing_pads): Do not update frequency.
1013 * final.c (compute_alignments): Use to_frequency.
1014 (dump_basic_block_info): Do not dump frequency.
1015 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
1016 (dump_gimple_bb_header): Do not dump frequency.
1017 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
1019 * gimple-streamer-in.c (input_bb): Do not stream frequency.
1020 * gimple-streamer-out.c (output_bb): Do not stream frequency.
1021 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
1022 (init_before_recovery): Do not update frequency.
1023 (sched_create_recovery_edges): Do not update frequency.
1024 * hsa-gen.c (convert_switch_statements): Do not update frequency.
1025 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1026 (ipa_cp_c_finalize): Set max_count to uninitialized.
1027 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1028 (param_change_prob): Use counts.
1029 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1031 * ipa-split.c (consider_split): Use to_frequency.
1032 (split_function): Use to_frequency.
1033 * ira-build.c (loop_compare_func): Likewise.
1034 (mark_loops_for_removal): Likewise.
1035 (mark_all_loops_for_removal): Likewise.
1036 * loop-doloop.c (doloop_modify): Do not update frequency.
1037 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1039 * lto-streamer-in.c (input_function): Update count_max.
1040 * omp-expand.c (expand_omp_taskreg): Update count_max.
1041 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1042 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1043 (maybe_hot_count_p): Use ipa counts only.
1044 (maybe_hot_bb_p): Simplify.
1045 (maybe_hot_edge_p): Simplify.
1046 (probably_never_executed): Do not take frequency argument.
1047 (probably_never_executed_bb_p): Do not pass frequency.
1048 (probably_never_executed_edge_p): Likewise.
1049 (combine_predictions_for_bb): Check that profile is nonzero.
1050 (propagate_freq): Do not set frequency.
1051 (drop_profile): Simplify.
1052 (counts_to_freqs): Simplify.
1053 (expensive_function_p): Use to_frequency.
1054 (propagate_unlikely_bbs_forward): Simplify.
1055 (determine_unlikely_bbs): Simplify.
1056 (estimate_bb_frequencies): Add hack to silence graphite issues.
1057 (compute_function_frequency): Use ipa counts.
1058 (pass_profile::execute): Update.
1059 (rebuild_frequencies): Use counts only.
1060 (force_edge_cold): Use counts only.
1061 * profile-count.c (profile_count::dump): Dump new count types.
1062 (profile_count::differs_from_p): Check compatiblity.
1063 (profile_count::to_frequency): New function.
1064 (profile_count::to_cgraph_frequency): New function.
1065 * profile-count.h (struct function): Declare.
1066 (enum profile_quality): Add profile_guessed_local and
1067 profile_guessed_global0.
1068 (class profile_proability): Decrease number of bits to 29;
1069 update from_reg_br_prob_note and to_reg_br_prob_note.
1070 (class profile_count: Update comment; decrease number of bits
1071 to 61. Check compatibility.
1072 (profile_count::compatible_p): New private member function.
1073 (profile_count::ipa_p): New member function.
1074 (profile_count::operator<): Handle global zero correctly.
1075 (profile_count::operator>): Handle global zero correctly.
1076 (profile_count::operator<=): Handle global zero correctly.
1077 (profile_count::operator>=): Handle global zero correctly.
1078 (profile_count::nonzero_p): New member function.
1079 (profile_count::force_nonzero): New member function.
1080 (profile_count::max): New member function.
1081 (profile_count::apply_scale): Handle IPA scalling.
1082 (profile_count::guessed_local): New member function.
1083 (profile_count::global0): New member function.
1084 (profile_count::ipa): New member function.
1085 (profile_count::to_frequency): Declare.
1086 (profile_count::to_cgraph_frequency): Declare.
1087 * profile.c (OVERLAP_BASE): Delete.
1088 (compute_frequency_overlap): Delete.
1089 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1090 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1091 * sched-ebb.c (rank): Use counts only.
1092 * shrink-wrap.c (handle_simple_exit): Use counts only.
1093 (try_shrink_wrapping): Use counts only.
1094 (place_prologue_for_one_component): Use counts only.
1095 * tracer.c (find_best_predecessor): Use to_frequency.
1096 (find_trace): Use to_frequency.
1097 (tail_duplicate): Use to_frequency.
1098 * trans-mem.c (expand_transaction): Do not update frequency.
1099 * tree-call-cdce.c: Do not update frequency.
1100 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1101 (gimple_merge_blocks): Likewise.
1102 (gimple_split_edge): Likewise.
1103 (gimple_duplicate_sese_region): Likewise.
1104 (gimple_duplicate_sese_tail): Likewise.
1105 (move_sese_region_to_fn): Likewise.
1106 (gimple_account_profile_record): Likewise.
1107 (insert_cond_bb): Likewise.
1108 * tree-complex.c (expand_complex_div_wide): Likewise.
1109 * tree-eh.c (lower_resx): Update profile.
1110 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1112 (initialize_cfun): Do not initialize frequencies
1113 (freqs_to_counts): Delete.
1114 (copy_cfg_body): Ignore count parameter.
1115 (copy_body): Update.
1116 (expand_call_inline): Update count_max.
1117 (optimize_inline_calls): Update count_max.
1118 (tree_function_versioning): Update count_max.
1119 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1120 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1122 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1123 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1124 (try_peel_loop): Likewise.
1125 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1127 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1128 (tree_transform_and_unroll_loop): Do not use frequencies
1129 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1130 Use reliable prediction only.
1131 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1132 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1133 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1134 probability scaling.
1135 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1136 not update frequency
1137 (any_remaining_duplicated_blocks): Likewise.
1138 (update_profile): Likewise.
1139 (estimated_freqs_path): Delete.
1140 (freqs_to_counts_path): Delete.
1141 (clear_counts_path): Delete.
1142 (ssa_fix_duplicate_block_edges): Likewise.
1143 (duplicate_thread_path): Likewise.
1144 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1145 * tree-tailcall.c (decrease_profile): Do not update frequency.
1146 (eliminate_tail_call): Likewise.
1147 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1148 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1149 (optimize_mask_stores): Likewise.
1150 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1151 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1152 (ubsan_expand_ptr_ifn): Update profile.
1153 * value-prof.c (gimple_ic): Simplify.
1154 * value-prof.h (gimple_ic): Update prototype.
1155 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1156 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1158 (want_inline_self_recursive_call_p): Likewise.
1159 (resolve_noninline_speculation): Only cummulate defined counts.
1160 (inline_small_functions): Use nonzero_p.
1161 (ipa_inline): Do not access freed node.
1163 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1165 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1166 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1168 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1170 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1171 non-legitimate address.
1173 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1175 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1176 (*lt0_<mode>di, *lt0_<mode>si): New.
1178 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1180 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1181 TARGET_PAIRED_FLOAT.
1183 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1184 Jim Wilson <jim.wilson@linaro.org>
1186 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1187 * config/aarch64/aarch64-tune.md: Regenerated.
1188 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1189 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1191 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1193 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1194 (arc_expand_prologue): Restore blink for millicode.
1195 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1197 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1200 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1201 gen_vec_duplicate after forcing the scalar into a register.
1203 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1205 * combine (try_combine): Print the insns input to try_combine to the
1208 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1211 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1212 Remove second argument from aarch64_process_target_attr call.
1213 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1215 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1217 (aarch64_handle_attr_arch): Remove second argument.
1218 (aarch64_handle_attr_cpu): Ditto.
1219 (aarch64_handle_attr_tune): Ditto.
1220 (aarch64_handle_attr_isa_flags): Ditto.
1221 (aarch64_process_one_target_attr): Ditto.
1222 (aarch64_process_target_attr): Ditto.
1223 (aarch64_option_valid_attribute_p): Remove second argument.
1224 on aarch64_process_target_attr call.
1226 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1228 * diagnostic.c: Include "selftest-diagnostic.h".
1229 (selftest::assert_location_text): New function.
1230 (selftest::test_diagnostic_get_location_text): New function.
1231 (selftest::diagnostic_c_tests): Call it.
1233 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1235 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1236 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1237 (class selftest::test_diagnostic_context): Move to...
1238 * selftest-diagnostic.c: New file.
1239 * selftest-diagnostic.h: New file.
1241 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1243 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1244 offset range for FT32B.
1245 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1246 * config/ft32/ft32.md: Add TARGET_NOPM.
1247 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1248 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1250 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1252 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1254 2017-11-02 Jeff Law <law@redhat.com>
1256 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1257 virtual keyword on FINAL OVERRIDE members.
1259 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1260 virtuals together. Add virtual destructor.
1261 (substitute_and_fold_engine): Similarly.
1263 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1265 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1267 2017-11-02 Richard Biener <rguenther@suse.de>
1269 PR tree-optimization/82795
1270 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1272 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1274 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1276 (gcc_GAS_CHECK_FEATURE): Remove.
1277 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1278 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1279 * configure: Regenerate.
1281 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1283 * config/arc/arc.c (hwloop_optimize): Account for empty
1286 2017-11-02 Richard Biener <rguenther@suse.de>
1289 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1290 Truncate ARRAY_REF index and element size.
1292 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1294 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1296 2017-11-01 Jeff Law <law@redhat.com>
1298 * tree-ssa-ccp.c (ccp_folder): New class derived from
1299 substitute_and_fold_engine.
1300 (ccp_folder::get_value): New member function.
1301 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1302 (ccp_fold_stmt): Remove prototype.
1303 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1304 * tree-ssa-copy.c (copy_folder): New class derived from
1305 substitute_and_fold_engine.
1306 (copy_folder::get_value): Renamed from get_value.
1307 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1308 * tree-vrp.c (vrp_folder): New class derived from
1309 substitute_and_fold_engine.
1310 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1311 (vrp_folder::get_value): New member function.
1312 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1313 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1314 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1315 provide a class interface to folder/substitute routines.
1316 (ssa_prop_fold_stmt_fn): Remove typedef.
1317 (ssa_prop_get_value_fn): Likewise.
1318 (subsitute_and_fold): Remove prototype.
1319 (replace_uses_in): Likewise.
1320 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1321 Renamed from replace_uses_in. Call the virtual member function
1322 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1323 (substitute_and_fold_dom_walker): Remove initialization of
1324 data member entries for calbacks. Add substitute_and_fold_engine
1325 member and initialize it.
1326 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1327 member functions for get_value, replace_phi_args_in c
1328 replace_uses_in, and fold_stmt calls.
1329 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1330 substitute_and_fold. Remove assert. Update ctor call.
1332 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1333 (ssa_prop_visit_phi_fn): Likewise.
1334 (class ssa_propagation_engine): New class to provide an interface
1336 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1338 (ssa_prop_visit_phi): Likewise.
1339 (ssa_propagation_engine::simulate_stmt): Moved into class.
1340 Call visit_phi/visit_stmt from the class rather than via
1341 file scoped static variables.
1342 (ssa_propagation_engine::simulate_block): Moved into class.
1343 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1344 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1345 set file scoped statics for the visit_stmt/visit_phi callbacks.
1346 * tree-complex.c (complex_propagate): New class derived from
1347 ssa_propagation_engine.
1348 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1349 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1350 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1352 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1353 ssa_propagation_engine.
1354 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1355 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1356 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1357 * tree-ssa-copy.c (copy_prop): New class derived from
1358 ssa_propagation_engine.
1359 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1360 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1361 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1362 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1363 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1364 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1365 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1367 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1369 PR rtl-optimization/82778
1370 PR rtl-optimization/82597
1371 * compare-elim.c (struct comparison): Add in_a_setter field.
1372 (find_comparison_dom_walker::before_dom_children): Remove killed
1373 bitmap and df_simulate_find_defs call, instead walk the defs.
1374 Compute last_setter and initialize in_a_setter. Merge definitions
1375 with first initialization for a few variables.
1376 (try_validate_parallel): Use insn_invalid_p instead of
1377 recog_memoized. Return insn rather than just the pattern.
1378 (try_merge_compare): Fix up comment. Don't uselessly test if
1379 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1381 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1384 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1385 Alan Hayward <alan.hayward@arm.com>
1386 David Sherwood <david.sherwood@arm.com>
1388 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1389 aarch64_hard_regno_nregs to get the number of registers
1392 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1393 Alan Hayward <alan.hayward@arm.com>
1394 David Sherwood <david.sherwood@arm.com>
1396 * config/aarch64/constraints.md (Upl): Rename to...
1398 * config/aarch64/aarch64.md
1399 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1402 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1403 Alan Hayward <alan.hayward@arm.com>
1404 David Sherwood <david.sherwood@arm.com>
1406 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1407 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
1410 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1411 Alan Hayward <alan.hayward@arm.com>
1412 David Sherwood <david.sherwood@arm.com>
1414 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
1415 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1416 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1418 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1419 of a single element.
1420 * config/aarch64/iterators.md: Add a comment above the permute
1421 unspecs to say that they are generated directly by
1422 aarch64_expand_vec_perm_const.
1423 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1425 2017-11-01 Nathan Sidwell <nathan@acm.org>
1427 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1429 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1431 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1432 and medany code models, and describe what they do.
1434 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1436 Revert accidental duplicate:
1438 * combine.c (can_change_dest_mode): Reject changes in
1439 REGMODE_NATURAL_SIZE.
1441 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1443 PR rtl-optimization/64682
1444 PR rtl-optimization/69567
1445 PR rtl-optimization/69737
1446 PR rtl-optimization/82683
1447 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1448 register mentioned in the note, drop the note, unless it came from I3,
1449 in which case it should go to I3 again.
1451 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1453 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1454 and return false if not.
1455 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1457 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1459 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1460 range is known to be empty.
1462 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1463 Alan Hayward <alan.hayward@arm.com>
1464 David Sherwood <david.sherwood@arm.com>
1466 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1467 and CONST_VECTOR_NUNITS instead of computing the number of units from
1468 the byte sizes of the vector and element.
1469 (simplify_binary_operation_1): Likewise.
1470 (simplify_const_binary_operation): Likewise.
1471 (simplify_ternary_operation): Likewise.
1473 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1474 Alan Hayward <alan.hayward@arm.com>
1475 David Sherwood <david.sherwood@arm.com>
1477 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1478 (int_mem_offset): ...this new function.
1479 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1480 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1481 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1484 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1485 Alan Hayward <alan.hayward@arm.com>
1486 David Sherwood <david.sherwood@arm.com>
1488 * lower-subreg.c (interesting_mode_p): New function.
1489 (compute_costs, find_decomposable_subregs, decompose_register)
1490 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1491 (resolve_clobber, dump_choices): Use it.
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 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
1499 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1500 Alan Hayward <alan.hayward@arm.com>
1501 David Sherwood <david.sherwood@arm.com>
1503 * alias.c (find_base_value, find_base_term): Only process integer
1504 truncations. Check the precision rather than the size.
1506 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1507 Alan Hayward <alan.hayward@arm.com>
1508 David Sherwood <david.sherwood@arm.com>
1510 * machmode.h (is_narrower_int_mode): New function
1511 * optabs.c (expand_float, expand_fix): Use it.
1512 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1514 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1515 Alan Hayward <alan.hayward@arm.com>
1516 David Sherwood <david.sherwood@arm.com>
1518 * rtl.h (narrower_subreg_mode): New function.
1519 * ira-color.c (update_costs_from_allocno): Use it.
1521 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1522 Alan Hayward <alan.hayward@arm.com>
1523 David Sherwood <david.sherwood@arm.com>
1525 * optabs-query.h (convert_optab_p): New function, split out from...
1526 (convert_optab_handler): ...here.
1527 (widening_optab_handler): Delete.
1528 (find_widening_optab_handler): Remove permit_non_widening parameter.
1529 (find_widening_optab_handler_and_mode): Likewise. Provide an
1530 override that operates on mode class wrappers.
1531 * optabs-query.c (widening_optab_handler): Delete.
1532 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1533 parameter. Assert that the two modes are the same class and that
1534 the "from" mode is narrower than the "to" mode. Use
1535 convert_optab_handler instead of widening_optab_handler.
1536 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1537 instead of widening_optab_handler.
1538 * expr.c (expand_expr_real_2): Update calls to
1539 find_widening_optab_handler.
1540 * optabs.c (expand_widen_pattern_expr): Likewise.
1541 (expand_binop_directly): Take the insn_code as a parameter.
1542 (expand_binop): Only call find_widening_optab_handler for
1543 conversion optabs; use optab_handler otherwise. Update calls
1544 to find_widening_optab_handler and expand_binop_directly.
1545 Use convert_optab_handler instead of widening_optab_handler.
1546 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1547 find_widening_optab_handler and use scalar_mode rather than
1549 (convert_plusminus_to_widen): Likewise.
1551 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1552 Alan Hayward <alan.hayward@arm.com>
1553 David Sherwood <david.sherwood@arm.com>
1555 * machmode.h (fixed_size_mode): New class.
1556 * rtl.h (get_pool_mode): Return fixed_size_mode.
1557 * gengtype.c (main): Add fixed_size_mode.
1558 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1559 (get_raw_arg_mode): Likewise.
1560 * doc/tm.texi: Regenerate.
1561 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1562 * targhooks.c (default_get_reg_raw_mode): Likewise.
1563 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1564 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1565 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1566 (msp430_get_raw_result_mode): Likewise.
1567 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1568 * dbxout.c (dbxout_parms): Require fixed-size modes.
1569 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1570 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1571 * omp-low.c (lower_oacc_reductions): Likewise.
1572 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1573 (simplify_subreg): Update accordingly.
1574 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1575 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1576 that aren't fixed-size.
1577 (get_pool_mode): Return a fixed_size_mode.
1578 (output_constant_pool_2): Take a fixed_size_mode.
1580 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1581 Alan Hayward <alan.hayward@arm.com>
1582 David Sherwood <david.sherwood@arm.com>
1584 * doc/rtl.texi (vec_series): Document.
1585 (const): Say that the operand can be a vec_series.
1586 * rtl.def (VEC_SERIES): New rtx code.
1587 * rtl.h (const_vec_series_p_1): Declare.
1588 (const_vec_series_p): New function.
1589 * emit-rtl.h (gen_const_vec_series): Declare.
1590 (gen_vec_series): Likewise.
1591 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1592 (gen_vec_series): Likewise.
1593 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1594 * simplify-rtx.c (simplify_unary_operation): Handle negations
1596 (simplify_binary_operation_series): New function.
1597 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1598 (test_vector_ops_series): New function.
1599 (test_vector_ops): Call it.
1600 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1601 gen_const_vec_series.
1602 (altivec_lvsr): Likewise.
1603 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1605 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1606 Alan Hayward <alan.hayward@arm.com>
1607 David Sherwood <david.sherwood@arm.com>
1609 * doc/rtl.texi (const): Update description of address constants.
1610 Say that vector constants are allowed too.
1611 * common.md (E, F): Use CONSTANT_P instead of checking for
1613 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1614 checking for CONST_VECTOR.
1615 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1617 * expr.c (expand_expr_real_2): Check for vector modes instead
1618 of checking for CONST_VECTOR.
1619 * rtl.h (const_vec_p): New function.
1620 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1621 (unwrap_const_vec_duplicate): Handle them here too.
1623 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1624 David Malcolm <dmalcolm@redhat.com>
1625 Alan Hayward <alan.hayward@arm.com>
1626 David Sherwood <david.sherwood@arm.com>
1628 * rtl.h (vec_duplicate_p): New function.
1629 * selftest-rtl.c (assert_rtx_eq_at): New function.
1630 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1631 (assert_rtx_eq_at): Declare.
1632 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1633 * selftest-run-tests.c (selftest::run_tests): Call it.
1634 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1635 (simplify_unary_operation_1): Recursively handle vector duplicates.
1636 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1638 (simplify_subreg): Handle subregs of vector duplicates.
1639 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1640 (selftest::simplify_rtx_c_tests): New functions.
1642 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1643 Alan Hayward <alan.hayward@arm.com>
1644 David Sherwood <david.sherwood@arm.com>
1646 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1647 (gen_vec_duplicate): Likewise.
1648 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1650 (gen_const_vector): ...here.
1651 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1652 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1653 whose elements are all equal.
1654 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1655 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1656 (simplify_relational_operation): Likewise.
1657 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1659 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1660 (aarch64_expand_vector_init): Likewise.
1661 * config/arm/arm.c (neon_vdup_constant): Likewise.
1662 (neon_expand_vector_init): Likewise.
1663 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1664 (arm_block_set_unaligned_vect): Likewise.
1665 (arm_block_set_aligned_vect): Likewise.
1666 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1667 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1668 (expand_vec_perm_even_odd_pack): Likewise.
1669 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1670 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1671 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1672 gen_const_vec_duplicate.
1673 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1674 * config/mips/mips.c (mips_gen_const_int_vector): Use
1675 gen_const_vec_duplicate.
1676 (mips_expand_vector_init): Use CONST0_RTX.
1677 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1678 (define_split): Use gen_const_vec_duplicate.
1679 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1680 (define_split): Use gen_const_vec_duplicate.
1681 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1682 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1683 * config/spu/spu.c (spu_const): Likewise.
1685 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1686 Alan Hayward <alan.hayward@arm.com>
1687 David Sherwood <david.sherwood@arm.com>
1689 * combine.c (can_change_dest_mode): Reject changes in
1690 REGMODE_NATURAL_SIZE.
1692 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1694 * configure.ac (--enable-libssp): New.
1695 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1696 trying to determine target-specific default. Adjust indentation.
1697 * configure: Regenerated.
1698 * doc/install.texi (Configuration): Expand --disable-libssp
1701 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1703 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1706 2017-10-31 Martin Jambor <mjambor@suse.cz>
1709 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1711 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1713 * auto-profile.c (autofdo_source_profile::read): Use
1714 UNKNOWN_LOCATION rather than 0.
1715 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1716 (warning_at): ...this overload.
1717 (warning_at_rich_loc_n): Rename to...
1718 (warning_n): ...this overload.
1719 (error_at_rich_loc): Rename to...
1720 (error_at): ...this overload.
1721 (pedwarn_at_rich_loc): Rename to...
1722 (pedwarn): ...this overload.
1723 (permerror_at_rich_loc): Rename to...
1724 (permerror): ...this overload.
1725 (inform_at_rich_loc): Rename to...
1726 (inform): ...this overload.
1727 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1728 (diagnostic_n_impl_richloc): Rename to...
1729 (diagnostic_n_impl): ...this rich_location *-based decl.
1730 (inform_at_rich_loc): Rename to...
1731 (inform): ...this, and add an assertion.
1732 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1733 (warning_at_rich_loc): Rename to...
1734 (warning_at): ...this, and add an assertion.
1735 (warning_at_rich_loc_n): Rename to...
1736 (warning_n): ...this, and add an assertion.
1737 (warning_n): Update location_t-based implementation for removal of
1738 location_t-based diagnostic_n_impl.
1739 (pedwarn_at_rich_loc): Rename to...
1740 (pedwarn): ...this, and add an assertion.
1741 (permerror_at_rich_loc): Rename to...
1742 (permerror): ...this, and add an assertion.
1743 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1744 (error_at_rich_loc): Rename to...
1745 (error_at): ...this, and add an assertion.
1746 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1747 (driver::do_spec_on_infiles): Likewise.
1748 * substring-locations.c (format_warning_va): Update for renaming
1749 of inform_at_rich_loc.
1751 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1753 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1754 _Float<N>X built-in functions so that the variant without the
1755 "__builtin_" prefix is only enabled for the GNU C and Objective C
1756 languages when they are in non-strict ANSI/ISO mode.
1757 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1758 * target.def (floatn_builtin_p): Add a target hook to control
1759 whether _Float<N> and _Float<N>X built-in functions without the
1760 "__builtin_" prefix are enabled, and return true for C and
1761 Objective C in the default hook. Include langhooks.h in
1763 * targhooks.h (default_floatn_builtin_p): Likewise.
1764 * targhooks.c (default_floatn_builtin_p): Likewise.
1765 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1766 floatn_builtin_p target hook.
1767 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1769 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1770 Eric Botcazou <ebotcazou@adacore.com>
1772 PR rtl-optimization/81803
1773 * lra-constraints.c (curr_insn_transform): Also reload the whole
1774 register for a strict subreg no wider than a word if this is for
1775 a WORD_REGISTER_OPERATIONS target.
1777 2017-10-31 Jason Merrill <jason@redhat.com>
1779 * gdbinit.in: Skip over inlines from timevar.h.
1781 2017-10-31 Martin Liska <mliska@suse.cz>
1783 * doc/gcov.texi: Document new option.
1784 * gcov.c (print_usage): Likewise print it.
1785 (process_args): Support the argument.
1786 (format_count): New function.
1787 (format_gcov): Use the function.
1789 2017-10-31 Martin Liska <mliska@suse.cz>
1791 * gcov.c (struct name_map): do not use typedef.
1792 Define operator== and operator<.
1793 (name_search): Remove.
1794 (name_sort): Remove.
1795 (main): Do not allocate names.
1796 (process_file): Add vertical space.
1797 (generate_results): Use std::find.
1798 (release_structures): Do not release memory.
1799 (find_source): Use std::find.
1801 2017-10-31 Martin Liska <mliska@suse.cz>
1803 * gcov.c (struct line_info): Remove it's typedef.
1804 (line_info::line_info): Add proper ctor.
1805 (line_info::has_block): Do not use a typedef.
1806 (struct source_info): Do not use typedef.
1807 (circuit): Likewise.
1808 (get_cycles_count): Likewise.
1809 (output_intermediate_file): Iterate via vector iterator.
1810 (add_line_counts): Use std::vector methods.
1811 (accumulate_line_counts): Likewise.
1812 (output_lines): Likewise.
1814 2017-10-31 Martin Liska <mliska@suse.cz>
1816 * gcov.c (struct source_info): Remove typedef.
1817 (source_info::source_info): Add proper ctor.
1818 (accumulate_line_counts): Use struct, not it's typedef.
1819 (output_gcov_file): Likewise.
1820 (output_lines): Likewise.
1821 (main): Do not allocate an array.
1822 (output_intermediate_file): Use size of vector container.
1823 (process_file): Resize the vector.
1824 (generate_results): Do not preallocate, use newly added vector
1826 (release_structures): Do not release sources.
1827 (find_source): Use vector methods.
1828 (add_line_counts): Do not use typedef.
1830 2017-10-31 Martin Liska <mliska@suse.cz>
1832 * doc/gcov.texi: Document that.
1833 * gcov.c (add_line_counts): Mark lines with a non-executed
1835 (output_line_beginning): Handle such lines.
1836 (output_lines): Pass new argument.
1837 (output_intermediate_file): Print it in intermediate format.
1839 2017-10-31 Martin Liska <mliska@suse.cz>
1841 * color-macros.h: New file.
1842 * diagnostic-color.c: Factor out color related to macros to
1844 * doc/gcov.texi: Document -k option.
1845 * gcov.c (INCLUDE_STRING): Include string.h.
1846 (print_usage): Add -k option.
1847 (process_args): Parse it.
1848 (pad_count_string): New function.
1849 (output_line_beginning): Likewise.
1850 (DEFAULT_LINE_START): New macro.
1851 (output_lines): Support color output.
1853 2017-10-31 Martin Liska <mliska@suse.cz>
1855 PR gcov-profile/82633
1856 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1857 their interaction with GCOV infrastructure.
1858 * configure.ac: Add -fkeep-{inline,static}-functions to
1860 * configure: Regenerate.
1862 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1865 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1867 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1870 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1871 into a register if it does not fit into an immediate offset field.
1873 2017-10-31 Olivier Hainque <hainque@adacore.com>
1875 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1877 2017-10-31 Julia Koval <julia.koval@intel.com>
1879 * config.gcc: Add gfniintrin.h.
1880 * config/i386/gfniintrin.h: New.
1881 * config/i386/i386-builtin-types.def
1882 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1883 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1884 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1885 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1886 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1887 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1888 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1889 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1890 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1891 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1892 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1893 * config/i386/immintrin.h: Include gfniintrin.h.
1894 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1896 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1898 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1900 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1902 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1903 (arm_ashldi3_1bit): Remove pattern.
1904 (ashrdi3): Remove shift by 1 expansion.
1905 (arm_ashrdi3_1bit): Remove pattern.
1906 (lshrdi3): Remove shift by 1 expansion.
1907 (arm_lshrdi3_1bit): Remove pattern.
1908 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1909 cost of ashldi3 by 1.
1910 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1911 (<shift>di3_neon): Likewise.
1913 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1915 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1916 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1917 and (*aarch64_simd_mov<VQ:mode>).
1918 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1919 pattern alternative.
1920 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1921 attributes to match pattern alternative.
1923 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1925 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1926 * config/rs6000/emmintrin.h: New file.
1927 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1929 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1931 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1932 * (movdi_vfp_cortexa8): Remove pattern.
1934 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1936 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1938 (Specific, alpha*-dec-osf5.1): Remove.
1939 (Specific, mips-sgi-irix5): Remove.
1940 (Specific, mips-sgi-irix6): Remove.
1942 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1945 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1946 arguments to clear_bit_region_be.
1948 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1950 * gimplify.c: Include memmodel.h.
1952 2017-10-30 Martin Jambor <mjambor@suse.cz>
1954 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1955 condition whether loop should be executed at all.
1957 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1959 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1960 gimple folding of vec_madd() intrinsics.
1961 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1962 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1963 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1965 2017-10-30 Richard Biener <rguenther@suse.de>
1967 PR tree-optimization/82762
1969 2017-10-23 Richard Biener <rguenther@suse.de>
1971 PR tree-optimization/82129
1973 2017-08-01 Richard Biener <rguenther@suse.de>
1975 PR tree-optimization/81181
1976 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1977 (compute_antic): ... end of iteration here.
1979 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1981 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1982 -std=iso9899:2017 and -std=gnu17.
1983 * doc/standards.texi (C Language): Document C17 support.
1984 * doc/cpp.texi (Overview): Mention -std=c17.
1985 (Standard Predefined Macros): Document C11 and C17 values of
1986 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1987 * doc/extend.texi (Inline): Do not list individual options for
1988 standards newer than C99.
1989 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1991 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1994 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1996 * asan.c (asan_finish_file): Align asan globals array by shadow
1999 2017-10-30 Jakub Jelinek <jakub@redhat.com>
2002 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
2003 (struct store_immediate_info): Add bitregion_start and bitregion_end
2005 (store_immediate_info::store_immediate_info): Add brs and bre
2006 arguments and initialize bitregion_{start,end} from those.
2007 (struct merged_store_group): Add bitregion_start, bitregion_end,
2008 align_base and mask fields. Drop unnecessary struct keyword from
2009 struct store_immediate_info. Add do_merge method.
2010 (clear_bit_region_be): Use memset instead of loop storing zeros.
2011 (merged_store_group::do_merge): New method.
2012 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
2013 stores as long as the surrounding bitregions have no gaps.
2014 (merged_store_group::merge_overlapping): Use do_merge.
2015 (merged_store_group::apply_stores): Test that bitregion_{start,end}
2016 is byte aligned, rather than requiring that start and width are
2017 byte aligned. Drop unnecessary struct keyword from
2018 struct store_immediate_info. Allocate and populate also mask array.
2019 Make start of the arrays relative to bitregion_start rather than
2020 start and size them according to bitregion_{end,start} difference.
2021 (struct imm_store_chain_info): Drop unnecessary struct keyword from
2022 struct store_immediate_info.
2023 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
2024 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
2025 struct keyword from struct store_immediate_info.
2026 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2027 between stores as long as the surrounding bitregions have no gaps.
2029 (struct split_store): Add orig non-static data member.
2030 (split_store::split_store): Initialize orig to false.
2031 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2032 if there is exactly a single original stmt. Change stmts argument
2033 to pointer from reference, if NULL, don't push anything to it. Add
2034 first argument, use it to optimize skipping over orig stmts that
2035 are known to be before bitpos already. Simplify.
2036 (split_group): Return unsigned int count how many stores are or
2037 would be needed rather than a bool. Add allow_unaligned argument.
2038 Change split_stores argument from reference to pointer, if NULL,
2039 only do a dry run computing how many stores would be produced.
2040 Rewritten algorithm to use both alignment and misalign if
2041 !allow_unaligned and handle bitfield stores with gaps.
2042 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2043 from bitregion_start instead of start. Compute allow_unaligned
2044 here, if true, do 2 split_group dry runs to compute which one
2045 produces fewer stores and prefer aligned if equal. Punt if
2046 new count is bigger or equal than original before emitting any
2047 statements, rather than during that. Remove no longer needed
2048 new_ssa_names tracking. Replace num_stmts with
2049 split_stores.length (). Use 32-bit stack allocated entries
2050 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2051 if possible. Handle bitfields with gaps.
2052 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2053 Compute bitregion_{start,end} for the stores and construct
2054 store_immediate_info with that. Formatting fixes.
2056 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2059 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2060 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2062 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2064 * gimplify.c: Include tm_p.h.
2066 * common.opt (gcoff): Re-add as ignored option.
2067 (gcoff1, gcoff2, gcoff3): Likewise.
2069 * Makefile.in (OBJS): Delete sdbout.o.
2070 (GTFILES): Delete $(srcdir)/sdbout.c.
2071 * debug.h: Delete sdb_debug_hooks.
2072 * final.c: Delete sdbout.h include.
2073 (final_scan_insn): Delete SDB_DEBUG check.
2074 (rest_of_clean_state): Likewise.
2075 * output.h: Delete sdb_begin_function_line.
2078 * toplev.c: Delete sdbout.h include.
2079 (process_options): Delete SDB_DEBUG check.
2080 * tree-core.h (tree_type_common): Delete pointer field of
2082 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2083 TYPE_SYMTAB_POINTER.
2084 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2085 (TYPE_SYMTAB_IS_POINTER): Delete.
2086 (TYPE_SYMTAB_IS_DIE): Renumber.
2087 * xcoffout.c: Refer to former sdbout.c file.
2088 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2090 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2091 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2092 (Debugging Options): Delete -gcoff.
2093 (-gstabs): Delete SDB reference.
2095 (-gcoff@var{level}): Delete.
2096 * doc/passes.texi (Debugging information output): Delete SDB and
2097 sdbout.c references.
2098 * doc/tm.texi: Regenerate.
2099 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2100 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2102 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2103 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2104 (SDB_DEBUGGING_INFO): Delete.
2105 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2106 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2107 * target.def (output_source_filename): Delete COFF reference.
2109 * common.opt (gcoff): Delete.
2110 (gxcoff+): Update Negative chain.
2111 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2113 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2114 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2115 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2117 (expand_function_start): Change sdb reference to past tense.
2118 (expand_function_end): Change sdb reference to past tense.
2119 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2120 * opts.c (debug_type_names): Delete coff entry.
2121 (common_handle_option): Delete OPT_gcoff case.
2122 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2124 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2125 * config/cris/cris.h: Delete SDB reference in comment.
2126 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2127 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2128 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2129 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2131 (ix86_expand_prologue): Likewise.
2132 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2133 * config/ia64/ia64.h: Likewise.
2134 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2135 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2137 * config/mmix/mmix.h: Likewise.
2138 * config/nds32/nds32.c: Likewise.
2139 * config/stormy/storym16.h: Likewise.
2140 * config/visium/visium.h: Likewise.
2141 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2143 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2145 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2146 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2147 FRAME_POINTER_REGNUM point at high end of local var area.
2149 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2151 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2152 Move comment around. Do not reset best_edge for a copiable
2153 destination if the copy would cause a partition change.
2154 (better_edge_p): Remove redundant check.
2156 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2158 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2160 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2162 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2163 for math functions that have _Float<N> and _Float<N>X variants.
2164 (mathfn_built_in_2): Add support for math functions that have
2165 _Float<N> and _Float<N>X variants.
2166 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2167 (expand_builtin_mathfn_ternary): Add support for fma with
2168 _Float<N> and _Float<N>X variants.
2169 (expand_builtin): Likewise.
2170 (fold_builtin_3): Likewise.
2171 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2172 create math function _Float<N> and _Float<N>X variants as external
2174 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2175 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2176 the __builtin_ prefix and if not strict ansi, without the prefix.
2177 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2178 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2179 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2180 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2181 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2182 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2183 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2184 function signatures for fma _Float<N> and _Float<N>X variants.
2185 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2186 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2187 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2188 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2189 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2190 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2191 * gencfn-macros.c (print_case_cfn): Add support for math functions
2192 that have _Float<N> and _Float<N>X variants.
2193 (print_define_operator_list): Likewise.
2194 (fltfn_suffixes): Likewise.
2196 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2197 for math functions that have _Float<N> and _Float<N>X variants.
2198 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2199 and _Float<N>X variants.
2200 (COPYSIGN): Likewise.
2203 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2204 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2206 (integer_valued_read_call_p): Likewise.
2207 * fold-const-call.c (fold_const_call_ss): Likewise.
2208 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2209 _Float<N> and _Float<N>X variants.
2210 (fold_const_call_ssss): Add support for fma _Float<N> and
2211 _Float<N>X variants.
2212 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2213 support for copysign and fma _Float<N> and _Float<N>X variants.
2214 (backprop::process_builtin_call_use): Likewise.
2215 * tree-call-cdce.c (can_test_argument_range); Add support for
2216 sqrt _Float<N> and _Float<N>X variants.
2217 (edom_only_function): Likewise.
2218 (get_no_error_domain): Likewise.
2219 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2220 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2221 copysign _Float<N> and _Float<N>X variants.
2222 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2223 handled by machine independent code.
2224 (FMAF128): Likewise.
2225 * doc/cpp.texi (Common Predefined Macros): Document defining
2226 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2227 fma _Float<N> and _Float<N>X variants.
2229 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2232 * config/i386/i386-modes.def (CCFPU): Remove definition.
2233 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2234 (ix86_cc_modes_compatible): Ditto.
2235 (ix86_expand_carry_flag_compare): Ditto.
2236 (ix86_expand_int_movcc): Ditto.
2237 (ix86_expand_int_addcc): Ditto.
2238 (ix86_reverse_condition): Ditto.
2239 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2240 Return true/false for unordered/ordered fp comparisons.
2241 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2242 (ix86_prepare_fp_compare_args): Update for rename.
2243 (ix86_expand_fp_compare): Update for rename. Generate unordered
2244 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2245 (ix86_expand_sse_compare_and_jump): Ditto.
2246 * config/i386/predicates.md (fcmov_comparison_operator):
2247 Remove CCFPU mode handling.
2248 (ix86_comparison_operator): Ditto.
2249 (ix86_carry_flag_operator): Ditto.
2250 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2251 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2252 (*cmpu<mode>_cc_i387): Ditto.
2253 (FPCMP): Remove mode iterator.
2254 (unord): Remove mode attribute.
2255 (unord_subst): New define_subst transformation
2256 (unord): New define_subst attribute.
2258 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2259 (*cmpi<unord>xf_i387): Ditto.
2260 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2261 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2262 using unord_subst transformation.
2263 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2264 (round_saeonly): Also handle CCFP mode.
2265 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2266 Remove UNSPEC_SAHF unspec handling.
2268 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2270 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2272 2017-10-27 Jeff Law <law@redhat.com>
2274 * gimple-ssa-sprintf.c: Include domwalk.h.
2275 (class sprintf_dom_walker): New class, derived from dom_walker.
2276 (sprintf_dom_walker::before_dom_children): New function.
2277 (struct call_info): Moved into sprintf_dom_walker class
2278 (compute_formath_length, handle_gimple_call): Likewise.
2279 (sprintf_length::execute): Call the dominator walker rather
2280 than walking the statements.
2282 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2283 gimple statement locations.
2284 (check_array_bounds): Corresponding changes. Get the statement's
2285 location directly from wi->stmt.
2287 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2290 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2292 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2294 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2295 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2297 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2300 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2301 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2302 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2303 maybe_get_pool_constant.
2304 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2307 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2309 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2310 2.26 caveat. Update gas and gld versions.
2311 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2314 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2316 * cgraph.h (set_malloc_flag): Declare.
2317 * cgraph.c (set_malloc_flag_1): New function.
2318 (set_malloc_flag): Likewise.
2319 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2320 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2322 (read_ipa_call_summary): Add support for reading is_return_callee.
2323 (write_ipa_call_summary): Stream is_return_callee.
2324 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2325 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2326 ipa-prop.h, ipa-fnsummary.h.
2327 (pure_const_names): Change to static.
2328 (malloc_state_e): Define.
2329 (malloc_state_names): Define.
2330 (funct_state_d): Add field malloc_state.
2331 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2332 (check_retval_uses): New function.
2333 (malloc_candidate_p): Likewise.
2334 (analyze_function): Add support for malloc attribute.
2335 (pure_const_write_summary): Stream malloc_state.
2336 (pure_const_read_summary): Add support for reading malloc_state.
2337 (dump_malloc_lattice): New function.
2338 (propagate_malloc): New function.
2339 (warn_function_malloc): New function.
2340 (ipa_pure_const::execute): Call propagate_malloc and
2341 ipa_free_fn_summary.
2342 (pass_local_pure_const::execute): Add support for malloc attribute.
2343 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2344 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2346 2017-10-27 Martin Liska <mliska@suse.cz>
2348 PR gcov-profile/82457
2349 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2350 for fork-like functions to be properly instrumented.
2352 2017-10-27 Richard Biener <rguenther@suse.de>
2355 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2356 info when we redirected EH.
2358 2017-10-26 Michael Collison <michael.collison@arm.com>
2360 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2362 (<optab>_trunchf<GPI:mode>2: New pattern.
2363 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2364 * config/aarch64/iterators.md (wv): New mode attribute.
2365 (vf, VF): New mode attributes.
2366 (vgp, VGP): New mode attributes.
2367 (s): Update attribute with SImode and DImode prefixes.
2369 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2371 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2372 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2373 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2374 (nios2_option_overide): Initialize it. Don't allow R0-relative
2375 addressing with PIC.
2376 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2377 (nios2_symbolic_constant_p): Likewise.
2378 (nios2_legitimate_address_p): Likewise.
2379 (nios2_r0rel_section_name_p): New.
2380 (nios2_symbol_ref_in_r0rel_data_p): New.
2381 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2382 (r0rel_constant_p): New.
2383 (nios2_print_operand_address): Handle r0rel_constant_p.
2384 (nios2_cdx_narrow_form_p): Likewise.
2385 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2386 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2387 (Nios II Options): Document -mr0rel-sec.
2389 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2391 * config/nios2/nios2.c: Include xregex.h.
2392 (nios2_gprel_sec_regex): New.
2393 (nios2_option_overide): Initialize it. Don't allow GP-relative
2394 addressing with PIC.
2395 (nios2_small_section_name_p): Check for regex match.
2396 * config/nios2/nios2.opt (mgprel-sec=): New option.
2397 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2398 (Nios II Options): Document -mgprel-sec.
2400 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2402 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2404 2017-10-26 Tom de Vries <tom@codesourcery.com>
2406 PR tree-optimization/82707
2407 * gimple.c (gimple_copy): Fix unsharing of
2408 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2410 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2412 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2413 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2414 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2415 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2416 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2417 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2418 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2419 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2420 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2421 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2422 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2423 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2424 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2425 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2426 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2427 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2428 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2430 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2432 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2434 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2435 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2436 warning to rs6000.c. Remove the Undocumented flag, since it has
2438 (-mabi=ibmlongdouble): Likewise.
2439 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2440 already set, set the default format for long double.
2441 (rs6000_debug_reg_global): Print whether long double is IBM or
2443 (rs6000_option_override_internal): Rework setting long double
2444 format. Only warn if the user is changing the long double default
2445 and they did not use -Wno-psabi.
2446 * doc/invoke.texi (PowerPC options): Update the documentation for
2447 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2449 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2450 Alan Hayward <alan.hayward@arm.com>
2451 David Sherwood <david.sherwood@arm.com>
2453 * rtl.h (wider_subreg_mode): New function.
2454 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2455 rather than an unsigned int *.
2456 * ira-color.c (regno_max_ref_width): Replace with...
2457 (regno_max_ref_mode): ...this new variable.
2458 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2459 Use wider_subreg_mode.
2460 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2461 rather than an unsigned int *.
2462 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2463 (process_alt_operands): Likewise.
2464 (invariant_p): Likewise.
2465 * lra-spills.c (assign_mem_slot): Likewise.
2466 (add_pseudo_to_slot): Likewise.
2467 * lra.c (collect_non_operand_hard_regs): Likewise.
2468 (add_regs_to_insn_regno_info): Likewise.
2469 * reload1.c (regno_max_ref_width): Replace with...
2470 (regno_max_ref_mode): ...this new variable.
2471 (reload): Update accordingly. Update call to
2472 ira_sort_regnos_for_alter_reg.
2473 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2474 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2475 (scan_paradoxical_subregs): Likewise.
2477 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2479 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2480 (aarch64_frame): Add emit_frame_chain boolean.
2481 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2482 Move eh_return case to aarch64_layout_frame.
2483 (aarch64_layout_frame): Initialize emit_frame_chain.
2484 (aarch64_expand_prologue): Use emit_frame_chain.
2486 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2488 * config/aarch64/aarch64.c (aarch64_layout_frame):
2489 Ensure LR is always stored at the bottom of the callee-saves.
2490 Remove rarely used frame layout which saves callee-saves at top of
2491 frame, so the store of LR can be used as a valid probe in all cases.
2493 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2495 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2496 Improve unaligned TImode/TFmode base/offset split.
2498 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2499 Alan Hayward <alan.hayward@arm.com>
2500 David Sherwood <david.sherwood@arm.com>
2502 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2503 * combine.c (find_single_use_1): Likewise.
2504 (expand_field_assignment): Likewise.
2505 (move_deaths): Likewise.
2506 * lra-constraints.c (simplify_operand_subreg): Likewise.
2507 (curr_insn_transform): Likewise.
2508 * lra.c (collect_non_operand_hard_regs): Likewise.
2509 (add_regs_to_insn_regno_info): Likewise.
2510 * rtlanal.c (reg_referenced_p): Likewise.
2511 (covers_regno_no_parallel_p): Likewise.
2513 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2515 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2516 Don't print any bits outside the precision of the value.
2517 * wide-int.cc (test_printing): Add some new tests.
2519 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2521 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2522 supports -xbrace_comment option.
2523 * configure: Regenerate.
2524 * config.in: Regenerate.
2525 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2526 (ASM_CPU_SPEC): Use it.
2528 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2530 * target.def (static_rtx_alignment): New hook.
2531 * targhooks.h (default_static_rtx_alignment): Declare.
2532 * targhooks.c (default_static_rtx_alignment): New function.
2533 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2534 * doc/tm.texi: Regenerate.
2535 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2536 instead of targetm.constant_alignment. Remove call to
2538 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2539 (cris_preferred_mininum_alignment): New function, split out from...
2540 (cris_constant_alignment): ...here.
2541 (cris_static_rtx_alignment): New function.
2542 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2544 (ix86_constant_alignment): ...here.
2545 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2546 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2547 (mmix_static_rtx_alignment): New function.
2548 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2549 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2551 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2554 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2555 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2557 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2559 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2560 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2561 mark as a sign-extending load.
2562 (local_pic_load_u): Define.
2564 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2567 * fold-const.c (operand_equal_for_comparison_p): Also return true
2568 if ARG0 is a simple variant of ARG1 with narrower precision.
2569 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2571 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2573 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2575 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2576 scatter_static, scatter_per_elt.
2577 * x86-tune-costs.h: Add new cost entries.
2579 2017-10-25 Richard Biener <rguenther@suse.de>
2581 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2582 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2583 class pass_fre): Move to ...
2584 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2585 class pass_fre): ... here and adjust for statistics.
2587 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2590 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2591 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2592 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2593 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2596 2017-10-25 Richard Biener <rguenther@suse.de>
2598 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2599 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2600 eliminate_push_avail, eliminate_insert): Move inside...
2601 (class eliminate_dom_walker): ... this class in preparation
2603 (fini_eliminate): Remove by merging with ...
2604 (eliminate): ... this function. Adjust for class changes.
2605 (pass_pre::execute): Remove fini_eliminate call.
2606 (pass_fre::execute): Likewise.
2608 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2611 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2612 (VPERMI2, VPERMI2I): New mode iterators.
2613 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2614 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2616 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2617 mode iterator. Remove 3 old define_insn patterns.
2618 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2619 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2620 VPERMI2 mode iterator, remove the other two expanders.
2621 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2622 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2623 instructions, remove the other two patterns.
2624 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2625 mode iterator, remove the other two patterns.
2626 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2627 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2628 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2629 and adjust argument order accordingly.
2630 (ix86_expand_vec_perm): Adjust caller.
2631 (expand_vec_perm_1): Likewise.
2632 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2633 (expand_vec_perm_vpermt2_vpshub2): ... this.
2634 (ix86_expand_vec_perm_const_1): Adjust caller.
2635 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2638 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2639 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2640 (vec_shl_<mode>): Remove unused expander.
2641 (avx512bw_<shift_insn><mode>3): New define_insn.
2642 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2643 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2645 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2648 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2651 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2653 PR rtl-optimization/82396
2654 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2655 (autopref_multipass_init): Simplify initialization.
2656 (autopref_rank_data): Simplify sort order.
2657 * gcc/sched-int.h (autopref_multipass_data_): Remove
2658 multi_mem_insn_p, min_offset and max_offset.
2660 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2663 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2664 Check special value of flag_omit_frame_pointer.
2665 (aarch64_can_eliminate): Likewise.
2666 (aarch64_override_options_after_change_1): Simplify handling of
2667 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2669 2017-10-24 Richard Biener <rguenther@suse.de>
2671 PR tree-optimization/82697
2672 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2673 zero for conditional load and unconditional store.
2675 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2677 * doc/install.texi: Document bootstrap-cet.
2679 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2682 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2683 ENDBR instruction at function entrance if function is only
2686 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2689 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2690 patterns to better describe from which operation the CF is computed.
2691 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2692 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2693 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2694 is 0, use _0 suffixed expanders instead of emitting a comparison
2697 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2699 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2700 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2702 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2705 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2706 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2707 * loop-iv.c (iv_get_reaching_def): Likewise.
2708 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2709 variable is promoted and the partition contains undefined values.
2711 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2713 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2715 (nios2_address_cost): Define.
2716 (nios2_legitimize_address): Recognize (exp + constant) directly.
2717 (TARGET_ADDRESS_COST): Define.
2719 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2721 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2722 (nios2_symbolic_memory_operand_p): Declare.
2723 (nios2_split_large_constant): Declare.
2724 (nios2_split_symbolic_memory_operand): Declare.
2725 * config/nios2/nios2.c: Adjust includes.
2726 (nios2_symbolic_constant_allowed): New.
2727 (nios2_symbolic_constant_p): New.
2728 (nios2_plus_symbolic_constant_p): New.
2729 (nios2_valid_addr_expr_p): Recognize addresses involving
2731 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2732 (nios2_symbolic_memory_operand_p): New.
2733 (nios2_large_constant_p): New.
2734 (nios2_split_large_constant): New.
2735 (nios2_split_plus_large_constant): New.
2736 (nios2_split_symbolic_memory_operand): New.
2737 (nios2_legitimize_address): Code refactoring. Handle addresses
2738 involving symbolic constants.
2739 (nios2_emit_move_sequence): Likewise.
2740 (nios2_print_operand): Improve error output.
2741 (nios2_print_operand_address): Handle LO_SUM.
2742 (nios2_cdx_narrow_form_p): Likewise.
2743 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2744 operands involving symbolic constants.
2745 (movhi_internal, movsi_internal): Likewise.
2746 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2747 (extendhisi2, extendqi<mode>2): Likewise.
2749 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2751 * tree-pass.h (PROP_rtl_split_insns): Define.
2752 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2754 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2756 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2758 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2761 * target.def (const_not_ok_for_debug_p): Default to
2762 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2763 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2764 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2765 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2766 which targetm.const_not_ok_for_debug_p returned true.
2767 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2769 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2771 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2772 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2773 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2775 (ix86_const_not_ok_for_debug_p): New function.
2776 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2777 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2779 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2782 * system.h: Conditionally include "unique-ptr.h" if
2783 INCLUDE_UNIQUE_PTR is defined.
2784 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2785 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2787 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2789 * config/rl78/rl78.md: New define_expand "subdi3".
2791 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2794 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2795 DF_REF_INSN if DF_REF_INSN_INFO is false.
2797 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2799 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2800 xmm_move instead of sse_move.
2801 (sse_store_index): New function.
2802 (ix86_register_move_cost): Be more sensible about mismatch stall;
2803 model AVX moves correctly; make difference between sse->integer and
2805 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2806 moves; make difference between SSE and AVX.
2807 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2808 and zmm_move. Increase size of sse load and store tables;
2809 add unaligned load and store tables; add ssemmx_to_integer.
2810 * x86-tune-costs.h: Update all entries according to real
2811 move latencies from Agner Fog's manual and chip documentation.
2813 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2816 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2817 * config/i386/constraints.md (Wf): New constraint.
2818 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2819 (cmp<dwi>_doubleword): Removed.
2820 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2821 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2822 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2823 expand with cmp<dwi>_doubleword. For LTU and GEU use
2824 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2826 * common.opt (gcolumn-info): Enable by default.
2827 * doc/invoke.texi (gcolumn-info): Document new default.
2829 2017-10-23 Richard Biener <rguenther@suse.de>
2831 PR tree-optimization/82672
2832 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2833 Fold the stmt if we propagated into it.
2835 2017-10-23 Richard Biener <rguenther@suse.de>
2837 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2838 (bitmap_remove_expr_from_set): ... this. All callers call this
2839 for non-constant values.
2840 (bitmap_set_subtract): Rename to...
2841 (bitmap_set_subtract_expressions): ... this. Adjust and
2843 (bitmap_set_contains_value): Remove superfluous check.
2844 (bitmap_set_replace_value): Inline into single caller ...
2845 (bitmap_value_replace_in_set): ... here and simplify.
2846 (dependent_clean): Merge into ...
2847 (clean): ... this using an overload. Adjust.
2848 (prune_clobbered_mems): Adjust.
2849 (compute_antic_aux): Likewise.
2850 (compute_partial_antic_aux): Likewise.
2852 2017-10-23 Richard Biener <rguenther@suse.de>
2854 PR tree-optimization/82129
2856 2017-08-01 Richard Biener <rguenther@suse.de>
2858 PR tree-optimization/81181
2859 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2860 (compute_antic): ... end of iteration here.
2862 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2864 * target.def (starting_frame_offset): New hook.
2865 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2866 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2867 * doc/tm.texi.in: Regenerate.
2868 * hooks.h (hook_hwi_void_0): Declare.
2869 * hooks.c (hook_hwi_void_0): New function.
2870 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2871 STARTING_FRAME_OFFSET.
2872 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2873 * reload1.c (reload): Likewise.
2874 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2875 instead of STARTING_FRAME_OFFSET.
2876 * function.c (try_fit_stack_local): Likewise.
2877 (assign_stack_local_1): Likewise
2878 (instantiate_virtual_regs): Likewise.
2879 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2880 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2881 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2882 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2883 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2884 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2885 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2886 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2887 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2888 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2889 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2890 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2891 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2892 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2893 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2894 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2895 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2896 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2897 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2898 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2899 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2900 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2901 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2902 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2903 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2904 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2905 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2906 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2907 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2908 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2909 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2910 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2911 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2912 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2913 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2914 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2915 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2916 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2917 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2918 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2919 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2920 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2921 return a HOST_WIDE_INT.
2922 (avr_builtin_setjmp_frame_value): Use it instead of
2923 STARTING_FRAME_OFFSET.
2924 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2925 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2926 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2928 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2929 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2930 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2931 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2932 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2933 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2934 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2935 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2936 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2937 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2938 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2939 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2941 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2942 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2943 * config/mips/mips.c (mips_compute_frame_info): Refer to
2944 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2945 (mips_starting_frame_offset): New function.
2946 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2947 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2948 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2949 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2950 and return a HOST_WIDE_INT.
2951 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2952 (mmix_initial_elimination_offset): Refer to
2953 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2954 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2955 * config/pa/pa.c (pa_starting_frame_offset): New function.
2956 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2957 (pa_expand_prologue): Likewise.
2958 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2959 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2960 !FRAME_GROWS_DOWNWARD handling to...
2961 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2962 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2963 !FRAME_GROWS_DOWNWARD handling to...
2964 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2965 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2966 !FRAME_GROWS_DOWNWARD handling to...
2967 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2968 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2970 (rs6000_starting_frame_offset): New function.
2971 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2972 !FRAME_GROWS_DOWNWARD handling to...
2973 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2974 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2975 !FRAME_GROWS_DOWNWARD handling to...
2976 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2977 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2978 !FRAME_GROWS_DOWNWARD handling to...
2979 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2980 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2981 (rs6000_starting_frame_offset): New function.
2982 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2983 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2984 * config/vax/vax.c (vax_starting_frame_offset): New function.
2985 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2986 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2987 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2988 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2989 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2990 * system.h (STARTING_FRAME_OFFSET): Poison.
2992 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2994 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2995 SCALAR_TYPE_MODE instead of 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 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
3003 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3004 Alan Hayward <alan.hayward@arm.com>
3005 David Sherwood <david.sherwood@arm.com>
3007 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
3009 2017-10-23 Richard Biener <rguenther@suse.de>
3011 PR tree-optimization/82129
3012 * tree-ssa-pre.c (bitmap_set_and): Remove.
3013 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
3014 canonicalizing expressions in the set to those with lowest
3015 ID rather than taking that from the first edge.
3017 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
3019 * combine.c (rtx_equal_for_field_assignment_p): Use
3020 byte_lowpart_offset.
3022 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3023 Alan Hayward <alan.hayward@arm.com>
3024 David Sherwood <david.sherwood@arm.com>
3026 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3027 to a SUBREG_PROMOTED_VAR.
3029 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3030 Alan Hayward <alan.hayward@arm.com>
3031 David Sherwood <david.sherwood@arm.com>
3033 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3034 (expand_debug_source_expr): Likewise.
3035 * combine.c (combine_simplify_rtx): Likewise.
3036 * cse.c (fold_rtx): Likewise.
3037 * optabs.c (expand_float): Likewise.
3038 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3039 (simplify_binary_operation_1): Likewise.
3041 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3042 Alan Hayward <alan.hayward@arm.com>
3043 David Sherwood <david.sherwood@arm.com>
3045 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3046 (record_promoted_value): Likewise.
3047 * expr.c (expand_expr_real_2): Likewise.
3048 * ree.c (update_reg_equal_equiv_notes): Likewise.
3049 (combine_set_extension): Likewise.
3050 * rtlanal.c (low_bitmask_len): Likewise.
3051 * simplify-rtx.c (neg_const_int): Likewise.
3052 (simplify_binary_operation_1): Likewise.
3054 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3055 Alan Hayward <alan.hayward@arm.com>
3056 David Sherwood <david.sherwood@arm.com>
3058 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3059 * regcprop.c (maybe_mode_change): Likewise.
3060 * reload1.c (alter_reg): Likewise.
3062 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3064 * inchash.h (inchash::hash::add_wide_int): New function.
3065 * lto-streamer-out.c (hash_tree): Use it.
3067 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3069 * inchash.h (inchash::hash::add_wide_int): Rename to...
3070 (inchash::hash::add_hwi): ...this.
3071 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3072 (polymorphic_call_target_hasher::hash): Likewise.
3073 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3074 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3075 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3076 * lto-streamer-out.c (hash_tree): Likewise.
3077 * optc-save-gen.awk: Likewise.
3078 * tree.c (add_expr): Likewise.
3080 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3083 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3084 for ordered inequality comparisons even with TARGET_IEEE_FP.
3086 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3089 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3090 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3091 Expand with cmp<dwi>_doubleword.
3093 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3095 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3096 List CET intrinsics.
3097 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3098 specific to -fcf-protection option.
3100 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3102 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3103 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3104 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3105 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3106 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3107 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3108 (extra_objs): Add cet.o for Linux/x86 targets.
3109 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3110 * config/i386/cet.c: New file.
3111 * config/i386/cetintrin.h: Likewise.
3112 * config/i386/t-cet: Likewise.
3113 * config/i386/cpuid.h (bit_SHSTK): New.
3114 (bit_IBT): Likewise.
3115 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3116 pass IBT and SHSTK bits.
3117 * config/i386/i386-builtin-types.def
3118 (VOID_FTYPE_UNSIGNED_PVOID): New.
3119 (VOID_FTYPE_UINT64_PVOID): Likewise.
3120 * config/i386/i386-builtin.def: Add CET intrinsics.
3121 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3122 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3123 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3124 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3126 * config/i386/i386.c (rest_of_insert_endbranch): New.
3127 (pass_data_insert_endbranch): Likewise.
3128 (pass_insert_endbranch): Likewise.
3129 (make_pass_insert_endbranch): Likewise.
3130 (ix86_notrack_prefixed_insn_p): Likewise.
3131 (ix86_target_string): Add -mibt, -mshstk flags.
3132 (ix86_option_override_internal): Add flag_cf_protection
3134 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3135 (ix86_print_operand): Add 'notrack' prefix output.
3136 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3137 (ix86_expand_builtin): Expand CET intrinsics.
3138 (x86_output_mi_thunk): Add 'endbranch' instruction.
3139 * config/i386/i386.h (TARGET_IBT): New.
3140 (TARGET_IBT_P): Likewise.
3141 (TARGET_SHSTK): Likewise.
3142 (TARGET_SHSTK_P): Likewise.
3143 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3144 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3145 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3146 (builtin_setjmp_setup): New pattern.
3147 (builtin_longjmp): Likewise.
3148 (rdssp<mode>): Likewise.
3149 (incssp<mode>): Likewise.
3150 (saveprevssp): Likewise.
3151 (rstorssp): Likewise.
3152 (wrss<mode>): Likewise.
3153 (wruss<mode>): Likewise.
3154 (setssbsy): Likewise.
3155 (clrssbsy): Likewise.
3156 (nop_endbr): Likewise.
3157 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3159 * config/i386/immintrin.h: Include <cetintrin.h>.
3160 * config/i386/linux-common.h
3161 (file_end_indicate_exec_stack_and_cet): New prototype.
3162 (TARGET_ASM_FILE_END): New.
3164 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3166 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3167 latencies instead of having separate table; make difference between
3168 integer and float costs.
3169 * i386.h (processor_costs): Remove scalar_stmt_cost,
3170 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3171 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3173 * x86-tune-costs.h: Remove entries which has been removed in
3174 procesor_costs from all tables; make cond_taken_branch_cost
3175 and cond_not_taken_branch_cost COST_N_INSNS based.
3177 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3179 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3181 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3183 * config/i386/i386.md (isa): Remove fma_avx512f.
3184 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3185 <avx512>_fmadd_<mode>_mask3<round_name>,
3186 <avx512>_fmsub_<mode>_mask<round_name>,
3187 <avx512>_fmsub_<mode>_mask3<round_name>,
3188 <avx512>_fnmadd_<mode>_mask<round_name>,
3189 <avx512>_fnmadd_<mode>_mask3<round_name>,
3190 <avx512>_fnmsub_<mode>_mask<round_name>,
3191 <avx512>_fnmsub_<mode>_mask3<round_name>,
3192 <avx512>_fmaddsub_<mode>_mask<round_name>,
3193 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3194 <avx512>_fmsubadd_<mode>_mask<round_name>,
3195 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3196 (*vec_widen_umult_even_v16si<mask_name>,
3197 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3198 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3200 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3202 * extend.texi: Add 'nocf_check' documentation.
3203 * gimple.texi: Add second parameter to
3204 gimple_build_call_from_tree.
3205 * invoke.texi: Add -fcf-protection documentation.
3206 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3208 2017-10-20 Richard Biener <rguenther@suse.de>
3210 PR tree-optimization/82473
3211 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3212 the largest input type.
3214 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3216 * c-attribs.c (handle_nocf_check_attribute): New function.
3217 (c_common_attribute_table): Add 'nocf_check' handling.
3218 * gimple-parser.c: Add second argument NULL to
3219 gimple_build_call_from_tree.
3220 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3221 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3223 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3224 * common.opt: Add fcf-protection flag.
3225 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3226 * flag-types.h: Add enum cf_protection_level.
3227 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3228 Add 'nocf_check' attribute propagation to gimple call.
3229 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3230 (gimple_build_call_from_tree): Update prototype.
3231 (gimple_call_nocf_check_p): New function.
3232 (gimple_call_set_nocf_check): Likewise.
3233 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3234 * ipa-icf.c: Add nocf_check attribute in statement hash.
3235 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3236 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3237 * toplev.c (process_options): Add flag_cf_protection handling.
3239 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3241 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3243 2017-10-20 Richard Biener <rguenther@suse.de>
3245 PR tree-optimization/82603
3246 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3247 remove false predicated stores.
3249 2017-10-20 Richard Biener <rguenther@suse.de>
3251 * graphite-isl-ast-to-gimple.c
3252 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3253 Remove return value and simplify, dump copied stmt after lhs
3255 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3256 Reduce dump verbosity.
3257 (gsi_insert_earliest): Likewise.
3258 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3259 * graphite.c (print_global_statistics): Adjust dumping.
3260 (print_graphite_scop_statistics): Likewise.
3261 (print_graphite_statistics): Do not dump loops here.
3262 (graphite_transform_loops): But here.
3264 2017-10-20 Nicolas Roche <roche@adacore.com>
3266 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3267 * configure: Regenerate.
3269 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3272 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3273 functions when optimizing replace GIMPLE_RETURN stmts with
3274 calls to __builtin_unreachable ().
3277 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3278 for -fsanitize=thread link of executables.
3279 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3280 link of executables.
3283 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3285 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3286 define_insns for logical vector shifts to use VI248_AVX512BW
3287 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3288 condition, useless isa and prefix attributes. Change the first
3289 2 of these define_insns to ...
3290 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3291 define_insn for avx512vl.
3292 (<shift_insn><mode>3): ... and this, new define_insn without
3293 masking for non-avx512vl.
3296 * config/i386/sse.md (*andnot<mode>3,
3297 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3298 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3299 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3300 not applied use empty suffix even for TARGET_AVX512VL.
3301 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3302 is applied, supply evex,evex or evex,evex,evex instead of just
3305 2017-10-20 Julia Koval <julia.koval@intel.com>
3307 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3308 (OPTION_MASK_ISA_GFNI_UNSET): New.
3309 (ix86_handle_option): Handle OPT_mgfni.
3310 * config/i386/cpuid.h (bit_GFNI): New.
3311 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3312 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3313 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3314 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3315 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3316 * config/i386/i386.opt: Add mgfni.
3318 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3320 * config/msp430/msp430.c (msp430_option_override): Disable
3321 -fdelete-null-pointer-checks.
3322 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3324 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3326 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3327 of x87 and SSE instructions.
3329 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3331 * asan.c (create_cond_insert_point): Do not update edge count.
3332 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3333 (afdo_propagate_circuit): Likewise.
3334 (afdo_calculate_branch_prob): Likewise.
3335 (afdo_annotate_cfg): Likewise.
3336 * basic-block.h (struct edge_def): Remove count.
3337 (edge_def::count): New accessor.
3338 * bb-reorder.c (rotate_loop): Update.
3339 (find_traces_1_round): Update.
3340 (connect_traces): Update.
3341 (sanitize_hot_paths): Update.
3342 * cfg.c (unchecked_make_edge): Update.
3343 (make_single_succ_edge): Update.
3344 (check_bb_profile): Update.
3345 (dump_edge_info): Update.
3346 (update_bb_profile_for_threading): Update.
3347 (scale_bbs_frequencies_int): Update.
3348 (scale_bbs_frequencies_gcov_type): Update.
3349 (scale_bbs_frequencies_profile_count): Update.
3350 (scale_bbs_frequencies): Update.
3351 * cfganal.c (connect_infinite_loops_to_exit): Update.
3352 * cfgbuild.c (compute_outgoing_frequencies): Update.
3353 (find_many_sub_basic_blocks): Update.
3354 * cfgcleanup.c (try_forward_edges): Update.
3355 (try_crossjump_to_edge): Update
3356 * cfgexpand.c (expand_gimple_cond): Update
3357 (expand_gimple_tailcall): Update
3358 (construct_exit_block): Update
3359 * cfghooks.c (verify_flow_info): Update
3360 (redirect_edge_succ_nodup): Update
3361 (split_edge): Update
3362 (make_forwarder_block): Update
3363 (duplicate_block): Update
3364 (account_profile_record): Update
3365 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3366 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3367 * cfgloopmanip.c (scale_loop_profile): Update.
3369 (lv_adjust_loop_entry_edge): Update.
3370 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3371 (force_nonfallthru_and_redirect): Update.
3372 (purge_dead_edges): Update.
3373 (rtl_flow_call_edges_add): Update.
3374 * cgraphunit.c (init_lowered_empty_function): Update.
3375 (cgraph_node::expand_thunk): Update.
3376 * gimple-pretty-print.c (dump_probability): Update.
3377 (dump_edge_probability): Update.
3378 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3379 * haifa-sched.c (sched_create_recovery_edges): Update.
3380 * hsa-gen.c (convert_switch_statements): Update.
3381 * ifcvt.c (dead_or_predicable): Update.
3382 * ipa-inline-transform.c (inline_transform): Update.
3383 * ipa-split.c (split_function): Update.
3384 * ipa-utils.c (ipa_merge_profiles): Update.
3385 * loop-doloop.c (add_test): Update.
3386 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3387 * lto-streamer-in.c (input_cfg): Update.
3388 (input_function): Update.
3389 * lto-streamer-out.c (output_cfg): Update.
3390 * modulo-sched.c (sms_schedule): Update.
3391 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3392 * predict.c (maybe_hot_edge_p): Update.
3393 (unlikely_executed_edge_p): Update.
3394 (probably_never_executed_edge_p): Update.
3395 (dump_prediction): Update.
3396 (drop_profile): Update.
3397 (propagate_unlikely_bbs_forward): Update.
3398 (determine_unlikely_bbs): Update.
3399 (force_edge_cold): Update.
3400 * profile.c (compute_branch_probabilities): Update.
3401 * reg-stack.c (better_edge): Update.
3402 * shrink-wrap.c (handle_simple_exit): Update.
3403 * tracer.c (better_p): Update.
3404 * trans-mem.c (expand_transaction): Update.
3405 (split_bb_make_tm_edge): Update.
3406 * tree-call-cdce.c: Update.
3407 * tree-cfg.c (gimple_find_sub_bbs): Update.
3408 (gimple_split_edge): Update.
3409 (gimple_duplicate_sese_region): Update.
3410 (gimple_duplicate_sese_tail): Update.
3411 (gimple_flow_call_edges_add): Update.
3412 (insert_cond_bb): Update.
3413 (execute_fixup_cfg): Update.
3414 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3415 * tree-complex.c (expand_complex_div_wide): Update.
3416 * tree-eh.c (lower_resx): Update.
3417 (unsplit_eh): Update.
3418 (cleanup_empty_eh_move_lp): Update.
3419 * tree-inline.c (copy_edges_for_bb): Update.
3420 (freqs_to_counts): Update.
3421 (copy_cfg_body): Update.
3422 * tree-ssa-dce.c (remove_dead_stmt): Update.
3423 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3424 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3425 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3426 (unloop_loops): Update.
3427 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3428 * tree-ssa-loop-split.c (connect_loops): Update.
3429 (split_loop): Update.
3430 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3431 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3432 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3433 * tree-ssa-reassoc.c (branch_fixup): Update.
3434 * tree-ssa-tail-merge.c (replace_block_by): Update.
3435 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3436 (compute_path_counts): Update.
3437 (update_profile): Update.
3438 (recompute_probabilities): Update.
3439 (update_joiner_offpath_counts): Update.
3440 (estimated_freqs_path): Update.
3441 (freqs_to_counts_path): Update.
3442 (clear_counts_path): Update.
3443 (ssa_fix_duplicate_block_edges): Update.
3444 (duplicate_thread_path): Update.
3445 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3446 (case_bit_test_cmp): Update.
3447 (collect_switch_conv_info): Update.
3448 (gen_inbound_check): Update.
3449 (do_jump_if_equal): Update.
3450 (emit_cmp_and_jump_insns): Update.
3451 * tree-tailcall.c (decrease_profile): Update.
3452 (eliminate_tail_call): Update.
3453 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3454 (vect_do_peeling): Update.
3455 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3456 * ubsan.c (ubsan_expand_null_ifn): Update.
3457 (ubsan_expand_ptr_ifn): Update.
3458 * value-prof.c (gimple_divmod_fixed_value): Update.
3459 (gimple_mod_pow2): Update.
3460 (gimple_mod_subtract): Update.
3461 (gimple_ic): Update.
3462 (gimple_stringop_fixed_value): Update.
3464 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3467 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3469 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3471 PR rtl-optimization/82395
3472 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3473 based on non_spilled_static_chain_regno_p.
3475 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3477 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3478 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3479 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3481 2017-10-19 Martin Sebor <msebor@redhat.com>
3483 PR tree-optimization/82596
3484 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3486 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3488 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3489 (get_mem_refs_of_builtin_call): Likewise.
3490 * builtins.c (expand_builtin_apply): Adjust call to
3491 allocate_dynamic_stack_space.
3492 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3493 the third argument to allocate_dynamic_stack_space, otherwise -1.
3494 (expand_builtin): Deal with all alloca variants.
3495 (is_inexpensive_builtin): Likewise.
3496 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3497 * calls.c (special_function_p): Deal with all alloca variants.
3498 (initialize_argument_information): Adjust call to
3499 allocate_dynamic_stack_space.
3500 (expand_call): Likewise.
3501 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3502 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3503 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3504 use it for the stack usage computation.
3505 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3506 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3507 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3508 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3509 (in_loop_p): Remove first argument and useless check.
3510 (pass_walloca::execute): Remove useless test and adjust call to above.
3511 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3512 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3513 (gimplify_call_expr): Deal with all alloca variants.
3514 * hsa-gen.c (gen_hsa_alloca): Likewise.
3515 (gen_hsa_insns_for_call): Likewise.
3516 * ipa-pure-const.c (special_builtin_state): Likewise.
3517 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3518 * tree-object-size.c (alloc_object_size): Likewise.
3519 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3520 (call_may_clobber_ref_p_1): Likewise.
3521 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3522 (ccp_fold_stmt): Likewise.
3523 (optimize_stack_restore): Likewise.
3524 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3525 (mark_all_reaching_defs_necessary_1): Likewise.
3526 (propagate_necessity): Likewise.
3527 (eliminate_unnecessary_stmts): Likewise.
3528 * tree.c (build_common_builtin_nodes): Build
3529 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3530 (build_alloca_call_expr): New function.
3531 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3532 (CASE_BUILT_IN_ALLOCA): Likewise.
3533 (build_alloca_call_expr): Declare.
3534 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3536 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3539 * dwarf2out.c (new_die_raw): New static inline function.
3540 (new_die): Use it to create the DIE.
3541 (add_AT_external_die_ref): Likewise.
3542 (clone_die): Likewise.
3543 (clone_as_declaration): Likewise.
3544 (dwarf2out_vms_debug_main_pointer): Likewise.
3545 (base_type_die): Likewise. Remove early return for corner cases.
3546 Do not call add_pubtype on the DIE here.
3547 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3548 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3549 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3550 native order exists for base types, attach the DIE manually and call
3551 add_pubtype on it. Do not equate a reverse order DIE to the type.
3553 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3555 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3556 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3558 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3560 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3562 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3563 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3564 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3565 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3566 * opts.c (sanitizer_opts): Add builtin.
3567 * ubsan.c (instrument_builtin): New function.
3568 (pass_ubsan::execute): Call it.
3569 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3570 * doc/invoke.texi: Document -fsanitize=builtin.
3572 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3573 builtins, store max (log2 (align), 0) into uchar field instead of
3574 align into uptr field.
3575 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3576 store uchar 0 field instead of uptr 0 field.
3577 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3578 instead of passing one address of struct with 2 locations pass
3579 two addresses of structs with 1 location each.
3580 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3581 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3582 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3583 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3584 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3585 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3586 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3587 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3589 2017-10-19 Martin Liska <mliska@suse.cz>
3592 * file-find.c (remove_prefix): Remove.
3593 * file-find.h (remove_prefix): Likewise.
3594 * gcc-ar.c: Remove smartness of lookup.
3596 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3598 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3599 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3600 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3602 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3605 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3606 (setcc + and to xor + setcc): New peephole2.
3608 2017-10-19 Tom de Vries <tom@codesourcery.com>
3610 * doc/sourcebuild.texi (Test Directives, Variants of
3611 dg-require-support): Add dg-require-stack-size.
3613 2017-10-19 Martin Liska <mliska@suse.cz>
3616 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3617 that have a large alignment.
3618 (gimplify_target_expr): Likewise.
3620 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3622 PR rtl-optimization/82602
3623 * ira.c (rtx_moveable_p): Return false for volatile asm.
3625 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3628 * config/i386/i386-modes.def (CCGZ): New CC mode.
3629 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3630 * config/i386/predicates.md (ix86_comparison_operator):
3632 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3633 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3634 with double-word subtraction.
3635 (put_condition_code): Handle CCGZmode.
3637 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3639 * wide-int.cc (debug (const wide_int &)): New.
3640 (debug (const wide_int *)): New.
3641 (debug (const widest_int &)): New.
3642 (debug (const widest_int *)): New.
3644 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3647 * lra-constraints.c (curr_insn_transform): Use non-input operand
3648 instead of output one for matched reload.
3650 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3652 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3653 (tree-ssa-loop-ivopts.h): New header file.
3654 (struct builtin_info): New fields.
3655 (classify_builtin_1): Compute and record base and offset parts for
3656 memset builtin partition by calling strip_offset.
3657 (offset_cmp, fuse_memset_builtins): New functions.
3658 (finalize_partitions): Fuse adjacent memset partitions by calling
3660 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3661 Expose the interface.
3662 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3664 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3666 PR tree-optimization/82574
3667 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3668 that data reference must be executed exactly once per iteration
3669 against the outermost loop in nest.
3670 (classify_partition): Update call to above function.
3672 2017-10-18 Richard Biener <rguenther@suse.de>
3674 PR tree-optimization/82591
3675 * graphite.c (graphite_transform_loops): Move code gen message
3677 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3678 Here. Handle scop_to_isl_ast failing.
3679 (scop_to_isl_ast): Limit the number of ISL operations.
3681 2017-10-18 Richard Biener <rguenther@suse.de>
3683 * graphite-isl-ast-to-gimple.c
3684 (translate_isl_ast_to_gimple::set_rename): Simplify.
3685 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3686 (graphite_copy_stmts_from_block): ... here.
3687 (copy_bb_and_scalar_dependences): Simplify.
3688 (add_parameters_to_ivs_params): Canonicalize.
3689 (generate_entry_out_of_ssa_copies): Simplify.
3690 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3691 by passing in ISL dimension.
3692 (parameter_index_in_region_1): Rename to ...
3693 (parameter_index_in_region): ... this.
3694 (extract_affine): Adjust assert, pass down parameter index.
3695 (add_param_constraints): Use range-info when available.
3696 (build_scop_context): Adjust.
3697 * sese.c (new_sese_info): Adjust.
3698 (free_sese_info): Likewise.
3699 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3700 Remove unused typedefs.
3701 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3703 2017-10-18 Martin Liska <mliska@suse.cz>
3705 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3707 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3709 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3710 (s390_sched_init): Do not reset s390_sched_state if we entered the
3711 current basic block via a fallthru edge and all others are unlikely.
3713 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3715 * config/s390/s390.c (NUM_SIDES): New variable.
3716 (LONGRUNNING_THRESHOLD): New variable.
3717 (LATENCY_FACTOR): New variable.
3718 (s390_sched_score): Decrease score for long-running instructions on
3720 (s390_sched_variable_issue): Perform bookkeeping for long-running
3723 2017-10-18 Richard Biener <rguenther@suse.de>
3725 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3726 Simplify with removal of the parameter rename map.
3727 (set_rename): Likewise.
3728 (should_copy_to_new_region): Likewise.
3729 (graphite_copy_stmts_from_block): Likewise.
3730 (copy_bb_and_scalar_dependences): Remove initialization of
3731 unused copied_bb_map.
3733 (copy_internal_parameters): Likewise.
3734 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3735 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3736 Use INTEGRAL_TYPE_P.
3737 (parameter_index_in_region_1): Rename to ...
3738 (assign_parameter_index_in_region): ... this. Assert we have
3739 a parameter we handle.
3740 (scan_tree_for_params): Adjust.
3741 * sese.h (parameter_rename_map_t): Remove.
3742 (struct sese_info_t): Remove unused parameter_rename_map and
3743 copied_bb_map members.
3744 * sese.c (new_sese_info): Adjust.
3745 (free_sese_info): Likewise.
3747 2017-10-18 Martin Liska <mliska@suse.cz>
3750 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3751 on an abnormal edge.
3753 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3755 * doc/invoke.texi (ffunction-sections and fdata-sections):
3758 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3760 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3761 the use statement can throw internally.
3763 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3765 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3766 any RTX present on the RHS of a SET.
3767 * compare-elim.c (try_eliminate_compare): Restore comment.
3769 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3771 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3772 may be also called on tcc_type nodes.
3773 * langhooks.c (lhd_tree_size): Likewise.
3775 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3777 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3778 format_warning_at_substring.
3779 (maybe_warn): Convert source_range * param to a location_t. Pass
3780 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3781 (format_directive): Remove code to extract source_ranges and
3782 source_range * in favor of just a location_t.
3783 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3785 * substring-locations.c (format_warning_va): Convert
3786 source_range * param to a location_t.
3787 (format_warning_at_substring): Likewise.
3788 * substring-locations.h (format_warning_va): Likewise.
3789 (format_warning_at_substring): Likewise.
3791 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3793 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3795 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3796 and scatter/gather ops.
3798 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3799 vec_gather_load and vec_scatter_store.
3800 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3801 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3802 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3803 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3804 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3805 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3807 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3809 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3810 Detect FTST instruction and handle its register pops. Only pop
3811 second operand if can_pop_second_op is true.
3812 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3813 set can_pop_second_op to false in the compare_for_stack_reg call.
3815 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3816 output_fp_compare for stack register operands.
3817 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3818 instructions here. Do not emit stack register pops here. Assert
3819 that FCOMPP pops next to top stack register. Rewrite function.
3821 2017-10-17 Nathan Sidwell <nathan@acm.org>
3824 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3825 use DECL_ASSEMBLER_NAME_RAW.
3828 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3831 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3832 Wilco Dijkstra <wilco.dijkstra@arm.com>
3834 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3835 converstion to Pmode from the buf_addr.
3837 2017-10-17 Richard Biener <rguenther@suse.de>
3839 * graphite-dependences.c (scop_get_reads_and_writes): Change
3840 output parameters to references.
3842 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3844 PR 71026/tree-optimization
3845 * fold-const.c (distribute_real_division): Removed.
3846 (fold_binary_loc): Remove calls to distribute_real_divison.
3848 2017-10-17 Richard Biener <rguenther@suse.de>
3850 * graphite-scop-detection.c
3851 (scop_detection::stmt_has_simple_data_refs_p): Always use
3852 the full nest as region.
3853 (try_generate_gimple_bb): Likewise.
3854 * sese.c (scalar_evolution_in_region): Simplify now that
3855 SCEV can handle instantiation in regions.
3856 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3857 in the non-loop part of a function if requested.
3859 2017-10-17 Richard Biener <rguenther@suse.de>
3861 PR tree-optimization/82563
3862 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3864 (graphite_regenerate_ast_isl): Call it.
3865 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3867 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3869 PR tree-optimization/82549
3870 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3871 Formatting fixes. Instead of calling make_bit_field_ref with negative
3874 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3876 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3877 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3878 _mm_maskz_reduce_ss): New.
3879 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3880 __builtin_ia32_reducess_mask): Ditto..
3881 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3882 * config/i386/sse.md (reduces<mode>): Renamed to ...
3883 (reduces<mode><mask_scalar_name>): ... this.
3884 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3886 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3887 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3889 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3891 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3892 * selftest-run-tests.c (selftest::run_tests): Call
3893 selftest::unique_ptr_tests_cc_tests.
3894 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3895 * unique-ptr-tests.cc: New file.
3897 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3900 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3902 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3903 (make_hard_regno_born, make_hard_regno_dead): Update
3904 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3906 2017-10-16 Jeff Law <law@redhat.com>
3908 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3910 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3912 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3914 2017-10-16 Olivier Hainque <hainque@adacore.com>
3916 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3917 with_cpu if we were configured for an e500v2 target cpu name.
3919 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3921 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3922 * doc/invoke.texi: Document +nodsp as a valid extension for
3925 2017-10-16 Martin Liska <mliska@suse.cz>
3927 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3928 (test_set_range): Likewise.
3929 (test_range_functions): Rename to ...
3930 (test_bit_in_range): ... this.
3931 (sbitmap_c_tests): Add new test.
3933 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3935 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3937 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3938 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3940 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3942 * config/aarch64/aarch64-builtins.c
3943 (aarch64_types_quadopu_lane_qualifiers): New.
3944 (TYPES_QUADOPU_LANE): New.
3945 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3946 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3947 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3948 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3949 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3950 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3951 (Vdottype, DOTPROD): New.
3952 (sur): Add SDOT and UDOT.
3954 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3956 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3957 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3958 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3960 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3961 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3962 Enable TARGET_DOTPROD.
3963 (cortex-a75.cortex-a55): Likewise.
3964 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3966 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3968 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3969 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3971 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3973 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3974 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3975 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3976 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3977 * config/arm/types.md (neon_dot, neon_dot_q): New.
3978 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3980 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3982 * config/arm/arm.h (TARGET_DOTPROD): New.
3983 * config/arm/arm.c (arm_arch_dotprod): New.
3984 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3985 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3986 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3987 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3988 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3989 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3990 * doc/invoke.texi (armv8.2-a): Document dotprod
3992 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3994 * i386.c (ix86_vec_cost): New function.
3995 (ix86_rtx_costs): Handle vector operations better.
3996 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3997 * x86-tune-costs.h: Add new costs to all tables.
3999 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
4001 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
4003 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
4004 divsd, sqrtss and sqrtsd
4005 * x86-tune-costs.h: Add new entries to all costs.
4006 (znver1_cost): Fix to match real instruction latencies.
4008 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4009 Michael Collison <michael.collison@arm.com>
4011 * compare-elim.c: Include emit-rtl.h.
4012 (can_merge_compare_into_arith): New function.
4013 (try_validate_parallel): Likewise.
4014 (try_merge_compare): Likewise.
4015 (try_eliminate_compare): Call the above when no previous clobber
4017 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
4020 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4024 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
4025 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4029 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4030 to be any operand_equal_p operands. For & (B - 1) require
4031 B to be power of 2. Recognize
4032 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4034 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4037 * optabs.c (expand_memory_blockage): Fix call of
4038 targetm.have_memory_blockage.
4040 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4043 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4044 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4045 objects to extra_objs instead of overwriting it.
4047 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4049 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4050 Use any_fp_register_operand as operand[3] predicate. Simplify
4051 equality test for operands[2] and operands[4] memory location.
4052 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4053 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4054 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4055 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4056 any_fp_register_operand as operand[1] predicate. Simplify
4057 equality test for operands[0] and operands[3] memory location.
4058 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4059 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4060 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4062 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4064 * target-insns.def: Add memory_blockage.
4065 * optabs.c (expand_memory_blockage): New function.
4066 (expand_asm_memory_barrier): Rename ...
4067 (expand_asm_memory_blockage): ... to this.
4068 (expand_mem_thread_fence): Call expand_memory_blockage
4069 instead of expand_asm_memory_barrier.
4070 (expand_mem_singnal_fence): Ditto.
4071 (expand_atomic_load): Ditto.
4072 (expand_atomic_store): Ditto.
4073 * doc/md.texi (Standard Pattern Names For Generation):
4074 Document memory_blockage instruction pattern.
4076 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4078 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4079 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4080 * config/rl78/rl78.md: New define_expand "adddi3".
4082 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4084 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4087 2017-10-13 Jeff Law <law@redhat.com>
4089 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4091 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4094 * internal-fn.c (expand_mul_overflow): If both operands have
4095 the same highpart of -1 or 0 and the topmost bit of lowpart
4096 is different, overflow is if res <= 0 rather than res < 0.
4098 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4100 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4101 TARGET_P9_VECTOR code for unaligned_load case.
4103 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4105 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4107 2017-10-13 Nathan Sidwell <nathan@acm.org>
4109 * tree-core.h (tree_contains_struct): Make bool.
4110 * tree.c (tree_contains_struct): Likewise.
4111 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4112 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4113 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4114 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4115 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4117 2017-10-13 Richard Biener <rguenther@suse.de>
4119 * graphite-isl-ast-to-gimple.c
4120 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4121 parameters and dominance check.
4122 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4123 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4124 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4125 Do not update SSA form here or do intermediate IL verification.
4126 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4127 (graphite_initialize): Remove check on the number of loops in
4128 the function and inline into graphite_transform_loops.
4129 (graphite_finalize): Inline into graphite_transform_loops.
4130 (graphite_transform_loops): Perform SSA update and IL verification
4132 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4134 2017-10-13 Richard Biener <rguenther@suse.de>
4136 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4137 graphite_expression_type_precision): Avoid global constructor
4139 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4140 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4141 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4142 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4143 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4144 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4146 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4149 * config/i386/i386.h (ix86_red_zone_size): New.
4150 * config/i386/i386.md (push peephole2s): Replace
4151 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4153 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4154 Alan Hayward <alan.hayward@arm.com>
4155 David Sherwood <david.sherwood@arm.com>
4157 * combine.c (can_change_dest_mode): Reject changes in
4158 REGMODE_NATURAL_SIZE.
4160 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4161 Alan Hayward <alan.hayward@arm.com>
4162 David Sherwood <david.sherwood@arm.com>
4164 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4165 (expand_debug_source_expr): Likewise.
4166 * combine.c (combine_simplify_rtx): Likewise.
4167 * cse.c (fold_rtx): Likewise.
4168 * fwprop.c (canonicalize_address): Likewise.
4169 * targhooks.c (default_shift_truncation_mask): Likewise.
4171 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4172 Alan Hayward <alan.hayward@arm.com>
4173 David Sherwood <david.sherwood@arm.com>
4175 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4176 (widened_mode): Likewise.
4177 (expand_unop): Likewise.
4178 * ree.c (transform_ifelse): Likewise.
4179 (merge_def_and_ext): Likewise.
4180 (combine_reaching_defs): Likewise.
4181 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4183 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4184 Alan Hayward <alan.hayward@arm.com>
4185 David Sherwood <david.sherwood@arm.com>
4187 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4188 * combine.c (gen_lowpart_for_combine): Likewise.
4189 * dwarf2out.c (rtl_for_decl_location): Likewise.
4190 * final.c (alter_subreg): Likewise.
4191 * rtlhooks.c (gen_lowpart_general): Likewise.
4192 (gen_lowpart_if_possible): Likewise.
4194 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4195 Alan Hayward <alan.hayward@arm.com>
4196 David Sherwood <david.sherwood@arm.com>
4198 * calls.c (expand_call): Use subreg_lowpart_offset.
4199 * cse.c (cse_insn): Likewise.
4200 * regcprop.c (copy_value): Likewise.
4201 (copyprop_hardreg_forward_1): Likewise.
4203 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4206 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4207 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4208 =Q constraints to +Q and into insn condition add check
4209 that operands[0] and operands[1] are equal.
4210 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4211 =Q constraints to +Q and into insn condition add check
4212 that operands[0] is equal to either operands[1] or operands[2].
4215 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4216 instead of handling MINUS_EXPR twice (once for each argument),
4217 canonicalize operand order and handle just once, use rtype where
4218 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4221 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4222 any values of __C while still being pattern recognizable as a simple
4225 2017-10-13 Richard Biener <rguenther@suse.de>
4227 PR tree-optimization/82451
4229 2017-10-02 Richard Biener <rguenther@suse.de>
4231 PR tree-optimization/82355
4232 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4233 a mapping for the enclosing loop but avoid generating one for
4235 (copy_bb_and_scalar_dependences): Remove premature codegen
4236 error on PHIs in blocks duplicated into multiple places.
4237 * graphite-scop-detection.c
4238 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4239 in the region use it as loop and nest to analyze the DR in.
4240 (try_generate_gimple_bb): Likewise.
4241 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4242 (add_loop_constraints): For blocks in a loop not in the region
4243 create a dimension with a single iteration.
4244 * sese.h (gbb_loop_at_index): Remove assert.
4246 * cfgloop.c (loop_preheader_edge): For the loop tree root
4247 return the single successor of the entry block.
4248 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4249 Reset the SCEV hashtable and niters.
4250 * graphite-scop-detection.c
4251 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4252 assert that we only have POLYNOMIAL_CHREC that vary in loops
4253 contained in the region.
4254 (scop_detection::graphite_can_represent_expr): Adjust.
4255 (scop_detection::stmt_has_simple_data_refs_p): For loops
4256 not in the region set loop to NULL. The nest is now the
4257 entry edge to the region.
4258 (try_generate_gimple_bb): Likewise.
4259 * sese.c (scalar_evolution_in_region): Adjust for
4260 instantiate_scev change.
4261 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4262 Make nest parameter the edge into the region.
4263 (create_data_ref): Likewise.
4264 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4265 entry edge into a region and adjust instantiate_scev calls.
4266 (create_data_ref): Likewise.
4267 (graphite_find_data_references_in_stmt): Likewise.
4268 (find_data_references_in_stmt): Pass the loop preheader edge
4269 from the nest argument.
4270 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4271 parameter the edge into the region.
4272 (instantiate_parameters): Use the loop preheader edge as entry.
4273 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4275 (get_instantiated_value_entry): Make instantiate_below parameter
4276 the edge into the region.
4277 (instantiate_scev_name): Likewise. Adjust dominance checks,
4278 when we cannot use loop-based instantiation instantiate by
4279 walking use-def chains.
4280 (instantiate_scev_poly): Adjust.
4281 (instantiate_scev_binary): Likewise.
4282 (instantiate_scev_convert): Likewise.
4283 (instantiate_scev_not): Likewise.
4284 (instantiate_array_ref): Remove.
4285 (instantiate_scev_3): Likewise.
4286 (instantiate_scev_2): Likewise.
4287 (instantiate_scev_1): Likewise.
4288 (instantiate_scev_r): Do not blindly handle N-operand trees.
4289 Do not instantiate array-refs. Handle all constants and invariants.
4290 (instantiate_scev): Make instantiate_below parameter
4291 the edge into the region.
4292 (resolve_mixers): Use the loop preheader edge for the region
4293 parameter to instantiate_scev_r.
4294 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4296 2017-10-13 Richard Biener <rguenther@suse.de>
4298 PR tree-optimization/82525
4299 * graphite-isl-ast-to-gimple.c
4300 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4302 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4303 Fail code generation when we cannot represent the isl integer.
4304 (binary_op_to_tree): Elide modulo operations that are no-ops
4305 in the type we code generate. Remove now superfluous code
4308 2017-10-13 Richard Biener <rguenther@suse.de>
4310 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4311 (scop_detection::harmful_loop_in_region): Remove premature
4312 IV type restriction.
4313 (scop_detection::graphite_can_represent_scev): We can handle
4314 pointer IVs just fine.
4316 2017-10-13 Alan Modra <amodra@gmail.com>
4318 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4319 "Clobbers and Scratch Registers". Add paragraph on
4320 alternative to clobbers for scratch registers and OpenBLAS
4323 2017-10-13 Alan Modra <amodra@gmail.com>
4325 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4326 example of a memory input for a string of known length. Move
4327 commentary out of table. Add a number of new examples
4328 covering array memory inputs.
4330 2017-10-12 Martin Liska <mliska@suse.cz>
4332 PR tree-optimization/82493
4333 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4334 (test_range_functions): New function.
4335 (sbitmap_c_tests): Likewise.
4336 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4337 * selftest.h (sbitmap_c_tests): New function.
4339 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4341 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4343 * config/rs6000/amo.h: Fix spacing issue.
4345 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4348 * config/i386/i386.md (*ashl<mode>3_mask_1,
4349 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4350 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4353 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4355 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4356 (profile_probability): Set max_probability
4357 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4359 * profile-count.c (profile_probability::differs_from_p): Do not
4360 rely on max_probaiblity == 10000
4362 2017-10-12 Jeff Law <law@redhat.com>
4364 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4367 2017-10-12 Martin Sebor <msebor@redhat.com>
4371 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4372 (handle_alias_pairs): Call it.
4373 * common.opt (-Wattribute-alias): New option.
4374 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4375 * doc/invoke.texi (-Wattribute-alias): Document.
4377 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4380 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4382 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4384 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4385 (make_hard_regno_born, make_hard_regno_dead): Update
4386 bb_killed_pseudos and bb_gen_pseudos.
4388 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4390 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4392 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4394 * config/alpha/alpha.c (alpha_split_conditional_move):
4395 Use std::swap instead of manually swapping.
4396 (alpha_stdarg_optimize_hook): Ditto.
4397 (alpha_canonicalize_comparison): Ditto.
4399 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4401 * tree-loop-distribution.c (struct builtin_info): New struct.
4402 (struct partition): Refactor fields into struct builtin_info.
4403 (partition_free): Free struct builtin_info.
4404 (build_size_arg_loc, build_addr_arg_loc): Delete.
4405 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4406 information from struct builtin_info.
4407 (find_single_drs): New function refactored from classify_partition.
4408 Also moved builtin validity checks to this function.
4409 (compute_access_range, alloc_builtin): New functions.
4410 (classify_builtin_st, classify_builtin_ldst): New functions.
4411 (classify_partition): Refactor code into functions find_single_drs,
4412 classify_builtin_st and classify_builtin_ldst.
4413 (distribute_loop): Don't do runtime alias check when distributing
4415 (find_seed_stmts_for_distribution): New function.
4416 (pass_loop_distribution::execute): Refactor code finding seed
4417 stmts into above function. Support distribution for the innermost
4418 two-level loop nest. Adjust dump information.
4420 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4422 * tree-loop-distribution.c: Adjust the general comment.
4423 (NUM_PARTITION_THRESHOLD): New macro.
4424 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4425 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4426 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4427 in call to build_partition_graph.
4428 (finalize_partitions): New parameter. Make loop distribution more
4429 conservative by fusing more partitions.
4430 (distribute_loop): Don't do runtime alias check in case of loop nest
4432 (find_seed_stmts_for_distribution): New function.
4433 (prepare_perfect_loop_nest): New function.
4434 (pass_loop_distribution::execute): Refactor code finding seed stmts
4435 and loop nest into above functions. Support loop nest distribution.
4436 Adjust dump information accordingly.
4438 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4440 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4441 and set PTYPE_SEQUENTIAL for merged partition.
4443 2017-10-12 Richard Biener <rguenther@suse.de>
4445 PR tree-optimization/69728
4447 2017-09-19 Richard Biener <rguenther@suse.de>
4449 PR tree-optimization/69728
4450 * graphite-sese-to-poly.c (schedule_error): New global.
4451 (add_loop_schedule): Handle empty domain by failing the
4453 (build_original_schedule): Handle schedule_error.
4455 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4456 domain by returning an unchanged schedule.
4458 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4460 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4461 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4463 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4465 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4468 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4471 * expr.c (store_field): Don't optimize away bitsize == 0 store
4472 from CALL_EXPR with addressable return type.
4474 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4476 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4477 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4478 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4479 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4480 TARGET_ISEL instead of TARGET_ISEL<sel>.
4482 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4484 * config/rs6000/rs6000.c
4485 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4487 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4489 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4491 (reg_or_cint_operand): Fix comment.
4492 (reg_or_zero_operand): New predicate.
4493 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4494 * config/rs6000/rs6000.c (output_isel): Delete.
4495 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4496 instead of reg_or_cint_operand. Output instruction directly (not via
4498 (isel_unsigned_<mode>): Ditto.
4499 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4500 gpc_reg_operand. Add an instruction alternative for this. Output
4501 instruction directly.
4502 (*isel_reversed_unsigned_<mode>): Ditto.
4504 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4506 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4507 (TARGET_CANONICALIZE_COMPARISON): Define.
4509 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4512 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4513 Check whether the dest is REG before adding REG_EQUIV note.
4515 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4518 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4520 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4521 (make_hard_regno_born, make_hard_regno_dead): Update
4522 bb_killed_pseudos and bb_gen_pseudos.
4524 2017-10-11 Nathan Sidwell <nathan@acm.org>
4526 * incpath.h (enum incpath_kind): Name enum, prefix values.
4527 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4528 * incpath.c (heads, tails): Use INC_MAX.
4529 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4530 (merge_include_chains, split_quote_chain,
4531 register_include_chains): Update incpath_kind names.
4532 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4533 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4535 (add_framework_path, darwin_register_objc_includes): Likewise.
4536 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4538 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4540 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4541 Do not use float_operator operator predicate.
4542 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4543 * config/i386/predicates.md (float_operator): Remove predicate.
4545 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4547 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4548 (*jccxf_i387): Ditto.
4549 (*jcc<mode>_i387): Ditto.
4550 (*jccu<mode>_i387): Ditto.
4551 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4552 (*jcc_*_i387 splitters): Remove.
4553 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4554 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4555 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4558 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4560 * profile-count.h (slow_safe_scale_64bit): New function.
4561 (safe_scale_64bit): New inline.
4562 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4563 * profile-count.c: Include wide-int.h
4564 (slow_safe_scale_64bit): New.
4566 2017-10-11 Nathan Sidwell <nathan@acm.org>
4568 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4569 HAS_DECL_ASSEMBLER_NAME_P.
4570 * gimple-expr.c (gimple_decl_printable_name: Check
4571 HAS_DECL_ASSEMBLER_NAME_P too.
4572 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4573 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4574 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4575 * passes.c (rest_of_decl_compilation): Check
4576 HAS_DECL_ASSEMBLER_NAME_P too.
4577 * recog.c (verify_changes): Likewise.
4578 * tree-pretty-print.c (dump_decl_name): Likewise.
4579 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4581 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4582 (SET_DECL_ASSEMBLER_NAME): Use it.
4583 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4584 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4585 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4587 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4589 * config.gcc (i386, x86_64): Add extra objects.
4590 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4591 (ix86_min_insn_size): Declare.
4592 (ix86_issue_rate): Declare.
4593 (ix86_adjust_cost): Declare.
4594 (ia32_multipass_dfa_lookahead): Declare.
4595 (ix86_macro_fusion_p): Declare.
4596 (ix86_macro_fusion_pair_p): Declare.
4597 (ix86_bd_has_dispatch): Declare.
4598 (ix86_bd_do_dispatch): Declare.
4599 (ix86_core2i7_init_hooks): Declare.
4600 (ix86_atom_sched_reorder): Declare.
4601 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4602 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4603 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4604 (rip_relative_addr_p): Rename to ...
4605 (ix86_rip_relative_addr_p): ... this one; export.
4606 (memory_address_length): Update.
4607 (ix86_issue_rate): Move to x86-tune-sched.c.
4608 (ix86_flags_dependent): Move to x86-tune-sched.c.
4609 (ix86_agi_dependent): Move to x86-tune-sched.c.
4610 (exact_dependency_1): Move to x86-tune-sched.c.
4611 (exact_store_load_dependency): Move to x86-tune-sched.c.
4612 (ix86_adjust_cost): Move to x86-tune-sched.c.
4613 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4614 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4615 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4616 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4617 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4618 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4619 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4620 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4621 (min_insn_size): Rename to ...
4622 (ix86_min_insn_size): ... this one; export.
4623 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4624 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4625 (core2i7_first_cycle_multipass_backtrack): Move to
4626 x86-tune-sched-core.c.
4627 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4628 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4629 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4630 (ix86_avoid_jump_mispredicts): Update.
4631 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4632 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4633 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4634 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4635 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4636 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4637 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4638 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4639 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4640 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4641 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4642 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4643 (BIG): Move to ix86-tune-sched-bd.c.
4644 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4645 (enum insn_path): Move to ix86-tune-sched-bd.c.
4646 (get_mem_group): Move to ix86-tune-sched-bd.c.
4647 (is_cmp): Move to ix86-tune-sched-bd.c.
4648 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4649 (is_branch): Move to ix86-tune-sched-bd.c.
4650 (is_prefetch): Move to ix86-tune-sched-bd.c.
4651 (init_window): Move to ix86-tune-sched-bd.c.
4652 (allocate_window): Move to ix86-tune-sched-bd.c.
4653 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4654 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4655 (process_end_window): Move to ix86-tune-sched-bd.c.
4656 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4657 (find_constant): Move to ix86-tune-sched-bd.c.
4658 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4659 (has_immediate): Move to ix86-tune-sched-bd.c.
4660 (get_insn_path): Move to ix86-tune-sched-bd.c.
4661 (get_insn_group): Move to ix86-tune-sched-bd.c.
4662 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4663 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4664 (add_insn_window): Move to ix86-tune-sched-bd.c.
4665 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4666 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4667 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4668 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4669 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4670 (do_dispatch): Move to ix86-tune-sched-bd.c.
4671 (has_dispatch): Move to ix86-tune-sched-bd.c.
4672 * config/i386/t-i386: Add new object files.
4673 * config/i386/x86-tune-costs.h: New file.
4674 * config/i386/x86-tune-sched-atom.c: New file.
4675 * config/i386/x86-tune-sched-bd.c: New file.
4676 * config/i386/x86-tune-sched-core.c: New file.
4677 * config/i386/x86-tune-sched.c: New file.
4679 2017-10-11 Liu Hao <lh_mouse@126.com>
4681 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4682 the generic version below instead.
4683 (should_colorize): Recognize Windows consoles as terminals
4685 * pretty-print.c [__MINGW32__] (write_all): New function.
4686 [__MINGW32__] (find_esc_head): Likewise.
4687 [__MINGW32__] (find_esc_terminator): Likewise.
4688 [__MINGW32__] (eat_esc_sequence): Likewise.
4689 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4691 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4694 2017-10-11 Richard Biener <rguenther@suse.de>
4696 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4697 Properly call analyze_scalar_evolution with the loop of the stmt.
4699 2017-10-11 Richard Biener <rguenther@suse.de>
4701 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4702 * tree-core.h (tree_base): Add chrec_var union member.
4703 * tree.h (CHREC_VAR): Remove.
4704 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4705 * tree-chrec.h (build_polynomial_chrec): Adjust.
4706 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4707 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4709 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4711 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4712 * match.pd: ... here.
4713 ((T) X == (T) Y): Relax condition.
4715 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4717 PR tree-optimization/82472
4718 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4720 (break_alias_scc_partitions): Update postorder number.
4722 2017-10-11 Martin Liska <mliska@suse.cz>
4725 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4727 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4729 2017-10-11 Martin Liska <mliska@suse.cz>
4731 * print-rtl.c (print_insn): Move declaration of idbuf
4732 to same scope as name.
4734 2017-10-11 Martin Liska <mliska@suse.cz>
4739 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4742 2017-10-11 Martin Liska <mliska@suse.cz>
4745 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4748 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4750 * config.gcc (default_gnu_indirect_function): Default to yes for
4751 arm*-*-linux* with glibc.
4753 2017-10-11 Richard Biener <rguenther@suse.de>
4755 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4756 default-defs and types we do not want to analyze.
4757 (interpret_loop_phi): Replace unreachable code with an assert.
4758 (compute_scalar_evolution_in_loop): Remove and inline ...
4759 (analyze_scalar_evolution_1): ... here, replacing condition with
4760 what makes the intent clearer. Remove handling of cases
4761 get_scalar_evolution now handles.
4763 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4765 PR rtl-optimization/81434
4766 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4767 comment for main loop. In sched_group_found if, also add checks for
4768 pass and min_cost_group.
4770 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4772 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4773 (rs6000_insn_cost): New function.
4774 * config/rs6000/rs6000.md (cost): New attribute.
4776 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4777 H.J. Lu <hongjiu.lu@intel.com>
4781 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4782 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4783 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4784 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4785 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4786 OPTION_MASK_ISA_MMX and some other bit set require both
4787 mmx and the ISAs without the mmx bit.
4788 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4789 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4790 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4791 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4792 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4793 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4794 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4795 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4796 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4797 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4798 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4799 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4800 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4801 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4802 Add OPTION_MASK_ISA_MMX.
4804 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4806 * config.gcc (armv7*-*-freebsd*): New target.
4807 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4809 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4811 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4812 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4815 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4817 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4818 a template parameter.
4819 (WIDE_INT_REF_FOR): Update accordingly.
4820 * tree.h (wi::int_traits <const_tree>): Delete.
4821 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4822 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4823 (wi::tree_to_wide_ref): New typedef.
4824 (wi::to_wide): New function.
4825 * calls.c (get_size_range): Use wi::to_wide when operating on
4827 * cgraph.c (cgraph_node::create_thunk): Likewise.
4828 * config/i386/i386.c (ix86_data_alignment): Likewise.
4829 (ix86_local_alignment): Likewise.
4830 * dbxout.c (stabstr_O): Likewise.
4831 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4832 * expr.c (const_vector_from_tree): Likewise.
4833 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4834 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4835 (fold_negate_expr_1, int_const_binop_1, const_binop)
4836 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4837 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4838 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4839 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4840 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4841 (fold_not_const, round_up_loc): Likewise.
4842 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4843 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4844 (alloca_call_type): Likewise.
4845 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4846 * godump.c (go_output_typedef): Likewise.
4847 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4848 * internal-fn.c (get_min_precision): Likewise.
4849 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4850 * ipa-polymorphic-call.c
4851 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4852 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4853 (ipa_modify_call_arguments): Likewise.
4854 * match.pd: Likewise.
4855 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4856 * print-tree.c (print_node_brief, print_node): Likewise.
4857 * stmt.c (expand_case): Likewise.
4858 * stor-layout.c (layout_type): Likewise.
4859 * tree-affine.c (tree_to_aff_combination): Likewise.
4860 * tree-cfg.c (group_case_labels_stmt): Likewise.
4861 * tree-data-ref.c (dr_analyze_indices): Likewise.
4862 (prune_runtime_alias_test_list): Likewise.
4863 * tree-dump.c (dequeue_and_dump): Likewise.
4864 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4865 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4866 * tree-pretty-print.c (dump_generic_node): Likewise.
4867 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4868 (simple_iv_with_niters): Likewise.
4869 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4870 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4871 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4872 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4873 (refine_value_range_using_guard, number_of_iterations_ne_max)
4874 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4875 (get_cst_init_from_scev, record_nonwrapping_iv)
4876 (scev_var_range_cant_overflow): Likewise.
4877 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4878 * tree-ssa-pre.c (compute_avail): Likewise.
4879 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4880 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4881 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4882 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4883 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4884 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4885 (dump_case_nodes, try_switch_expansion): Likewise.
4886 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4887 (vect_do_peeling): Likewise.
4888 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4889 * tree-vect-stmts.c (vectorizable_load): Likewise.
4890 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4891 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4892 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4893 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4894 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4895 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4896 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4897 (evrp_dom_walker::before_dom_children): Likewise.
4898 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4899 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4900 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4901 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4902 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4903 (get_range_pos_neg): Likewise.
4904 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4905 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4906 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4907 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4908 * config/avr/avr.c (avr_fold_builtin): Likewise.
4909 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4910 * config/msp430/msp430.c (msp430_attr): Likewise.
4911 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4912 * config/powerpcspe/powerpcspe-c.c
4913 (altivec_resolve_overloaded_builtin): Likewise.
4914 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4915 (rs6000_expand_ternop_builtin): Likewise.
4916 * config/rs6000/rs6000-c.c
4917 (altivec_resolve_overloaded_builtin): Likewise.
4918 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4919 (rs6000_expand_ternop_builtin): Likewise.
4920 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4922 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4924 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4925 when copying loop nest with only one inner loop.
4927 2017-10-10 Richard Biener <rguenther@suse.de>
4929 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4930 blocks if SCEV is active.
4931 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4933 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4934 (scev_initialize): Assert we are not yet initialized.
4936 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4938 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4939 inner loop's exit stmt by making it always exit the loop, otherwise
4940 we would generate an infinite empty loop.
4942 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4944 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4945 renaming variables in new preheader if it's deleted.
4947 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4949 * tree-loop-distribution.c (struct partition): Remove unused field
4950 loops of the structure.
4951 (partition_alloc, partition_free): Ditto.
4952 (build_rdg_partition_for_vertex): Ditto.
4954 2017-10-09 Jeff Law <law@redhat.com>
4956 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4957 return type to match prototype and documentation.
4959 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4961 * config/rs6000/rs6000.c (processor_costs): Move to ...
4962 * config/rs6000/rs6000.h: ... here.
4963 (rs6000_cost): Declare.
4965 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4967 * except.c (setjmp_fn): New global variable.
4968 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4969 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4970 if DONT_USE_BUILTIN_SETJMP is defined.
4972 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4974 * target.def (insn_cost): New hook.
4975 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4976 * doc/tm.texi: Regenerate.
4977 * rtlanal.c (insn_cost): Use the new hook.
4979 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4981 * combine.c (combine_validate_cost): Compute the new insn_cost,
4982 not just pattern_cost.
4983 (try_combine): Adjust comment.
4985 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4987 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4989 * combine.c (uid_insn_cost): Adjust comment.
4990 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4992 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4993 * dse.c (find_shift_sequence): Ditto.
4994 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4995 (bb_valid_for_noce_process_p): Use pattern_cost.
4996 * rtl.h (insn_rtx_cost): Delete.
4997 (pattern_cost): New prototype.
4998 (insn_cost): New prototype.
4999 * rtlanal.c (insn_rtx_cost): Rename to...
5000 (pattern_cost): ... this.
5003 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
5005 * config/i386/i386.md (*jcc_2): Remove insn pattern.
5006 (*jcc<mode>_0_r_i387): Ditto.
5007 (*jccxf_r_i387): Ditto.
5008 (*jcc<mode>_r_i387): Ditto.
5009 (*jccu<mode>_r_i387): Ditto.
5010 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
5011 (*jcc): Rename from *jcc_1.
5013 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5015 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
5016 deferred rescans after the lvx/stvx recombination pre-pass.
5018 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
5020 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
5021 memory operation instruction support.
5022 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
5023 (rs6000-ibm-aix[789]*): Likewise.
5024 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
5025 Document new functions.
5027 2017-10-09 Richard Biener <rguenther@suse.de>
5029 PR tree-optimization/82397
5030 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5031 equality only for semantically equal trees.
5033 2017-10-09 Richard Biener <rguenther@suse.de>
5035 PR tree-optimization/82449
5036 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5037 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5038 allow constant addresses.
5039 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5042 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5044 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5047 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5050 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5053 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5056 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5058 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5061 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5062 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5064 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5066 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5067 (WI_BINARY_PREDICATE_RESULT): Likewise.
5068 (wi::binary_traits::operator_result): New type.
5069 (wi::binary_traits::predicate_result): Likewise.
5070 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5071 (generic_wide_int::operator==, generic_wide_int::operator!=)
5072 (generic_wide_int::operator&, generic_wide_int::and_not)
5073 (generic_wide_int::operator|, generic_wide_int::or_not)
5074 (generic_wide_int::operator^, generic_wide_int::operator+
5075 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5077 (operator~, unary operator-, operator==, operator!=, operator&)
5078 (operator|, operator^, operator+, binary operator-, operator*): New
5080 * expr.c (get_inner_reference): Use wi::bit_and_not.
5081 * fold-const.c (fold_binary_loc): Likewise.
5082 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5083 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5084 (bit_value_binop): Likewise.
5085 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5086 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5087 (extract_range_from_binary_expr_1): Likewise.
5088 (masked_increment): Likewise.
5089 (simplify_bit_ops_using_ranges): Likewise.
5091 2017-10-09 Martin Jambor <mjambor@suse.cz>
5094 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5095 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5096 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5098 (hsa_fixup_mov_insn_type): New function.
5099 (hsa_op_with_type::get_in_type): Use it.
5100 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5101 immediates in an assert.
5102 (hsa_op_with_type::extend_int_to_32bit): New method.
5103 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5104 types. Convert to dest type if necessary.
5105 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5106 (reg_for_gimple_ssa): Pass false as min32int to
5107 hsa_type_for_scalar_tree_type.
5108 (gen_hsa_addr): Fixup type when creating addresable temporary.
5109 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5110 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5111 necessary. Call hsa_fixup_mov_insn_type.
5112 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5113 extend operands and convert to dest type if necessary.
5114 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5115 to dest type if necessary.
5116 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5117 if conversion nt necessary and size matches.
5118 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5119 to dest type if necessary.
5120 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5121 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5123 (gen_hsa_clrsb): Likewise.
5124 (gen_hsa_ffs): Likewise.
5125 (gen_hsa_divmod): Extend operands and convert to dest type if
5127 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5129 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5131 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5132 Remove empty default arguments. Use a brace block as output
5134 (conditional return): Ditto.
5136 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5137 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5139 (group_ending_nop): Ditto.
5140 (doloop_end): Ditto.
5141 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5142 (splitters for those): Ditto.
5144 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5146 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5147 a conditional jump (and the compare for it) so that pc_rtx is the
5149 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5150 for the deleted and renamed ctr<mode>_internal[234] patterns.
5151 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5152 Delete second conditional return pattern.
5153 (ctr<mode>_internal2): Delete this second bdnz pattern.
5154 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5155 (ctr<mode>_internal4): Delete this second bdz pattern.
5157 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5159 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5160 (always_initialized_rtx_for_ssa_name_p): New predicate.
5161 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5162 (finish_out_of_ssa): Free new field of SA.
5163 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5164 * tree-ssa-coalesce.c: Include tree-ssa.h.
5165 (get_parm_default_def_partitions): Remove extern keyword.
5166 (get_undefined_value_partitions): New function.
5167 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5168 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5170 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5172 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5174 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5176 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5178 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5179 for targets that preffer 128bit.
5181 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5183 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5185 2017-10-08 Olivier Hainque <hainque@adacore.com>
5187 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5188 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5190 (thumb_set_return_address): Likewise.
5192 2017-10-08 Olivier Hainque <hainque@adacore.com>
5194 * common/config/arm/arm-common.c (arm_except_unwind_info):
5195 Handle DWARF2_UNWIND_INFO.
5197 2017-10-07 Michael Collison <michael.collison@arm.com>
5199 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5202 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5204 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5205 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5206 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5207 defined, force the creation of a new block for a dispatch label.
5209 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5211 * invoke.texi (Wsuggest-attribute=cold): Document.
5212 * common.opt (Wsuggest-attribute=cold): New
5213 * ipa-pure-const.c (warn_function_cold): New function.
5214 * predict.c (compute_function_frequency): Use it.
5215 * predict.h (warn_function_cold): Declare.
5217 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5219 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5222 2017-10-06 Martin Liska <mliska@suse.cz>
5224 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5225 keyword for member functions.
5226 (struct sanopt_tree_couple): New struct.
5227 (struct sanopt_tree_couple_hash): New function.
5228 (struct sanopt_ctx): Add new hash_map.
5229 (has_dominating_ubsan_ptr_check): New function.
5230 (record_ubsan_ptr_check_stmt): Likewise.
5231 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5232 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5233 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5235 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5238 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5239 aarch64_simd_valid_immediate on CONST_VECTORs.
5240 (aarch64_reg_or_bic_imm): Likewise.
5242 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5244 PR rtl-optimization/82396
5245 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5247 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5249 * graphite-dependences.c (scop_get_reads): Move code to...
5250 (scop_get_must_writes): Move code to...
5251 (scop_get_may_writes): Move code to...
5252 (scop_get_reads_and_writes): ... here.
5253 (scop_get_dependences): Call scop_get_reads_and_writes.
5255 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5257 PR tree-optimization/82434
5258 * fold-const.h (can_native_encode_type_p,
5259 can_native_encode_string_p): Remove.
5260 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5261 don't encode anything, just return what would be otherwise returned.
5262 (native_encode_fixed, native_encode_complex, native_encode_vector):
5264 (native_encode_string): Likewise. Inline by hand
5265 can_native_encode_string_p.
5266 (can_native_encode_type_p): Remove.
5267 (can_native_encode_string_p): Remove.
5268 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5269 STRING_CSTs using can_native_encode_string_p, test all
5270 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5271 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5272 argument from native_encode_expr.
5273 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5274 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5277 2017-10-06 Richard Biener <rguenther@suse.de>
5279 PR tree-optimization/82397
5280 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5281 operand_equal_p but rely on data_ref_compare_tree for detecting
5283 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5284 to match up with dr_group_sort_cmp.
5286 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5289 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5291 * config/s390/s390-builtin-types.def: Regenerate.
5293 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5296 * config/s390/s390-builtin-types.def: Regenerate.
5297 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5298 Change flag from B_VXE to B_VX.
5299 (s390_vec_min_dbl): Remove B_VXE flag.
5301 2017-10-06 Richard Biener <rguenther@suse.de>
5303 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5304 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5305 translate_isl_ast_to_gimple::is_valid_rename,
5306 translate_isl_ast_to_gimple::get_rename,
5307 translate_isl_ast_to_gimple::get_def_bb_for_const,
5308 translate_isl_ast_to_gimple::get_new_name,
5309 translate_isl_ast_to_gimple::collect_all_ssa_names,
5310 translate_isl_ast_to_gimple::copy_loop_phi_args,
5311 translate_isl_ast_to_gimple::collect_all_ssa_names,
5312 translate_isl_ast_to_gimple::copy_loop_phi_args,
5313 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5314 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5315 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5316 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5317 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5318 translate_isl_ast_to_gimple::copy_cond_phi_args,
5319 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5320 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5321 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5322 translate_isl_ast_to_gimple::rename_uses,
5323 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5324 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5325 (set_rename_for_each_def): Likewise.
5326 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5327 here. Handle rewriting SCEV analyzable uses here.
5328 (copy_bb_and_scalar_dependences): Generate code for PHI
5330 (graphite_regenerate_ast_isl): Adjust.
5331 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5332 (add_write, add_read): New functions.
5333 (build_cross_bb_scalars_def): Use it and simplify.
5334 (build_cross_bb_scalars_use): Likewise.
5335 (graphite_find_cross_bb_scalar_vars): Inline into...
5336 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5337 simulating out-of-SSA. Compute liveout and add dependencies.
5338 (build_scops): Force an empty entry block.
5339 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5341 (sese_build_liveouts): Declare.
5342 (sese_trivially_empty_bb_p): Likewise.
5343 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5344 compute liveout and debug_liveout.
5345 (sese_bad_liveouts_use): Remove.
5346 (sese_reset_debug_liveouts_bb): Likewise.
5347 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5348 (sese_build_liveouts): Build liveout and debug_liveout and store
5350 (new_sese_info): Adjust.
5351 (free_sese_info): Likewise.
5352 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5353 do not build liveout here.
5354 (move_sese_in_condition): Adjust region entry.
5355 (scev_analyzable_p): Match up with chrec_apply requirements.
5356 (sese_trivially_empty_bb_p): New.
5357 * tree-into-ssa.c (get_reaching_def): Properly support generating
5358 default-defs for incremental rewrite of anonymous names.
5360 2017-10-06 Richard Biener <rguenther@suse.de>
5362 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5363 precision do not perform modulo reduction.
5365 2017-10-06 Richard Biener <rguenther@suse.de>
5367 PR tree-optimization/82436
5368 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5369 conservatively choose the vectorization factor when checking
5370 whether we can perform the required load permutation.
5371 (vect_transform_slp_perm_load): Assert when we may not fail.
5373 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5375 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5376 message for incompatible -msdata=* and -mcall-* options.
5378 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5380 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5381 rate for post-reload scheduling.
5383 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5385 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5387 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5389 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5390 to improve monte carlo in scimark.
5392 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5394 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5395 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5396 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5397 pentium4_cost, nocona_cost): Set reassociation width to 1.
5398 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5399 width to 2 for fp operations and 1 otherwise.
5400 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5402 (atom_cost): Set reassociation width to 2.
5403 (slm_cost, generic_cost): Set fp reassociation width
5404 to 2 and 1 otherwise.
5405 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5406 (core_cost): Set fp reassociation width to 4 and vector to 2.
5407 (ix86_reassociation_width): Rewrite using cost table; special case
5408 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5409 and TARGET_AVX128_OPTIMAL.
5410 * config/i386/i386.h (processor_costs): Add
5411 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5412 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5413 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5414 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5415 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5416 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5418 2017-10-05 Nathan Sidwell <nathan@acm.org>
5420 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5422 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5424 * config/arm/arm.c (arm_test_fpu_data): New.
5425 (arm_run_selftests): Call arm_test_fpu_data.
5427 2017-10-04 Nathan Sidwell <nathan@acm.org>
5429 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5431 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5433 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5435 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5436 check type for aarch64_simd_valid_immediate.
5437 (aarch64_output_simd_mov_immediate): Update prototype.
5438 (aarch64_simd_valid_immediate): Update prototype.
5439 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5440 support for ORR-immediate.
5441 (and<mode>3): modified pattern to add support for BIC-immediate.
5442 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5443 now checks for valid immediate for BIC and ORR based on new enum
5445 (aarch64_output_simd_mov_immediate): Function now used to output
5446 BIC/ORR imm as well based on new enum argument.
5447 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5449 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5450 (aarch64_reg_or_bic_imm): Likewise.
5452 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5454 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5455 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5457 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5461 PR rtl-optimization/82396
5462 * haifa-sched.c (autopref_multipass_init): Simplify
5464 (autopref_rank_data): Simplify sort order.
5465 * sched-int.h (autopref_multipass_data_): Remove
5466 multi_mem_insn_p, min_offset and max_offset.
5468 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5470 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5472 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5474 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5477 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5479 * doc/sourcebuild.texi: Document vect_long_mult.
5481 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5483 PR tree-optimization/82413
5484 * fold-const.c (build_range_check): Use widest_int when comparing
5485 the maximum ETYPE value with HIGH.
5487 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5489 PR rtl-optimization/82396
5490 * haifa-sched.c (autopref_multipass_init): Simplify
5492 (autopref_rank_data): Simplify sort order.
5493 * sched-int.h (autopref_multipass_data_): Remove
5494 multi_mem_insn_p, min_offset and max_offset.
5496 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5498 PR tree-optimization/82381
5499 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5500 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5503 PR tree-optimization/82374
5504 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5505 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5506 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5507 current_function_decl to the new decl.
5509 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5511 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5512 helper macro for IEEE float128 hardware built-in functions.
5513 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5515 (TRUNCF128_ODD): Likewise.
5516 (ADDF128_ODD): Likewise.
5517 (SUBF128_ODD): Likewise.
5518 (MULF128_ODD): Likewise.
5519 (DIVF128_ODD): Likewise.
5520 (FMAF128_ODD): Likewise.
5521 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5522 UNSPEC_TRUNC_ROUND_TO_ODD.
5523 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5524 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5525 floating point round to odd instructions.
5526 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5527 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5528 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5529 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5530 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5531 (trunc<mode>sf2_hw): Change the truncate with round to odd
5532 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5533 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5534 to odd hardware instructions.
5535 (sub<mode>3_odd): Likewise.
5536 (mul<mode>3_odd): Likewise.
5537 (div<mode>3_odd): Likewise.
5538 (sqrt<mode>2_odd): Likewise.
5539 (fma<mode>4_odd): Likewise.
5540 (fms<mode>4_odd): Likewise.
5541 (nfma<mode>4_odd): Likewise.
5542 (nfms<mode>4_odd): Likewise.
5543 (trunc<mode>df2_odd): Change the truncate with round to odd
5544 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5546 * doc/extend.texi (PowerPC built-in functions): Update documentation
5547 for existing IEEE float128-bit built-in functions. Add built-in
5548 functions that generate the IEEE 128-bit floating point round to
5551 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5553 PR rtl-optimization/77729
5554 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5555 to (X&(C1&~C2))|C2 transformations.
5557 2017-10-03 Martin Jambor <mjambor@suse.cz>
5559 PR tree-optimization/82363
5560 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5561 mismatch, mark lacc written regardless of racc.
5563 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5565 PR tree-optimization/82381
5566 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5567 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5568 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5569 fallthrough into reassoc_stmt_dominates_stmt_p.
5572 * combine.c (combine_instructions): Don't combine in unreachable
5575 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5578 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5579 function to not use the have_cpu variable. Do not set cpu_index,
5580 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5582 (rs6000_valid_attribute_p): Remove duplicate initializations of
5583 old_optimize and func_optimize.
5584 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5585 (rs6000_activate_target_options): Make global.
5586 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5589 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5591 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5592 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5593 Set *poffset to 0 and *psize and *pmax_size to -1 if
5594 *poffset + *psize overflows in HOST_WIDE_INT.
5596 PR tree-optimization/82387
5597 PR tree-optimization/82388
5598 PR tree-optimization/82389
5599 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5600 instead of live_bytes non-NULL.
5602 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5605 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5608 2017-10-02 Richard Biener <rguenther@suse.de>
5610 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5611 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5612 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5614 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5616 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5617 requested precision matches the type's.
5618 * calls.c (alloc_max_size): Calculate the new candidate size as
5619 a widest_int and use wi::to_widest when comparing it with the
5620 current candidate size.
5621 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5622 zero rather than integer_zero_node.
5623 * match.pd: Check for a no-op conversion before using wi::add
5624 rather than after. Use tree_to_uhwi when summing small shift
5625 counts into an unsigned int.
5627 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5628 Alan Hayward <alan.hayward@arm.com>
5629 David Sherwood <david.sherwood@arm.com>
5632 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5633 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5634 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5635 POINTER_AND_FP_REGS.
5637 2017-10-02 Richard Biener <rguenther@suse.de>
5639 PR tree-optimization/82355
5640 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5641 a mapping for the enclosing loop but avoid generating one for
5643 (copy_bb_and_scalar_dependences): Remove premature codegen
5644 error on PHIs in blocks duplicated into multiple places.
5645 * graphite-scop-detection.c
5646 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5647 in the region use it as loop and nest to analyze the DR in.
5648 (try_generate_gimple_bb): Likewise.
5649 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5650 (add_loop_constraints): For blocks in a loop not in the region
5651 create a dimension with a single iteration.
5652 * sese.h (gbb_loop_at_index): Remove assert.
5654 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5656 * omp-expand.c (adjust_context_scope): New function.
5657 (expand_parallel_call): Call adjust_context_scope.
5659 2017-10-01 Jeff Law <law@redhat.com>
5661 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5662 dom_opt_dom_walker class with direct access to private members.
5663 Add comments. Call test_for_singularity.
5664 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5665 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5666 m_dummy_cond anymore.
5667 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5669 (pass_dominator:execute): Build the dummy_cond here and pass it
5670 to the dom_opt_dom_walker ctor.
5671 (test_for_singularity): New function.
5673 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5674 Maya Rashish <coypu@sdf.org>
5676 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5677 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5678 (alpha*-*-netbsd*) Use nbsd_tm_file.
5679 (arm*-*-netbsdelf*) Likewise.
5680 (i[34567]86-*-netbsdelf*) Likewise.
5681 (x86_64-*-netbsd*) Likewise.
5682 (mips*-*-netbsd*) Likewise.
5683 (powerpc-*-netbsd*) Likewise.
5684 (sh*-*-netbsd*) Likewise.
5685 (sparc-*-netbsdelf*) Likewise.
5686 (sparc64-*-netbsd*) Likewise.
5687 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5689 (vax-*-netbsdelf*) Likewise.
5690 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5691 (UINT_FAST8_TYPE) Likewise.
5692 (INT_FAST16_TYPE) Check CHAR_FAST16.
5693 (UINT_FAST16_TYPE) Likewise.
5695 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5698 * config/i386/i386.md
5699 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5700 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5701 *divmodsi4_zext_2): New define_insn_and_split.
5702 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5703 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5704 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5705 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5706 New define_insn_and_split.
5707 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5708 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5709 operands[1] having DImode when mode is SImode.
5711 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5712 always SImode for DIV and MOD in REG_EQUAL notes.
5714 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5717 * match.pd: Fix handling of NaNs in pattern.
5719 2017-09-29 Jeff Law <law@redhat.com>
5721 * sbitmap.c (bitmap_bit_in_range_p): New function.
5722 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5723 * tree-ssa-dse.c (live_bytes_read): New function.
5724 (dse_classify_store): Ignore reads of dead bytes.
5726 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5727 typos and whitespace errors.
5728 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5729 * config/s390/s390.c (s390_emit_prologue): Likewise.
5731 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5734 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5735 with a symbol for LRA.
5737 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5739 PR rtl-optimization/82338
5740 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5742 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5744 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5745 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5746 (qsort_chk): Declare.
5747 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5748 (qsort_chk): New function.
5750 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5752 PR tree-optimization/82337
5753 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5754 phi definition if the PHI result appears in an abnormal PHI.
5755 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5756 basis appears in an abnormal PHI.
5758 2017-09-29 Richard Biener <rguenther@suse.de>
5760 * graphite-isl-ast-to-gimple.c
5761 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5762 (binary_op_to_tree): Use it.
5763 (get_rename_from_scev): Likewise.
5764 (copy_loop_phi_nodes): Likewise.
5765 (copy_bb_and_scalar_dependences): Likewise.
5766 (translate_pending_phi_nodes): Likewise.
5768 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5771 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5772 for movabsq $(i32 << shift), r64.
5774 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5776 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5777 index when encoding %esp as %rsp to avoid 0x67 prefix.
5779 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5781 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5782 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5784 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5786 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5787 Extensions with more than 16 double VFP registers.
5788 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5789 to_clear_mask and all code related to it. Replace the remaining
5790 entry by a sbitmap and adapt code accordingly.
5792 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5794 * brig-builtins.def: Change pure attributes to const.
5796 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5798 * config.gcc (default_gnu_indirect_function): Default to yes for
5799 sparc*-*-linux* with glibc.
5801 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5803 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5804 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5805 when creating .init_array and .fini_array sections with priority
5808 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5811 * config/aarch64/aarch64.c
5812 (aarch64_builtin_support_vector_misalignment): Always return false
5813 when misalignment is unknown.
5815 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5817 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5818 this function to return false if the definition used by the swap
5819 instruction is artificial, or if the memory address from which the
5820 constant value is loaded is not represented by a base address held
5821 in a register or if the base address register is a frame or stack
5822 pointer. Additionally, return false if the base address of the
5823 loaded constant is a SYMBOL_REF but is not considered to be a
5825 (replace_swapped_load_constant): New function.
5826 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5827 loaded constant vector with a load of a swapped constant vector.
5829 2017-09-27 Carl Love <cel@us.ibm.com>
5831 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5832 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5833 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5836 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5838 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5839 first, always call autopref_rank_data otherwise.
5841 2017-09-27 Richard Biener <rguenther@suse.de>
5843 * graphite-scop-detection.c (find_scop_parameters): Move
5844 loop bound handling ...
5845 (gather_bbs::before_dom_children): ... here, avoiding the need
5846 to build scop_info->loop_nest.
5847 (record_loop_in_sese): Remove.
5848 * sese.h (sese_info_t::loop_nest): Remove.
5849 * sese.c (new_sese_info): Do not allocate loop_nest.
5850 (free_sese_info): Do not free loop_nest.
5852 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5855 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5856 lhs from calls if the lhs has addressable type.
5858 2017-09-27 Richard Biener <rguenther@suse.de>
5860 * graphite.h (scop::max_alias_set): New member.
5861 * graphite-scop-detection.c: Remove references to non-existing
5862 --param in comments.
5863 (build_alias_sets): Record the maximum alias set used for drs.
5864 (build_scops): Support zero as unlimited for
5865 --param graphite-max-arrays-per-scop.
5866 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5868 (build_poly_sr_1): ... here. Compute alias set based on the
5869 maximum alias set used for drs rather than
5870 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5872 2017-09-27 Richard Biener <rguenther@suse.de>
5874 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5875 --param loop-block-tile-size=0 to disable tiling.
5877 2017-09-27 Richard Biener <rguenther@suse.de>
5879 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5880 (graphite-max-nb-scop-params): Document special value zero.
5881 * domwalk.h (dom_walker::STOP): New symbolical constant.
5882 (dom_walker::dom_walker): Add optional parameter for bb to
5884 (dom_walker::~dom_walker): Declare.
5885 (dom_walker::before_dom_children): Document STOP return value.
5886 (dom_walker::m_user_bb_to_rpo): New member.
5887 (dom_walker::m_bb_to_rpo): Likewise.
5888 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5889 mapping here if not provided by the user.
5890 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5891 provided by the user.
5892 (dom_walker::STOP): Define.
5893 (dom_walker::walk): Do not compute bb to RPO mapping here.
5894 Support STOP return value from before_dom_children to stop
5896 * graphite-optimize-isl.c (optimize_isl): If the schedule
5897 is the same still generate code if -fgraphite-identity
5898 or -floop-parallelize-all are given.
5899 * graphite-scop-detection.c: Include cfganal.h.
5900 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5902 (gather_bbs::before_dom_children): Return STOP for BBs
5904 (build_scops): Compute bb to RPO mapping and pass it to
5905 the domwalk. Treat --param graphite-max-nb-scop-params=0
5906 as not limiting the number of params.
5907 * graphite.c (graphite_initialize): Remove limit on the
5908 number of basic-blocks in a function.
5909 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5910 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5911 default value of 10.
5913 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5915 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5916 Adjust code to eliminate needing to do the shift right 32-bits
5917 operation after XSCVDPSPN.
5919 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5921 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5922 ((X / Y) != 0 -> X >= Y): Likewise.
5924 2017-09-26 Carl Love <cel@us.ibm.com>
5926 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5927 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5928 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5929 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5930 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5931 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5932 definitions and overloading.
5933 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5934 statement for P9V_BUILTIN_XST_LEN_R.
5935 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5936 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5937 define_expand and define_insn for the instructions and builtins.
5938 * doc/extend.texi: Update the built-in documentation file for the new
5940 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5941 define_insn for the instructions
5943 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5946 * gcc/config/netbsd-protos.h: New file.
5947 * gcc/config/netbsd.c: New file.
5948 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5949 * gcc/config/t-netbsd: New file.
5950 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5951 (tmake_file) Add t-netbsd.
5952 (extra_objs) Add netbsd.o.
5954 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5958 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5960 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5962 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5963 sign extension from a vector register to a GPR by doing a 32-bit
5964 direct move and then an EXTSW.
5965 (extendsi<mode>2 splitter): Likewise.
5966 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5967 right or vector extract after doing XSCVDPSPN. Use
5968 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5970 (movdi_from_sf_zero_ext): Likewise.
5971 (reload_gpr_from_vsxsf): Likewise.
5972 (p8_mfvsrd_4_disf): Delete, no longer used.
5973 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5974 then needing to move the SFmode to a GPR to use the XSCVDPSP
5975 instruction instead of FRSP and XSCVDPSPN.
5976 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5977 it is adjacent to the other XSCVSPDP insns.
5978 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5979 SFmode to be in traditional Altivec registers.
5980 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5981 (vsx_xscvspdpn): Likewise.
5982 (vsx_xscvspdpn_scalar): Likewise.
5984 2017-09-26 Martin Jambor <mjambor@suse.cz>
5986 * tree-sra.c (compare_access_positions): Put integral types first,
5987 stabilize sorting of integral types, remove conditions putting
5988 non-full-precision integers last.
5989 (sort_and_splice_var_accesses): Disable scalarization if a
5990 non-integert would be represented by a non-full-precision integer.
5992 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5994 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5995 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5996 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5997 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5998 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5999 conditionals inside the function instead of around it. Call
6000 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
6001 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
6003 2017-09-26 Richard Biener <rguenther@suse.de>
6005 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
6007 (scop_detection::build_scop_breadth): ... this. Removed.
6008 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
6009 (scop_detection::harmful_stmt_in_bb): Likewise.
6010 (scop_detection::graphite_can_represent_stmt): Likewise.
6011 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
6012 (scop_detection::can_represent_loop): Remove recursion, fold in ...
6013 (scop_detection::can_represent_loop_1): ... this. Removed.
6014 (scop_detection::harmful_loop_in_region): Simplify after inlining
6015 the above and remove more quadraticness.
6016 (build_scops): Adjust.
6017 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
6020 2017-09-26 Jakub Jelinek <jakub@redhat.com>
6023 * config/i386/i386.c (ix86_print_operand_address_as): Only test
6024 REGNO (base) == SP_REG if base is a REG.
6027 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6028 if it is different SSA_NAME.
6029 (optimize_range_tests_cmp_bitwise): New function.
6030 (optimize_range_tests): Call it.
6032 2017-09-26 Richard Biener <rguenther@suse.de>
6034 PR tree-optimization/82321
6035 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6036 for the def being inside the loop.
6038 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6040 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6042 * config/s390/s390-builtins.def: Fix constraint on op4.
6044 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6046 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6047 independent expanders.
6048 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6049 ("vec_ordered", "vec_unordered"): New expanders.
6051 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6053 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6056 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6058 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6059 vec_unpacks_lo_v16qi.
6060 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6062 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6064 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6065 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6066 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6068 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6070 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6072 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6074 ("*vec_slb<mode>"): New insn pattern.
6075 ("vec_shr_<mode>"): New expander.
6076 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6077 and force the shift count operand to V16QImode.
6078 ("vec_srb<mode>"): Set shift count mode to V16QI.
6080 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6082 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6083 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6084 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6086 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6089 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6091 2017-09-26 Richard Biener <rguenther@suse.de>
6093 PR tree-optimization/82320
6094 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6097 2017-09-25 Jeff Law <law@redhat.com>
6099 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6100 prototype for new argument.
6101 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6102 mostly extracted from rs6000_emit_allocate_stack.
6103 (rs6000_emit_probe_stack_range_stack_clash): New function.
6104 (rs6000_emit_allocate_stack): Call
6105 rs6000_emit_probe_stack_range_stack_clash as needed.
6106 (rs6000_emit_probe_stack_range): Add additional argument
6107 to call to gen_probe_stack_range{si,di}.
6108 (output_probe_stack_range): New.
6109 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6110 (output_probe_stack_range_stack_clash): New.
6111 (rs6000_emit_prologue): Emit notes into dump file as requested.
6112 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6113 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6114 Add additional operand and pass it to output_probe_stack_range.
6116 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6118 PR tree-optimization/82163
6119 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6120 (checking_verify_loop_closed_ssa): New parameter.
6121 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6122 (check_loop_closed_ssa_stmt): Delete.
6123 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6124 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6125 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6128 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6130 * brig-builtins.def: Treat HSAIL barrier builtins as
6131 setjmp/longjump style functions.
6133 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6135 * target.def (constant_alignment): New hook.
6136 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6137 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6138 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6139 * doc/tm.texi: Regenerate.
6140 * targhooks.h (default_constant_alignment): Declare.
6141 (constant_alignment_word_strings): Likewise.
6142 * targhooks.c (default_constant_alignment): New function.
6143 (constant_alignment_word_strings): Likewise.
6144 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6145 instead of CONSTANT_ALIGNMENT.
6146 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6147 (force_const_mem): Likewise.
6148 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6149 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6150 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6151 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6152 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6154 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6155 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6156 constant_alignment_word_strings.
6157 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6158 (CONSTANT_ALIGNMENT): Likewise.
6159 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6160 (arm_constant_alignment): New function.
6161 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6162 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6163 constant_alignment_word_strings.
6164 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6165 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6166 constant_alignment_word_strings.
6167 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6168 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6169 (cris_constant_alignment): New function.
6170 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6171 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6172 (epiphany_constant_alignment): New function.
6173 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6174 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6175 constant_alignment_word_strings.
6176 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6177 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6178 constant_alignment_word_strings.
6179 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6180 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6181 constant_alignment_word_strings.
6182 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6183 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6184 * config/i386/i386.c (ix86_constant_alignment): Make static.
6185 Use the same interface as the target hook.
6186 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6187 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6188 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6189 constant_alignment_word_strings.
6190 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6191 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6192 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6193 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6194 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6195 constant_alignment_word_strings.
6196 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6197 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6198 constant_alignment_word_strings.
6199 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6200 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6201 constant_alignment_word_strings.
6202 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6203 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6205 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6206 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6207 * config/mips/mips.c (mips_constant_alignment): New function.
6208 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6209 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6210 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6211 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6212 (mmix_constant_alignment): Make static. Use the same interface
6214 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6215 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6216 constant_alignment_word_strings.
6217 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6218 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6219 constant_alignment_word_strings.
6220 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6221 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6222 constant_alignment_word_strings.
6223 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6224 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6225 (rs6000_constant_alignment): New function.
6226 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6227 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6228 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6229 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6230 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6231 (rs6000_constant_alignment): New function.
6232 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6233 * config/s390/s390.c (s390_constant_alignment): New function.
6234 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6235 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6236 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6237 constant_alignment_word_strings.
6238 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6239 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6240 (sparc_constant_alignment): New function.
6241 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6242 * config/spu/spu.c (spu_constant_alignment): New function.
6243 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6244 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6245 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6246 constant_alignment_word_strings.
6247 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6248 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6249 constant_alignment_word_strings.
6250 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6251 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6252 constant_alignment_word_strings.
6253 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6254 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6255 (visium_constant_alignment): New function.
6256 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6257 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6258 (xtensa_constant_alignment): New function.
6259 * system.h (CONSTANT_ALIGNMENT): Poison.
6261 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6263 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6264 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6265 (rs6000_builtin_valid_without_lhs): New helper function.
6266 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6267 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6269 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6271 * target.h (vec_perm_indices): Use unsigned short rather than
6273 (auto_vec_perm_indices): Likewise.
6274 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6275 Use unsigned int rather than unsigned char.
6276 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6278 2017-09-25 Richard Biener <rguenther@suse.de>
6280 * cfgloop.h (sort_sibling_loops): Declare.
6281 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6282 (sort_sibling_loops): New function sorting the sibling loop list
6284 * graphite.c (graphite_transform_loops): Sort sibling loops.
6286 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6288 * target.def (vec_perm_const_ok): Change sel parameter to
6290 * optabs-query.c (can_vec_perm_p): Update accordingly.
6291 * doc/tm.texi: Regenerate.
6292 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6293 auto_vec_perm_indices and remove separate nelt field.
6294 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6295 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6296 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6297 (aarch64_expand_vec_perm_const): Update accordingly.
6298 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6299 to vec_perm_indices.
6300 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6301 auto_vec_perm_indices and remove separate nelt field.
6302 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6303 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6304 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6306 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6307 to vec_perm_indices.
6308 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6309 sel to vec_perm_indices.
6310 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6311 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6312 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6314 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6317 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6320 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6321 on the FUNCTION_DECL function context if it has a DIE that is a
6324 2017-09-25 Richard Biener <rguenther@suse.de>
6326 PR tree-optimization/82285
6327 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6330 2017-09-25 Tom de Vries <tom@codesourcery.com>
6334 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6337 2017-09-25 Richard Biener <rguenther@suse.de>
6339 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6340 ISL errors other than isl_error_quota happen. Dump if the
6341 schedule is the same.
6342 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6343 errors instead of aborting inside ISL.
6345 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6348 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6349 of libgcc_eh for m64.
6350 * config/i386/darwin64.h: Likewise.
6352 2017-09-25 Richard Biener <rguenther@suse.de>
6355 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6356 attribute for incomplete types nor twice for complete ones.
6358 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6361 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6362 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6365 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6368 * config/i386/i386.opt (mprefer-avx256): Use
6369 ix86_target_flags variable.
6370 * config/i386/i386.c (ix86_target_string): Move
6371 -mprefer-avx256 to flag2_opts.
6373 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6376 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6377 and x != -1 | y != -1 into (x & y) != -1.
6379 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6381 * config.gcc: Add new case statement to set
6382 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6383 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6384 s390x-*-linux* case statements. Added aarch64 to the list of
6385 supported architectures.
6387 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6389 PR tree-optimization/82289
6390 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6391 STMT_VINFO_RELEVANT_P.
6393 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6394 Alan Hayward <alan.hayward@arm.com>
6395 David Sherwood <david.sherwood@arm.com>
6397 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6398 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6399 (extract_range_from_binary_expr_1): Don't call
6400 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6402 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6403 Alan Hayward <alan.hayward@arm.com>
6404 David Sherwood <david.sherwood@arm.com>
6406 * target.def (preferred_vector_alignment): New hook.
6407 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6409 * doc/tm.texi: Regenerate.
6410 * targhooks.h (default_preferred_vector_alignment): Declare.
6411 * targhooks.c (default_preferred_vector_alignment): New function.
6412 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6414 (DR_TARGET_ALIGNMENT): New macro.
6415 (aligned_access_p): Update commentary.
6416 (vect_known_alignment_in_bytes): New function.
6417 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6419 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6420 Calculate the misalignment based on the target alignment rather than
6422 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6423 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6424 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6425 the target alignment, rather than masking the element misalignment
6426 with the number of elements in a vector. Also use the target
6427 alignment when calculating the maximum number of peels.
6428 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6429 instead of TYPE_ALIGN_UNIT.
6430 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6431 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6432 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6433 (vect_create_data_ref_ptr): Likewise.
6434 (vect_setup_realignment): Realign by ANDing with
6435 -DR_TARGET_MISALIGNMENT.
6436 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6437 the number of peels based on DR_TARGET_ALIGNMENT.
6438 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6439 with the guaranteed alignment boundary when deciding whether
6441 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6442 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6443 (ensure_base_align): Remove stmt_info parameter. Get the
6444 target base alignment from DR_TARGET_ALIGNMENT.
6445 (vectorizable_store): Update call accordingly. Interpret
6446 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6448 (vectorizable_load): Likewise.
6450 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6451 Alan Hayward <alan.hayward@arm.com>
6452 David Sherwood <david.sherwood@arm.com>
6454 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6455 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6456 (vect_enhance_data_refs_alignment): Likewise.
6458 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6460 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6461 error. Only quit immediately if parsing is complete.
6462 (BEGIN): Initialize fatal_err and parse_done.
6463 (begin fpu, end fpu): Check number of arguments.
6464 (begin arch, end arch): Likewise.
6465 (begin cpu, end cpu): Likewise.
6466 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6467 (optalias): Likewise.
6469 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6471 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6472 * config/arm/arm-isa.h: Delete. Move definitions to ...
6473 * arm-cpus.in: ... here. Use new feature and fgroup values.
6474 * config/arm/arm.c (arm_option_override): Use lower case for feature
6476 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6477 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6478 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6480 (print_isa_bits_for): New function.
6481 (gen_isa): New function.
6482 (gen_comm_data): Use print_isa_bits_for.
6483 (define feature): New keyword.
6484 (define fgroup): New keyword.
6485 * config/arm/t-arm (TM_H): Remove.
6486 (GTM_H): Add arm-isa.h.
6487 (arm-isa.h): Add rule to generate file.
6488 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6489 case for feature bit names.
6491 2017-09-22 Richard Biener <rguenther@suse.de>
6493 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6495 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6496 print of no dependency loops ...
6497 * graphite.c (graphite_transform_loops): ... here.
6498 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6500 (same_close_phi_node, remove_duplicate_close_phi,
6501 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6502 (canonicalize_loop_closed_ssa): ... here and simplify.
6503 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6504 (optimize_isl): Use dump_printf_loc to tell when we stopped
6505 optimizing because of an ISL timeout.
6507 2017-09-22 Richard Biener <rguenther@suse.de>
6509 PR tree-optimization/82291
6510 * tree-if-conv.c (predicate_mem_writes): Make sure to
6511 remove writes in blocks predicated with false.
6513 2017-09-22 Richard Biener <rguenther@suse.de>
6515 * sese.c: Include cfganal.h.
6516 (if_region_set_false_region): Remove.
6517 (create_if_region_on_edge): Likewise.
6518 (move_sese_in_condition): Re-implement without destroying
6521 2017-09-22 Richard Biener <rguenther@suse.de>
6523 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6524 Verify both BBs contain loop PHI nodes before dispatching to
6526 (graphite_regenerate_ast_isl): Do not recompute dominators,
6527 do not verify three times. Restructure for clarity.
6528 * graphite-scop-detection.c (same_close_phi_node,
6529 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6530 defined_in_loop_p, canonicalize_loop_closed_ssa,
6531 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6532 checking and SSA rewrite, move to ...
6533 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6534 (graphite_initialize): Do not pass in ctx, do not reset the
6535 SCEV cache, compute only dominators.
6536 (graphite_transform_loops): Allocate ISL ctx after
6537 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6538 Maintain post-dominators only around build_scops.
6539 * sese.c (if_region_set_false_region): Make static. Free
6540 and recompute dominators.
6541 (move_sese_in_condition): Assert we don't get called with
6542 post-dominators computed.
6543 * sese.h (if_region_set_false_region): Remove.
6545 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6547 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6548 mode attribute for TARGET_AVX512VL.
6550 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6552 * config/i386/i386.opt (mprefer-avx256): New option.
6553 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6555 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6556 for TARGET_PREFER_AVX256.
6557 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6559 2017-09-21 Jeff Law <law@redhat.com>
6561 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6562 Fix dump output if the only stack space is for pushed registers.
6564 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6566 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6569 2017-09-21 Martin Sebor <msebor@redhat.com>
6572 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6573 code (in C++) or code that triggers warnings.
6575 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6577 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6579 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6581 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6582 * sched-rgn.c: Ditto.
6583 * sel-sched-ir.c: Ditto.
6585 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6587 * toplev.h (set_random_seed): Adjust return type.
6588 * toplev.c (init_local_tick): Move eager initialization of random_seed
6589 to get_random_seed. Adjust comment.
6590 (init_random_seed): Inline to get_random_seed, delete.
6591 (get_random_seed): Initialize random_seed lazily.
6592 (set_random_seed): Do not return previous value.
6593 (print_switch_value): Do not call get_random_seed.
6595 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6597 * cgraph.c (delete_function_version): New, broken out from...
6598 (cgraph_node::delete_function_version): ...here. Rename to
6599 cgraph_node::delete_function_version_by_decl. Update all uses.
6600 (cgraph_node::remove): Call delete_function_version.
6602 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6605 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6606 VAR_DECLs to which addressable non-volatile parameters are mapped
6607 and for id->retvar after the return value assignment. Clear
6608 id->retval and id->retbnd after inlining.
6610 2017-09-21 Richard Biener <rguenther@suse.de>
6612 PR tree-optimization/82276
6613 PR tree-optimization/82244
6614 * tree-vrp.c (build_assert_expr_for): Set
6615 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6617 (remove_range_assertions): Revert earlier change.
6619 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6622 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6624 2017-09-21 Richard Biener <rguenther@suse.de>
6626 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6627 Restore valid IL after code generation errors.
6628 * graphite.c (graphite_transform_loops): Diagnose code
6629 generation issues as MSG_MISSED_OPTIMIZATION and continue
6630 with processing SCOPs.
6632 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6633 Alan Hayward <alan.hayward@arm.com>
6634 David Sherwood <david.sherwood@arm.com>
6636 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6637 rather than choosing between plus_constant and gen_rtx_<CODE>.
6638 * expr.c (emit_push_insn): Likewise.
6639 (expand_expr_real_2): Likewise.
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 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6647 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6648 Alan Hayward <alan.hayward@arm.com>
6649 David Sherwood <david.sherwood@arm.com>
6651 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6652 calling tree_to_uhwi.
6654 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6655 Alan Hayward <alan.hayward@arm.com>
6656 David Sherwood <david.sherwood@arm.com>
6658 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6659 INTEGER_CST rather than a negative test for ADDR_EXPR.
6661 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6662 Alan Hayward <alan.hayward@arm.com>
6663 David Sherwood <david.sherwood@arm.com>
6665 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6666 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6668 2017-09-21 Richard Biener <rguenther@suse.de>
6670 PR tree-optimization/71351
6671 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6672 graphite_create_new_loop_guard): Remove, fold remaining parts
6674 (translate_isl_ast_node_for): ... here and simplify.
6676 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6679 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6680 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6681 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6682 alternative always use QI mode, for -Os imov (=R,R) alternative
6683 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6686 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6687 Jeff Law <law@redhat.com>
6689 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6690 (allocate_stack_space): New function, partially extracted from
6692 (s390_emit_prologue): Track offset to most recent stack probe.
6693 Code to allocate space moved into allocate_stack_space.
6694 Dump actions when no stack is allocated.
6695 (s390_prologue_plus_offset): New function.
6696 (s390_emit_stack_probe): Likewise.
6698 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6700 * common.opt (Wa, Wl, Wp, g, gz=): Add
6702 (gno-column-info): Remove.
6703 (gcolumn-info): Drop RejectNegative.
6705 (gno-record-gcc-switches): Remove.
6706 (grecord-gcc-switches): Drop RejectNegative.
6707 (gno-split-dwarf): Remove.
6708 (gsplit-dwarf): Drop RejectNegative.
6709 (gno-strict-dwarf): Remove.
6710 (gstrict-dwarf): Drop RejectNegative.
6711 * config/darwin.opt (gfull, gused): Add RejectNegative.
6712 * dwarf2out.c (gen_producer_string): Drop
6713 gno-record-gcc-switches handler.
6714 * optc-gen.awk: Add g to prefixes with negative forms.
6715 * opts-common.c (remapping_prefix_p): New.
6716 (find_opt): Check it.
6717 (generate_canonical_option): Test g prefix.
6718 (option_map): Add -gno- mapping.
6719 (add_misspelling_candidates): Check remapping_prefix_p.
6721 2017-09-20 Jeff Law <law@redhat.com>
6723 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6724 thinko in stack clash protection support.
6726 * explow.c (compute_stack_clash_protection_loop_data): Use
6727 CONST_INT_P instead of explicit test. Verify object is a
6728 CONST_INT_P before looking at INTVAL.
6729 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6730 instead of explicit test.
6732 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6735 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6736 address instead of to r1 and r11.
6738 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6740 * config.gcc: Support "knm".
6741 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6742 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6744 * config/i386/i386.c (m_KNM): Define.
6745 (processor_target_table): Add "knm".
6747 (ix86_option_override_internal): Add "knm".
6748 (ix86_issue_rate): Add PROCESSOR_KNM.
6749 (ix86_adjust_cost): Ditto.
6750 (ia32_multipass_dfa_lookahead): Ditto.
6751 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6752 (fold_builtin_cpu): Add M_INTEL_KNM.
6753 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6754 (processor_type): Add PROCESSOR_KNM.
6755 * config/i386/x86-tune.def: Add m_KNM.
6756 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6758 2017-09-20 Richard Biener <rguenther@suse.de>
6760 PR tree-optimization/80213
6761 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6762 are allowed in empty BBs as well.
6763 (canonicalize_loop_closed_ssa): Also look for other complex
6765 (scop_detection::get_sese): Include the loop-closed PHI block
6767 (scop_detection::merge_sese): Remove code adding extra blocks.
6768 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6769 (build_scops): Assert the final returned scop is invalid.
6771 2017-09-20 Richard Biener <rguenther@suse.de>
6773 PR tree-optimization/82264
6774 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6776 (vn_phi_lookup): Likewise.
6777 (vn_phi_insert): Likewise.
6779 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6781 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6782 that fits into uhwi or shwi, add DW_AT_const_value regardless
6783 of early_dwarf without going through RTL, using add_AT_unsigned
6786 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6787 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6788 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6789 (DEBUG_MACRO_SECTION): Likewise.
6790 (DEBUG_LTO_MACRO_SECTION): Likewise.
6791 (DEBUG_STR_DWO_SECTION): Likewise.
6792 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6793 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6794 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6795 (DEBUG_LTO_LINE_STR_SECTION): Define.
6796 (init_sections_and_labels): Initialize debug_line_str_section
6797 variable. Initialize debug_loc_section for -gdwarf-5 to
6798 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6800 2017-09-20 Richard Biener <rguenther@suse.de>
6802 * graphite-sese-to-poly.c (extract_affine): Properly handle
6803 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6805 2017-09-20 Richard Biener <rguenther@suse.de>
6807 PR tree-optimization/81373
6808 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6809 Force SESE live-out defs to be handled even if they are
6812 2017-09-19 Jeff Law <law@redhat.com>
6814 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6815 nothing for stack adjustments with REG_STACK_CHECK.
6816 * sched-deps.c (parse_add_or_inc): Reject insns with
6817 REG_STACK_CHECK from dependency breaking.
6818 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6819 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6820 * reg-notes.def (STACK_CHECK): New note.
6822 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6823 (ix86_expand_prologue): Dump stack clash info as needed.
6824 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6826 * function.c (dump_stack_clash_frame_info): New function.
6827 * function.h (dump_stack_clash_frame_info): Prototype.
6828 (enum stack_clash_probes): New enum.
6830 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6831 flag_stack_clash_protection.
6832 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6833 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6834 (arm_frame_pointer_required): Likewise.
6835 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6836 (ia64_expand_prologue): Likewise.
6837 * config/mips/mips.c (mips_expand_prologue): Likewise.
6838 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6839 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6840 (sparc_flat_expand_prologue): Likewise.
6841 * config/spu/spu.c (spu_expand_prologue): Likewise.
6843 * explow.c: Include "params.h".
6844 (anti_adjust_stack_and_probe_stack_clash): New function.
6845 (get_stack_check_protect): Likewise.
6846 (compute_stack_clash_protection_loop_data): Likewise.
6847 (emit_stack_clash_protection_loop_start): Likewise.
6848 (emit_stack_clash_protection_loop_end): Likewise.
6849 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6850 Use anti_adjust_stack_and_probe_stack_clash.
6851 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6852 (emit_stack_clash_protection_loop_start): Likewise.
6853 (emit_stack_clash_protection_loop_end): Likewise.
6854 * rtl.h (get_stack_check_protect): Prototype.
6855 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6856 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6857 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6859 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6861 * doc/tm.texi: Rebuilt.
6862 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6863 get_stack_check_protect.
6864 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6865 * config/arm/arm.c (arm_expand_prologue): Likewise.
6866 (arm_frame_pointer_required): Likewise.
6867 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6868 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6869 * config/mips/mips.c (mips_expand_prologue): Likewise.
6870 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6871 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6872 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6873 (sparc_flat_expand_prologue): Likewise.
6875 * common.opt (-fstack-clash-protection): New option.
6876 * flag-types.h (enum stack_check_type): Note difference between
6877 -fstack-check= and -fstack-clash-protection.
6878 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6879 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6880 * toplev.c (process_options): Issue warnings/errors for cases
6881 not handled with -fstack-clash-protection.
6882 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6883 (-fstack-check): Note additional problem with -fstack-check=generic.
6884 Note that -fstack-check is primarily for Ada and refer users
6885 to -fstack-clash-protection for stack-clash-protection.
6886 Document new params for stack clash protection.
6888 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6890 * config/i386/i386.c (ix86_split_long_move): Do not handle
6891 address used for LEA in a special way.
6893 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6895 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6897 2017-09-19 Martin Sebor <msebor@redhat.com>
6900 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6901 of incompatible types.
6903 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6905 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6906 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6907 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6908 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6910 2017-09-19 Richard Biener <rguenther@suse.de>
6912 PR tree-optimization/82244
6913 * tree-vrp.c (remove_range_assertions): Do not propagate
6914 a constant to abnormals but replace the assert with a copy.
6916 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6918 PR rtl-optimization/57878
6919 PR rtl-optimization/68988
6920 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6921 avoidance test involving non_reload_pseudos. Move frequency test
6922 below the general fragmentation avoidance test.
6924 2017-09-19 Richard Biener <rguenther@suse.de>
6926 PR tree-optimization/69728
6927 * graphite-sese-to-poly.c (schedule_error): New global.
6928 (add_loop_schedule): Handle empty domain by failing the
6930 (build_original_schedule): Handle schedule_error.
6932 2017-09-19 Richard Biener <rguenther@suse.de>
6934 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6935 Do not iterate to sibling loops but only to siblings of inner
6938 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6941 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6942 that the registers are different.
6944 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6946 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6947 to processor_model and "amdfam17h" to arch_names_table.
6948 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6950 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6953 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6955 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6956 Alan Hayward <alan.hayward@arm.com>
6957 David Sherwood <david.sherwood@arm.com>
6959 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6961 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6963 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6964 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6965 vect_schedule_slp_instance.
6966 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6967 Update call to vect_slp_analyze_node_operations. Simplify return
6969 (vect_slp_analyze_bb_1): Update call accordingly.
6970 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6971 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6972 (vect_schedule_slp): Update call accordingly.
6974 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6975 Alan Hayward <alan.hayward@arm.com>
6976 David Sherwood <david.sherwood@arm.com>
6978 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6979 with types that aren't in fact scalar.
6981 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6983 * tree-vect-slp.c (vect_record_max_nunits): New function,
6985 (vect_build_slp_tree_1): ...here.
6986 (vect_build_slp_tree_2): Call it for phis too.
6988 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6990 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6991 to vect_get_vec_def_for_operand when getting the mask operand.
6993 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6994 Alan Hayward <alan.hayward@arm.com>
6995 David Sherwood <david.sherwood@arm.com>
6997 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
7000 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7001 Alan Hayward <alan.hayward@arm.com>
7002 David Sherwood <david.sherwood@arm.com>
7004 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
7005 calculation for vector booleans.
7007 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
7008 Alan Hayward <alan.hayward@arm.com>
7009 David Sherwood <david.sherwood@arm.com>
7011 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
7013 (vect_transform_stmt): ...here.
7014 (vect_analyze_stmt): Use it instead of calling
7015 vectorizable_live_operation directly.
7017 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
7019 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
7020 non-SIMT targets in acc vector loops.
7022 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
7024 * configure.ac: Add arc and check if assembler supports gdwarf2.
7025 * configure: Regenerate.
7027 2017-09-18 Richard Biener <rguenther@suse.de>
7029 PR tree-optimization/82220
7030 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7031 epilogue niters from the min_profitable_iters compute.
7033 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7036 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7037 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7038 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7040 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7043 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7044 switching to a new text section.
7046 2017-09-18 Richard Biener <rguenther@suse.de>
7048 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7050 (build_alias_set): Reject aliases with no access function.
7052 2017-09-18 Richard Biener <rguenther@suse.de>
7054 PR tree-optimization/79622
7055 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7057 (build_cross_bb_scalars_use): Likewise.
7059 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7061 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7063 2017-09-18 Alan Modra <amodra@gmail.com>
7066 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7067 stack_pointer_rtx for count 0. Update comments. Break up
7068 large rtl expression.
7070 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7072 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7073 Increase to 20 bytes.
7074 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7075 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7076 or avx version of the stub.
7078 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7081 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7082 compute the minimum stack alignment. Also update preferred stack
7083 boundary for leaf functions.
7085 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7087 PR tree-optimization/82228
7088 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7091 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7093 * common/config/nds32/nds32-common.c
7094 (nds32_option_optimization_table): Refine formatting.
7095 (nds32_option_optimization_table): Use -fsched-pressure and
7096 -fomit-frame-pointer for specific optimization level.
7098 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7100 * config/nds32/nds32.c: Refine formatting and comments.
7101 * config/nds32/nds32.h: Likewise.
7102 * config/nds32/nds32.md: Likewise.
7103 * config/nds32/nds32-cost.c: Likewise.
7104 * config/nds32/nds32-isr.c: Likewise.
7105 * config/nds32/nds32-md-auxiliary.c: Likewise.
7106 * config/nds32/nds32-multiple.md: Likewise.
7107 * config/nds32/nds32-predicates.c: Likewise.
7109 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7110 Jakub Jelinek <jakub@redhat.com>
7112 Add support for -std=c++2a.
7113 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7115 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7117 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7120 * doc/extend.texi (Common Function Attributes): Add
7121 references to ARM, AArch64, and S/390 specific attributes.
7122 (Function Specific Option Pragmas): Add AArch64 and S/390
7123 to list of back ends that support the target pragma.
7125 2017-09-15 Nathan Sidwell <nathan@acm.org>
7127 * doc/standards.texi: Fix C++17 description. Update URLs for
7130 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7132 * common.opt (Wcast-align=strict): New warning option.
7133 * doc/invoke.texi: Document -Wcast-align=strict.
7135 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7137 * cgraph.h (cgraph_thunk_info): Add comments.
7138 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7139 assert for VIRTUAL_* arguments stricter.
7141 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7143 PR tree-optimization/71026
7144 * match.pd: Move RDIV patterns from fold-const.c
7145 * fold-const.c (distribute_real_division): Removed.
7146 (fold_binary_loc): Remove calls to distribute_real_divison.
7148 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7150 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7151 c++1z and gnu++1z as deprecated. Change other references to
7152 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7153 Change -Wc++1z-compat to -Wc++17-compat.
7154 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7155 * dwarf2out.c (highest_c_language): Handle C++17.
7156 (gen_compile_unit_die): Likewise.
7158 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7160 PR rtl-optimization/82192
7161 * combine.c (make_extraction): Don't look through non-paradoxical
7162 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7165 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7166 Alan Hayward <alan.hayward@arm.com>
7167 David Sherwood <david.sherwood@arm.com>
7169 * target.def (function_arg_offset): New hook.
7170 * targhooks.h (default_function_arg_offset): Declare.
7171 * targhooks.c (default_function_arg_offset): New function.
7172 * function.c (locate_and_pad_parm): Use
7173 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7174 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7175 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7176 * doc/tm.texi: Regenerate.
7177 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7178 * config/spu/spu.c (spu_function_arg_offset): New function.
7179 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7180 * system.h (FUNCTION_ARG_OFFSET): Poison.
7182 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7183 Alan Hayard <alan.hayward@arm.com>
7184 David Sherwood <david.sherwood@arm.com>
7186 * target.def (truly_noop_truncation): New hook.
7187 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7188 than TRULY_NOOP_TRUNCATION.
7189 * hooks.h (hook_bool_uint_uint_true): Declare.
7190 * hooks.c (hook_bool_uint_uint_true): New function.
7191 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7192 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7193 * doc/tm.texi: Regenerate.
7194 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7195 rather than TRULY_NOOP_TRUNCATION in comments.
7196 (simplify_comparison): Likewise.
7197 (record_truncated_value): Likewise.
7198 * expmed.c (extract_bit_field_1): Likewise.
7199 (extract_split_bit_field): Likewise.
7200 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7201 instead of TRULY_NOOP_TRUNCATION.
7202 * function.c (assign_parm_setup_block): Likewise.
7203 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7204 * rtlhooks.c: Include target.h.
7205 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7206 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7207 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7208 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7209 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7210 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7211 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7212 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7213 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7214 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7215 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7216 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7217 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7218 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7219 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7220 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7221 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7222 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7223 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7224 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7225 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7226 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7227 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7228 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7229 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7230 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7231 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7232 rather than TRULY_NOOP_TRUNCATION in comments.
7233 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7234 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7235 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7236 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7237 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7238 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7239 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7240 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7241 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7242 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7243 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7244 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7245 rather than TRULY_NOOP_TRUNCATION in comments.
7246 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7247 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7248 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7249 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7250 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7251 TRULY_NOOP_TRUNCATION condition.
7252 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7253 (TRULY_NOOP_TRUNCATION): Delete.
7254 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7255 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7256 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7257 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7258 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7259 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7260 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7261 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7262 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7263 rather than TRULY_NOOP_TRUNCATION in comments.
7264 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7265 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7266 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7267 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7268 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7269 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7271 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7274 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7275 (*cmp_ior): Likewise.
7276 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7277 (*ior_scc_scc_cmp): Likewise.
7278 (*and_scc_scc): Likewise.
7279 (*and_scc_scc_cmp): Likewise.
7281 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7282 Alan Hayard <alan.hayward@arm.com>
7283 David Sherwood <david.sherwood@arm.com>
7285 * target.def (can_change_mode_class): New hook.
7286 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7287 (hard_regno_nregs): Likewise.
7288 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7289 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7290 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7291 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7292 (LOAD_EXTEND_OP): Update accordingly.
7293 * doc/tm.texi: Regenerate.
7294 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7295 CANNOT_CHANGE_MODE_CLASS.
7296 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7297 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7298 * combine.c (simplify_set): Update accordingly.
7299 * emit-rtl.c (validate_subreg): Likewise.
7300 * recog.c (general_operand): Likewise.
7301 * regcprop.c (mode_change_ok): Likewise.
7302 * reload1.c (choose_reload_regs): Likewise.
7303 (inherit_piecemeal_p): Likewise.
7304 * rtlanal.c (simplify_subreg_regno): Likewise.
7305 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7306 instead of CANNOT_CHANGE_MODE_CLASS.
7307 (reload_cse_simplify_operands): Likewise.
7308 * reload.c (push_reload): Use targetm.can_change_mode_class
7309 instead of CANNOT_CHANGE_MODE_CLASS.
7310 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7311 REG_CANNOT_CHANGE_MODE_P.
7312 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7313 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7314 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7315 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7316 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7317 (arm_can_change_mode_class): New function.
7318 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7319 than CANNOT_CHANGE_MODE_CLASS in comments.
7320 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7321 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7322 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7323 (ix86_can_change_mode_class): ...this new function, inverting the
7324 sense of the return value.
7325 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7326 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7327 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7328 (ia64_can_change_mode_class): New function.
7329 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7330 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7331 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7332 (m32c_can_change_mode_class): ...this new function, inverting the
7333 sense of the return value.
7334 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7335 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7336 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7337 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7338 (mips_can_change_mode_class): ...this new function, inverting the
7339 sense of the return value.
7340 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7341 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7342 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7343 (msp430_can_change_mode_class): New function.
7344 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7345 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7346 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7347 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7348 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7349 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7350 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7351 (pa_cannot_change_mode_class): Replace with...
7352 (pa_can_change_mode_class): ...this new function, inverting the
7353 sense of the return value.
7354 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7355 than CANNOT_CHANGE_MODE_CLASS in comments.
7356 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7357 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7358 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7359 (pdp11_cannot_change_mode_class): Replace with...
7360 (pdp11_can_change_mode_class): ...this new function, inverting the
7361 sense of the return value.
7362 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7363 * config/powerpcspe/powerpcspe-protos.h
7364 (rs6000_cannot_change_mode_class_ptr): Delete.
7365 * config/powerpcspe/powerpcspe.c
7366 (rs6000_cannot_change_mode_class_ptr): Delete.
7367 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7368 (rs6000_option_override_internal): Assign to
7369 targetm.can_change_mode_class instead of
7370 rs6000_cannot_change_mode_class_ptr.
7371 (rs6000_cannot_change_mode_class): Replace with...
7372 (rs6000_can_change_mode_class): ...this new function, inverting the
7373 sense of the return value.
7374 (rs6000_debug_cannot_change_mode_class): Replace with...
7375 (rs6000_debug_can_change_mode_class): ...this new function.
7376 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7377 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7378 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7379 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7380 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7382 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7383 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7384 (rs6000_option_override_internal): Assign to
7385 targetm.can_change_mode_class instead of
7386 rs6000_cannot_change_mode_class_ptr.
7387 (rs6000_cannot_change_mode_class): Replace with...
7388 (rs6000_can_change_mode_class): ...this new function, inverting the
7389 sense of the return value.
7390 (rs6000_debug_cannot_change_mode_class): Replace with...
7391 (rs6000_debug_can_change_mode_class): ...this new function.
7392 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7393 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7394 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7395 (s390_can_change_mode_class): ...this new function, inverting the
7396 sense of the return value.
7397 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7398 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7399 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7400 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7401 (sh_cannot_change_mode_class): Replace with...
7402 (sh_can_change_mode_class): ...this new function, inverting the
7403 sense of the return value.
7404 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7405 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7406 (sparc_can_change_mode_class): New function.
7407 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7408 * config/spu/spu.c (spu_can_change_mode_class): New function.
7409 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7410 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7411 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7412 (visium_can_change_mode_class): New function.
7413 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7415 2017-09-15 Richard Biener <rguenther@suse.de>
7417 PR tree-optimization/82217
7418 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7419 but not undefined case.
7421 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7424 * postreload.c (reload_cse_simplify_operands): Skip
7425 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7427 2017-09-15 Richard Biener <rguenther@suse.de>
7429 PR tree-optimization/68823
7430 * graphite-scop-detection.c (build_alias_set): If we have a
7431 possible dependence check whether we can handle them by just
7432 looking at the DRs DR_ACCESS_FNs.
7433 (build_scops): If build_alias_set fails, fail the SCOP.
7435 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7437 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7438 to support float128 built-in functions that require the ISA 3.0
7440 (BU_FLOAT128_3_HW): Likewise.
7441 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7443 (FMAF128): Likewise.
7445 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7446 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7447 floating point instructions.
7448 (rs6000_invalid_builtin): Likewise.
7449 (rs6000_builtin_mask_names): Likewise.
7450 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7451 (RS6000_BTM_FLOAT128_HW): Likewise.
7452 (RS6000_BTM_COMMON): Likewise.
7453 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7455 * doc/extend.texi (RS/6000 built-in functions): Document the
7456 IEEE 128-bit floating point square root and fused multiply-add
7459 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7461 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7462 reg (r2) isn't in the set of registers defined in the prologue.
7464 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7465 Alan Hayward <alan.hayward@arm.com>
7466 David Sherwood <david.sherwood@arm.com>
7468 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7469 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7470 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7471 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7472 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7474 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7475 max_vectorization_factor.
7476 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7478 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7479 Alan Hayward <alan.hayward@arm.com>
7480 David Sherwood <david.sherwood@arm.com>
7482 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7483 (vect_worthwhile_without_simd_p): Declare.
7484 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7485 (vectorizable_reduction): Use it.
7486 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7487 (vectorizable_operation): Likewise.
7489 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7490 Alan Hayward <alan.hayward@arm.com>
7491 David Sherwood <david.sherwood@arm.com>
7493 * tree-vectorizer.h (vect_get_num_copies): New function.
7494 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7495 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7496 (vectorizable_induction): Likewise.
7497 (vectorizable_live_operation): Likewise.
7498 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7499 (vectorizable_bswap): Likewise.
7500 (vectorizable_call): Likewise.
7501 (vectorizable_conversion): Likewise.
7502 (vectorizable_assignment): Likewise.
7503 (vectorizable_shift): Likewise.
7504 (vectorizable_operation): Likewise.
7505 (vectorizable_store): Likewise.
7506 (vectorizable_load): Likewise.
7507 (vectorizable_condition): Likewise.
7508 (vectorizable_comparison): Likewise.
7509 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7511 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7512 Alan Hayward <alan.hayward@arm.com>
7513 David Sherwood <david.sherwood@arm.com>
7515 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7516 of vect_init_vector.
7518 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7519 Alan Hayward <alan.hayward@arm.com>
7520 David Sherwood <david.sherwood@arm.com>
7522 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7523 an inline wrapper that provides a location.
7524 (gimple_build_vector): Likewise.
7525 * gimple-fold.c (gimple_build_vector_from_val): New function.
7526 (gimple_build_vector): Likewise.
7527 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7528 functions to build the initial value. Always return a gimple value.
7529 (get_initial_defs_for_reduction): Likewise. Only compute
7531 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7532 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7533 (vectorizable_induction): Use gimple_build_vector rather than
7536 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7537 Alan Hayward <alan.hayward@arm.com>
7538 David Sherwood <david.sherwood@arm.com>
7540 * target.h (vec_perm_indices): New typedef.
7541 (auto_vec_perm_indices): Likewise.
7542 * optabs-query.h: Include target.h
7543 (can_vec_perm_p): Take a vec_perm_indices *.
7544 * optabs-query.c (can_vec_perm_p): Likewise.
7545 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7546 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7547 * tree-vect-generic.c (lower_vec_perm): Likewise.
7548 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7549 (vect_grouped_load_supported): Likewise.
7550 (vect_shift_permute_load_chain): Likewise.
7551 (vect_permute_store_chain): Use auto_vec_perm_indices.
7552 (vect_permute_load_chain): Likewise.
7553 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7554 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7555 Update uses of can_vec_perm_p.
7556 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7557 mode with a number of elements. Take a vec_perm_indices *.
7558 (vect_create_epilog_for_reduction): Update accordingly.
7559 Use auto_vec_perm_indices.
7560 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7561 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7562 (vect_transform_slp_perm_load): Likewise.
7563 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7564 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7565 (vect_gen_perm_mask_checked): Likewise.
7566 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7567 (vect_gen_perm_mask_checked): Likewise.
7568 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7569 (vectorizable_store): Likewise.
7570 (vectorizable_load): Likewise.
7571 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7572 (vectorizable_bswap): Likewise.
7574 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7575 Alan Hayward <alan.hayward@arm.com>
7576 David Sherwood <david.sherwood@arm.com>
7578 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7579 * tree.c (build_vector): Likewise.
7580 (build_vector_from_ctor): Update accordingly.
7581 (build_vector_from_val): Likewise.
7582 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7583 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7584 * tree-vect-generic.c (add_rshift): Likewise.
7585 (expand_vector_divmod): Likewise.
7586 (optimize_vector_constructor): Likewise.
7587 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7588 (vect_transform_slp_perm_load): Likewise.
7589 (vect_schedule_slp_instance): Likewise.
7590 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7591 (vectorizable_call): Likewise.
7592 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7593 * expmed.c (make_tree): Likewise.
7594 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7595 a vector passed to build_vector.
7596 (fold_convert_const): Likewise.
7597 (exact_inverse): Likewise.
7598 (fold_ternary_loc): Likewise.
7599 (fold_relational_const): Likewise.
7600 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7601 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7602 (const_unop): Likewise. Store the reduction accumulator in a
7603 variable rather than an array.
7604 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7605 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7606 the new vector, rather than constructing it after the input arrays.
7607 (native_interpret_vector): Use auto_vec<tree> when building
7608 a vector passed to build_vector. Add elements in order.
7609 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7610 auto_vec<tree> when building a vector passed to build_vector.
7611 (vect_create_epilog_for_reduction): Likewise.
7612 (vectorizable_induction): Likewise.
7613 (get_initial_def_for_reduction): Likewise. Fix indentation of
7615 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7617 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7618 passed to build_vector.
7620 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7621 Alan Hayward <alan.hayward@arm.com>
7622 David Sherwood <david.sherwood@arm.com>
7624 * tree-core.h (tree_base::u): Add an "nelts" field.
7625 (tree_vector): Use VECTOR_CST_NELTS as the length.
7626 * tree.c (tree_size): Likewise.
7627 (make_vector): Initialize VECTOR_CST_NELTS.
7628 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7629 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7630 TYPE_VECTOR_SUBPARTS.
7631 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7632 as the number of units, setting it from VECTOR_CST_NELTS.
7633 (const_vector_from_tree): Likewise.
7634 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7635 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7636 (fold_negate_expr_1): Likewise.
7637 (fold_convert_const): Likewise.
7638 (const_binop): Likewise. Differentiate the number of output and
7640 (const_unop): Likewise.
7641 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7642 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7643 in cases that did the opposite.
7645 2017-09-14 Richard Biener <rguenther@suse.de>
7647 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7650 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7652 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7654 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7657 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7658 if and where to split a bb, except for splitting before debug insn
7659 sequences followed by non-label real insn. Delete debug insns
7660 in between basic blocks.
7662 * combine.c (make_compound_operation_int): Formatting fixes.
7664 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7665 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7666 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7667 * config/sol2.h (LINK_EH_SPEC): Likewise.
7668 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7669 * config/s390/linux.h (LINK_SPEC): Likewise.
7670 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7671 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7672 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7673 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7674 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7675 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7676 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7677 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7679 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7681 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7683 (ENDFILE_LINUX_SPEC): Likewise.
7684 (LINK_EH_SPEC): Likewise.
7685 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7686 (LINK_OS_LINUX_SPEC32): Likewise.
7687 (LINK_OS_LINUX_SPEC64): Likewise.
7688 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7689 (LINK_OS_LINUX_SPEC): Likewise.
7691 2017-09-13 Martin Liska <mliska@suse.cz>
7694 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7695 CASE_HIGH is NULL_TREE.
7697 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7698 Alan Hayward <alan.hayward@arm.com>
7699 David Sherwood <david.sherwood@arm.com>
7701 * target.def (secondary_memory_needed): New hook.
7702 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7703 instead of SECONDARY_MEMORY_NEEDED.
7704 (secondary_memory_needed_mode): Likewise.
7705 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7706 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7707 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7708 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7709 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7710 * doc/tm.texi: Regenerate.
7711 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7712 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7713 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7714 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7715 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7716 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7717 mode argument first and change the reg_class arguments to reg_class_t.
7718 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7719 Make static. Update the call to inline_secondary_memory_needed.
7720 (ix86_register_move_cost): Update the call to
7721 inline_secondary_memory_needed.
7722 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7723 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7725 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7726 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7728 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7729 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7730 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7731 and match hook interface. Add comment from mips.h.
7732 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7733 * config/mmix/mmix.md (truncdfsf2): Refer to
7734 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7736 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7737 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7738 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7739 (pa_secondary_memory_needed): New function.
7740 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7741 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7742 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7743 (pdp11_secondary_memory_needed): Make static and match hook interface.
7744 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7745 * config/powerpcspe/powerpcspe-protos.h
7746 (rs6000_secondary_memory_needed_ptr): Delete.
7747 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7749 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7750 (rs6000_option_override_internal): Assign to
7751 targetm.secondary_memory_needed rather than
7752 rs6000_secondary_memory_needed_ptr.
7753 (rs6000_secondary_memory_needed): Match hook interface.
7754 (rs6000_debug_secondary_memory_needed): Likewise.
7755 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7756 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7757 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7758 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7759 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7760 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7762 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7763 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7764 (rs6000_option_override_internal): Assign to
7765 targetm.secondary_memory_needed rather than
7766 rs6000_secondary_memory_needed_ptr.
7767 (rs6000_secondary_memory_needed): Match hook interface.
7768 (rs6000_debug_secondary_memory_needed): Likewise.
7769 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7770 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7771 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7772 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7773 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7774 (sparc_secondary_memory_needed): New function.
7775 * lra-constraints.c (check_and_process_move): Refer to
7776 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7778 (curr_insn_transform): Likewise.
7779 (process_alt_operands): Use targetm.secondary_memory_needed
7780 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7781 (check_secondary_memory_needed_p): Likewise.
7782 (choose_split_class): Likewise.
7783 * reload.c: Unconditionally include code that was previously
7784 conditional on SECONDARY_MEMORY_NEEDED.
7785 (push_secondary_reload): Use targetm.secondary_memory_needed
7786 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7787 (push_reload): Likewise.
7788 * reload1.c: Unconditionally include code that was previously
7789 conditional on SECONDARY_MEMORY_NEEDED.
7790 (choose_reload_regs): Use targetm.secondary_memory_needed
7791 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7792 (gen_reload): Likewise.
7793 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7795 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7796 Alan Hayward <alan.hayward@arm.com>
7797 David Sherwood <david.sherwood@arm.com>
7799 * target.def (secondary_memory_needed_mode): New hook:
7800 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7801 * targhooks.h (default_secondary_memory_needed_mode): New function.
7802 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7803 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7804 * doc/tm.texi: Regenerate.
7805 * lra-constraints.c (check_and_process_move): Use
7806 targetm.secondary_memory_needed_mode instead of
7807 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7808 (curr_insn_transform): Likewise.
7809 * reload.c (get_secondary_mem): Likewise.
7810 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7811 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7813 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7814 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7815 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7816 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7817 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7819 * config/powerpcspe/powerpcspe-protos.h
7820 (rs6000_secondary_memory_needed_mode): Delete.
7821 * config/powerpcspe/powerpcspe.c
7822 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7823 (rs6000_secondary_memory_needed_mode): Make static.
7824 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7825 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7827 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7829 (rs6000_secondary_memory_needed_mode): Make static.
7830 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7831 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7832 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7833 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7834 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7836 (sparc_secondary_memory_needed_mode): New function.
7837 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7839 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7841 * config/aarch64/constraints.md (Umq): New constraint.
7842 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7844 (mov<mode>): Update condition.
7846 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7848 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7849 when bitposition is the same.
7851 2017-09-13 Richard Biener <rguenther@suse.de>
7853 * dwarf2out.c (output_die_symbol): Remove.
7854 (output_die): Do not output a DIEs symbol.
7856 2017-09-13 Richard Biener <rguenther@suse.de>
7859 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7860 default-def to avoid breaking iterator update with the weird
7861 interaction with cgraph_update_edges_for_call_stmt_node.
7863 2017-09-13 Richard Biener <rguenther@suse.de>
7865 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7866 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7867 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7868 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7870 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7872 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7873 Disable pc relative literal load irrespective of
7874 TARGET_FIX_ERR_A53_84341 for default.
7876 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7878 * config/sparc/sparc.c (output_return): Output the source location of
7879 the insn in the delay slot, if any.
7880 (output_sibcall): Likewise.
7882 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7885 * common.opt (-static-pie): New alias.
7886 (shared): Negate static-pie.
7887 (-no-pie): Update help text.
7889 (static-pie): New option.
7890 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7891 -static-pie support.
7892 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7893 (LINK_EH_SPEC): Likewise.
7894 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7895 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7896 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7897 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7898 (init_gcc_specs): Likewise.
7899 (init_spec): Likewise.
7900 (display_help): Update help message for -pie.
7901 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7904 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7906 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7907 (movdi_aarch64): Likewise.
7908 (movti_aarch64): Likewise.
7910 2017-09-12 Simon Wright <simon@pushface.org>
7913 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7914 calculation of the minor version, always output as 0.
7916 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7919 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7920 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7921 on it early, rather than manual conversion late. For
7922 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7923 instead of performing manual conversion.
7925 2017-09-12 Carl Love <cel@us.ibm.com>
7927 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7928 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7930 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7931 VMULOSW): Add definitions.
7932 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7933 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7934 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7935 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7936 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7938 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7940 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7942 (movti_aarch64): Likewise.
7943 (movdf_aarch64): Likewise.
7944 (movtf_aarch64): Likewise.
7945 (load_pairdi): Likewise.
7946 (store_pairdi): Likewise.
7947 (load_pairdf): Likewise.
7948 (store_pairdf): Likewise.
7949 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7950 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7951 (ldr_got_small_<mode>): Likewise.
7952 (ldr_got_small_28k_<mode>): Likewise.
7953 (ldr_got_tiny): Likewise.
7954 * config/aarch64/iterators.md (ldst_sz): New.
7955 (ldpstp_sz): Likewise.
7956 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7958 (thunderx_load): Split load_8 to load_16.
7959 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7961 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7962 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7963 (xgene1_store_pair): Split store_8 to store_16.
7964 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7965 (falkor_st_0_st_sd): Split store_8 to store_16.
7967 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7969 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7970 and store1/2/3/4 to store_4/8/12/16.
7971 * config/aarch64/aarch64.md: Update for rename.
7972 * config/arm/arm.md: Likewise.: Likewise.
7973 * config/arm/arm.c: Likewise.
7974 * config/arm/thumb1.md: Likewise.
7975 * config/arm/thumb2.md: Likewise.
7976 * config/arm/vfp.md: Likewise.
7977 * config/arm/arm-generic.md: Likewise.
7978 * config/arm/arm1020e.md: Likewise.
7979 * config/arm/arm1026ejs.md: Likewise.
7980 * config/arm/arm1136jfs.md: Likewise.
7981 * config/arm/arm926ejs.md: Likewise.
7982 * config/arm/cortex-a15.md: Likewise.
7983 * config/arm/cortex-a17.md: Likewise.
7984 * config/arm/cortex-a5.md: Likewise.
7985 * config/arm/cortex-a53.md: Likewise.
7986 * config/arm/cortex-a57.md: Likewise.
7987 * config/arm/cortex-a7.md: Likewise.
7988 * config/arm/cortex-a8.md: Likewise.
7989 * config/arm/cortex-a9.md: Likewise.
7990 * config/arm/cortex-m4.md: Likewise.
7991 * config/arm/cortex-m7.md: Likewise.
7992 * config/arm/cortex-r4.md: Likewise.
7993 * config/arm/exynos-m1.md: Likewise.
7994 * config/arm/fa526.md: Likewise.
7995 * config/arm/fa606te.md: Likewise.
7996 * config/arm/fa626te.md: Likewise.
7997 * config/arm/fa726te.md: Likewise.
7998 * config/arm/fmp626.md: Likewise.
7999 * config/arm/iwmmxt.md: Likewise.
8000 * config/arm/ldmstm.md: Likewise.
8001 * config/arm/marvell-pj4.md: Likewise.
8002 * config/arm/xgene1.md: Likewise.
8003 * config/aarch64/thunderx.md: Likewise.
8004 * config/aarch64/thunderx2t99.md: Likewise.
8005 * config/aarch64/falkor.md: Likewise.
8007 2017-09-12 Martin Liska <mliska@suse.cz>
8009 * attribs.c (private_lookup_attribute): New function.
8010 * attribs.h (private_lookup_attribute): Declared here.
8011 (lookup_attribute): Called from this place.
8013 2017-09-12 Richard Biener <rguenther@suse.de>
8015 PR tree-optimization/82157
8016 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
8017 stmts with side-effects.
8019 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8020 Alan Hayward <alan.hayward@arm.com>
8021 David Sherwood <david.sherwood@arm.com>
8023 * target.def (hard_regno_nregs): New hook.
8024 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
8025 * targhooks.h (default_hard_regno_nregs): Declare.
8026 * targhooks.c (default_hard_regno_nregs): New function.
8027 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8028 (TARGET_HARD_REGNO_NREGS): ...this hook.
8029 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8030 (CLASS_MAX_NREGS): Likewise.
8031 * doc/tm.texi: Regenerate.
8032 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8033 instead of HARD_REGNO_NREGS.
8034 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8035 HARD_REGNO_NREGS in the comment.
8036 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8037 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8038 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8039 Return an unsigned int.
8040 (TARGET_HARD_REGNO_NREGS): Redefine.
8041 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8042 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8043 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8044 (arc_hard_regno_nregs): New function.
8045 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8046 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8047 (arm_hard_regno_nregs): New function.
8048 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8049 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8050 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8051 (TARGET_HARD_REGNO_NREGS): Redefine.
8052 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8053 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8054 (HARD_REGNO_NREGS): Delete.
8055 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8056 (cr16_hard_regno_nregs): New function.
8057 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8058 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8059 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8060 (cris_hard_regno_nregs): New function.
8061 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8062 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8063 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8064 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8065 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8066 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8067 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8068 (frv_hard_regno_nregs): Make static. Take and return an
8070 (frv_class_max_nregs): Remove outdated copy of documentation.
8071 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8072 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8073 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8074 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8075 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8076 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8077 (TARGET_HARD_REGNO_NREGS): Redefine.
8078 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8079 (CLASS_MAX_NREGS): Update comment.
8080 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8081 (ia64_hard_regno_nregs): New function.
8082 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8083 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8084 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8085 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8086 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8088 (m32c_hard_regno_nregs): Likewise. Make static.
8089 (TARGET_HARD_REGNO_NREGS): Redefine.
8090 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8091 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8092 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8093 (m68k_hard_regno_nregs): New function.
8094 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8095 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8096 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8097 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8098 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8099 Take and return an unsigned int.
8100 (TARGET_HARD_REGNO_NREGS): Redefine.
8101 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8102 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8103 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8104 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8105 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8106 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8107 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8108 (msp430_hard_regno_nregs): Make static. Take and return an
8110 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8111 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8112 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8113 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8114 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8115 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8116 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8117 (TARGET_HARD_REGNO_NREGS): Redefine.
8118 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8119 (PA_HARD_REGNO_NREGS): ...this.
8120 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8121 (PA_HARD_REGNO_NREGS): ...this.
8122 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8123 (pa_hard_regno_nregs): New function.
8124 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8125 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8126 (pdp11_hard_regno_nregs): New function.
8127 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8128 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8129 (rs6000_hard_regno_nregs_hook): New function.
8130 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8131 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8132 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8133 Take and return an unsigned int. Move earlier in file.
8134 (TARGET_HARD_REGNO_NREGS): Redefine.
8135 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8136 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8137 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8138 (rl78_hard_regno_nregs): Make static. Take and return an
8140 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8141 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8142 (rs6000_hard_regno_nregs_hook): New function.
8143 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8144 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8145 (TARGET_HARD_REGNO_NREGS): Redefine.
8146 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8147 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8148 instead of HARD_REGNO_NREGS.
8149 (s390_hard_regno_nregs): New function.
8150 (s390_hard_regno_mode_ok): Add comment from s390.h.
8151 (TARGET_HARD_REGNO_NREGS): Redefine.
8152 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8153 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8154 (sh_hard_regno_nregs): New function.
8155 (sh_pass_in_reg_p): Use it.
8156 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8157 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8158 (sparc_hard_regno_nregs): New function.
8159 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8160 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8161 (spu_function_arg_advance): Use it, supplying a valid register number.
8162 (TARGET_HARD_REGNO_NREGS): Redefine.
8163 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8164 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8165 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8166 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8167 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8168 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8169 (CLASS_MAX_NREGS): Remove copy of old documentation.
8170 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8171 (visium_hard_regno_nregs): New function.
8172 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8173 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8174 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8175 xtensa_hard_regno_nregs): New function.
8176 * system.h (HARD_REGNO_NREGS): Poison.
8178 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8180 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8181 hard_regno_nregs instead of HARD_REGNO_NREGS.
8182 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8183 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8184 (c6x_expand_epilogue): Likewise.
8185 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8186 (frv_read_iacc_argument): Likewise.
8187 * config/sh/sh.c: Include regs.h.
8188 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8189 (regs_used): Likewise.
8190 (output_stack_adjust): Likewise.
8191 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8192 * expmed.c: Include regs.h.
8193 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8194 * ree.c: Include regs.h.
8195 (combine_reaching_defs): Use hard_regno_nregs instead of
8197 (add_removable_extension): Likewise.
8199 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8201 * regs.h (hard_regno_nregs): Turn into a function.
8202 (end_hard_regno): Update accordingly.
8203 * caller-save.c (setup_save_areas): Likewise.
8204 (save_call_clobbered_regs): Likewise.
8205 (replace_reg_with_saved_mem): Likewise.
8206 (insert_restore): Likewise.
8207 (insert_save): Likewise.
8208 * combine.c (can_change_dest_mode): Likewise.
8209 (move_deaths): Likewise.
8210 (distribute_notes): Likewise.
8211 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8212 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8213 (rs6000_split_multireg_move): Likewise.
8214 (rs6000_register_move_cost): Likewise.
8215 (rs6000_memory_move_cost): Likewise.
8216 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8217 (rs6000_split_multireg_move): Likewise.
8218 (rs6000_register_move_cost): Likewise.
8219 (rs6000_memory_move_cost): Likewise.
8220 * cselib.c (cselib_reset_table): Likewise.
8221 (cselib_lookup_1): Likewise.
8222 * emit-rtl.c (set_mode_and_regno): Likewise.
8223 * function.c (aggregate_value_p): Likewise.
8224 * ira-color.c (setup_profitable_hard_regs): Likewise.
8225 (check_hard_reg_p): Likewise.
8226 (calculate_saved_nregs): Likewise.
8227 (assign_hard_reg): Likewise.
8228 (improve_allocation): Likewise.
8229 (calculate_spill_cost): Likewise.
8230 * ira-emit.c (modify_move_list): Likewise.
8231 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8232 (ira_hard_reg_in_set_p): Likewise.
8233 * ira.c (setup_reg_mode_hard_regset): Likewise.
8234 (clarify_prohibited_class_mode_regs): Likewise.
8235 (check_allocation): Likewise.
8236 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8237 (lra_setup_reg_renumber): Likewise.
8238 (setup_try_hard_regno_pseudos): Likewise.
8239 (spill_for): Likewise.
8240 (assign_hard_regno): Likewise.
8241 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8242 * lra-constraints.c (in_class_p): Likewise.
8243 (lra_constraint_offset): Likewise.
8244 (simplify_operand_subreg): Likewise.
8245 (lra_constraints): Likewise.
8246 (split_reg): Likewise.
8247 (split_if_necessary): Likewise.
8248 (invariant_p): Likewise.
8249 (inherit_in_ebb): Likewise.
8250 * lra-lives.c (process_bb_lives): Likewise.
8251 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8252 (get_hard_regs): Likewise.
8253 (do_remat): Likewise.
8254 * lra-spills.c (assign_spill_hard_regs): Likewise.
8255 * mode-switching.c (create_pre_exit): Likewise.
8256 * postreload.c (reload_combine_recognize_pattern): Likewise.
8257 * recog.c (peep2_find_free_register): Likewise.
8258 * regcprop.c (kill_value_regno): Likewise.
8259 (set_value_regno): Likewise.
8260 (copy_value): Likewise.
8261 (maybe_mode_change): Likewise.
8262 (find_oldest_value_reg): Likewise.
8263 (copyprop_hardreg_forward_1): Likewise.
8264 * regrename.c (check_new_reg_p): Likewise.
8265 (regrename_do_replace): Likewise.
8266 * reload.c (push_reload): Likewise.
8267 (combine_reloads): Likewise.
8268 (find_dummy_reload): Likewise.
8269 (operands_match_p): Likewise.
8270 (find_reloads): Likewise.
8271 (find_equiv_reg): Likewise.
8272 (reload_adjust_reg_for_mode): Likewise.
8273 * reload1.c (count_pseudo): Likewise.
8274 (count_spilled_pseudo): Likewise.
8275 (find_reg): Likewise.
8276 (clear_reload_reg_in_use): Likewise.
8277 (free_for_value_p): Likewise.
8278 (allocate_reload_reg): Likewise.
8279 (choose_reload_regs): Likewise.
8280 (reload_adjust_reg_for_temp): Likewise.
8281 (emit_reload_insns): Likewise.
8282 (delete_output_reload): Likewise.
8283 * rtlanal.c (subreg_get_info): Likewise.
8284 * sched-deps.c (sched_analyze_reg): Likewise.
8285 * sel-sched.c (init_regs_for_mode): Likewise.
8286 (mark_unavailable_hard_regs): Likewise.
8287 (choose_best_reg_1): Likewise.
8288 (verify_target_availability): Likewise.
8289 * valtrack.c (dead_debug_insert_temp): Likewise.
8290 * var-tracking.c (track_loc_p): Likewise.
8291 (emit_note_insn_var_location): Likewise.
8292 * varasm.c (make_decl_rtl): Likewise.
8293 * reginfo.c (choose_hard_reg_mode): Likewise.
8294 (init_reg_modes_target): Refer directly to
8295 this_target_regs->x_hard_regno_nregs.
8297 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8299 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8300 instead of hard_regno_nregs.
8302 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8304 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8305 end_hard_regno instead of hard_regno_nregs.
8306 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8307 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8308 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8309 * ira-color.c (improve_allocation): Likewise.
8310 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8311 * lra-lives.c (mark_regno_live): Likewise.
8312 (mark_regno_dead): Likewise.
8313 * lra-remat.c (operand_to_remat): Likewise.
8314 * lra.c (collect_non_operand_hard_regs): Likewise.
8315 * postreload.c (reload_combine_note_store): Likewise.
8316 (move2add_valid_value_p): Likewise.
8317 * reload.c (regno_clobbered_p): Likewise.
8319 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8321 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8323 * config/v850/v850.c (v850_reorg): Likewise.
8324 * reload.c (refers_to_regno_for_reload_p): Likewise.
8325 (find_equiv_reg): Likewise.
8326 * reload1.c (reload_reg_reaches_end_p): Likewise.
8328 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8330 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8332 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8333 * config/arm/arm.c (output_move_neon): Likewise.
8334 (arm_attr_length_move_neon): Likewise.
8335 (neon_split_vcombine): Likewise.
8336 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8337 (c6x_mark_reg_written): Likewise.
8338 (c6x_dwarf_register_span): Likewise.
8339 * config/i386/i386.c (ix86_save_reg): Likewise.
8340 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8341 (rws_access_reg): Likewise.
8342 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8343 * mode-switching.c (create_pre_exit): Likewise.
8344 * ree.c (combine_reaching_defs): Likewise.
8345 (add_removable_extension): Likewise.
8346 * regcprop.c (find_oldest_value_reg): Likewise.
8347 (copyprop_hardreg_forward_1): Likewise.
8348 * reload.c (reload_inner_reg_of_subreg): Likewise.
8349 (push_reload): Likewise.
8350 (combine_reloads): Likewise.
8351 (find_dummy_reload): Likewise.
8352 (reload_adjust_reg_for_mode): Likewise.
8353 * reload1.c (find_reload_regs): Likewise.
8354 (forget_old_reloads_1): Likewise.
8355 (reload_reg_free_for_value_p): Likewise.
8356 (reload_adjust_reg_for_temp): Likewise.
8357 (emit_reload_insns): Likewise.
8358 (delete_output_reload): Likewise.
8359 * sel-sched.c (choose_best_reg_1): Likewise.
8360 (choose_best_pseudo_reg): Likewise.
8362 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8363 Alan Hayward <alan.hayward@arm.com>
8364 David Sherwood <david.sherwood@arm.com>
8366 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8367 * target.def (slow_unaligned_access): New hook.
8368 * targhooks.h (default_slow_unaligned_access): Declare.
8369 * targhooks.c (default_slow_unaligned_access): New function.
8370 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8371 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8372 * doc/tm.texi: Regenerate.
8373 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8374 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8375 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8377 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8378 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8380 (rs6000_slow_unaligned_access): New function.
8381 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8382 (expand_block_compare): Likewise.
8383 (expand_strn_compare): Likewise.
8384 (rs6000_rtx_costs): Likewise.
8385 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8386 (riscv_slow_unaligned_access): Likewise.
8387 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8388 (riscv_slow_unaligned_access_p): ...this and make static.
8389 (riscv_option_override): Update accordingly.
8390 (riscv_slow_unaligned_access): New function.
8391 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8392 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8393 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8394 (rs6000_slow_unaligned_access): New function.
8395 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8396 (rs6000_rtx_costs): Likewise.
8397 * config/rs6000/rs6000-string.c (expand_block_compare)
8398 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8399 of SLOW_UNALIGNED_ACCESS.
8400 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8401 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8402 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8403 of SLOW_UNALIGNED_ACCESS.
8404 * expmed.c (simple_mem_bitfield_p): Likewise.
8405 * expr.c (alignment_for_piecewise_move): Likewise.
8406 (emit_group_load_1): Likewise.
8407 (emit_group_store): Likewise.
8408 (copy_blkmode_from_reg): Likewise.
8409 (emit_push_insn): Likewise.
8410 (expand_assignment): Likewise.
8411 (store_field): Likewise.
8412 (expand_expr_real_1): Likewise.
8413 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8414 * lra-constraints.c (simplify_operand_subreg): Likewise.
8415 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8416 * gimple-ssa-store-merging.c: Likewise in block comment at start
8418 * tree-ssa-strlen.c: Include target.h.
8419 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8420 of SLOW_UNALIGNED_ACCESS.
8421 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8423 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8425 PR rtl-optimization/82185
8426 * expmed.c (emit_store_flag_int): Only test tem if it has been
8429 2017-09-12 Richard Biener <rguenther@suse.de>
8432 * match.pd ((FTYPE) N CMP CST): Fix typo.
8434 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8436 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8438 (mips_near_type_p): Add 'short_call' attribute as a synonym
8440 * doc/extend.texi (short_call): Document new function attribute.
8442 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8445 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8446 assertion check that in the condition.
8447 (get_atomic_generic_size): Likewise. Before testing if parameter
8448 has pointer type, if it has array type, call for C++
8449 default_conversion to perform array-to-pointer conversion.
8451 2017-09-12 Richard Biener <rguenther@suse.de>
8453 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8454 for operations we cannot scalarize.
8456 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8458 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8459 vectors heap vectors. Clean up comments.
8460 Make visited_bbs a reference.
8461 (profitable_jump_thread_path): Make GC
8462 vectors heap vectors. Clean up comments.
8464 (convert_and_register_jump_thread_path): Make GC vectors heap
8466 (check_subpath_and_update_thread_path): Same. Clean up comments.
8467 Make visited_bbs a reference.
8468 (handle_phi): Abstract common code to to
8469 register_jump_thread_path_if_profitable.
8470 Rename VAR_BB to DEF_BB.
8472 Make GC vectors heap vectors.
8473 Make visited_bbs a reference.
8474 (handle_assignment): Same.
8475 (register_jump_thread_path_if_profitable): New.
8476 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8478 Make GC vectors heap vectors. Clean up comments.
8479 Make visited_bbs a reference.
8480 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8481 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8484 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8487 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8488 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8490 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8492 Revert r251800 and r251799.
8494 2017-09-11 Martin Jambor <mjambor@suse.cz>
8497 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8498 arguments in advance.
8499 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8500 use it to find predecessor edges.
8501 (naive_outof_ssa): Collect vector of predecessors.
8503 2017-09-08 Jason Merrill <jason@redhat.com>
8505 PR c++/70029 - ICE with ref-qualifier and -flto
8506 * langhooks.h (struct lang_hooks_for_types): Add
8507 copy_lang_qualifiers.
8508 * attribs.c (build_type_attribute_qual_variant): Use it.
8509 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8511 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8512 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8514 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8517 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8518 (*mulsi3_sp64): New instruction.
8519 (mulsi3): New expander.
8521 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8523 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8525 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8527 * sancov.c: Include memmodel.h.
8529 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8532 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8535 2017-09-07 Carl Love <cel@us.ibm.com>
8537 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8538 the sldi instruction.
8540 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8542 * sancov.c: Include tm_p.h.
8544 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8547 * output.h (switch_to_other_text_partition): New declaration.
8548 * varasm.c (switch_to_other_text_partition): New function.
8549 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8550 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8551 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8552 to the other text partition before emitting LCL label and switch back
8553 after emitting the word after it.
8555 2017-09-07 Richard Biener <rguenther@suse.de>
8557 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8558 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8559 critical edges here, after loop init.
8560 (pass_data_pre): Remove PROP_no_crit_edges flags.
8561 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8562 for valueization of call args to avoid leaking VN_TOP.
8563 (visit_use): Assert we do not visit default defs.
8564 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8565 Use error_mark_node to more easily detect leaking VN_TOP.
8566 All default-defs are varying, not VN_TOP. Mark them visited.
8567 (run_scc_vn): Make code match comment.
8569 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8571 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8572 OPTION_MASK_FLOAT128_KEYWORD.
8573 (POWERPC_MASKS): Likewise.
8574 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8575 support for the -mfloat128-type option, and make -mfloat128
8576 default on PowerPC Linux systems. Define or undefine
8577 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8578 Define __float128 to be __ieee128 if IEEE 128-bit support is
8579 enabled, or undefine it.
8580 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8581 Delete defining __FLOAT128_TYPE__.
8582 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8583 -mfloat128-type option and make -mfloat128 default on PowerPC
8585 (TARGET_FLOAT128_TYPE): Likewise.
8586 (-mfloat128-type): Likewise.
8587 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8588 Delete the -mfloat128-type option and make -mfloat128 default on
8589 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8590 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8591 128-bit floating point is enabled. Change tests from using
8592 -mfloat128-type to -mfloat128.
8593 (rs6000_mangle_type): Use the correct mangling for the __float128
8594 type even if normal long double is restricted to 64-bits.
8595 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8597 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8598 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8599 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8601 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8602 documentation for -mfloat128.
8604 2017-09-06 Olivier Hainque <hainque@adacore.com>
8606 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8608 2017-09-06 Wish Wu <wishwu007@gmail.com>
8609 Jakub Jelinek <jakub@redhat.com>
8611 * asan.c (initialize_sanitizer_builtins): Add
8612 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8613 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8614 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8615 BT_FN_VOID_UINT64_PTR variables.
8616 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8617 (BT_FN_VOID_UINT16_UINT16): Likewise.
8618 (BT_FN_VOID_UINT32_UINT32): Likewise.
8619 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8620 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8621 (BT_FN_VOID_UINT64_PTR): Likewise.
8622 * common.opt (flag_sanitize_coverage): New variable.
8623 (fsanitize-coverage=trace-pc): Remove.
8624 (fsanitize-coverage=): Add.
8625 * flag-types.h (enum sanitize_coverage_code): New enum.
8626 * fold-const.c (fold_range_test): Disable non-short-circuit
8627 optimization if flag_sanitize_coverage.
8628 (fold_truth_andor): Likewise.
8629 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8630 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8631 (coverage_sanitizer_opts): New array.
8632 (get_closest_sanitizer_option): Add OPTS argument, handle also
8633 OPT_fsanitize_coverage_.
8634 (parse_sanitizer_options): Adjusted to also handle
8635 OPT_fsanitize_coverage_.
8636 (common_handle_option): Add OPT_fsanitize_coverage_.
8637 * sancov.c (instrument_comparison, instrument_switch): New function.
8638 (sancov_pass): Add trace-cmp support.
8639 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8640 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8641 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8642 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8643 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8644 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8645 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8646 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8647 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8648 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8650 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8652 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8653 error. Only quit immediately if parsing is complete.
8654 (BEGIN): Initialize fatal_err and parse_done.
8655 (begin fpu, end fpu): Check number of arguments.
8656 (begin arch, end arch): Likewise.
8657 (begin cpu, end cpu): Likewise.
8658 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8659 (optalias): Likewise.
8661 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8663 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8664 * config/arm/arm-isa.h: Delete. Move definitions to ...
8665 * arm-cpus.in: ... here. Use new feature and fgroup values.
8666 * config/arm/arm.c (arm_option_override): Use lower case for feature
8668 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8669 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8670 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8672 (print_isa_bits_for): New function.
8673 (gen_isa): New function.
8674 (gen_comm_data): Use print_isa_bits_for.
8675 (define feature): New keyword.
8676 (define fgroup): New keyword.
8677 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8678 (arm-isa.h): Add rule to generate file.
8679 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8680 case for feature bit names.
8682 2017-09-06 Richard Biener <rguenther@suse.de>
8684 * tree-ssa-pre.c (NECESSARY): Remove.
8685 (create_expression_by_pieces): Do not touch pass-local flags.
8686 (insert_into_preds_of_block): Likewise.
8687 (do_pre_regular_insertion): Likewise.
8688 (eliminate_insert): Likewise.
8689 (eliminate_dom_walker::before_dom_children): Likewise.
8690 (fini_eliminate): Do not look at inserted_exprs.
8691 (mark_operand_necessary): Remove.
8692 (remove_dead_inserted_code): Replace with simple work-list
8693 algorithm based on inserted_exprs and SSA uses.
8694 (pass_pre::execute): Re-order fini_eliminate and
8695 remove_dead_inserted_code.
8697 2017-09-06 Olivier Hainque <hainque@adacore.com>
8699 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8700 for VxWorks 7. Adjust surrounding comments.
8702 2017-09-06 Richard Biener <rguenther@suse.de>
8704 * gimple-ssa-strength-reduction.c
8705 (find_candidates_dom_walker::before_dom_children): Also allow
8708 2017-09-06 Richard Biener <rguenther@suse.de>
8710 PR tree-optimization/82108
8711 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8712 for gap in the non-permutation SLP case.
8714 2017-09-06 Martin Jambor <mjambor@suse.cz>
8716 PR tree-optimization/82078
8717 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8718 add_access_to_work_queue...
8719 (build_accesses_from_assign): ...here.
8720 (propagate_all_subaccesses): Make sure racc is the group
8721 representative, if there is one.
8723 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8726 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8729 2017-09-06 Richard Biener <rguenther@suse.de>
8731 * gimple-ssa-strength-reduction.c
8732 (find_candidates_dom_walker::before_doom_children): Use a
8733 type and not a mode check.
8735 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8738 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8740 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8742 2017-09-05 Jeff Law <law@redhat.com>
8744 PR tree-optimization/64910
8745 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8746 cases where we have 3 or more operands.
8748 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8751 * omp-low.c (lower_omp_for): Recompute tree invariant if
8752 gimple_omp_for_initial/final is ADDR_EXPR.
8755 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8756 into gimple val before gimplification fo the COND_EXPR.
8758 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8760 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8761 REGION_COPY argument.
8762 (thread_through_all_blocks): Remove unused argument to
8763 duplicate_thread_path.
8765 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8766 Alan Hayward <alan.hayward@arm.com>
8767 David Sherwood <david.sherwood@arm.com>
8769 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8770 Take a scalar_mode rather than a machine_mode.
8771 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8772 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8773 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8774 (aarch64_gen_adjusted_ldpstp): Likewise.
8775 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8777 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8778 Alan Hayward <alan.hayward@arm.com>
8779 David Sherwood <david.sherwood@arm.com>
8781 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8782 Take a scalar_int_mode instead of a machine_mode.
8783 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8784 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8785 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8786 (aarch64_simd_attr_length_rglist): Delete.
8787 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8788 a scalar_int_mode instead of a machine_mode.
8789 (aarch64_add_offset): Likewise.
8790 (aarch64_internal_mov_immediate): Likewise
8791 (aarch64_add_constant_internal): Likewise.
8792 (aarch64_add_constant): Likewise.
8793 (aarch64_movw_imm): Likewise.
8794 (aarch64_rtx_arith_op_extract_p): Likewise.
8795 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8796 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8797 Remove assert that the mode isn't a vector.
8798 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8799 (aarch64_expand_mov_immediate): Update calls after above changes.
8800 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8801 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8802 (aarch64_move_imm): Likewise.
8803 (aarch64_can_const_movi_rtx_p): Likewise.
8804 (aarch64_strip_extend): Likewise.
8805 (aarch64_extr_rtx_p): Likewise.
8806 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8807 a CONST_INT when the mode parameter is VOIDmode.
8808 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8810 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8812 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8813 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8814 (bitwise_type_for_mode): Update accordingly.
8816 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8818 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8819 * stor-layout.c (mode_for_size_tree): Likewise.
8820 (mode_for_array): Update accordingly.
8821 (layout_decl): Likewise.
8822 (compute_record_mode): Likewise. Only set the mode once.
8824 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8826 * target.def (get_mask_mode): Change return type to opt_mode.
8828 * doc/tm.texi: Regenerate.
8829 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8830 * targhooks.c (default_get_mask_mode): Likewise.
8831 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8832 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8833 targetm.get_mask_mode.
8834 * tree.c (build_truth_vector_type): Likewise.
8836 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8838 * machmode.h (mode_for_vector): Return an opt_mode.
8839 * stor-layout.c (mode_for_vector): Likewise.
8840 (mode_for_int_vector): Update accordingly.
8841 (layout_type): Likewise.
8842 * config/i386/i386.c (emit_memmov): Likewise.
8843 (ix86_expand_set_or_movmem): Likewise.
8844 (ix86_expand_vector_init): Likewise.
8845 (ix86_get_mask_mode): Likewise.
8846 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8848 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8849 * expmed.c (extract_bit_field_1): Likewise.
8850 * expr.c (expand_expr_real_2): Likewise.
8851 * optabs-query.c (can_vec_perm_p): Likewise.
8852 (can_vec_mask_load_store_p): Likewise.
8853 * optabs.c (expand_vec_perm): Likewise.
8854 * targhooks.c (default_get_mask_mode): Likewise.
8855 * tree-vect-stmts.c (vectorizable_store): Likewise.
8856 (vectorizable_load): Likewise.
8857 (get_vectype_for_scalar_type_and_size): Likewise.
8859 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8861 * machmode.h (mode_for_int_vector): New function.
8862 * stor-layout.c (mode_for_int_vector): Likewise.
8863 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8864 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8865 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8866 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8867 (s390_expand_vcond): Likewise.
8869 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8871 * machmode.h (opt_machine_mode): New type.
8872 (opt_mode<T>): Allow construction from anything that can be
8874 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8875 (mode_for_size): Return an opt_machine_mode.
8876 * stor-layout.c (mode_for_size): Likewise.
8877 (mode_for_size_tree): Update call accordingly.
8878 (bitwise_mode_for_mode): Likewise.
8879 (make_fract_type): Likewise.
8880 (make_accum_type): Likewise.
8881 * caller-save.c (replace_reg_with_saved_mem): Update call
8883 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8884 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8885 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8886 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8887 * expmed.c (extract_bit_field_1): Likewise.
8888 * reload.c (get_secondary_mem): Likewise.
8889 * varasm.c (assemble_integer): Likewise.
8890 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8893 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8895 * machmode.h (decimal_float_mode_for_size): New function.
8896 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8897 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8898 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8899 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8900 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8901 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8903 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8905 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8906 (get_builtin_sync_mode): Likewise.
8907 (expand_ifn_atomic_compare_exchange): Likewise.
8908 (expand_builtin_atomic_clear): Likewise.
8909 (expand_builtin_atomic_test_and_set): Likewise.
8910 (fold_builtin_atomic_always_lock_free): Likewise.
8911 * calls.c (compute_argument_addresses): Likewise.
8912 (emit_library_call_value_1): Likewise.
8913 (store_one_arg): Likewise.
8914 * combine.c (combine_instructions): Likewise.
8915 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8916 * config/arm/arm.c (arm_function_value): Likewise.
8917 (aapcs_allocate_return_reg): Likewise.
8918 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8919 * config/i386/i386.c (construct_container): Likewise.
8920 (ix86_gimplify_va_arg): Likewise.
8921 (ix86_expand_sse_cmp): Likewise.
8922 (emit_memmov): Likewise.
8923 (emit_memset): Likewise.
8924 (expand_small_movmem_or_setmem): Likewise.
8925 (ix86_expand_pextr): Likewise.
8926 (ix86_expand_pinsr): Likewise.
8927 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8928 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8930 * config/mips/mips.c (mips_function_value_1) Likewise.
8931 (mips_block_move_straight): Likewise.
8932 (mips_expand_ins_as_unaligned_store): Likewise.
8933 * config/powerpcspe/powerpcspe.c
8934 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8935 (rs6000_darwin64_record_arg_flush): Likewise.
8936 * config/rs6000/rs6000.c
8937 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8938 (rs6000_darwin64_record_arg_flush): Likewise.
8939 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8940 (sparc_function_value_1): Likewise.
8941 * config/spu/spu.c (adjust_operand): Likewise.
8942 (spu_emit_branch_or_set): Likewise.
8943 (arith_immediate_p): Likewise.
8944 * emit-rtl.c (gen_lowpart_common): Likewise.
8945 * expr.c (expand_expr_real_1): Likewise.
8946 * function.c (assign_parm_setup_block): Likewise.
8947 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8948 * reload1.c (alter_reg): Likewise.
8949 * stor-layout.c (mode_for_vector): Likewise.
8950 (layout_type): Likewise.
8952 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8954 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8955 (spu_convert_move): Likewise.
8956 * lower-subreg.c (resolve_simple_move): Likewise.
8958 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8961 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8962 define_insn to a define_expand.
8963 (altivec_vsum2sws_direct): New define_insn.
8964 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8966 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8968 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8971 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8976 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8977 realign stack if stack alignment needed is less than incoming
8980 2017-09-05 Marek Polacek <polacek@redhat.com>
8983 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8986 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8988 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8990 2017-09-05 Richard Biener <rguenther@suse.de>
8992 PR tree-optimization/82084
8993 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8995 2017-09-05 Richard Biener <rguenther@suse.de>
8997 PR tree-optimization/82102
8998 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
9000 2017-09-05 Martin Liska <mliska@suse.cz>
9002 PR tree-optimization/82032
9003 * tree-cfg.c (generate_range_test): New function.
9004 * tree-cfg.h (generate_range_test): Declared here.
9005 * tree-cfgcleanup.c (convert_single_case_switch): New function.
9006 (cleanup_control_expr_graph): Use it.
9007 * tree-switch-conversion.c (try_switch_expansion): Remove
9009 (emit_case_nodes): Use generate_range_test.
9011 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9014 * config/i386/i386.md (*<btsc><mode>_mask): Add
9015 TARGET_USE_BT to insn constraint.
9016 (*btr<mode>_mask): Ditto.
9018 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
9020 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
9021 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
9023 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9026 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9027 TARGET_NEON and TARGET_IWMMXT.
9028 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9029 TARGET_NEON and TARGET_IWMMXT.
9030 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9032 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9034 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9035 (ix86_rewrite_tls_address): Ditto.
9036 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9037 (ix86_rewrite_tls_address_1): Ditto.
9038 (ix86_rewrite_tls_address): Ditto.
9039 * config/i386/predicates.md (tls_address_pattern): New predicate.
9040 * config/i386/i386.md (TLS address splitter): New splitter.
9042 2017-09-04 Richard Biener <rguenther@suse.de>
9044 PR tree-optimization/82084
9045 * fold-const.h (can_native_encode_string_p): Declare.
9046 * fold-const.c (can_native_encode_string_p): Factor out from ...
9047 (native_encode_string): ... here.
9048 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9049 vectorizing stores from constants we later cannot handle.
9051 2017-09-04 Marek Polacek <polacek@redhat.com>
9054 * doc/invoke.texi: Update -Wtautological-compare documentation.
9056 2017-09-04 Jeff Law <law@redhat.com>
9058 PR tree-optimization/64910
9059 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9060 swap the first and last operand if the last is a constant.
9062 2017-09-04 Marek Polacek <polacek@redhat.com>
9065 * convert.c (do_narrow): When sanitizing signed integer overflows,
9066 bail out for signed types.
9067 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9069 2017-09-04 Richard Biener <rguenther@suse.de>
9071 PR tree-optimization/82060
9072 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9073 Move devirtualization after stmt folding and before EH/AB/noreturn
9074 cleanup to get the stmt refs canonicalized. Use a bool instead
9075 of gimple_modified_p since that doesn't work for NOPs. Schedule
9076 NOPs generated by folding for removal.
9078 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9079 Alan Hayward <alan.hayward@arm.com>
9080 David Sherwood <david.sherwood@arm.com>
9082 * coretypes.h (pad_direction): New enum.
9083 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9084 (FUNCTION_ARG_PADDING): Likewise.
9085 * target.def (function_arg_padding): New hook.
9086 * targhooks.h (default_function_arg_padding): Declare.
9087 * targhooks.c (default_function_arg_padding): New function.
9088 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9089 (TARGET_FUNCTION_ARG_PADDING): ...this.
9090 * doc/tm.texi: Regenerate.
9091 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9092 instead of direction.
9093 (compute_argument_addresses): Likewise.
9094 (load_register_parameters): Likewise.
9095 (emit_library_call_value_1): Likewise.
9096 (store_one_arg): Use targetm.calls.function_arg_padding instead
9097 of FUNCTION_ARG_PADDING.
9098 (must_pass_in_stack_var_size_or_pad): Likewise.
9099 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9100 (emit_group_store): Likewise.
9101 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9102 instead of FUNCTION_ARG_PADDING.
9103 (emit_push_insn): Likewise, and propagate enum change throughout
9105 * function.h (direction): Delete.
9106 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9108 * function.c (assign_parm_find_stack_rtl): Likewise.
9109 (assign_parm_setup_block_p): Likewise.
9110 (assign_parm_setup_block): Likewise.
9111 (gimplify_parameters): Likewise.
9112 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9113 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9115 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9116 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9117 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9118 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9119 (aarch64_function_arg_padding): ...this new function.
9120 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9121 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9122 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9123 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9124 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9125 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9126 (arm_pad_arg_upward): Replace with...
9127 (arm_function_arg_padding): ...this new function.
9128 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9130 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9131 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9132 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9133 (ia64_hpux_function_arg_padding): Replace with...
9134 (ia64_function_arg_padding): ...this new function. Use pad_direction
9135 instead of direction. Check for TARGET_HPUX.
9136 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9137 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9138 (iq2000_function_arg_padding): New function.
9139 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9140 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9141 (mips_function_arg_padding): ...this new function.
9142 (mips_pad_reg_upward): Update accordingly.
9143 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9144 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9145 targetm.calls.function_arg_padding.
9146 (FUNCTION_ARG_PADDING): Delete.
9147 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9148 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9149 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9150 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9151 (nios2_block_reg_padding): Return pad_direction instead of direction.
9152 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9153 instead of direction.
9154 (nios2_function_arg_padding): Likewise. Make static.
9155 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9156 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9157 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9158 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9159 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9160 (pa_function_arg_padding): Make static. Return pad_direction instead
9162 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9163 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9164 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9165 instead of direction. Use targetm.calls.function_arg_padding.
9166 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9167 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9168 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9169 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9170 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9172 (function_arg_padding): Rename to...
9173 (rs6000_function_arg_padding): ...this. Make static. Return
9174 pad_direction instead of direction.
9175 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9176 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9177 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9178 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9179 instead of direction. Use targetm.calls.function_arg_padding.
9180 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9181 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9182 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9183 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9184 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9185 (function_arg_padding): Rename to...
9186 (rs6000_function_arg_padding): ...this. Make static. Return
9187 pad_direction instead of direction.
9188 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9189 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9190 * config/s390/s390.c (s390_function_arg_padding): New function.
9191 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9192 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9193 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9194 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9195 (function_arg_padding): Rename to...
9196 (sparc_function_arg_padding): ...this. Make static. Return
9197 pad_direction instead of direction.
9198 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9199 * config/spu/spu.c (spu_function_arg_padding): New function.
9200 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9201 * system.h (FUNCTION_ARG_PADDING): Poison.
9203 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9204 Alan Hayward <alan.hayward@arm.com>
9205 David Sherwood <david.sherwood@arm.com>
9207 * target.def (modes_tieable_p): New hook.
9208 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9209 (TARGET_MODES_TIEABLE_P): ...this.
9210 * doc/tm.texi.in: Regenerate.
9211 * hooks.h (hook_bool_mode_mode_true): Declare.
9212 * hooks.c (hook_bool_mode_mode_true): New function.
9213 * combine.c (subst): Use targetm.modes_tieable_p instead of
9215 * dse.c (find_shift_sequence): Likewise.
9216 * expmed.c (extract_low_bits): Likewise.
9217 * lower-subreg.c: Include target.h.
9218 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9220 * rtlanal.c (rtx_cost): Likewise.
9221 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9222 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9223 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9224 (TARGET_MODES_TIEABLE_P): Redefine.
9225 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9226 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9227 (TARGET_MODES_TIEABLE_P): Redefine.
9228 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9229 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9230 (arc_modes_tieable_p): New function.
9231 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9232 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9233 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9234 (arm_modes_tieable_p): Make static.
9235 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9236 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9237 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9238 (TARGET_MODES_TIEABLE_P): Redefine.
9239 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9240 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9241 (TARGET_MODES_TIEABLE_P): Redefine.
9242 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9243 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9244 (cr16_modes_tieable_p): New function.
9245 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9246 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9247 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9248 (TRULY_NOOP_TRUNCATION): Update comment.
9249 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9250 (TRULY_NOOP_TRUNCATION): Update comment.
9251 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9252 (frv_modes_tieable_p): New function.
9253 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9254 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9255 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9256 (TARGET_MODES_TIEABLE_P): Redefine.
9257 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9258 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9259 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9260 (TARGET_MODES_TIEABLE_P): Redefine.
9261 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9262 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9263 (ia64_modes_tieable_p): New function.
9264 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9265 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9266 (iq2000_modes_tieable_p): New function.
9267 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9268 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9269 (lm32_modes_tieable_p): New function.
9270 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9271 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9272 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9273 (TARGET_MODES_TIEABLE_P): Redefine.
9274 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9275 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9276 (m32r_modes_tieable_p): New function.
9277 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9278 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9279 (m68k_modes_tieable_p): New function.
9280 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9281 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9282 (mcore_modes_tieable_p): New function.
9283 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9284 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9286 (TARGET_MODES_TIEABLE_P): Redefine.
9287 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9288 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9289 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9290 (TARGET_MODES_TIEABLE_P): Redefine.
9291 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9292 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9293 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9294 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9295 (mn10300_modes_tieable_p): ...this and make static.
9296 (TARGET_MODES_TIEABLE_P): Redefine.
9297 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9298 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9299 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9300 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9301 (msp430_modes_tieable_p): Make static.
9302 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9303 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9304 (TARGET_MODES_TIEABLE_P): Redefine.
9305 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9306 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9307 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9308 (TARGET_MODES_TIEABLE_P): Redefine.
9309 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9310 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9311 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9312 (TARGET_MODES_TIEABLE_P): Redefine.
9313 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9314 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9315 (pdp11_modes_tieable_p): New function.
9316 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9317 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9318 (rs6000_modes_tieable_p): New function.
9319 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9320 * config/powerpcspe/powerpcspe.md: Update comment.
9321 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9322 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9323 (TARGET_MODES_TIEABLE_P): Redefine.
9324 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9325 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9326 (rl78_modes_tieable_p): New function.
9327 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9328 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9329 (rs6000_modes_tieable_p): New function.
9330 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9331 * config/rs6000/rs6000.md: Update comment.
9332 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9333 * config/rx/rx.c (rx_modes_tieable_p): New function.
9334 (TARGET_MODES_TIEABLE_P): Redefine.
9335 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9336 * config/s390/s390.c (s390_modes_tieable_p): New function.
9337 (TARGET_MODES_TIEABLE_P): Redefine.
9338 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9339 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9340 (sh_modes_tieable_p): New function.
9341 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9342 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9343 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9344 (sparc_modes_tieable_p): Make static.
9345 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9346 * config/spu/spu.c (spu_modes_tieable_p): New function.
9347 (TARGET_MODES_TIEABLE_P): Redefine.
9348 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9349 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9350 (TARGET_MODES_TIEABLE_P): Redefine.
9351 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9352 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9353 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9354 * config/v850/v850.c (v850_modes_tieable_p): New function.
9355 (TARGET_MODES_TIEABLE_P): Redefine.
9356 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9357 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9358 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9359 (visium_modes_tieable_p): New function.
9360 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9361 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9362 (xtensa_modes_tieable_p): New function.
9363 * system.h (MODES_TIEABLE_P): Poison.
9365 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9366 Alan Hayward <alan.hayward@arm.com>
9367 David Sherwood <david.sherwood@arm.com>
9369 * target.def (hard_regno_mode_ok): New hook.
9370 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9371 (TARGET_HARD_REGNO_MODE_OK): ...this.
9372 * doc/tm.texi.in: Regenerate.
9373 * hooks.h (hook_bool_uint_mode_true): Declare.
9374 * hooks.c (hook_bool_uint_mode_true): New function.
9375 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9377 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9378 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9379 instead of HARD_REGNO_MODE_OK.
9380 * caller-save.c: Include target.h.
9381 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9383 * combine.c (can_combine_p): Likewise.
9384 (combinable_i3pat): Likewise.
9385 (can_change_dest_mode): Likewise.
9386 * expr.c (init_expr_target): Likewise.
9387 (convert_move): Likewise.
9388 (convert_modes): Likewise.
9389 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9390 (setup_prohibited_mode_move_regs): Likewise.
9391 * ira.h (target_ira): Likewise.
9392 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9393 * lra-constraints.c (process_alt_operands): Likewise.
9394 (split_reg): Likewise.
9395 * recog.c (peep2_find_free_register): Likewise.
9396 * ree.c (combine_reaching_defs): Likewise.
9397 * regcprop.c (maybe_mode_change): Likewise.
9398 * reginfo.c (init_reg_sets_1): Likewise.
9399 (choose_hard_reg_mode): Likewise.
9400 (simplifiable_subregs): Likewise.
9401 * regrename.c (check_new_reg_p): Likewise.
9402 * reload.c (find_valid_class): Likewise.
9403 (find_valid_class_1): Likewise.
9404 (reload_inner_reg_of_subreg): Likewise.
9405 (push_reload): Likewise.
9406 (combine_reloads): Likewise.
9407 (find_dummy_reload): Likewise.
9408 (find_reloads): Likewise.
9409 * reload1.c (find_reg): Likewise.
9410 (set_reload_reg): Likewise.
9411 (allocate_reload_reg): Likewise.
9412 (choose_reload_regs): Likewise.
9413 (reload_adjust_reg_for_temp): Likewise.
9414 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9415 (simplify_subreg_regno): Likewise.
9416 * sel-sched.c (init_regs_for_mode): Likewise.
9417 * varasm.c (make_decl_rtl): Likewise.
9418 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9419 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9421 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9422 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9423 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9424 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9425 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9426 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9427 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9428 (arc_mode_class): Delete.
9429 (HARD_REGNO_MODE_OK): Delete.
9430 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9431 (arc_hard_regno_mode_ok): Rename old array to...
9432 (arc_hard_regno_mode_ok_modes): ...this.
9433 (arc_conditional_register_usage): Update accordingly.
9434 (arc_mode_class): Make static.
9435 (arc_hard_regno_mode_ok): New function.
9436 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9437 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9438 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9439 (arm_hard_regno_mode_ok): Make static.
9440 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9442 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9443 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9444 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9446 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9447 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9448 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9449 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9450 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9451 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9452 * config/bfin/predicates.md (valid_reg_operand): Use
9453 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9454 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9455 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9456 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9457 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9458 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9459 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9460 (cr16_hard_regno_mode_ok): Make static and return a bool.
9461 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9462 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9463 (cris_hard_regno_mode_ok): New function.
9464 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9465 (epiphany_mode_class): Delete.
9466 (HARD_REGNO_MODE_OK): Delete.
9467 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9468 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9469 (hard_regno_mode_ok): Rename to...
9470 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9471 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9472 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9474 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9475 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9476 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9477 (frv_hard_regno_mode_ok): Make static and return a bool.
9478 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9480 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9481 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9482 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9483 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9485 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9486 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9487 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9488 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9490 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9491 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9492 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9493 (ia64_hard_regno_mode_ok): New function.
9494 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9495 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9496 (iq2000_hard_regno_mode_ok): New function.
9497 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9498 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9499 (lm32_hard_regno_mode_ok): New function.
9500 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9501 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9502 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9503 instead of HARD_REGNO_MODE_OK.
9504 (m32c_hard_regno_ok): Rename to...
9505 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9506 (m32c_cannot_change_mode_class): Update accordingly.
9507 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9508 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9509 (m32r_mode_class): Delete.
9510 (HARD_REGNO_MODE_OK): Delete.
9511 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9512 (m32r_hard_regno_mode_ok): Rename to...
9513 (m32r_hard_regno_modes): ...this.
9514 (m32r_mode_class): Make static.
9515 (m32r_hard_regno_mode_ok): New function.
9516 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9517 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9518 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9519 (m68k_hard_regno_mode_ok): Make static.
9520 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9521 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9522 (mcore_hard_regno_mode_ok): New function.
9523 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9524 (HARD_REGNO_MODE_OK): Delete.
9525 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9527 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9528 (microblaze_hard_regno_mode_ok): New function.
9529 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9530 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9531 (mips_hard_regno_mode_ok): Delete.
9532 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9533 (mips_hard_regno_mode_ok_p): ...this and make static.
9534 (mips_hard_regno_mode_ok_p): Rename to...
9535 (mips_hard_regno_mode_ok_uncached): ...this.
9536 (mips_hard_regno_mode_ok): New function.
9537 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9538 of HARD_REGNO_MODE_OK.
9539 (mips_option_override): Update after above name changes.
9540 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9541 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9542 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9543 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9544 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9545 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9546 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9547 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9548 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9549 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9550 (msp430_hard_regno_mode_ok): Make static and return a bool.
9551 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9552 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9553 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9555 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9556 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9557 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9558 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9559 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9560 (PA_HARD_REGNO_MODE_OK): ...this
9561 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9562 (PA_HARD_REGNO_MODE_OK): ...this.
9563 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9564 (pa_hard_regno_mode_ok): New function.
9565 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9566 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9567 (pdp11_hard_regno_mode_ok): New function.
9568 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9569 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9571 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9573 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9574 (rs6000_hard_regno_mode_ok): Rename to...
9575 (rs6000_hard_regno_mode_ok_uncached): ...this.
9576 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9577 (rs6000_hard_regno_mode_ok): New function.
9578 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9579 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9580 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9581 (riscv_hard_regno_mode_ok): ...this and make static.
9582 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9583 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9584 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9585 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9586 (rl78_hard_regno_mode_ok): Make static and return bool.
9587 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9588 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9590 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9591 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9592 (rs6000_hard_regno_mode_ok): Rename to...
9593 (rs6000_hard_regno_mode_ok_uncached): ...this.
9594 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9595 (rs6000_hard_regno_mode_ok): New function.
9596 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9597 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9598 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9599 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9600 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9601 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9602 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9603 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9604 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9605 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9606 (sh_hard_regno_mode_ok): Make static.
9607 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9608 instead of HARD_REGNO_MODE_OK.
9609 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9610 (sparc_mode_class): Delete.
9611 (HARD_REGNO_MODE_OK): Delete.
9612 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9613 (hard_regno_mode_classes): Make static.
9614 (sparc_mode_class): Likewise.
9615 (sparc_hard_regno_mode_ok): New function.
9616 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9617 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9618 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9620 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9621 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9622 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9623 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9624 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9625 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9626 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9627 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9628 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9629 (visium_hard_regno_mode_ok): New function.
9630 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9631 instead of HARD_REGNO_MODE_OK.
9632 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9633 (HARD_REGNO_MODE_OK): Delete.
9634 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9635 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9636 (xtensa_option_override): Update accordingly.
9637 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9638 (xtensa_hard_regno_mode_ok): New function.
9639 * system.h (HARD_REGNO_MODE_OK): Poison.
9641 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9642 Alan Hayward <alan.hayward@arm.com>
9643 David Sherwood <david.sherwood@arm.com>
9645 * target.def (hard_regno_call_part_clobbered): New hook.
9646 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9647 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9648 * doc/tm.texi: Regenerate.
9649 * hooks.h (hook_bool_uint_mode_false): Declare.
9650 * hooks.c (hook_bool_uint_mode_false): New function.
9651 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9652 * cselib.c (cselib_process_insn): Use
9653 targetm.hard_regno_call_part_clobbered instead of
9654 HARD_REGNO_CALL_PART_CLOBBERED.
9655 * ira-conflicts.c (ira_build_conflicts): Likewise.
9656 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9657 * lra-constraints.c (need_for_call_save_p): Likewise.
9658 * lra-lives.c: Include target.h.
9659 (check_pseudos_live_through_calls): Use
9660 targetm.hard_regno_call_part_clobbered instead of
9661 HARD_REGNO_CALL_PART_CLOBBERED.
9662 * regcprop.c: Include target.h.
9663 (copyprop_hardreg_forward_1): Use
9664 targetm.hard_regno_call_part_clobbered instead of
9665 HARD_REGNO_CALL_PART_CLOBBERED.
9666 * reginfo.c (choose_hard_reg_mode): Likewise.
9667 * regrename.c (check_new_reg_p): Likewise.
9668 * reload.c (find_equiv_reg): Likewise.
9669 * reload1.c (emit_reload_insns): Likewise.
9670 * sched-deps.c (deps_analyze_insn): Likewise.
9671 * sel-sched.c (init_regs_for_mode): Likewise.
9672 (mark_unavailable_hard_regs): Likewise.
9673 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9674 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9675 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9677 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9678 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9679 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9681 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9683 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9684 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9685 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9687 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9688 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9689 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9691 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9692 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9694 * config/powerpcspe/powerpcspe.c
9695 (rs6000_hard_regno_call_part_clobbered): New function.
9696 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9697 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9698 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9700 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9701 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9702 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9704 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9705 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9706 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9708 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9709 Alan Hayward <alan.hayward@arm.com>
9710 David Sherwood <david.sherwood@arm.com>
9712 * rtl.h (subreg_memory_offset): Declare.
9713 * emit-rtl.c (subreg_memory_offset): New function.
9714 * expmed.c (store_bit_field_1): Use it.
9715 * expr.c (undefined_operand_subword_p): Likewise.
9716 * simplify-rtx.c (simplify_subreg): Likewise.
9718 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9720 PR rtl-optimization/57448
9723 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9724 using atomic_load pattern.
9725 (expand_atomic_store): Likewise.
9727 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9730 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9731 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9732 handling. Use replace_call_with_value with NULL instead of
9733 gsi_replace, unlink_stmt_vdef and release_defs.
9735 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9738 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9740 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9743 * rtl.h (emit_library_call_value_1): Declare.
9744 (emit_library_call): Replace declaration with a series of overloads.
9745 Remove the parameter count argument.
9746 (emit_library_call_value): Likewise.
9747 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9748 with an "rtx_mode_t *".
9749 (emit_library_call_value): Delete.
9750 (emit_library_call): Likewise.
9751 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9752 (asan_emit_allocas_unpoison): Likewise.
9753 * builtins.c (expand_builtin_powi): Likewise.
9754 (expand_asan_emit_allocas_unpoison): Likewise.
9755 * cfgexpand.c (expand_main_function): Likewise.
9756 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9757 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9758 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9759 * config/arm/arm.c (arm_trampoline_init): Likewise.
9760 (arm_call_tls_get_addr): Likewise.
9761 (arm_expand_divmod_libfunc): Likewise.
9762 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9763 (smulsi3_highpart): Likewise.
9764 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9765 (c6x_expand_compare): Likewise.
9766 (c6x_expand_movmem): Likewise.
9767 * config/frv/frv.c (frv_trampoline_init): Likewise.
9768 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9769 (ix86_expand_divmod_libfunc): Likewise.
9770 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9771 (ia64_expand_compare): Likewise.
9772 (ia64_profile_hook): Likewise.
9773 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9774 (nonlocal_goto): Likewise.
9775 (restore_stack_nonlocal): Likewise.
9776 * config/m32r/m32r.c (block_move_call): Likewise.
9777 (m32r_trampoline_init): Likewise.
9778 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9779 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9780 (m68k_call_m68k_read_tp): Likewise.
9781 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9782 (microblaze_expand_divide): Likewise.
9783 * config/mips/mips.h (mips_args): Likewise.
9784 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9785 (MIPS_ICACHE_SYNC): Likewise.
9786 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9787 (nios2_trampoline_init): Likewise.
9788 * config/pa/pa.c (hppa_tls_call): Likewise.
9789 (pa_trampoline_init): Likewise.
9790 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9791 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9792 (expand_strn_compare): Likewise.
9793 (rs6000_generate_compare): Likewise.
9794 (rs6000_expand_float128_convert): Likewise.
9795 (output_profile_hook): Likewise.
9796 (rs6000_trampoline_init): Likewise.
9797 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9798 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9799 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9800 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9801 (rs6000_generate_compare): Likewise.
9802 (rs6000_expand_float128_convert): Likewise.
9803 (output_profile_hook): Likewise.
9804 (rs6000_trampoline_init): Likewise.
9805 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9806 * config/sh/sh.c (sh_trampoline_init): Likewise.
9807 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9808 (sparc_emit_float_lib_cmp): Likewise.
9809 (sparc32_initialize_trampoline): Likewise.
9810 (sparc64_initialize_trampoline): Likewise.
9811 (sparc_profile_hook): Likewise.
9812 * config/spu/spu.c (ea_load_store): Likewise.
9813 * config/spu/spu.md (floatunssidf2): Likewise.
9814 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9815 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9816 * config/visium/visium.c (expand_block_move_4): Likewise.
9817 (expand_block_move_2): Likewise.
9818 (expand_block_move_1): Likewise.
9819 (expand_block_set_4): Likewise.
9820 (expand_block_set_2): Likewise.
9821 (expand_block_set_1): Likewise.
9822 (visium_trampoline_init): Likewise.
9823 (visium_profile_hook): Likewise.
9824 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9825 (xtensa_setup_frame_addresses): Likewise.
9826 (xtensa_trampoline_init): Likewise.
9827 * except.c (sjlj_emit_function_enter): Likewise.
9828 (sjlj_emit_function_exit): Likewise.
9829 * explow.c (allocate_dynamic_stack_space): Likewise.
9830 (probe_stack_range): Likewise.
9831 * expr.c (convert_mode_scalar): Likewise.
9832 * optabs.c (expand_binop): Likewise.
9833 (expand_twoval_binop_libfunc): Likewise.
9834 (expand_unop): Likewise.
9835 (prepare_cmp_insn): Likewise.
9836 (prepare_float_lib_cmp): Likewise.
9837 (expand_float): Likewise.
9838 (expand_fix): Likewise.
9839 (expand_fixed_convert): Likewise.
9840 (maybe_emit_sync_lock_test_and_set): Likewise.
9841 (expand_atomic_compare_and_swap): Likewise.
9842 (expand_mem_thread_fence): Likewise.
9843 (expand_atomic_fetch_op): Likewise.
9845 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9847 * doc/generic.texi (OpenACC): Adjust URL.
9848 * doc/invoke.texi (C Dialect Options): Ditto.
9850 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9852 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9853 predicate for operand 1. Add (m,<S>) constraint.
9854 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9855 Prevent memory operand 1 with register operand 2.
9857 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9859 PR rtl-optimization/82024
9860 * combine.c (try_combine): If the combination result is a PARALLEL,
9861 and we only need to retain the SET in there that would be placed
9862 at I2, check that we can place that at I3 instead, before doing so.
9864 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9867 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9869 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9870 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9873 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9874 Jeff Law <law@redhat.com>
9876 * varasm.c (bss_initializer_p): Do not put constants into .bss
9877 (categorize_decl_for_section): Handle bss_initializer_p returning
9878 false when DECL_INITIAL is NULL.
9880 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9883 * config/s390/s390.c (s390_can_inline_p): New function.
9885 2017-09-01 Jeff Law <law@redhat.com>
9887 PR tree-optimization/82052
9888 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9889 Always initialize the returned slot after a hash table miss
9890 when INSERT is true.
9892 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9894 * config/s390/s390.md (mem_signal_fence): Remove.
9895 * doc/md.texi (mem_signal_fence): Remove.
9896 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9898 * target-insns.def (mem_signal_fence): Remove.
9900 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9903 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9906 * asan.c (create_odr_indicator): Strip name encoding from assembler
9907 name before appending it after __odr_asan_.
9909 2017-09-01 Martin Liska <mliska@suse.cz>
9911 PR tree-optimization/82059
9912 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9913 frequency only when an edge is redirected.
9915 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9917 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9918 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9919 (arc_conditional_register_usage): Remove ARC600 lp_count
9921 (arc_file_start): Emit Tag_ARC_CPU_variation.
9922 (arc_can_use_doloop_p): New conditions to use ZOLs.
9923 (hwloop_fail): New function.
9924 (hwloop_optimize): Likewise.
9925 (hwloop_pattern_reg): Likewise.
9926 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9927 (arc_reorg_loops): New function, calls reorg_loops.
9928 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9929 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9931 (arc_loop_hazard): Remove function, functionality moved into
9933 (arc_hazard): Remove arc_loop_hazard call.
9934 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9935 into hwloop_optimize.
9936 (arc_label_align): Remove ZOL handling.
9937 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9938 * config/arc/arc.md (doloop_begin): Remove pattern.
9939 (doloop_begin_i): Likewise.
9940 (doloop_end_i): Likewise.
9941 (doloop_fallback): Likewise.
9942 (doloop_fallback_m): Likewise.
9943 (doloop_end): Reimplement expand.
9944 (arc_lp): New pattern for LP instruction.
9945 (loop_end): New pattern.
9946 (loop_fail): Likewise.
9947 (decrement_and_branch_until_zero): Likewise.
9948 * config/arc/arc.opt (mlpc-width): New option.
9949 * doc/invoke.texi (mlpc-width): Document option.
9951 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9953 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9954 (arc_ccfsm_advance): Fix checking for delay slots.
9955 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9957 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9959 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9961 (movhi_insn): Update store instruction constraint which are saving
9962 6-bit short immediates.
9963 (movsi_insn): Consider also short scaled load operations.
9964 (zero_extendhisi2_i): Use Usd constraint instead of T.
9965 (extendhisi2_i): Add q constraint.
9966 (arc_clzsi2): Add type and length attributes.
9967 (arc_ctzsi2): Likewise.
9968 * config/arc/constraints.md (Usc): Update constraint, the
9969 assembler can parse two relocations for a single instruction.
9971 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9973 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9974 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9976 2017-08-31 Olivier Hainque <hainque@adacore.com>
9978 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9979 match as powerpc-wrs-vxworks*.
9981 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9983 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9984 register constraint for by-element operand.
9985 (aarch64_mls_elt_merge<mode>): Likewise.
9987 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9989 * config/arc/arc.c (arc_can_follow_jump): Check for short
9992 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9994 * config.gcc: Use g.opt for arc.
9995 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9996 functionality moved to ...
9997 (legitimate_scaled_address_p): New function, ...here.
9998 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9999 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
10000 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
10002 (arc_override_options): Handle G option.
10003 (arc_output_pic_addr_const): Correct function definition.
10004 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
10005 (arc_decl_anon_ns_mem_p): Delete.
10006 (arc_in_small_data_p): Overhaul this function to take into
10007 consideration the value given via G option.
10008 (arc_rewrite_small_data_1): Renamed and corrected old
10009 arc_rewrite_small_data function.
10010 (arc_rewrite_small_data): New function.
10011 (small_data_pattern): Don't use pic_offset_table_rtx.
10012 * config/arc/arc.h (CC1_SPEC): Recognize G option.
10013 * config/arc/simdext.md (movmisalignv2hi): Use
10014 prepare_move_operands function.
10016 (movmisalign*): Likewise.
10017 * doc/invoke.texi (ARC options): Document -G option.
10019 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
10021 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
10023 * config/arc/arc.c (arc_print_operand): Output scalled address for
10024 sdata whenever is possible.
10025 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10026 load/stores are available.
10027 (compact_sda_memory_operand): Check for the alignment required by
10028 code density instructions.
10029 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10031 * config/arc/constraints.md (Usd): Update constraint.
10032 (Us0): New constraint.
10033 (Usc): Update constraint.
10035 2017-08-31 Richard Biener <rguenther@suse.de>
10037 PR middle-end/82054
10038 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10039 function only once.
10041 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10043 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10044 Resize type_signature.
10046 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10047 Alan Hayward <alan.hayward@arm.com>
10048 David Sherwood <david.sherwood@arm.com>
10050 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10051 subregs whose inner modes can be stored in GPRs.
10052 (aarch64_classify_index): Likewise.
10054 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10055 Alan Hayward <alan.hayward@arm.com>
10056 David Sherwood <david.sherwood@arm.com>
10058 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10059 (V_INT_EQUIV): ...this.
10060 (v_cmp_result): Rename to...
10061 (v_int_equiv): ...this.
10062 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10063 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10064 (copysign<mode>3): Likewise.
10065 (aarch64_simd_bsl<mode>_internal): Likewise.
10066 (aarch64_simd_bsl<mode>): Likewise.
10067 (vec_cmp<mode><mode>): Likewise.
10068 (vcond<mode><mode>): Likewise.
10069 (vcond<v_cmp_mixed><mode>): Likewise.
10070 (vcondu<mode><v_cmp_mixed>): Likewise.
10071 (aarch64_cm<optab><mode>): Likewise.
10072 (aarch64_cmtst<mode>): Likewise.
10073 (aarch64_fac<optab><mode>): Likewise.
10074 (vec_perm_const<mode>): Likewise.
10075 (vcond_mask_<mode><v_cmp_result>): Rename to...
10076 (vcond_mask_<mode><v_int_equiv>): ...this.
10077 (vec_cmp<mode><v_cmp_result>): Rename to...
10078 (vec_cmp<mode><v_int_equiv>): ...this.
10080 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10081 Alan Hayward <alan.hayward@arm.com>
10082 David Sherwood <david.sherwood@arm.com>
10084 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
10086 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10087 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10088 (UNSPEC_LD4_DREG): New unspecs.
10089 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10090 (aarch64_ld2<mode>_dreg_be): Replace with...
10091 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10093 (aarch64_ld3<mode>_dreg_le)
10094 (aarch64_ld3<mode>_dreg_be): Replace with...
10095 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10097 (aarch64_ld4<mode>_dreg_le)
10098 (aarch64_ld4<mode>_dreg_be): Replace with...
10099 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10102 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10104 PR tree-optimization/81987
10105 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10106 insert an initializer in a location not dominated by the stride
10109 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10111 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10112 on the entire header of the finally block in the fallthru case.
10114 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10116 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10118 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10120 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10121 rs6000_emit_move_from_cr and call renamed function.
10122 (rs6000_emit_prologue): Call renamed functions.
10123 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10124 movesi_from_cr, remove volatile CRs.
10126 2017-08-30 Jon Beniston <jon@beniston.com>
10127 Richard Biener <rguenther@suse.de>
10129 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10130 of VECTOR_MODE_P check.
10131 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10132 element vector types.
10134 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10136 * df.h (df_read_modify_subreg_p): Remove in favor of...
10137 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10138 const_rtx instead of an rtx.
10139 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10140 * df-problems.c (df_word_lr_mark_ref): Likewise.
10141 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10142 (mark_pseudo_reg_dead): Likewise.
10143 (mark_ref_dead): Likewise.
10144 * reginfo.c (init_subregs_of_mode): Likewise.
10145 * sched-deps.c (sched_analyze_1): Likewise.
10146 * df-scan.c (df_def_record_1): Likewise.
10147 (df_uses_record): Likewise.
10148 (df_read_modify_subreg_p): Remove in favor of...
10149 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10150 const_rtx instead of an rtx.
10152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10153 Alan Hayward <alan.hayward@arm.com>
10154 David Sherwood <david.sherwood@arm.com>
10156 * rtl.h (partial_subreg_p): New function.
10157 * caller-save.c (save_call_clobbered_regs): Use it.
10158 * calls.c (expand_call): Likewise.
10159 * combine.c (combinable_i3pat): Likewise.
10160 (simplify_set): Likewise.
10161 (make_extraction): Likewise.
10162 (make_compound_operation_int): Likewise.
10163 (gen_lowpart_or_truncate): Likewise.
10164 (force_to_mode): Likewise.
10165 (make_field_assignment): Likewise.
10166 (reg_truncated_to_mode): Likewise.
10167 (record_truncated_value): Likewise.
10168 (move_deaths): Likewise.
10169 * cse.c (record_jump_cond): Likewise.
10170 (cse_insn): Likewise.
10171 * cselib.c (cselib_lookup_1): Likewise.
10172 * expmed.c (extract_bit_field_using_extv): Likewise.
10173 * function.c (assign_parm_setup_reg): Likewise.
10174 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10175 * ira-build.c (create_insn_allocnos): Likewise.
10176 * lra-coalesce.c (merge_pseudos): Likewise.
10177 * lra-constraints.c (match_reload): Likewise.
10178 (simplify_operand_subreg): Likewise.
10179 (curr_insn_transform): Likewise.
10180 * lra-lives.c (process_bb_lives): Likewise.
10181 * lra.c (new_insn_reg): Likewise.
10182 (lra_substitute_pseudo): Likewise.
10183 * regcprop.c (mode_change_ok): Likewise.
10184 (maybe_mode_change): Likewise.
10185 (copyprop_hardreg_forward_1): Likewise.
10186 * reload.c (push_reload): Likewise.
10187 (find_reloads): Likewise.
10188 (find_reloads_subreg_address): Likewise.
10189 * reload1.c (alter_reg): Likewise.
10190 (eliminate_regs_1): Likewise.
10191 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10193 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10195 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10196 back to if statements, including unpack.
10198 2017-08-30 Martin Liska <mliska@suse.cz>
10200 PR inline-asm/82001
10201 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10203 (func_checker::compare_asm_inputs_outputs): ... this function.
10204 (func_checker::compare_gimple_asm): Use the function to compare
10205 also ASM constrains.
10206 * ipa-icf-gimple.h: Rename the function.
10208 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10209 Alan Hayward <alan.hayward@arm.com>
10210 David Sherwood <david.sherwood@arm.com>
10212 * coretypes.h (complex_mode): New type.
10213 * gdbhooks.py (build_pretty_printer): Handle it.
10214 * machmode.h (complex_mode): New class.
10215 (complex_mode::includes_p): New function.
10216 (is_complex_int_mode): Likewise.
10217 (is_complex_float_mode): Likewise.
10218 * genmodes.c (get_mode_class): Handle complex mode classes.
10219 * function.c (expand_function_end): Use is_complex_int_mode.
10221 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10222 Alan Hayward <alan.hayward@arm.com>
10223 David Sherwood <david.sherwood@arm.com>
10225 * coretypes.h (scalar_mode_pod): New typedef.
10226 * gdbhooks.py (build_pretty_printer): Handle it.
10227 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10228 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10229 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10230 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10231 as_a <scalar_mode>.
10233 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10234 Alan Hayward <alan.hayward@arm.com>
10235 David Sherwood <david.sherwood@arm.com>
10237 * machmode.h (mode_for_vector): Take a scalar_mode instead
10239 * stor-layout.c (mode_for_vector): Likewise.
10240 * explow.c (promote_mode): Use as_a <scalar_mode>.
10241 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10243 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10244 Alan Hayward <alan.hayward@arm.com>
10245 David Sherwood <david.sherwood@arm.com>
10247 * target.def (preferred_simd_mode): Take a scalar_mode
10248 instead of a machine_mode.
10249 * targhooks.h (default_preferred_simd_mode): Likewise.
10250 * targhooks.c (default_preferred_simd_mode): Likewise.
10251 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10252 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10253 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10254 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10255 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10256 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10257 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10258 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10260 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10261 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10262 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10263 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10264 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10265 * doc/tm.texi: Regenerate.
10266 * optabs-query.c (can_vec_mask_load_store_p): Return false for
10269 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10270 Alan Hayward <alan.hayward@arm.com>
10271 David Sherwood <david.sherwood@arm.com>
10273 * target.def (scalar_mode_supported_p): Take a scalar_mode
10274 instead of a machine_mode.
10275 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10276 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10277 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10278 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10279 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10280 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10281 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10282 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10283 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10284 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10285 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10286 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10287 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10288 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10290 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10291 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10292 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10293 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10294 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10296 * doc/tm.texi: Regenerate.
10298 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10299 Alan Hayward <alan.hayward@arm.com>
10300 David Sherwood <david.sherwood@arm.com>
10302 * coretypes.h (opt_scalar_mode): New typedef.
10303 * gdbhooks.py (build_pretty_printers): Handle it.
10304 * machmode.h (mode_iterator::get_2xwider): Add overload for
10306 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10308 * expr.c (convert_mode_scalar): Likewise.
10309 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10310 * optabs.c (expand_float): Likewise.
10311 (expand_fix): Likewise.
10312 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10314 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10315 Alan Hayward <alan.hayward@arm.com>
10316 David Sherwood <david.sherwood@arm.com>
10318 * optabs.c (expand_float): Explicitly check for scalars before
10319 using a branching expansion.
10320 (expand_fix): Likewise.
10322 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10323 Alan Hayward <alan.hayward@arm.com>
10324 David Sherwood <david.sherwood@arm.com>
10326 * expr.c (convert_mode): Split scalar handling out into...
10327 (convert_mode_scalar): ...this new function. Treat the modes
10330 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10331 Alan Hayward <alan.hayward@arm.com>
10332 David Sherwood <david.sherwood@arm.com>
10334 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10336 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10338 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10339 Alan Hayward <alan.hayward@arm.com>
10340 David Sherwood <david.sherwood@arm.com>
10342 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10343 rather than a machine_mode.
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 * fixed-value.c (fixed_from_double_int): Likewise.
10350 (fixed_from_string): Likewise.
10351 (fixed_convert): Likewise.
10352 (fixed_convert_from_int): Likewise.
10353 (fixed_convert_from_real): Likewise.
10354 (real_convert_from_fixed): Likewise.
10355 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10358 Alan Hayward <alan.hayward@arm.com>
10359 David Sherwood <david.sherwood@arm.com>
10361 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10362 of separate mode class checks. Do not allow vector modes here.
10363 (immed_wide_int_const): Use as_a <scalar_mode>.
10364 * explow.c (trunc_int_for_mode): Likewise.
10365 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10366 (wi::shwi): Likewise.
10367 (wi::min_value): Likewise.
10368 (wi::max_value): Likewise.
10369 * dwarf2out.c (loc_descriptor): Likewise.
10370 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10371 for CONST_WIDE_INT.
10373 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10374 Alan Hayward <alan.hayward@arm.com>
10375 David Sherwood <david.sherwood@arm.com>
10377 * tree.h (SCALAR_TYPE_MODE): New macro.
10378 * expr.c (expand_expr_addr_expr_1): Use it.
10379 (expand_expr_real_2): Likewise.
10380 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10381 (fold_convert_const_fixed_from_int): Likewise.
10382 (fold_convert_const_fixed_from_real): Likewise.
10383 (native_encode_fixed): Likewise
10384 (native_encode_complex): Likewise
10385 (native_encode_vector): Likewise.
10386 (native_interpret_fixed): Likewise.
10387 (native_interpret_real): Likewise.
10388 (native_interpret_complex): Likewise.
10389 (native_interpret_vector): Likewise.
10390 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10391 (simd_clone_adjust_argument_types): Likewise.
10392 (simd_clone_init_simd_arrays): Likewise.
10393 (simd_clone_adjust): Likewise.
10394 * stor-layout.c (layout_type): Likewise.
10395 * tree.c (build_minus_one_cst): Likewise.
10396 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10397 * tree-inline.c (estimate_move_cost): Likewise.
10398 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10399 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10400 (vectorizable_reduction): Likewise.
10401 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10402 (vect_recog_mixed_size_cond_pattern): Likewise.
10403 (check_bool_pattern): Likewise.
10404 (adjust_bool_pattern): Likewise.
10405 (search_type_for_mask_1): Likewise.
10406 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10407 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10408 (vectorizable_load): Likewise.
10409 (vectorizable_store): Likewise.
10410 * ubsan.c (ubsan_encode_value): Likewise.
10411 * varasm.c (output_constant): Likewise.
10413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10414 Alan Hayward <alan.hayward@arm.com>
10415 David Sherwood <david.sherwood@arm.com>
10417 * coretypes.h (scalar_mode): New class.
10418 * machmode.h (scalar_mode): Likewise.
10419 (scalar_mode::includes_p): New function.
10420 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10421 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10422 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10423 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10424 * expmed.c (store_bit_field_1): Likewise.
10425 (extract_bit_field_1): Likewise.
10426 * expr.c (write_complex_part): Likewise.
10427 (read_complex_part): Likewise.
10428 (emit_move_complex_push): Likewise.
10429 (expand_expr_real_2): Likewise.
10430 * function.c (assign_parm_setup_reg): Likewise.
10431 (assign_parms_unsplit_complex): Likewise.
10432 * optabs.c (expand_binop): Likewise.
10433 * rtlanal.c (subreg_get_info): Likewise.
10434 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10435 * varasm.c (output_constant_pool_2): Likewise.
10437 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10438 Alan Hayward <alan.hayward@arm.com>
10439 David Sherwood <david.sherwood@arm.com>
10441 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10443 (expmed_mult_highpart_optab): Likewise.
10444 (expmed_mult_highpart): Likewise.
10446 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10447 Alan Hayward <alan.hayward@arm.com>
10448 David Sherwood <david.sherwood@arm.com>
10450 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10451 instead of a machine_mode.
10452 (builtin_memset_read_str): Likewise.
10453 * builtins.c (c_readstr): Likewise.
10454 (builtin_memcpy_read_str): Likewise.
10455 (builtin_strncpy_read_str): Likewise.
10456 (builtin_memset_read_str): Likewise.
10457 (builtin_memset_gen_str): Likewise.
10458 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10459 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10460 instead of a machine_mode.
10461 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10463 (make_extraction): Likewise.
10464 (try_widen_shift_mode): Take and return scalar_int_modes instead
10466 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10467 a scalar_int_mode instead of a machine_mode.
10468 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10469 (avr_addr_space_pointer_mode): Likewise.
10470 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10471 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10472 (msp430_unwind_word_mode): Likewise.
10473 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10474 (spu_addr_space_pointer_mode): Likewise.
10475 (spu_addr_space_address_mode): Likewise.
10476 (spu_libgcc_cmp_return_mode): Likewise.
10477 (spu_libgcc_shift_count_mode): Likewise.
10478 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10479 (rl78_addr_space_pointer_mode): Likewise.
10480 (fl78_unwind_word_mode): Likewise.
10481 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10483 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10484 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10485 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10486 (mips_valid_pointer_mode): Likewise.
10487 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10488 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10489 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10491 (ft32_addr_space_address_mode): Likewise.
10492 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10493 scalar_int_mode instead of a machine_mode.
10494 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10496 (m32c_addr_space_address_mode): Likewise.
10497 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10498 (rs6000_eh_return_filter_mode): Likewise.
10499 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10500 (rs6000_eh_return_filter_mode): Likewise.
10501 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10502 (s390_libgcc_shift_count_mode): Likewise.
10503 (s390_unwind_word_mode): Likewise.
10504 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10506 * target.def (mode_rep_extended): Likewise.
10507 (valid_pointer_mode): Likewise.
10508 (addr_space.valid_pointer_mode): Likewise.
10509 (eh_return_filter_mode): Return a scalar_int_mode rather than
10511 (libgcc_cmp_return_mode): Likewise.
10512 (libgcc_shift_count_mode): Likewise.
10513 (unwind_word_mode): Likewise.
10514 (addr_space.pointer_mode): Likewise.
10515 (addr_space.address_mode): Likewise.
10516 * doc/tm.texi: Regenerate.
10517 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10519 (do_jump): Use scalar_int_mode for local variables.
10520 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10521 rather than a machine_mode.
10522 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10523 (scompare_loc_descriptor_wide): Likewise.
10524 (scompare_loc_descriptor_narrow): Likewise.
10525 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10527 * except.c (sjlj_emit_dispatch_table): Likewise.
10528 (expand_builtin_eh_copy_values): Likewise.
10529 * explow.c (convert_memory_address_addr_space_1): Likewise.
10530 Take a scalar_int_mode rather than a machine_mode.
10531 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10532 than a machine_mode.
10533 (memory_address_addr_space): Use scalar_int_mode for local variables.
10534 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10535 rather than a machine_mode.
10536 * expmed.c (mask_rtx): Likewise.
10537 (init_expmed_one_conv): Likewise.
10538 (expand_mult_highpart_adjust): Likewise.
10539 (extract_high_half): Likewise.
10540 (expmed_mult_highpart_optab): Likewise.
10541 (expmed_mult_highpart): Likewise.
10542 (expand_smod_pow2): Likewise.
10543 (expand_sdiv_pow2): Likewise.
10544 (emit_store_flag_int): Likewise.
10545 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10547 (extract_low_bits): Likewise.
10548 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10550 * expr.c (pieces_addr::adjust): Likewise.
10551 (can_store_by_pieces): Likewise.
10552 (store_by_pieces): Likewise.
10553 (clear_by_pieces_1): Likewise.
10554 (expand_expr_addr_expr_1): Likewise.
10555 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10556 (expand_expr_real_1): Likewise.
10557 (try_casesi): Likewise.
10558 * final.c (shorten_branches): Likewise.
10559 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10560 type of "mode" to machine_mode.
10561 * internal-fn.c (expand_arith_overflow_result_store): Take a
10562 scalar_int_mode rather than a machine_mode.
10563 (expand_mul_overflow): Use scalar_int_mode for local variables.
10564 * loop-doloop.c (doloop_modify): Likewise.
10565 (doloop_optimize): Likewise.
10566 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10567 than a machine_mode.
10568 (expand_doubleword_shift_condmove): Likewise.
10569 (expand_doubleword_shift): Likewise.
10570 (expand_doubleword_clz): Likewise.
10571 (expand_doubleword_popcount): Likewise.
10572 (expand_doubleword_parity): Likewise.
10573 (expand_absneg_bit): Use scalar_int_mode for local variables.
10574 (prepare_float_lib_cmp): Likewise.
10575 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10576 rather than a machine_mode.
10577 (convert_memory_address_addr_space): Likewise.
10578 (get_mode_bounds): Likewise.
10579 (get_address_mode): Return a scalar_int_mode rather than a
10581 * rtlanal.c (get_address_mode): Likewise.
10582 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10583 than a machine_mode.
10584 * targhooks.c (default_mode_rep_extended): Likewise.
10585 (default_valid_pointer_mode): Likewise.
10586 (default_addr_space_valid_pointer_mode): Likewise.
10587 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10588 than a machine_mode.
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 * targhooks.h (default_eh_return_filter_mode): Likewise.
10595 (default_libgcc_cmp_return_mode): Likewise.
10596 (default_libgcc_shift_count_mode): Likewise.
10597 (default_unwind_word_mode): Likewise.
10598 (default_addr_space_pointer_mode): Likewise.
10599 (default_addr_space_address_mode): Likewise.
10600 (default_mode_rep_extended): Take a scalar_int_mode rather than
10602 (default_valid_pointer_mode): Likewise.
10603 (default_addr_space_valid_pointer_mode): Likewise.
10604 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10606 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10607 rather than a machine_mode.
10608 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10609 for local variables.
10610 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10611 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10612 than a machine_mode.
10614 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10615 Alan Hayward <alan.hayward@arm.com>
10616 David Sherwood <david.sherwood@arm.com>
10618 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10619 the mode argument to scalar_int_mode.
10620 (do_jump_by_parts_zero_rtx): Likewise.
10621 (do_jump_by_parts_equality_rtx): Likewise.
10622 (do_jump_by_parts_greater): Take a mode argument.
10623 (do_jump_by_parts_equality): Likewise.
10624 (do_jump_1): Update calls accordingly.
10626 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10627 Alan Hayward <alan.hayward@arm.com>
10628 David Sherwood <david.sherwood@arm.com>
10630 * is-a.h (safe_dyn_cast): New function.
10631 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10632 (jump_table_for_label): Likewise.
10633 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10634 instead of an rtx_insn *.
10635 (shorten_branches): Use dyn_cast instead of LABEL_P and
10636 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10637 rtx_jump_table_data::get_data_mode.
10638 (final_scan_insn): Likewise.
10640 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10641 Alan Hayward <alan.hayward@arm.com>
10642 David Sherwood <david.sherwood@arm.com>
10644 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10645 trying to combine a full-register integer set with a subreg
10648 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10649 Alan Hayward <alan.hayward@arm.com>
10650 David Sherwood <david.sherwood@arm.com>
10652 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10653 that is always either address_mode or pointer_mode.
10655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10656 Alan Hayward <alan.hayward@arm.com>
10657 David Sherwood <david.sherwood@arm.com>
10659 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10660 when the two are known to be equal.
10662 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10663 Alan Hayward <alan.hayward@arm.com>
10664 David Sherwood <david.sherwood@arm.com>
10666 * simplify-rtx.c (simplify_const_unary_operation): Use
10667 is_a <scalar_int_mode> instead of checking for a nonzero
10668 precision. Forcibly convert op_mode to a scalar_int_mode
10669 in that case. More clearly differentiate the operand and
10670 result modes and use the former when deciding what the value
10671 of a count-bits operation should be. Use is_int_mode instead
10672 of checking for a MODE_INT. Remove redundant check for whether
10673 this mode has a zero precision.
10675 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10676 Alan Hayward <alan.hayward@arm.com>
10677 David Sherwood <david.sherwood@arm.com>
10679 * optabs.c (widen_leading): Change the type of the mode argument
10680 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10681 (widen_bswap): Likewise.
10682 (expand_parity): Likewise.
10683 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10684 (expand_ffs): Likewise.
10685 (epand_unop): Check for scalar integer modes before calling the
10688 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10689 Alan Hayward <alan.hayward@arm.com>
10690 David Sherwood <david.sherwood@arm.com>
10692 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10694 (expand_expr_real_1): Update call accordingly.
10696 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10697 Alan Hayward <alan.hayward@arm.com>
10698 David Sherwood <david.sherwood@arm.com>
10700 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10701 value_mode arguments. Use scalar_int_mode internally.
10702 (store_bit_field_1): Rename the new integer mode from imode
10703 to op0_mode and use it instead of GET_MODE (op0). Update calls
10704 to store_split_bit_field, store_bit_field_using_insv and
10705 store_fixed_bit_field.
10706 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10707 Use scalar_int_mode internally. Use a bit count rather than a mode
10708 when calculating the largest bit size for get_best_mode.
10709 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10710 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10711 Remove assertion that OP0 has a scalar integer mode.
10712 (store_split_bit_field): Add op0_mode and value_mode arguments.
10713 Update calls to extract_fixed_bit_field.
10714 (extract_bit_field_using_extv): Add an op0_mode argument.
10715 Use scalar_int_mode internally.
10716 (extract_bit_field_1): Rename the new integer mode from imode to
10717 op0_mode and use it instead of GET_MODE (op0). Update calls to
10718 extract_split_bit_field, extract_bit_field_using_extv and
10719 extract_fixed_bit_field.
10720 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10721 to extract_split_bit_field and extract_fixed_bit_field_1.
10722 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10723 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10724 on the target mode.
10725 (extract_split_bit_field): Add an op0_mode argument. Update call
10726 to extract_fixed_bit_field.
10728 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10729 Alan Hayward <alan.hayward@arm.com>
10730 David Sherwood <david.sherwood@arm.com>
10732 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10733 * explow.c (hard_function_value): Likewise.
10734 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10735 convert_to_mode call outside the loop.
10736 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10737 for the mode iterator. Require the mode specified by max_pieces
10739 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10741 (copy_blkmode_to_reg): Likewise.
10742 (set_storage_via_setmem): Likewise.
10743 * optabs.c (prepare_cmp_insn): Likewise.
10744 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10745 * stor-layout.c (finish_bitfield_representative): Likewise.
10747 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10748 Alan Hayward <alan.hayward@arm.com>
10749 David Sherwood <david.sherwood@arm.com>
10751 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10752 * expr.c (convert_move): Use them.
10753 (convert_modes): Likewise.
10754 (store_expr_with_bounds): Likewise.
10756 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10757 Alan Hayward <alan.hayward@arm.com>
10758 David Sherwood <david.sherwood@arm.com>
10760 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10761 parameter for the mode of "x". Remove the "known_x", "known_mode"
10762 and "known_ret" arguments. Change the type of the mode argument
10763 to scalar_int_mode.
10764 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10765 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10766 (reg_num_sign_bit_copies_for_combine): Likewise.
10767 * rtlanal.c (nonzero_bits1): Likewise.
10768 (num_sign_bit_copies1): Likewise.
10769 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10770 (reg_num_sign_bit_copies_general): Likewise.
10771 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10772 (reg_nonzero_bits_general): Likewise.
10774 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10775 Alan Hayward <alan.hayward@arm.com>
10776 David Sherwood <david.sherwood@arm.com>
10778 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10779 than in subroutines. Return 1 for non-integer modes.
10780 (cached_num_sign_bit_copies): Change the type of the mode parameter
10781 to scalar_int_mode.
10782 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10783 classes. Handle CONST_INT_P first and then check whether X also
10784 has a scalar integer mode. Check the same thing for inner registers
10785 of a SUBREG and for values that are being extended or truncated.
10787 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10788 Alan Hayward <alan.hayward@arm.com>
10789 David Sherwood <david.sherwood@arm.com>
10791 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10792 in subroutines. Return the mode mask for non-integer modes.
10793 (cached_nonzero_bits): Change the type of the mode parameter
10794 to scalar_int_mode.
10795 (nonzero_bits1): Likewise. Remove early exit for other mode
10796 classes. Handle CONST_INT_P first and then check whether X
10797 also has a scalar integer mode.
10799 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10800 Alan Hayward <alan.hayward@arm.com>
10801 David Sherwood <david.sherwood@arm.com>
10803 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10804 Return a scalar_int_mode and assert that the size is greater than
10806 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10807 (op_by_pieces_d::op_by_pieces_d): Likewise.
10808 (op_by_pieces_d::run): Likewise.
10809 (can_store_by_pieces): Likewise.
10811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10812 Alan Hayward <alan.hayward@arm.com>
10813 David Sherwood <david.sherwood@arm.com>
10815 * combine.c (extract_left_shift): Add a mode argument and update
10817 (make_compound_operation_int): Change the type of the mode parameter
10818 to scalar_int_mode and update the call to extract_left_shift.
10820 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10821 Alan Hayward <alan.hayward@arm.com>
10822 David Sherwood <david.sherwood@arm.com>
10824 * combine.c (simplify_and_const_int): Change the type of the mode
10825 parameter to scalar_int_mode.
10826 (simplify_and_const_int_1): Likewise. Update recursive call.
10828 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10829 Alan Hayward <alan.hayward@arm.com>
10830 David Sherwood <david.sherwood@arm.com>
10832 * combine.c (simplify_compare_const): Check that the mode is a
10833 scalar_int_mode (rather than VOIDmode) before testing its
10835 (simplify_comparison): Move COMPARISON_P handling out of the
10836 loop and restrict the latter part of the loop to scalar_int_modes.
10837 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10838 and when considering SUBREG_REGs. Use is_int_mode instead of
10839 checking GET_MODE_CLASS against MODE_INT.
10841 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10842 Alan Hayward <alan.hayward@arm.com>
10843 David Sherwood <david.sherwood@arm.com>
10845 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10846 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10847 shift_unit_mode and for modes involved in scalar shifts.
10849 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10850 Alan Hayward <alan.hayward@arm.com>
10851 David Sherwood <david.sherwood@arm.com>
10853 * combine.c (force_int_to_mode): New function, split out from...
10854 (force_to_mode): ...here. Keep xmode up-to-date and use it
10855 instead of GET_MODE (x).
10857 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10858 Alan Hayward <alan.hayward@arm.com>
10859 David Sherwood <david.sherwood@arm.com>
10861 * optabs-query.h (extraction_insn::struct_mode): Change type to
10862 opt_scalar_int_mode and update comment.
10863 (extraction_insn::field_mode): Change type to scalar_int_mode.
10864 (extraction_insn::pos_mode): Likewise.
10865 * combine.c (make_extraction): Update accordingly.
10866 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10867 (get_optab_extraction_insn): Likewise.
10868 * recog.c (simplify_while_replacing): Likewise.
10869 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10870 parameter to opt_scalar_int_mode.
10872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10873 Alan Hayward <alan.hayward@arm.com>
10874 David Sherwood <david.sherwood@arm.com>
10876 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10877 to a scalar_int_mode instead of a machine_mode.
10878 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10879 (get_best_mode): Return a boolean and use a pointer argument to store
10880 the selected mode. Replace the limit mode parameter with a bit limit.
10881 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10882 for the values returned by bit_field_mode_iterator::next_mode.
10883 (store_bit_field): Update call to get_best_mode.
10884 (store_fixed_bit_field): Likewise.
10885 (extract_fixed_bit_field): Likewise.
10886 * expr.c (optimize_bitfield_assignment_op): Likewise.
10887 * fold-const.c (optimize_bit_field_compare): Likewise.
10888 (fold_truth_andor_1): Likewise.
10889 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10890 Update for new type of m_mode.
10891 (get_best_mode): As above.
10893 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10894 Alan Hayward <alan.hayward@arm.com>
10895 David Sherwood <david.sherwood@arm.com>
10897 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10898 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10899 (store_bit_field): Check is_a <scalar_int_mode> before calling
10900 strict_volatile_bitfield_p.
10901 (extract_bit_field): Likewise.
10903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10904 Alan Hayward <alan.hayward@arm.com>
10905 David Sherwood <david.sherwood@arm.com>
10907 * target.def (cstore_mode): Return a scalar_int_mode.
10908 * doc/tm.texi: Regenerate.
10909 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10910 * targhooks.h (default_cstore_mode): Likewise.
10911 * targhooks.c (default_cstore_mode): Likewise, using a forced
10913 * expmed.c (emit_cstore): Expect the target of the cstore to be
10916 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10917 Alan Hayward <alan.hayward@arm.com>
10918 David Sherwood <david.sherwood@arm.com>
10920 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10922 (niter_desc): Likewise mode.
10923 (iv_analyze): Add a mode parameter.
10925 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10926 and change its type to scalar_int_mode.
10927 * loop-iv.c: Update commentary at head of file.
10928 (iv_constant): Pass the mode paraeter before the rtx it describes
10929 and change its type to scalar_int_mode. Remove VOIDmode handling.
10930 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10931 (iv_extend): Likewise.
10932 (shorten_into_mode): Likewise.
10933 (iv_add): Use scalar_int_mode.
10934 (iv_mult): Likewise.
10935 (iv_shift): Likewise.
10936 (canonicalize_iv_subregs): Likewise.
10937 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10938 it describes and change its mode to scalar_int_mode. Also change
10939 the type of the returned inner_mode to scalar_int_mode.
10940 (get_biv_step): Likewise, turning outer_mode from a pointer
10941 into a direct parameter. Update call to get_biv_step_1.
10942 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10943 iv_constant and get_biv_step.
10944 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10945 and change its type to scalar_int_mode. Don't initialise iv->mode
10946 to VOIDmode and remove later checks for its still being VOIDmode.
10947 Update calls to iv_analyze_op and iv_analyze_expr. Check
10948 is_a <scalar_int_mode> when changing the mode under consideration.
10949 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10950 Update call to iv_analyze_expr.
10951 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10952 inner register is not also a scalar_int_mode. Update call to
10954 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10955 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10956 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10957 separate mode class checks. Update calls to iv_analyze. Remove
10958 fix-up of VOIDmodes after iv_analyze_biv.
10959 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10960 don't have a scalar_int_mode. Update call to biv_p.
10962 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10963 Alan Hayward <alan.hayward@arm.com>
10964 David Sherwood <david.sherwood@arm.com>
10966 * cfgexpand.c (convert_debug_memory_address): Use
10967 as_a <scalar_int_mode>.
10968 * combine.c (expand_compound_operation): Likewise.
10969 (make_extraction): Likewise.
10970 (change_zero_ext): Likewise.
10971 (simplify_comparison): Likewise.
10972 * cse.c (cse_insn): Likewise.
10973 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10974 (mem_loc_descriptor): Likewise.
10975 (loc_descriptor): Likewise.
10976 * expmed.c (init_expmed_one_mode): Likewise.
10977 (synth_mult): Likewise.
10978 (emit_store_flag_1): Likewise.
10979 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10980 of a comparison with size.
10981 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10982 (reduce_to_bit_field_precision): Likewise.
10983 * function.c (expand_function_end): Likewise.
10984 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10985 * loop-doloop.c (doloop_modify): Likewise.
10986 * optabs.c (expand_binop): Likewise.
10987 (expand_unop): Likewise.
10988 (expand_copysign_absneg): Likewise.
10989 (prepare_cmp_insn): Likewise.
10990 (maybe_legitimize_operand): Likewise.
10991 * recog.c (const_scalar_int_operand): Likewise.
10992 * rtlanal.c (get_address_mode): Likewise.
10993 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10994 (simplify_cond_clz_ctz): Likewise.
10995 * tree-nested.c (get_nl_goto_field): Likewise.
10996 * tree.c (build_vector_type_for_mode): Likewise.
10997 * var-tracking.c (use_narrower_mode): Likewise.
10999 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11000 Alan Hayward <alan.hayward@arm.com>
11001 David Sherwood <david.sherwood@arm.com>
11003 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
11004 * builtins.c (expand_builtin_signbit): Use it.
11005 * cfgexpand.c (expand_debug_expr): Likewise.
11006 * dojump.c (do_jump): Likewise.
11007 (do_compare_and_jump): Likewise.
11008 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
11009 * expmed.c (make_tree): Likewise.
11010 * expr.c (expand_expr_real_2): Likewise.
11011 (expand_expr_real_1): Likewise.
11012 (try_casesi): Likewise.
11013 * fold-const-call.c (fold_const_call_ss): Likewise.
11014 * fold-const.c (unextend): Likewise.
11015 (extract_muldiv_1): Likewise.
11016 (fold_single_bit_test): Likewise.
11017 (native_encode_int): Likewise.
11018 (native_encode_string): Likewise.
11019 (native_interpret_int): Likewise.
11020 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
11021 * internal-fn.c (expand_addsub_overflow): Likewise.
11022 (expand_neg_overflow): Likewise.
11023 (expand_mul_overflow): Likewise.
11024 (expand_arith_overflow): Likewise.
11025 * match.pd: Likewise.
11026 * stor-layout.c (layout_type): Likewise.
11027 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11028 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11029 * tree-ssanames.c (get_range_info): Likewise.
11030 * tree-switch-conversion.c (array_value_type) Likewise.
11031 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11032 (vect_recog_divmod_pattern): Likewise.
11033 (vect_recog_mixed_size_cond_pattern): Likewise.
11034 * tree-vrp.c (extract_range_basic): Likewise.
11035 (simplify_float_conversion_using_ranges): Likewise.
11036 * tree.c (int_fits_type_p): Likewise.
11037 * ubsan.c (instrument_bool_enum_load): Likewise.
11038 * varasm.c (mergeable_string_section): Likewise.
11039 (narrowing_initializer_constant_valid_p): Likewise.
11040 (output_constant): Likewise.
11042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11043 Alan Hayward <alan.hayward@arm.com>
11044 David Sherwood <david.sherwood@arm.com>
11046 * machmode.h (NARROWEST_INT_MODE): New macro.
11047 * expr.c (alignment_for_piecewise_move): Use it instead of
11048 GET_CLASS_NARROWEST_MODE (MODE_INT).
11049 (push_block): Likewise.
11050 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11052 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11054 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11055 Alan Hayward <alan.hayward@arm.com>
11056 David Sherwood <david.sherwood@arm.com>
11058 * postreload.c (move2add_valid_value_p): Change the type of the
11059 mode parameter to scalar_int_mode.
11060 (move2add_use_add2_insn): Add a mode parameter and use it instead
11062 (move2add_use_add3_insn): Likewise.
11063 (reload_cse_move2add): Update accordingly.
11065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11066 Alan Hayward <alan.hayward@arm.com>
11067 David Sherwood <david.sherwood@arm.com>
11069 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11071 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11072 * optabs.c (expand_unop): Likewise.
11074 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11075 Alan Hayward <alan.hayward@arm.com>
11076 David Sherwood <david.sherwood@arm.com>
11078 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11079 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11080 (popcount_loc_descriptor): Likewise.
11081 (bswap_loc_descriptor): Likewise.
11082 (rotate_loc_descriptor): Likewise.
11083 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11084 calling the functions above.
11086 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11087 Alan Hayward <alan.hayward@arm.com>
11088 David Sherwood <david.sherwood@arm.com>
11090 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11092 (try_combine): Likewise.
11093 (simplify_if_then_else): Likewise.
11094 * cse.c (cse_insn): Likewise.
11095 * dwarf2out.c (mem_loc_descriptor): Likewise.
11096 * emit-rtl.c (gen_lowpart_common): Likewise.
11097 * simplify-rtx.c (simplify_truncation): Likewise.
11098 (simplify_binary_operation_1): Likewise.
11099 (simplify_const_relational_operation): Likewise.
11100 (simplify_ternary_operation): Likewise.
11101 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11103 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11104 Alan Hayward <alan.hayward@arm.com>
11105 David Sherwood <david.sherwood@arm.com>
11107 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11108 * reload.c (push_reload): Likewise.
11109 (find_reloads): Likewise.
11111 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11112 Alan Hayward <alan.hayward@arm.com>
11113 David Sherwood <david.sherwood@arm.com>
11115 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11116 (make_compound_operation_int): Likewise.
11117 (change_zero_ext): Likewise.
11118 * expr.c (convert_move): Likewise.
11119 (convert_modes): Likewise.
11120 * fwprop.c (forward_propagate_subreg): Likewise.
11121 * loop-iv.c (get_biv_step_1): Likewise.
11122 * optabs.c (widen_operand): Likewise.
11123 * postreload.c (move2add_valid_value_p): Likewise.
11124 * recog.c (simplify_while_replacing): Likewise.
11125 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11126 (simplify_binary_operation_1): Likewise. Remove redundant
11127 mode equality check.
11129 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11130 Alan Hayward <alan.hayward@arm.com>
11131 David Sherwood <david.sherwood@arm.com>
11133 * combine.c (combine_simplify_rtx): Add checks for
11134 is_a <scalar_int_mode>.
11135 (simplify_if_then_else): Likewise.
11136 (make_field_assignment): Likewise.
11137 (simplify_comparison): Likewise.
11138 * ifcvt.c (noce_try_bitop): Likewise.
11139 * loop-invariant.c (canonicalize_address_mult): Likewise.
11140 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11142 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11143 Alan Hayward <alan.hayward@arm.com>
11144 David Sherwood <david.sherwood@arm.com>
11146 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11147 is_a <scalar_int_mode> instead of != BLKmode.
11149 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11150 Alan Hayward <alan.hayward@arm.com>
11151 David Sherwood <david.sherwood@arm.com>
11153 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11154 instead of != VOIDmode.
11155 * combine.c (if_then_else_cond): Likewise.
11156 (change_zero_ext): Likewise.
11157 * dwarf2out.c (mem_loc_descriptor): Likewise.
11158 (loc_descriptor): Likewise.
11159 * rtlanal.c (canonicalize_condition): Likewise.
11160 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11162 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11163 Alan Hayward <alan.hayward@arm.com>
11164 David Sherwood <david.sherwood@arm.com>
11166 * simplify-rtx.c (simplify_binary_operation_1): Use
11167 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11169 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11170 Alan Hayward <alan.hayward@arm.com>
11171 David Sherwood <david.sherwood@arm.com>
11173 * wide-int.h (int_traits<unsigned char>) New class.
11174 (int_traits<unsigned short>) Likewise.
11175 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11176 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11178 * combine.c (set_nonzero_bits_and_sign_copies): Use
11179 is_a <scalar_int_mode>.
11180 (find_split_point): Likewise.
11181 (combine_simplify_rtx): Likewise.
11182 (simplify_logical): Likewise.
11183 (expand_compound_operation): Likewise.
11184 (expand_field_assignment): Likewise.
11185 (make_compound_operation): Likewise.
11186 (extended_count): Likewise.
11187 (change_zero_ext): Likewise.
11188 (simplify_comparison): Likewise.
11189 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11190 (ucompare_loc_descriptor): Likewise.
11191 (minmax_loc_descriptor): Likewise.
11192 (mem_loc_descriptor): Likewise.
11193 (loc_descriptor): Likewise.
11194 * expmed.c (init_expmed_one_mode): Likewise.
11195 * lra-constraints.c (lra_constraint_offset): Likewise.
11196 * optabs.c (prepare_libcall_arg): Likewise.
11197 * postreload.c (move2add_note_store): Likewise.
11198 * reload.c (operands_match_p): Likewise.
11199 * rtl.h (load_extend_op): Likewise.
11200 * rtlhooks.c (gen_lowpart_general): Likewise.
11201 * simplify-rtx.c (simplify_truncation): Likewise.
11202 (simplify_unary_operation_1): Likewise.
11203 (simplify_binary_operation_1): Likewise.
11204 (simplify_const_binary_operation): Likewise.
11205 (simplify_const_relational_operation): Likewise.
11206 (simplify_subreg): Likewise.
11207 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11208 * var-tracking.c (adjust_mems): Likewise.
11209 (prepare_call_arguments): Likewise.
11211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11212 Alan Hayward <alan.hayward@arm.com>
11213 David Sherwood <david.sherwood@arm.com>
11215 * machmode.h (is_int_mode): New fuction.
11216 * combine.c (find_split_point): Use it.
11217 (combine_simplify_rtx): Likewise.
11218 (simplify_if_then_else): Likewise.
11219 (simplify_set): Likewise.
11220 (simplify_shift_const_1): Likewise.
11221 (simplify_comparison): Likewise.
11222 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11223 * cse.c (notreg_cost): Likewise.
11224 (cse_insn): Likewise.
11225 * cselib.c (cselib_lookup_1): Likewise.
11226 * dojump.c (do_jump_1): Likewise.
11227 (do_compare_rtx_and_jump): Likewise.
11228 * dse.c (get_call_args): Likewise.
11229 * dwarf2out.c (rtl_for_decl_init): Likewise.
11230 (native_encode_initializer): Likewise.
11231 * expmed.c (emit_store_flag_1): Likewise.
11232 (emit_store_flag): Likewise.
11233 * expr.c (convert_modes): Likewise.
11234 (store_field): Likewise.
11235 (expand_expr_real_1): Likewise.
11236 * fold-const.c (fold_read_from_constant_string): Likewise.
11237 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11238 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11239 * optabs.c (expand_binop): Likewise.
11240 (expand_unop): Likewise.
11241 (expand_abs_nojump): Likewise.
11242 (expand_one_cmpl_abs_nojump): Likewise.
11243 * simplify-rtx.c (mode_signbit_p): Likewise.
11244 (val_signbit_p): Likewise.
11245 (val_signbit_known_set_p): Likewise.
11246 (val_signbit_known_clear_p): Likewise.
11247 (simplify_relational_operation_1): Likewise.
11248 * tree.c (vector_type_mode): Likewise.
11250 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11251 Alan Hayward <alan.hayward@arm.com>
11252 David Sherwood <david.sherwood@arm.com>
11254 * machmode.h (smallest_mode_for_size): Fix formatting.
11255 (smallest_int_mode_for_size): New function.
11256 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11257 instead of smallest_mode_for_size.
11258 * combine.c (make_extraction): Likewise.
11259 * config/arc/arc.c (arc_expand_movmem): Likewise.
11260 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11261 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11262 * config/s390/s390.c (s390_expand_insv): Likewise.
11263 * config/sparc/sparc.c (assign_int_registers): Likewise.
11264 * config/spu/spu.c (spu_function_value): Likewise.
11265 (spu_function_arg): Likewise.
11266 * coverage.c (get_gcov_type): Likewise.
11267 (get_gcov_unsigned_t): Likewise.
11268 * dse.c (find_shift_sequence): Likewise.
11269 * expmed.c (store_bit_field_1): Likewise.
11270 * expr.c (convert_move): Likewise.
11271 (store_field): Likewise.
11272 * internal-fn.c (expand_arith_overflow): Likewise.
11273 * optabs-query.c (get_best_extraction_insn): Likewise.
11274 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11275 * stor-layout.c (layout_type): Likewise.
11276 (initialize_sizetypes): Likewise.
11277 * targhooks.c (default_get_mask_mode): Likewise.
11278 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11280 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11281 Alan Hayward <alan.hayward@arm.com>
11282 David Sherwood <david.sherwood@arm.com>
11284 * machmode.h (opt_mode::else_blk): New function.
11285 (int_mode_for_mode): Declare.
11286 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11287 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11289 * cfgexpand.c (expand_debug_expr): Likewise.
11290 * combine.c (gen_lowpart_or_truncate): Likewise.
11291 (gen_lowpart_for_combine): Likewise.
11292 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11293 * config/avr/avr.c (avr_to_int_mode): Likewise.
11294 (avr_out_plus_1): Likewise.
11295 (avr_out_plus): Likewise.
11296 (avr_out_round): Likewise.
11297 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11298 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11299 (s390_expand_vcond): Likewise.
11300 * config/spu/spu.c (spu_split_immediate): Likewise.
11301 (spu_expand_mov): Likewise.
11302 * dse.c (get_stored_val): Likewise.
11303 * expmed.c (store_bit_field_1): Likewise.
11304 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11306 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11307 (extract_low_bits): Likewise.
11308 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11309 handling rather than repeating the check.
11310 (emit_group_store): Likewise.
11311 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11312 * optabs.c (expand_absneg_bit): Likewise.
11313 (expand_copysign_absneg): Likewise.
11314 (expand_copysign_bit): Likewise.
11315 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11316 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11317 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11318 * var-tracking.c (prepare_call_arguments): Likewise.
11319 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11320 int_mode_for_mode instead of mode_for_size.
11321 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11323 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11324 Alan Hayward <alan.hayward@arm.com>
11325 David Sherwood <david.sherwood@arm.com>
11327 * machmode.h (int_mode_for_size): New function.
11328 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11329 instead of mode_for_size.
11330 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11332 * combine.c (expand_field_assignment): Use int_mode_for_size
11333 instead of mode_for_size.
11334 (make_extraction): Likewise.
11335 (simplify_shift_const_1): Likewise.
11336 (simplify_comparison): Likewise.
11337 * dojump.c (do_jump): Likewise.
11338 * dwarf2out.c (mem_loc_descriptor): Likewise.
11339 * emit-rtl.c (init_derived_machine_modes): Likewise.
11340 * expmed.c (flip_storage_order): Likewise.
11341 (convert_extracted_bit_field): Likewise.
11342 * expr.c (copy_blkmode_from_reg): Likewise.
11343 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11344 * internal-fn.c (expand_mul_overflow): Likewise.
11345 * lower-subreg.c (simple_move): Likewise.
11346 * optabs-libfuncs.c (init_optabs): Likewise.
11347 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11348 * tree.c (vector_type_mode): Likewise.
11349 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11350 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11351 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11352 * tree-vect-stmts.c (vectorizable_load): Likewise.
11353 (vectorizable_store): Likewise.
11355 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11356 Alan Hayward <alan.hayward@arm.com>
11357 David Sherwood <david.sherwood@arm.com>
11359 * coretypes.h (pod_mode): New type.
11360 (scalar_int_mode_pod): New typedef.
11361 * machmode.h (pod_mode): New class.
11362 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11363 * genmodes.c (emit_mode_int_n): Update accordingly.
11364 * lower-subreg.h (target_lower_subreg): Change type to
11365 scalar_int_mode_pod.
11366 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11367 scalar_int_mode_pod.
11369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11370 Alan Hayward <alan.hayward@arm.com>
11371 David Sherwood <david.sherwood@arm.com>
11373 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11374 machine_mode to scalar_int_mode.
11375 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11376 (rs6000_option_override_internal): Remove cast to int.
11377 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11378 machine_mode to scalar_int_mode.
11379 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11380 (rs6000_option_override_internal): Remove cast to int.
11381 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11382 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11384 * config/s390/s390.c (s390_expand_builtin): Likewise.
11385 * coretypes.h (scalar_int_mode): New type.
11386 (opt_scalar_int_mode): New typedef.
11387 * machmode.h (scalar_int_mode): New class.
11388 (scalar_int_mode::includes_p): New function.
11389 (byte_mode): Change type to scalar_int_mode.
11390 (word_mode): Likewise.
11391 (ptr_mode): Likewise.
11392 * emit-rtl.c (byte_mode): Likewise.
11393 (word_mode): Likewise.
11394 (ptr_mode): Likewise.
11395 (init_derived_machine_modes): Update accordingly.
11396 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11397 and MODE_PARTIAL_INT.
11398 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11399 opt_scalar_int_mode.
11401 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11402 Alan Hayward <alan.hayward@arm.com>
11403 David Sherwood <david.sherwood@arm.com>
11405 * target.def (libgcc_floating_mode_supported_p): Take a
11407 * doc/tm.texi: Regenerate.
11408 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11410 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11411 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11414 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11415 Alan Hayward <alan.hayward@arm.com>
11416 David Sherwood <david.sherwood@arm.com>
11418 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11419 * doc/tm.texi: Regenerate.
11420 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11421 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11422 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11423 * targhooks.h (default_floatn_mode): Likewise.
11424 * targhooks.c (default_floatn_mode): Likewise.
11425 * tree.c (build_common_tree_nodes): Update accordingly.
11427 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11428 Alan Hayward <alan.hayward@arm.com>
11429 David Sherwood <david.sherwood@arm.com>
11431 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11432 (mode_iterator::iterate_p): Likewise.
11433 (mode_iterator::get_wider): Likewise.
11434 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11436 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11437 Alan Hayward <alan.hayward@arm.com>
11438 David Sherwood <david.sherwood@arm.com>
11440 * coretypes.h (opt_scalar_float_mode): New typedef.
11441 * machmode.h (float_mode_for_size): New function.
11442 * emit-rtl.c (double_mode): Delete.
11443 (init_emit_once): Use float_mode_for_size.
11444 * stor-layout.c (layout_type): Likewise.
11445 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11447 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11448 Alan Hayward <alan.hayward@arm.com>
11449 David Sherwood <david.sherwood@arm.com>
11451 * output.h (assemble_real): Take a scalar_float_mode.
11452 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11453 * config/arm/arm.md (consttable_4): Likewise.
11454 (consttable_8): Likewise.
11455 (consttable_16): Likewise.
11456 * config/mips/mips.md (consttable_float): Likewise.
11457 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11458 * varasm.c (assemble_real): Take a scalar_float_mode.
11459 (output_constant_pool_2): Update accordingly.
11460 (output_constant): Likewise.
11462 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11463 Alan Hayward <alan.hayward@arm.com>
11464 David Sherwood <david.sherwood@arm.com>
11466 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11467 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11468 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11469 (native_encode_real): Likewise.
11470 (native_interpret_real): Likewise.
11471 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11472 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11474 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11475 Alan Hayward <alan.hayward@arm.com>
11476 David Sherwood <david.sherwood@arm.com>
11478 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11479 <scalar_float_mode>. Simplify.
11480 (gen_extend_conv_libfunc): Likewise.
11482 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11483 Alan Hayward <alan.hayward@arm.com>
11484 David Sherwood <david.sherwood@arm.com>
11486 * coretypes.h (scalar_float_mode): New type.
11487 * machmode.h (mode_traits::from_int): Use machine_mode if
11488 USE_ENUM_MODES is defined.
11489 (is_a): New function.
11491 (dyn_cast): Likewise.
11492 (scalar_float_mode): New class.
11493 (scalar_float_mode::includes_p): New function.
11494 (is_float_mode): Likewise.
11495 * gdbhooks.py (MachineModePrinter): New class.
11496 (build_pretty_printer): Use it for scalar_float_mode.
11497 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11498 (format_helper::format_helper): Turn into a template.
11499 * genmodes.c (get_mode_class): New function.
11500 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11501 or machine_mode if none.
11502 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11503 as_a <scalar_float_mode>.
11504 * dwarf2out.c (mem_loc_descriptor): Likewise.
11505 (insert_float): Likewise.
11506 (add_const_value_attribute): Likewise.
11507 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11508 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11509 (expand_unop): Update accordingly.
11510 (expand_abs_nojump): Likewise.
11511 (expand_copysign_absneg): Take a scalar_float_mode.
11512 (expand_copysign_bit): Likewise.
11513 (expand_copysign): Update accordingly.
11515 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11516 Alan Hayward <alan.hayward@arm.com>
11517 David Sherwood <david.sherwood@arm.com>
11519 * coretypes.h (opt_mode): New class.
11520 * machmode.h (opt_mode): Likewise.
11521 (opt_mode::else_void): New function.
11522 (opt_mode::require): Likewise.
11523 (opt_mode::exists): Likewise.
11524 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11525 (GET_MODE_2XWIDER_MODE): Likewise.
11526 (mode_iterator::get_wider): Update accordingly.
11527 (mode_iterator::get_2xwider): Likewise.
11528 (mode_iterator::get_known_wider): Likewise, turning into a template.
11529 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11530 forcing a wider mode to exist.
11531 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11532 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11533 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11534 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11535 * lower-subreg.c (init_lower_subreg): Likewise.
11536 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11537 on the final iteration.
11538 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11539 a wider mode exists before asking for a move pattern.
11540 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11541 forcing a wider mode to exist.
11542 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11543 returning false if no such mode exists.
11544 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11545 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11546 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11547 Avoid checking for a MODE_INT if we already know the mode is not a
11549 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11550 forcing a wider mode to exist.
11551 (expmed_mult_highpart_optab): Likewise.
11552 (expmed_mult_highpart): Likewise.
11553 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11555 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11556 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11557 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11558 * internal-fn.c (expand_mul_overflow): Update use of
11559 GET_MODE_2XWIDER_MODE.
11560 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11561 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11562 GET_MODE_WIDER_MODE.
11563 (convert_plusminus_to_widen): Likewise.
11564 * tree-switch-conversion.c (array_value_type): Likewise.
11565 * var-tracking.c (emit_note_insn_var_location): Likewise.
11566 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11567 Return false inside rather than outside the loop if no wider mode
11569 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11570 and GET_MODE_2XWIDER_MODE
11571 (can_compare_p): Use else_void.
11572 * gdbhooks.py (OptMachineModePrinter): New class.
11573 (build_pretty_printer): Use it for opt_mode.
11575 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11576 Alan Hayward <alan.hayward@arm.com>
11577 David Sherwood <david.sherwood@arm.com>
11579 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11580 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11582 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11583 Alan Hayward <alan.hayward@arm.com>
11584 David Sherwood <david.sherwood@arm.com>
11586 * machmode.h (mode_traits): New structure.
11587 (get_narrowest_mode): New function.
11588 (mode_iterator::start): Likewise.
11589 (mode_iterator::iterate_p): Likewise.
11590 (mode_iterator::get_wider): Likewise.
11591 (mode_iterator::get_known_wider): Likewise.
11592 (mode_iterator::get_2xwider): Likewise.
11593 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11594 (FOR_EACH_MODE): Likewise.
11595 (FOR_EACH_MODE_FROM): Likewise.
11596 (FOR_EACH_MODE_UNTIL): Likewise.
11597 (FOR_EACH_WIDER_MODE): Likewise.
11598 (FOR_EACH_2XWIDER_MODE): Likewise.
11599 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11600 * combine.c (simplify_comparison): Likewise
11601 * config/i386/i386.c (type_natural_mode): Likewise.
11602 * cse.c (cse_insn): Likewise.
11603 * dse.c (find_shift_sequence): Likewise.
11604 * emit-rtl.c (init_derived_machine_modes): Likewise.
11605 (init_emit_once): Likewise.
11606 * explow.c (hard_function_value): Likewise.
11607 * expmed.c (extract_fixed_bit_field_1): Likewise.
11608 (extract_bit_field_1): Likewise.
11609 (expand_divmod): Likewise.
11610 (emit_store_flag_1): Likewise.
11611 * expr.c (init_expr_target): Likewise.
11612 (convert_move): Likewise.
11613 (alignment_for_piecewise_move): Likewise.
11614 (widest_int_mode_for_size): Likewise.
11615 (emit_block_move_via_movmem): Likewise.
11616 (copy_blkmode_to_reg): Likewise.
11617 (set_storage_via_setmem): Likewise.
11618 (compress_float_constant): Likewise.
11619 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11620 * optabs-query.c (get_best_extraction_insn): Likewise.
11621 * optabs.c (expand_binop): Likewise.
11622 (expand_twoval_unop): Likewise.
11623 (expand_twoval_binop): Likewise.
11624 (widen_leading): Likewise.
11625 (widen_bswap): Likewise.
11626 (expand_parity): Likewise.
11627 (expand_unop): Likewise.
11628 (prepare_cmp_insn): Likewise.
11629 (prepare_float_lib_cmp): Likewise.
11630 (expand_float): Likewise.
11631 (expand_fix): Likewise.
11632 (expand_sfix_optab): Likewise.
11633 * postreload.c (move2add_use_add2_insn): Likewise.
11634 * reg-stack.c (reg_to_stack): Likewise.
11635 * reginfo.c (choose_hard_reg_mode): Likewise.
11636 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11637 * stor-layout.c (mode_for_size): Likewise.
11638 (smallest_mode_for_size): Likewise.
11639 (mode_for_vector): Likewise.
11640 (finish_bitfield_representative): Likewise.
11641 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11642 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11643 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11644 * var-tracking.c (prepare_call_arguments): Likewise.
11646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11647 Alan Hayward <alan.hayward@arm.com>
11648 David Sherwood <david.sherwood@arm.com>
11650 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11651 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11652 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11653 * machmode.h (mode_size): Move earlier in file.
11654 (mode_precision): Likewise.
11655 (mode_inner): Likewise.
11656 (mode_nunits): Likewise.
11657 (mode_unit_size): Likewise.
11658 (unit_unit_precision): Likewise.
11659 (mode_wider): Likewise.
11660 (mode_2xwider): Likewise.
11661 (machine_mode): New class.
11662 (mode_to_bytes): New function.
11663 (mode_to_bits): Likewise.
11664 (mode_to_precision): Likewise.
11665 (mode_to_inner): Likewise.
11666 (mode_to_unit_size): Likewise.
11667 (mode_to_unit_precision): Likewise.
11668 (mode_to_nunits): Likewise.
11669 (GET_MODE_SIZE): Use mode_to_bytes.
11670 (GET_MODE_BITSIZE): Use mode_to_bits.
11671 (GET_MODE_PRECISION): Use mode_to_precision.
11672 (GET_MODE_INNER): Use mode_to_inner.
11673 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11674 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11675 (GET_MODE_NUNITS): Use mode_to_nunits.
11676 * system.h (ALWAYS_INLINE): New macro.
11677 * config/powerpcspe/powerpcspe-c.c
11678 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11679 int for arg1_mode and arg2_mode.
11681 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11682 Alan Hayward <alan.hayward@arm.com>
11683 David Sherwood <david.sherwood@arm.com>
11685 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11686 Prefix mode names with E_ in case statements.
11687 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11688 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11689 (aarch64_split_simd_move): Likewise.
11690 (aarch64_gen_storewb_pair): Likewise.
11691 (aarch64_gen_loadwb_pair): Likewise.
11692 (aarch64_gen_store_pair): Likewise.
11693 (aarch64_gen_load_pair): Likewise.
11694 (aarch64_get_condition_code_1): Likewise.
11695 (aarch64_constant_pool_reload_icode): Likewise.
11696 (get_rsqrte_type): Likewise.
11697 (get_rsqrts_type): Likewise.
11698 (get_recpe_type): Likewise.
11699 (get_recps_type): Likewise.
11700 (aarch64_gimplify_va_arg_expr): Likewise.
11701 (aarch64_simd_container_mode): Likewise.
11702 (aarch64_emit_load_exclusive): Likewise.
11703 (aarch64_emit_store_exclusive): Likewise.
11704 (aarch64_expand_compare_and_swap): Likewise.
11705 (aarch64_gen_atomic_cas): Likewise.
11706 (aarch64_emit_bic): Likewise.
11707 (aarch64_emit_atomic_swap): Likewise.
11708 (aarch64_emit_atomic_load_op): Likewise.
11709 (aarch64_evpc_trn): Likewise.
11710 (aarch64_evpc_uzp): Likewise.
11711 (aarch64_evpc_zip): Likewise.
11712 (aarch64_evpc_ext): Likewise.
11713 (aarch64_evpc_rev): Likewise.
11714 (aarch64_evpc_dup): Likewise.
11715 (aarch64_gen_ccmp_first): Likewise.
11716 (aarch64_gen_ccmp_next): Likewise.
11717 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11718 (alpha_emit_xfloating_libcall): Likewise.
11719 (emit_insxl): Likewise.
11720 (alpha_arg_type): Likewise.
11721 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11722 (arc_preferred_simd_mode): Likewise.
11723 (arc_secondary_reload): Likewise.
11724 (get_arc_condition_code): Likewise.
11725 (arc_print_operand): Likewise.
11726 (arc_legitimate_constant_p): Likewise.
11727 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11728 * config/arc/arc.md (casesi_load): Likewise.
11729 (casesi_compact_jump): Likewise.
11730 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11731 (cc_use_register): Likewise.
11732 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11733 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11734 (arm_init_iwmmxt_builtins): Likewise.
11735 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11736 (neon_expand_vector_init): Likewise.
11737 (arm_attr_length_move_neon): Likewise.
11738 (maybe_get_arm_condition_code): Likewise.
11739 (arm_emit_vector_const): Likewise.
11740 (arm_preferred_simd_mode): Likewise.
11741 (arm_output_iwmmxt_tinsr): Likewise.
11742 (thumb1_output_casesi): Likewise.
11743 (thumb2_output_casesi): Likewise.
11744 (arm_emit_load_exclusive): Likewise.
11745 (arm_emit_store_exclusive): Likewise.
11746 (arm_expand_compare_and_swap): Likewise.
11747 (arm_evpc_neon_vuzp): Likewise.
11748 (arm_evpc_neon_vzip): Likewise.
11749 (arm_evpc_neon_vrev): Likewise.
11750 (arm_evpc_neon_vtrn): Likewise.
11751 (arm_evpc_neon_vext): Likewise.
11752 (arm_validize_comparison): Likewise.
11753 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11754 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11755 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11756 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11757 (c6x_preferred_simd_mode): Likewise.
11758 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11759 (epiphany_rtx_costs): Likewise.
11760 * config/epiphany/predicates.md (proper_comparison_operator):
11762 * config/frv/frv.c (condexec_memory_operand): Likewise.
11763 (frv_emit_move): Likewise.
11764 (output_move_single): Likewise.
11765 (output_condmove_single): Likewise.
11766 (frv_hard_regno_mode_ok): Likewise.
11767 (frv_matching_accg_mode): Likewise.
11768 * config/h8300/h8300.c (split_adds_subs): Likewise.
11769 (h8300_rtx_costs): Likewise.
11770 (h8300_print_operand): Likewise.
11771 (compute_mov_length): Likewise.
11772 (output_logical_op): Likewise.
11773 (compute_logical_op_length): Likewise.
11774 (compute_logical_op_cc): Likewise.
11775 (h8300_shift_needs_scratch_p): Likewise.
11776 (output_a_shift): Likewise.
11777 (compute_a_shift_length): Likewise.
11778 (compute_a_shift_cc): Likewise.
11779 (expand_a_rotate): Likewise.
11780 (output_a_rotate): Likewise.
11781 * config/i386/i386.c (classify_argument): Likewise.
11782 (function_arg_advance_32): Likewise.
11783 (function_arg_32): Likewise.
11784 (function_arg_64): Likewise.
11785 (function_value_64): Likewise.
11786 (ix86_gimplify_va_arg): Likewise.
11787 (ix86_legitimate_constant_p): Likewise.
11788 (put_condition_code): Likewise.
11789 (split_double_mode): Likewise.
11790 (ix86_avx256_split_vector_move_misalign): Likewise.
11791 (ix86_expand_vector_logical_operator): Likewise.
11792 (ix86_split_idivmod): Likewise.
11793 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11794 (ix86_build_const_vector): Likewise.
11795 (ix86_build_signbit_mask): Likewise.
11796 (ix86_match_ccmode): Likewise.
11797 (ix86_cc_modes_compatible): Likewise.
11798 (ix86_expand_branch): Likewise.
11799 (ix86_expand_sse_cmp): Likewise.
11800 (ix86_expand_sse_movcc): Likewise.
11801 (ix86_expand_int_sse_cmp): Likewise.
11802 (ix86_expand_vec_perm_vpermi2): Likewise.
11803 (ix86_expand_vec_perm): Likewise.
11804 (ix86_expand_sse_unpack): Likewise.
11805 (ix86_expand_int_addcc): Likewise.
11806 (ix86_split_to_parts): Likewise.
11807 (ix86_vectorize_builtin_gather): Likewise.
11808 (ix86_vectorize_builtin_scatter): Likewise.
11809 (avx_vpermilp_parallel): Likewise.
11810 (inline_memory_move_cost): Likewise.
11811 (ix86_tieable_integer_mode_p): Likewise.
11812 (x86_maybe_negate_const_int): Likewise.
11813 (ix86_expand_vector_init_duplicate): Likewise.
11814 (ix86_expand_vector_init_one_nonzero): Likewise.
11815 (ix86_expand_vector_init_one_var): Likewise.
11816 (ix86_expand_vector_init_concat): Likewise.
11817 (ix86_expand_vector_init_interleave): Likewise.
11818 (ix86_expand_vector_init_general): Likewise.
11819 (ix86_expand_vector_set): Likewise.
11820 (ix86_expand_vector_extract): Likewise.
11821 (emit_reduc_half): Likewise.
11822 (ix86_emit_i387_round): Likewise.
11823 (ix86_mangle_type): Likewise.
11824 (ix86_expand_round_sse4): Likewise.
11825 (expand_vec_perm_blend): Likewise.
11826 (canonicalize_vector_int_perm): Likewise.
11827 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11828 (expand_vec_perm_1): Likewise.
11829 (expand_vec_perm_interleave3): Likewise.
11830 (expand_vec_perm_even_odd_pack): Likewise.
11831 (expand_vec_perm_even_odd_1): Likewise.
11832 (expand_vec_perm_broadcast_1): Likewise.
11833 (ix86_vectorize_vec_perm_const_ok): Likewise.
11834 (ix86_expand_vecop_qihi): Likewise.
11835 (ix86_expand_mul_widen_hilo): Likewise.
11836 (ix86_expand_sse2_abs): Likewise.
11837 (ix86_expand_pextr): Likewise.
11838 (ix86_expand_pinsr): Likewise.
11839 (ix86_preferred_simd_mode): Likewise.
11840 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11841 * config/i386/sse.md (*andnot<mode>3): Likewise.
11842 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11843 (*<code><mode>3): Likewise.
11844 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11845 (ia64_expand_atomic_op): Likewise.
11846 (ia64_arg_type): Likewise.
11847 (ia64_mode_to_int): Likewise.
11848 (ia64_scalar_mode_supported_p): Likewise.
11849 (ia64_vector_mode_supported_p): Likewise.
11850 (expand_vec_perm_broadcast): Likewise.
11851 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11852 (iq2000_function_arg_advance): Likewise.
11853 (iq2000_function_arg): Likewise.
11854 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11855 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11856 (m68k_libcall_value): Likewise.
11857 (m68k_function_value): Likewise.
11858 (sched_attr_op_type): Likewise.
11859 * config/mcore/mcore.c (mcore_output_move): Likewise.
11860 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11862 (microblaze_function_arg): Likewise.
11863 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11864 (mips_print_operand): Likewise.
11865 (mips_mode_ok_for_mov_fmt_p): Likewise.
11866 (mips_vector_mode_supported_p): Likewise.
11867 (mips_preferred_simd_mode): Likewise.
11868 (mips_expand_vpc_loongson_even_odd): Likewise.
11869 (mips_expand_vec_unpack): Likewise.
11870 (mips_expand_vi_broadcast): Likewise.
11871 (mips_expand_vector_init): Likewise.
11872 (mips_expand_vec_reduc): Likewise.
11873 (mips_expand_msa_cmp): Likewise.
11874 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11875 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11876 (cc_flags_for_mode): Likewise.
11877 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11878 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11879 (nds32_output_casesi_pc_relative): Likewise.
11880 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11881 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11882 (nvptx_gen_unpack): Likewise.
11883 (nvptx_gen_pack): Likewise.
11884 (nvptx_gen_shuffle): Likewise.
11885 (nvptx_gen_wcast): Likewise.
11886 (nvptx_preferred_simd_mode): Likewise.
11887 * config/pa/pa.c (pa_secondary_reload): Likewise.
11888 * config/pa/predicates.md (base14_operand): Likewise.
11889 * config/powerpcspe/powerpcspe-c.c
11890 (altivec_resolve_overloaded_builtin): Likewise.
11891 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11893 (rs6000_preferred_simd_mode): Likewise.
11894 (output_vec_const_move): Likewise.
11895 (rs6000_expand_vector_extract): Likewise.
11896 (rs6000_split_vec_extract_var): Likewise.
11897 (reg_offset_addressing_ok_p): Likewise.
11898 (rs6000_legitimate_offset_address_p): Likewise.
11899 (rs6000_legitimize_address): Likewise.
11900 (rs6000_emit_set_const): Likewise.
11901 (rs6000_const_vec): Likewise.
11902 (rs6000_emit_move): Likewise.
11903 (spe_build_register_parallel): Likewise.
11904 (rs6000_darwin64_record_arg_recurse): Likewise.
11905 (swap_selector_for_mode): Likewise.
11906 (spe_init_builtins): Likewise.
11907 (paired_init_builtins): Likewise.
11908 (altivec_init_builtins): Likewise.
11909 (do_load_for_compare): Likewise.
11910 (rs6000_generate_compare): Likewise.
11911 (rs6000_expand_float128_convert): Likewise.
11912 (emit_load_locked): Likewise.
11913 (emit_store_conditional): Likewise.
11914 (rs6000_output_function_epilogue): Likewise.
11915 (rs6000_handle_altivec_attribute): Likewise.
11916 (rs6000_function_value): Likewise.
11917 (emit_fusion_gpr_load): Likewise.
11918 (emit_fusion_p9_load): Likewise.
11919 (emit_fusion_p9_store): Likewise.
11920 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11921 (fusion_gpr_mem_load): Likewise.
11922 (fusion_addis_mem_combo_load): Likewise.
11923 (fusion_addis_mem_combo_store): Likewise.
11924 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11925 (fusion_gpr_mem_load): Likewise.
11926 (fusion_addis_mem_combo_load): Likewise.
11927 (fusion_addis_mem_combo_store): Likewise.
11928 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11930 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11931 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11932 (rs6000_preferred_simd_mode): Likewise.
11933 (output_vec_const_move): Likewise.
11934 (rs6000_expand_vector_extract): Likewise.
11935 (rs6000_split_vec_extract_var): Likewise.
11936 (reg_offset_addressing_ok_p): Likewise.
11937 (rs6000_legitimate_offset_address_p): Likewise.
11938 (rs6000_legitimize_address): Likewise.
11939 (rs6000_emit_set_const): Likewise.
11940 (rs6000_const_vec): Likewise.
11941 (rs6000_emit_move): Likewise.
11942 (rs6000_darwin64_record_arg_recurse): Likewise.
11943 (swap_selector_for_mode): Likewise.
11944 (paired_init_builtins): Likewise.
11945 (altivec_init_builtins): Likewise.
11946 (rs6000_expand_float128_convert): Likewise.
11947 (emit_load_locked): Likewise.
11948 (emit_store_conditional): Likewise.
11949 (rs6000_output_function_epilogue): Likewise.
11950 (rs6000_handle_altivec_attribute): Likewise.
11951 (rs6000_function_value): Likewise.
11952 (emit_fusion_gpr_load): Likewise.
11953 (emit_fusion_p9_load): Likewise.
11954 (emit_fusion_p9_store): Likewise.
11955 * config/rx/rx.c (rx_gen_move_template): Likewise.
11956 (flags_from_mode): Likewise.
11957 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11958 (s390_slb_comparison): Likewise.
11959 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11960 (s390_vector_mode_supported_p): Likewise.
11961 (s390_cc_modes_compatible): Likewise.
11962 (s390_match_ccmode_set): Likewise.
11963 (s390_canonicalize_comparison): Likewise.
11964 (s390_emit_compare_and_swap): Likewise.
11965 (s390_branch_condition_mask): Likewise.
11966 (s390_rtx_costs): Likewise.
11967 (s390_secondary_reload): Likewise.
11968 (__SECONDARY_RELOAD_CASE): Likewise.
11969 (s390_expand_cs): Likewise.
11970 (s390_preferred_simd_mode): Likewise.
11971 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11972 * config/sh/sh.c (sh_print_operand): Likewise.
11973 (dump_table): Likewise.
11974 (sh_secondary_reload): Likewise.
11975 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11976 * config/sh/sh.md (casesi_worker_1): Likewise.
11977 (casesi_worker_2): Likewise.
11978 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11979 (fcc_comparison_operator): Likewise.
11980 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11981 (emit_soft_tfmode_cvt): Likewise.
11982 (sparc_preferred_simd_mode): Likewise.
11983 (output_cbranch): Likewise.
11984 (sparc_print_operand): Likewise.
11985 (sparc_expand_vec_perm_bmask): Likewise.
11986 (vector_init_bshuffle): Likewise.
11987 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11988 (spu_vector_mode_supported_p): Likewise.
11989 (spu_expand_insv): Likewise.
11990 (spu_emit_branch_or_set): Likewise.
11991 (spu_handle_vector_attribute): Likewise.
11992 (spu_builtin_splats): Likewise.
11993 (spu_builtin_extract): Likewise.
11994 (spu_builtin_promote): Likewise.
11995 (spu_expand_sign_extend): Likewise.
11996 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11997 (tilegx_simd_int): Likewise.
11998 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11999 (tilepro_simd_int): Likewise.
12000 * config/v850/v850.c (const_double_split): Likewise.
12001 (v850_print_operand): Likewise.
12002 (ep_memory_offset): Likewise.
12003 * config/vax/vax.c (vax_rtx_costs): Likewise.
12004 (vax_output_int_move): Likewise.
12005 (vax_output_int_add): Likewise.
12006 (vax_output_int_subtract): Likewise.
12007 * config/visium/predicates.md (visium_branch_operator): Likewise.
12008 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
12009 (visium_print_operand_address): Likewise.
12010 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
12011 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
12012 (xtensa_expand_conditional_branch): Likewise.
12013 (xtensa_copy_incoming_a7): Likewise.
12014 (xtensa_output_literal): Likewise.
12015 * dfp.c (decimal_real_maxval): Likewise.
12016 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
12018 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12019 Alan Hayward <alan.hayward@arm.com>
12020 David Sherwood <david.sherwood@arm.com>
12022 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
12023 (mode_nunits_inline): Likewise.
12024 (mode_inner_inline): Likewise.
12025 (mode_unit_size_inline): Likewise.
12026 (mode_unit_precision_inline): Likewise.
12027 (emit_insn_modes_h): Likewise. Also emit a #define of the
12029 (emit_mode_wider): Add an E_ prefix to mode names.
12030 (emit_mode_complex): Likewise.
12031 (emit_mode_inner): Likewise.
12032 (emit_mode_adjustments): Likewise.
12033 (emit_mode_int_n): Likewise.
12034 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12035 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12036 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12037 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12038 (CRC32_BUILTIN, ENTRY): Likewise.
12039 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12040 (aarch64_pop_regs): Likewise.
12041 (aarch64_process_components): Likewise.
12042 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12043 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12044 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12045 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12046 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12047 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12049 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12050 (pcmpistr): Likewise.
12051 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12052 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12053 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12055 * config/rl78/rl78.c (mduc_regs): Likewise.
12056 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12057 (htm_expand_builtin): Likewise.
12058 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12059 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12060 * config/xtensa/xtensa.c (print_operand): Likewise.
12061 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12062 (NUM_MODE_VECTOR_INT): Likewise.
12063 * genoutput.c (null_operand): Likewise.
12064 (output_operand_data): Likewise.
12065 * genrecog.c (print_parameter_value): Likewise.
12066 * lra.c (debug_operand_data): Likewise.
12068 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12070 * dwarf2out.c (scompare_loc_descriptor_wide)
12071 (scompare_loc_descriptor_narrow): New functions, split out from...
12072 (scompare_loc_descriptor): ...here.
12073 * expmed.c (emit_store_flag_int): New function, split out from...
12074 (emit_store_flag): ...here.
12076 2017-08-30 Richard Biener <rguenther@suse.de>
12078 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12079 (dwarf2out_early_finish): Move setting of AT_pubnames from
12080 early debug output to early finish.
12082 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12084 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12085 and -mdata-region to the assembler.
12087 2017-08-30 Richard Biener <rguenther@suse.de>
12089 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12091 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12092 (dwarf2out_early_global_decl): Only generate a DIE for the
12093 abstract origin if it doesn't already exist or is a declaration DIE.
12094 (resolve_addr): Do not add the linkage name twice when
12095 generating a stub DIE for the DW_TAG_GNU_call_site target.
12097 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12099 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12100 Use machine_mode rather than int for arg1_mode.
12102 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12105 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12106 that the second argument of the built-in functions to unpack
12107 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12108 switch statement instead a lot of if statements.
12109 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12110 Allow 64-bit values to be in Altivec registers as well as
12111 traditional floating point registers.
12112 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12114 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12116 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12117 MAX_REGS_PER_ADDRESS == 1.
12119 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12121 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12122 * config/i386/i386.c (ix86_option_override_internal): Simplify
12123 setting of opts->x_flag_entry.
12125 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12126 Jakub Jelinek <jakub@redhat.com>
12127 Richard Biener <rguenther@suse.de>
12129 PR tree-optimization/81503
12130 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12131 folded constant fits in the target type; reorder tests for clarity.
12133 2017-08-29 Martin Liska <mliska@suse.cz>
12135 * passes.def: Include pass_lower_switch.
12136 * stmt.c (dump_case_nodes): Remove and move to
12137 tree-switch-conversion.
12138 (case_values_threshold): Likewise.
12139 (expand_switch_as_decision_tree_p): Likewise.
12140 (emit_case_decision_tree): Likewise.
12141 (expand_case): Likewise.
12142 (balance_case_nodes): Likewise.
12143 (node_has_low_bound): Likewise.
12144 (node_has_high_bound): Likewise.
12145 (node_is_bounded): Likewise.
12146 (emit_case_nodes): Likewise.
12147 (struct simple_case_node): New struct.
12148 (add_case_node): Remove.
12149 (emit_case_dispatch_table): Use vector instead of case_list.
12150 (reset_out_edges_aux): Remove.
12151 (compute_cases_per_edge): Likewise.
12152 (expand_case): Build list of simple_case_node.
12153 (expand_sjlj_dispatch_table): Use it.
12154 * tree-switch-conversion.c (struct case_node): Moved from
12155 stmt.c and adjusted.
12156 (emit_case_nodes): Likewise.
12157 (node_has_low_bound): Likewise.
12158 (node_has_high_bound): Likewise.
12159 (node_is_bounded): Likewise.
12160 (case_values_threshold): Likewise.
12161 (reset_out_edges_aux): Likewise.
12162 (compute_cases_per_edge): Likewise.
12163 (add_case_node): Likewise.
12164 (dump_case_nodes): Likewise.
12165 (balance_case_nodes): Likewise.
12166 (expand_switch_as_decision_tree_p): Likewise.
12167 (emit_jump): Likewise.
12168 (emit_case_decision_tree): Likewise.
12169 (try_switch_expansion): Likewise.
12170 (do_jump_if_equal): Likewise.
12171 (emit_cmp_and_jump_insns): Likewise.
12172 (fix_phi_operands_for_edge): New function.
12173 (record_phi_operand_mapping): Likewise.
12174 (class pass_lower_switch): New pass.
12175 (pass_lower_switch::execute): New function.
12176 (make_pass_lower_switch): Likewise.
12177 (conditional_probability):
12178 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12179 * tree-pass.h: Add make_pass_lower_switch.
12181 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12184 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12187 2017-08-29 Richard Biener <rguenther@suse.de>
12189 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12190 we do not add a DW_AT_inline attribute twice.
12191 (gen_subprogram_die): Remove code setting DW_AT_inline on
12192 DECL_ABSTRACT_P nodes.
12194 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12196 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12197 calls to internal functions.
12198 (gimplify_modify_expr): Likewise.
12199 * tree-call-cdce.c (use_internal_fn): Likewise.
12200 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12201 (convert_to_divmod): Set the nothrow flag.
12202 * tree-if-conv.c (predicate_mem_writes): Likewise.
12203 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12204 (vectorizable_call): Likewise.
12205 (vectorizable_store): Likewise.
12206 (vectorizable_load): Likewise.
12207 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12208 (vect_recog_mask_conversion_pattern): Likewise.
12210 2017-08-29 Martin Liska <mliska@suse.cz>
12213 * gcc.c (driver_handle_option): Add new argument.
12214 * opts-common.c (handle_option): Pass
12215 target_option_override_hook.
12216 * opts-global.c (lang_handle_option): Add new option.
12217 (set_default_handlers): Add new argument.
12218 (decode_options): Likewise.
12219 * opts.c (target_handle_option): Likewise.
12220 (common_handle_option): Call target_option_override_hook.
12221 * opts.h (struct cl_option_handler_func): Add hook for
12222 target option override.
12223 (struct cl_option_handlers): Likewise.
12224 (set_default_handlers): Add new argument.
12225 (decode_options): Likewise.
12226 (common_handle_option): Likewise.
12227 (target_handle_option): Likewise.
12228 * toplev.c (toplev::main): Pass targetm.target_option.override
12231 2017-08-29 Richard Biener <rguenther@suse.de>
12232 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12234 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12235 life to the active subtree.
12237 2017-08-28 Jeff Law <law@redhat.com>
12239 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12240 derive_equivalences.
12241 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12242 Code moved into....
12243 (edge_info::derive_equivalences): New private member function
12245 * tree-ssa-dom.c (class edge_info): Changed from a struct
12246 to a class. Add ctor/dtor, methods and data members.
12247 (edge_info::edge_info): Renamed from allocate_edge_info.
12248 Initialize additional members.
12249 (edge_info::~edge_info): New.
12250 (free_dom_edge_info): Delete the edge info.
12251 (record_edge_info): Use new class & associated member functions.
12252 Tighten forms for testing for edge equivalences.
12253 (record_temporary_equivalences): Iterate over the simple
12254 equivalences rather than assuming there's only one per edge.
12255 (cprop_into_successor_phis): Iterate over the simple
12256 equivalences rather than assuming there's only one per edge.
12257 (optimize_stmt): Use operand_equal_p rather than pointer
12258 equality for mini-DSE code.
12260 2017-08-28 Nathan Sidwell <nathan@acm.org>
12262 * gcc.c (execute): Fold SIGPIPE handling into switch
12263 statement. Adjust internal error message.
12265 2017-08-28 Richard Biener <rguenther@suse.de>
12268 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12269 Do nothing for removed DIEs.
12271 2017-08-28 Richard Biener <rguenther@suse.de>
12273 PR tree-optimization/81977
12274 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12277 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12280 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12281 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12282 using targetm.gen_mem_thread_fence.
12284 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12287 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12288 predicate to register_operand. Reorder operands.
12289 (*btr<mode>): Ditto.
12290 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12291 (*btr<mode>_mask): Ditto.
12293 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12295 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12296 * config/rs6000/xmmintrin.h: New file.
12297 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12299 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12302 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12303 parameter and_insn and return it.
12304 (recombine_lvx_pattern): Insert a copy to ensure availability of
12305 the base register of the copied masking operation at the point of
12306 the instruction replacement.
12307 (recombine_stvx_pattern): Likewise.
12309 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12311 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12312 undocumented switches.
12313 (-mpower9-dform-vector): Likewise.
12314 (-mpower9-dform): Likewise.
12315 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12316 comments to delete references to -mpower9-dform* switches.
12317 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12318 Delete reference to -mpower9-dform* switches, test for
12319 -mpower9-vector instead.
12320 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12321 (OTHER_P9_VECTOR_MASKS): Likewise.
12322 (POWERPC_MASKS): Likewise.
12323 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12324 tests against -mpower9-dform* to -mpower9-vector. Delete code
12325 that checked for -mpower9-dform* consistancy with other options.
12326 Add test for -mpower9-misc to enable other power9 switches.
12327 (rs6000_init_hard_regno_mode_ok): Likewise.
12328 (rs6000_option_override_internal): Likewise.
12329 (rs6000_emit_prologue): Likewise.
12330 (rs6000_emit_epilogue): Likewise.
12331 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12332 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12333 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12335 (emit_fusion_p9_store): Likewise.
12336 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12337 resetting these macros if the assembler does not support ISA 3.0
12339 (TARGET_P9_DFORM_VECTOR): Likewise.
12340 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12341 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12343 2017-08-25 Alan Modra <amodra@gmail.com>
12345 PR rtl-optimization/81747
12346 * cse.c (cse_extended_basic_block): Don't attempt to record
12347 equivalences for degenerate conditional branches that branch
12348 to their fall-through.
12350 2017-08-24 Martin Sebor <msebor@redhat.com>
12352 PR middle-end/81908
12353 * gimple-fold.c (size_must_be_zero_p): New function.
12354 (gimple_fold_builtin_memory_op): Call it.
12356 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12358 * config/rs6000/mm_malloc.h: New file.
12360 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12362 PR tree-optimization/81913
12363 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12364 analysis when either IVs in condition can wrap.
12366 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12368 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12369 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12371 2017-08-24 Richard Biener <rguenther@suse.de>
12374 * targhooks.c (default_target_can_inline_p): Properly
12375 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12376 is present and always compare.
12377 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12378 infer -mfpmath=sse from TARGET_SSE_P.
12379 (ix86_can_inline_p): Properly use target_option_default_node when
12380 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12382 2017-08-24 Richard Biener <rguenther@suse.de>
12385 * dwarf2out.c (output_die): Handle flag_generate_offload like
12387 (output_comp_unit): Likewise.
12388 (gen_array_type_die): Likewise.
12389 (dwarf2out_early_finish): Likewise.
12390 (note_variable_value_in_expr): Likewise.
12391 (dwarf2out_finish): Likewise. Adjust assert.
12392 * cgraphunit.c (symbol_table::compile): Move setting of
12393 flag_generate_offload earlier ...
12394 (symbol_table::finalize_compilation_unit): ... here, before
12395 early debug finalization.
12397 2017-08-24 Richard Biener <rguenther@suse.de>
12399 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12400 and ipa-fnsummary.h.
12401 (ix86_can_inline_p): When ix86_fpmath flags do not match
12402 check whether the callee uses FP math at all.
12404 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12406 PR middle-end/81931
12407 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12408 instead of TYPE_PRECISION.
12410 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12411 Alan Hayward <alan.hayward@arm.com>
12412 David Sherwood <david.sherwood@arm.com>
12414 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12416 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12418 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12420 (ix86_frame::stack_realign_allocate): New field.
12421 (struct machine_frame_state): Modify comments.
12422 (machine_frame_state::sp_realigned_fp_end): New field.
12423 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12424 layout calculation.
12425 (sp_valid_at): Add assertion to assure no attempt to access invalid
12426 offset of a realigned stack.
12427 (fp_valid_at): Likewise.
12428 (choose_baseaddr): Modify comments.
12429 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12430 ix86_expand_prologue.
12431 (ix86_expand_prologue): Modify stack realignment and allocation.
12432 (ix86_expand_epilogue): Modify comments.
12433 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12434 avx2_runtime, avx512f, and avx512f_runtime.
12436 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12438 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12439 (mstackrealign): Do not init to -1.
12440 * config/i386/i386.c (ix86_option_override_internal):
12441 Check opts_set, not opts when setting default value of
12442 opts->x_ix86_force_align_arg_pointer.
12444 2017-08-23 Richard Biener <rguenther@suse.de>
12446 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12447 lang_hooks.decl_printable_name.
12448 * print-rtl-function.c (print_rtx_function): Likewise.
12449 * tree-pretty-print.c (dump_function_header): Likewise.
12451 2017-08-23 Richard Biener <rguenther@suse.de>
12454 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12455 -g0 at compile-time.
12457 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12459 PR middle-end/19706
12460 * doc/sourcebuild.texi (Other hardware attributes):
12463 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12465 PR middle-end/19706
12466 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12467 Fix single-use check.
12469 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12471 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12473 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12475 * doc/install.texi: Modify to add more details on running selected
12478 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12480 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12481 is combined with -mabi=ms.
12482 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12485 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12487 PR tree-optimization/81488
12488 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12489 and cached_basis fields.
12490 (MAX_SPREAD): New constant.
12491 (alloc_cand_and_find_basis): Initialize new fields.
12492 (clear_visited): New function.
12493 (create_phi_basis_1): Rename from create_phi_basis, set visited
12494 and cached_basis fields.
12495 (create_phi_basis): New wrapper function.
12496 (phi_add_costs_1): Rename from phi_add_costs, add spread
12497 parameter, set visited field, short-circuit when limits reached.
12498 (phi_add_costs): New wrapper function.
12499 (record_phi_increments_1): Rename from record_phi_increments, set
12501 (record_phi_increments): New wrapper function.
12502 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12503 (phi_incr_cost): New wrapper function.
12504 (all_phi_incrs_profitable_1): Rename from
12505 all_phi_incrs_profitable, set visited field.
12506 (all_phi_incrs_profitable): New wrapper function.
12508 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12509 Alan Hayward <alan.hayward@arm.com>
12510 David Sherwood <david.sherwood@arm.com>
12512 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12513 that takes the outer and inner modes.
12514 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12515 comparison as the canonical test for a paradoxical subreg.
12516 * combine.c (simplify_set): Use paradoxical_subreg_p.
12517 (make_extraction): Likewise.
12518 (force_to_mode): Likewise.
12519 (rtx_equal_for_field_assignment_p): Likewise.
12520 (gen_lowpart_for_combine): Likewise.
12521 (simplify_comparison): Likewise.
12522 * cse.c (equiv_constant): Likewise.
12523 * expmed.c (store_bit_field_1): Likewise.
12524 * final.c (alter_subreg): Likewise.
12525 * fwprop.c (propagate_rtx): Likewise.
12526 (forward_propagate_subreg): Likewise.
12527 * ira-conflicts.c (ira_build_conflicts): Likewise.
12528 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12529 * lra-constraints.c (curr_insn_transform): Likewise.
12530 (split_reg): Likewise.
12531 * lra-eliminations.c (move_plus_up): Likewise.
12532 (lra_eliminate_regs_1): Likewise.
12533 * recog.c (general_operand): Likewise.
12534 * ree.c (combine_reaching_defs): Likewise.
12535 * reload.c (push_reload): Likewise.
12536 (find_reloads): Likewise.
12537 * reload1.c (elimination_effects): Likewise.
12538 (compute_reload_subreg_offset): Likewise.
12539 (choose_reload_regs): Likewise.
12540 * rtlanal.c (subreg_lsb_1): Likewise.
12541 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12542 (simplify_subreg): Likewise.
12543 * var-tracking.c (track_loc_p): Likewise.
12544 * emit-rtl.c (byte_lowpart_offset): Likewise.
12545 (paradoxical_subreg_p): Delete out-of-line definition.
12547 2017-08-22 Jeff Law <law@redhat.com>
12549 PR tree-optimization/81741
12550 PR tree-optimization/71947
12551 * tree-ssa-dom.c: Include tree-inline.h.
12552 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12553 equivalences if one is more expensive to compute than the other.
12554 * tree-ssa-scopedtables.h (class const_or_copies): Make
12555 record_const_or_copy_raw method private.
12556 (class avail_exprs_stack): New method simplify_binary_operation.
12557 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12558 avail_exprs_stack::simplify_binary_operation as needed.
12559 (avail_exprs_stack::simplify_binary_operation): New function.
12561 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12563 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12564 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12565 (DOT_SYMBOLS): Likewise.
12566 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12567 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12568 (RS6000_ABI_NAME): Likewise.
12569 (TARGET_CMODEL): Likewise.
12570 (TOC_SECTION_ASM_OP): Likewise.
12571 (SET_CMODEL): New macro.
12572 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12574 2017-08-22 Richard Biener <rguenther@suse.de>
12576 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12577 to allow for free-lang-data replacements similar to verify_type_variant.
12579 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12580 Alan Hayward <alan.hayward@arm.com>
12581 David Sherwood <david.sherwood@arm.com>
12583 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12584 VOIDmode for the LABEL_REF.
12586 2017-08-22 Richard Biener <rguenther@suse.de>
12588 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12590 2017-08-22 Richard Biener <rguenther@suse.de>
12592 * common.opt (feliminate-dwarf2-dups): Ignore.
12593 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12594 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12595 same_die_p_wrap, compute_section_prefix,
12596 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12597 (comdat_symbol_id, comdat_symbol_number): Likewise.
12598 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12600 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12601 (output_die): Mark unreachable path unreachable.
12602 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12603 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12604 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12605 (dwarf2out_early_finish): Likewise.
12607 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12609 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12611 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12614 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12615 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12616 (avr_attribute_table) <io, io_low, address>: Initialize
12617 .decl_required with true.
12619 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12621 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12622 undocumented debugging options.
12623 (-mvsx-scalar-double): Likewise.
12624 (-mallow-df-permute): Likewise.
12625 (-mvectorize-builtins): Likewise.
12626 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12627 (rs6000_builtin_vectorized_function): Likewise.
12628 (rs6000_builtin_md_vectorized_function): Likewise.
12629 (rs6000_opt_vars): Likewise.
12631 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12634 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12635 (*btrq_imm): Rename from *btrq.
12636 (*btcq_imm): Rename from *btcq.
12637 (btsc): New code attribute.
12638 (*<btsc><mode>): New insn pattern.
12639 (*btr<mode>): Ditto.
12640 (*<btsc><mode>_mask): New insn_and_split pattern.
12641 (*btr<mode>_mask): Ditto.
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 * function.c (pad_below): Simplify padding calculation.
12649 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12650 Alan Hayward <alan.hayward@arm.com>
12651 David Sherwood <david.sherwood@arm.com>
12653 * target.def (function_prologue): Remove frame size argument.
12654 (function_epilogue): Likewise.
12655 * doc/tm.texi: Regenerate.
12656 * final.c (final_start_function): Update call to function_prologue.
12657 (final_end_function): Update call to function_epilogue.
12658 (default_function_pro_epilogue): Remove frame size argument.
12659 * output.h (default_function_pro_epilogue): Likewise.
12660 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12661 (arm_output_function_prologue): Likewise.
12662 * config/frv/frv.c (frv_function_prologue): Likewise.
12663 (frv_function_epilogue): Likewise.
12664 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12665 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12666 (ia64_output_function_epilogue): Likewise.
12667 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12668 (m32r_output_function_epilogue): Likewise.
12669 * config/microblaze/microblaze.c (microblaze_function_prologue)
12670 (microblaze_function_epilogue): Likewise.
12671 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12672 (mips_output_function_epilogue): Likewise.
12673 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12674 (mmix_target_asm_function_epilogue): Likewise.
12675 * config/msp430/msp430.c (msp430_start_function): Likewise.
12676 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12677 (nds32_asm_function_epilogue): Likewise.
12678 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12679 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12680 (pa_output_function_epilogue): Likewise.
12681 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12682 (rs6000_output_function_epilogue): Likewise.
12683 * config/rl78/rl78.c (rl78_start_function): Likewise.
12684 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12685 (rs6000_output_function_epilogue): Likewise.
12686 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12687 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12688 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12689 (sparc_asm_function_epilogue): Likewise.
12691 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12693 * tree.h (type_has_mode_precision_p): New function.
12694 * convert.c (convert_to_integer_1): Use it.
12695 * expr.c (expand_expr_real_2): Likewise.
12696 (expand_expr_real_1): Likewise.
12697 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12698 * match.pd: Likewise.
12699 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12700 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12701 * tree-tailcall.c (process_assignment): Likewise.
12702 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12703 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12704 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12705 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12706 (vectorizable_assignment): Likewise.
12707 (vectorizable_shift): Likewise.
12708 (vectorizable_operation): Likewise.
12709 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12711 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12713 * match.pd: Add pow (C, x) simplification.
12715 2017-08-21 Richard Biener <rguenther@suse.de>
12717 PR tree-optimization/81900
12718 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12719 for blocks with abnormal predecessors.
12720 (compute_antic): Do not set visited flag prematurely.
12722 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12725 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12727 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12729 * stor-layout.h (vector_type_mode): Move to...
12730 * tree.h (vector_type_mode): ...here.
12731 * stor-layout.c (vector_type_mode): Move to...
12732 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12734 2017-08-21 Richard Biener <rguenther@suse.de>
12736 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12737 register_external_die hooks.
12738 (debug_false_tree_charstarstar_uhwistar): Declare.
12739 (debug_nothing_tree_charstar_uhwi): Likewise.
12740 * debug.c (do_nothing_debug_hooks): Adjust.
12741 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12742 (debug_nothing_tree_charstar_uhwi): Likewise.
12743 * dbxout.c (dbx_debug_hooks): Adjust.
12744 (xcoff_debug_hooks): Likewise.
12745 * sdbout.c (sdb_debug_hooks): Likewise.
12746 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12747 * dwarf2out.c (macinfo_label_base): New global.
12748 (dwarf2out_register_external_die): New function for the
12749 register_external_die hook.
12750 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12751 (dwarf2_debug_hooks): Use them.
12752 (dwarf2_lineno_debug_hooks): Adjust.
12753 (struct die_struct): Add with_offset flag.
12754 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12755 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12756 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12757 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12758 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12759 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12760 defining section names for the early LTO debug variants.
12761 (reset_indirect_string): New helper.
12762 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12763 (print_dw_val): Add support for offsetted symbol references.
12764 (get_ultimate_context): Split out from is_cxx.
12765 (is_cxx): Use get_ultimate_context.
12766 (is_fortran): Add decl overload.
12767 (compute_comp_unit_symbol): Split out worker from
12768 compute_section_prefix.
12769 (compute_section_prefix): Call compute_comp_unit_symbol and
12770 set comdat_type_p here.
12771 (output_die): Skip DIE symbol output for the LTO added one.
12772 Handle DIE symbol references with offset.
12773 (output_comp_unit): Guard section name mangling properly.
12774 For LTO debug sections emit a symbol at the section beginning
12775 which we use to refer to its DIEs.
12776 (add_abstract_origin_attribute): For DIEs registered via
12777 dwarf2out_register_external_die directly refer to the early
12778 DIE rather than indirectly through the shadow one we created.
12779 Remove obsolete call to dwarf2out_abstract_function for
12780 non-function/block origins.
12781 (gen_array_type_die): When generating early LTO debug do
12782 not emit DW_AT_string_length.
12783 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12784 late when in LTO. As suggested place a gcc_unreachable for
12785 the DECL_ABSTRACT_P case.
12786 (gen_subprogram_die): Avoid another specification DIE
12787 for early built declarations/definitions for the late LTO case.
12788 (gen_variable_die): Add type references for late duplicated VLA dies
12790 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12791 we have the abstract instance already.
12792 (process_scope_var): Adjust decl DIE contexts in LTO which
12793 first puts them in limbo.
12794 (gen_decl_die): Do not generate type DIEs late apart from
12795 types for VLAs or for decls we do not yet have a DIE. Do not
12796 call dwarf2out_abstract_function late.
12797 (dwarf2out_early_global_decl): Make sure to create DIEs
12798 for abstract instances of a decl first.
12799 (dwarf2out_late_global_decl): Adjust comment.
12800 (output_macinfo_op): With multiple macro sections use
12801 macinfo_label_base to distinguish labels.
12802 (output_macinfo): Likewise. Update macinfo_label_base.
12803 Pass in the line info label.
12804 (note_variable_value_in_expr): When generating LTO resolve
12805 all variable values here by generating DIEs as needed.
12806 (init_sections_and_labels): Add early LTO debug flag parameter
12807 and generate different sections and names if set. Add generation
12808 counter for the labels so we can have multiple of them.
12809 (reset_dies): Helper to allow DIEs to be output multiple times.
12810 (dwarf2out_finish): When outputting DIEs to the fat part of an
12811 LTO object first reset DIEs.
12812 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12813 (modified_type_die): Check for decl_ultimate_origin being self
12815 (gen_type_die_with_usage): Likewise.
12816 (gen_typedef_die): Allow decl_ultimate_origin being self.
12817 (set_decl_abstract_flags): Remove.
12818 (set_block_abstract_flags): Likewise.
12819 (dwarf2out_abstract_function): Treat the early generated DIEs
12820 as the abstract copy and only add DW_AT_inline and
12821 DW_AT_artificial here and call set_decl_origin_self.
12822 If the DIE has an abstract origin don't do anything.
12823 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12824 if we have none yet (Go fails to build one, PR78628).
12825 (variably_modified_type_p): Prevent endless recursion for Ada
12826 cyclic pointer types.
12827 * lto-streamer-in.c: Include debug.h.
12828 (dref_queue): New global.
12829 (lto_read_tree_1): Stream in DIE references.
12830 (lto_input_tree): Register DIE references.
12831 (input_function): Stream DECL_DEBUG_ARGS.
12832 * lto-streamer-out.c: Include debug.h.
12833 (lto_write_tree_1): Output DIE references.
12834 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12835 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12836 (output_function): Stream DECL_DEBUG_ARGS.
12837 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12838 Stream DECL_ABSTRACT_ORIGIN.
12839 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12840 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12841 DECL_CONTEXT for file-scope decls.
12842 * lto-streamer.h (struct dref_entry): Declare.
12843 (dref_queue): Likewise.
12844 * cfgexpand.c (pass_expand::execute): Do not call the
12845 outlining_inline_function hook here.
12846 * lto-wrapper.c (debug_obj): New global.
12847 (tool_cleanup): Unlink it if required.
12848 (debug_objcopy): New function.
12849 (run_gcc): Handle early debug sections in the IL files by
12850 extracting them to separate files, partially linkin them and
12851 feeding the result back as result to the linker.
12852 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12853 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12854 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12855 sections into a separate segment.
12856 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12858 (darwin_asm_dwarf_section): Likewise.
12859 (darwin_asm_output_dwarf_offset): Likewise.
12860 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12862 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12863 Alan Hayward <alan.hayward@arm.com>
12864 David Sherwood <david.sherwood@arm.com>
12866 * read-md.h (md_reader::record_potential_iterator_use): Replace
12867 pointer argument with an rtx and an index.
12868 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12869 (apply_mode_iterator): Likewise.
12870 (apply_code_iterator): Likewise.
12871 (apply_int_iterator): Likewise.
12872 (apply_subst_iterator): Likewise.
12873 (record_iterator_use): Likewise.
12874 (record_attribute_use): Likewise.
12875 (md_reader::record_potential_iterator_use): Likewise. Update calls
12876 to record_iterator_use and apply_iterator.
12877 (iterator_use): Replace ptr with x and index.
12878 (attribute_use): Likewise.
12879 (apply_attribute_uses): Update calls to apply_iterator.
12880 (apply_iterators): Likewise. Update initialization of iterator_use.
12881 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12882 and record_potential_iterator_use.
12883 (rtx_reader::read_rtx_operand): Likewise.
12885 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12886 Alan Hayward <alan.hayward@arm.com>
12887 David Sherwood <david.sherwood@arm.com>
12889 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12892 2017-08-21 Richard Biener <rguenther@suse.de>
12894 PR middle-end/81884
12895 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12896 at struct end conservatively when comparing common bases.
12898 2017-08-21 Richard Biener <rguenther@suse.de>
12900 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12901 (mem_ref_in_stmt): Remove.
12902 (determine_max_movement): Use ref index to get at the reference.
12903 (invariantness_dom_walker::before_dom_children): Deal with
12904 lim data already initialized.
12905 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12907 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12909 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12910 (TARGET_ISA_ROUND): Ditto.
12911 (TARGET_ROUND): Ditto.
12912 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12913 * config/i386/i386.md: Ditto.
12914 * config/i386/sse.md: Ditto.
12915 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12916 with OPTION_MASK_ISA_SSE4_1.
12918 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12921 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12922 __builtin_ia32_lzcnt_u16.
12924 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12927 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12928 (rs6000_set_current_function): Rewrite function to use it.
12930 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12933 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12934 and TYPE_WARN_IF_NOT_ALIGN.
12935 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12936 (handle_warn_if_not_align): New.
12937 (place_union_field): Call handle_warn_if_not_align.
12938 (place_field): Call handle_warn_if_not_align.
12939 Copy TYPE_WARN_IF_NOT_ALIGN.
12940 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12941 (layout_type): Likewise.
12942 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12944 (tree_decl_common): Add warn_if_not_align.
12945 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12946 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12947 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12948 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12949 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12950 * doc/extend.texi: Document warn_if_not_aligned attribute.
12951 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12953 2017-08-17 Martin Liska <mliska@suse.cz>
12956 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12957 (get_data_dependence): Use it as pointer type.
12958 (distribute_loop): Likewise.
12960 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12962 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12963 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12964 (p8_vmrgow_<mode>_direct): New define_insn.
12965 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12966 handle endianness for vmrgew and vmrgow permute patterns.
12968 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12970 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12971 * config/rs6000/rs6000-cpus.def: Remove comment.
12972 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12973 (POWERPC_MASKS): Likewise.
12974 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12975 use of TARGET_VSX_TIMODE.
12976 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12977 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12978 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12979 (rs6000_option_override_internal): Remove dead code.
12980 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12981 (rs6000_legitimize_reload_address): Likewise.
12982 (rs6000_legitimate_address_p): Likewise.
12983 (rs6000_opt_masks): Delete "vsx-timode".
12984 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12985 from function comment.
12986 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12987 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12988 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12990 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12991 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12992 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12993 (VSX_TI): Likewise.
12995 (define_peephole2): Likewise.
12997 2017-08-17 Martin Sebor <msebor@redhat.com>
13000 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
13001 past the end of an array.
13002 (test_pp_format): Add test cases.
13004 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
13006 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
13007 missing ECF_NOTHROW flags.
13009 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
13012 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
13013 operands residing in integer registers.
13014 (*vsx_le_perm_load_<mode>): Likewise.
13015 (*vsx_le_perm_store_<mode>): Likewise.
13016 (define_peephole2): Add peepholes to optimize the above.
13018 2017-08-17 Marek Polacek <polacek@redhat.com>
13020 PR middle-end/81814
13021 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
13022 to mimic what shorten_compare did. Change the return type to bool.
13023 (fold_cond_expr_with_comparison): Update call to
13024 operand_equal_for_comparison_p.
13025 (fold_ternary_loc): Likewise.
13027 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13029 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13031 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13033 2017-08-17 Richard Biener <rguenther@suse.de>
13035 * tree-ssa-structalias.c (solve_graph): When propagating
13036 to successors update the graphs succ edges and avoid duplicate work.
13038 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13041 * config/i386/i386.c (ix86_option_override_internal): Save target
13042 specific options after ix86_stack_protector_guard_reg was changed.
13044 2017-08-17 Richard Biener <rguenther@suse.de>
13046 PR tree-optimization/81827
13047 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13048 (new_var_info): Initialize it conservatively.
13049 (get_call_vi): Mark register vars.
13050 (new_scalar_tmp_constraint_exp): Likewise.
13051 (handle_rhs_call): Likewise.
13052 (handle_const_call): Likewise.
13053 (create_function_info_for): Likewise.
13054 (solve_constraints): Sort varinfos to separate register from
13055 non-register vars to pack points-to solution bitmaps during
13058 2017-08-17 Marek Polacek <polacek@redhat.com>
13060 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13062 2017-08-17 Richard Biener <rguenther@suse.de>
13064 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13065 to true when overflow is undefined and we saturated the result.
13067 2017-08-17 Alan Modra <amodra@gmail.com>
13070 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13071 Don't use store multiple if only one reg needs saving.
13072 (interesting_frame_related_regno): New function.
13073 (rs6000_frame_related): Don't emit frame info for regs that
13075 (rs6000_emit_epilogue): Likewise.
13077 2017-08-16 Nathan Sidwell <nathan@acm.org>
13079 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13080 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13081 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13082 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13083 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13084 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13085 (verify_type): Adjust for TYPE_BINFO move.
13086 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13087 process TYPE_BINFO directly.
13088 (hash_tree): Likewise.
13089 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13091 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13094 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13096 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13098 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13101 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13102 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13103 Add ix86_binary_operator_ok to insn constraint.
13104 (*iordi_1_bts): Ditto.
13105 (*xordi_1_btc): Ditto.
13106 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13107 Update corresponding peephole2 pattern.
13111 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13113 PR tree-optimization/81832
13114 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13115 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13117 2017-08-16 Marek Polacek <polacek@redhat.com>
13120 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13121 regarding size_zero_node.
13123 2017-08-16 Martin Liska <mliska@suse.cz>
13126 * config.gcc: Respect previously set extra_objs in case
13129 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13131 PR tree-optimization/81835
13132 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13133 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13134 not depend on the phi.
13136 2017-08-16 Alan Modra <amodra@gmail.com>
13138 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13141 2017-08-16 Alan Modra <amodra@gmail.com>
13143 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13144 (save_reg_p): ..into this. Update all callers.
13145 (first_reg_to_save): Simplify.
13147 2017-08-16 Alan Modra <amodra@gmail.com>
13149 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13152 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13156 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13157 constant count if that count is less than 32.
13159 2017-08-15 Nathan Sidwell <nathan@acm.org>
13161 * gcc.c (execute): Emit friendlier message if inferior is killed
13162 by an external cause.
13164 2017-08-15 Richard Biener <rguenther@suse.de>
13166 PR tree-optimization/81790
13167 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13168 CONSTRUCTORs from simplifying and VN.
13170 2017-08-14 Martin Sebor <msebor@redhat.com>
13172 * builtin-attrs.def: Add comments.
13174 2017-08-14 Martin Sebor <msebor@redhat.com>
13177 * doc/extend.texi (attribute nonstring): Document new attribute.
13179 2017-08-14 Martin Sebor <msebor@redhat.com>
13182 * tree-diagnostic.c (default_tree_printer): Handle %G.
13183 * gimple-pretty-print.h (percent_G_format): Declare new function.
13184 * gimple-pretty-print.c (percent_G_format): Define.
13185 * tree-pretty-print.c (percent_K_format): Add argument.
13187 2017-08-14 Martin Sebor <msebor@redhat.com>
13189 PR translation/79998
13190 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13191 Remove a stray space.
13193 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13196 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13197 (*iordi_1_bts): Ditto.
13198 (*xordi_1_btc): Ditto.
13200 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13203 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13204 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13206 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13207 quoted strings, and make more translator-friendly.
13208 (darwin_rs6000_override_options): Likewise.
13209 (rs6000_option_override_internal): Likewise.
13210 (rs6000_return_in_memory): Fix overlong line.
13211 (init_cmulative_args): Use quoted strings, and make more
13212 translator-friendly.
13213 (rs6000_pass_by_reference): Fix overlong line.
13214 (def_builtin): Use quoted strings.
13215 (altivec_expand_predicate_builtin): Use quoted strings, and make
13216 more translator-friendly.
13217 (htm_expand_builtin): Use quoted strings.
13218 (cpu_expand_builtin): Use quoted strings, and make more
13219 translator-friendly.
13220 (altivec_expand_builtin): Likewise.
13221 (paired_expand_predicate_builtin): Likewise.
13222 (rs6000_invalid_builtin): Likewise.
13223 (builtin_function_type): Use quoted strings.
13224 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13225 more translator-friendly.
13226 (rs6000_trampoline_init): Likewise.
13227 (rs6000_handle_altivec_attribute): Likewise.
13228 (rs6000_inner_target_options): Use quoted strings.
13229 (rs6000_disable_incompatible_switches): Likewise.
13230 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13231 strings, and make more translator-friendly.
13232 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13234 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13236 PR tree-optimization/81799
13237 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13238 cond_expr to simple gimple operand.
13240 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13242 PR middle-end/46932
13243 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13245 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13249 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13250 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13251 TARGET_GASISR_PROLOGUES.
13252 * config/avr/avr.c (avr_option_override): Same.
13253 (avr_pass_pre_proep::execute): Same.
13255 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13258 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13259 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13261 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13263 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13264 define_insn_and_split. Split to a memory load from 0 in
13265 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13266 using PTR mode iterator.
13267 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13268 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13269 (*add_tp_<mode>): Redefine as define_insn_and_split.
13270 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13271 address space. Merge with *add_tp_x32 using PTR mode iterator.
13272 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13273 Split to an add with a memory load from 0 in
13274 DEFAULT_TLS_SEG_REG address space.
13276 2017-08-12 Andrew Pinski <apinski@cavium.com>
13278 * config/aarch64/aarch64-option-extensions.def (rdma):
13279 Fix feature string to what Linux prints out in /proc/cpuinfo.
13281 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13284 * dwarf2out.c (modified_type_die): For C typedef types that have
13285 an ultimate origin, process the ultimate origin instead of the
13287 (gen_typedef_die): Assert that input DECLs have no ultimate
13289 (gen_type_die_with_usage): For typedef variants that have an
13290 ultimate origin, just call gen_decl_die on the original DECL.
13291 (process_scope_var): Avoid creating DIEs for local typedefs and
13292 concrete static variables.
13294 2017-08-12 Alan Modra <amodra@gmail.com>
13298 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13299 match gnu-user.h startfile.
13300 (ENDFILE_LINUX_SPEC): Similarly.
13302 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13305 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13307 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13309 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13310 * config/aarch64/aarch64.md (mov<mode>): Change.
13311 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13312 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13313 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13315 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13317 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13318 for storage order barriers.
13320 2017-08-11 Martin Liska <mliska@suse.cz>
13323 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13324 variables of void type.
13326 2017-08-11 Martin Liska <mliska@suse.cz>
13328 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13329 TARGET_SUPPORTS_ALIASES.
13330 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13331 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13332 (optimize_weakref): Likewise.
13333 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13334 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13335 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13337 2017-08-11 Martin Liska <mliska@suse.cz>
13340 * config/i386/i386.c (make_resolver_func): Do complete
13341 refactoring of the function.
13343 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13346 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13347 * config/i386/i386.c (ix86_stack_protect_guard): Use
13348 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13349 * doc/invoke.texi (x86 Options): Document
13350 -mstack-protector-guard-symbol= option.
13352 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13354 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13355 * config/i386/i386.c (ix86_split_stack_guard): New function.
13356 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13357 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13358 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13359 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13360 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13361 (split_stack_space_check): Call ix86_split_stack_guard.
13363 2017-08-10 Martin Sebor <msebor@redhat.com>
13365 * print-tree.c (print_node): Print location using the established
13367 Replace spaces with colons.
13368 (debug_raw, debug): Ditto.
13370 2017-08-10 Martin Sebor <msebor@redhat.com>
13373 * pretty-print.c (pp_format): Correct the handling of %s precision.
13375 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13378 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13380 (ix86_finalize_stack_frame_flags): This. Also clear
13381 frame_pointer_needed if -fno-omit-frame-pointer is used without
13383 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13384 with ix86_finalize_stack_frame_flags.
13385 (ix86_expand_epilogue): Likewise.
13386 (ix86_expand_split_stack_prologue): Likewise.
13387 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13389 2017-08-10 Martin Liska <mliska@suse.cz>
13392 * c-attribs.c (handle_target_attribute):
13393 Report warning for an empty string argument of target attribute.
13395 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13398 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13400 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13401 or DECL_NONLOCAL labels.
13402 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13403 or DECL_NONLOCAL labels here.
13405 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13407 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13408 to indicate when early gimple folding has been disabled.
13409 (rs6000_gimple_fold_builtin): Add debug content.
13410 (rs6000_invalid_builtin): Fix whitespace.
13411 (rs6000_expand_builtin): Fix whitespace.
13412 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13414 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13417 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13418 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13420 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13422 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13423 (qdf24xx): Likewise.
13424 * config/aarch64/aarch64.md: Include falkor.md.
13425 * config/aarch64/falkor.md: New.
13427 2017-08-09 Marek Polacek <polacek@redhat.com>
13430 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13431 * diagnostic.c (emit_diagnostic): Add a comment.
13432 (emit_diagnostic_valist): New function.
13434 2017-08-09 Marek Polacek <polacek@redhat.com>
13437 * input.c (make_location): New overload.
13438 * input.h (make_location): Declare.
13440 2017-08-08 Alan Modra <amodra@gmail.com>
13441 H.J. Lu <hongjiu.lu@intel.com>
13444 * gcc.c (NO_PIE_SPEC): Delete.
13445 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13447 (LINK_PIE_SPEC): ..to here.
13448 (LINK_COMMAND_SPEC): Support -no-pie.
13449 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13450 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13452 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13453 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13454 (ENDFILE_CRTEND_SPEC): Similarly.
13456 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13459 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13460 (mstack-protector-guard-offset=): Ditto.
13461 * config/i386/i386.c (ix86_option_override): Handle
13462 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13464 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13465 ix86_stack_protect_guard_offset variables.
13466 (TARGET_STACK_PROTECT_GUARD): Always define.
13467 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13468 and -mstack-protector-guard-offset= options.
13470 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13472 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13473 boundary case for the last candidate.
13475 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13477 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13478 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13480 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13482 PR middle-end/19706
13483 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13484 * config/aarch64/aarch64-builtins.c
13485 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13486 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13487 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13489 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13490 Andrew Pinski <pinskia@gmail.com>
13492 PR middle-end/19706
13493 * internal-fn.def (XORSIGN): New.
13494 * optabs.def (xorsign_optab): New.
13495 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13496 (convert_expand_mult_copysign): New.
13497 (pass_optimize_widening_mul::execute): Call
13498 convert_expand_mult_copysign.
13500 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13502 PR tree-optimization/81354
13503 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13504 Insert on edges rather than explicitly creating landing pads.
13505 (analyze_candidates_and_replace): Commit edge inserts.
13507 2017-08-08 Richard Biener <rguenther@suse.de>
13509 PR middle-end/81719
13510 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13511 (expand_simple_operations): Also look through ADDR_EXPRs with
13512 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13514 2017-08-08 Richard Biener <rguenther@suse.de>
13516 PR tree-optimization/81723
13517 * tree-vect-slp.c (struct bst_traits): New hash traits.
13518 (bst_fail): New global.
13519 (vect_build_slp_tree_2): New worker, split out from ...
13520 (vect_build_slp_tree): ... this now wrapping it with using
13521 bst_fail set to cache SLP tree build fails. Properly handle
13523 (vect_analyze_slp_instance): Allocate and free bst_fail.
13525 2017-08-08 Martin Liska <mliska@suse.cz>
13528 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13529 LABEL_DECLs that can be from a different function.
13531 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13533 PR tree-optimization/81744
13534 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13535 loop's number of iterations.
13537 2017-08-08 Martin Liska <mliska@suse.cz>
13539 * asan.c: Include header files.
13540 * attribs.c (build_decl_attribute_variant): New function moved
13542 (build_type_attribute_qual_variant): Likewise.
13543 (cmp_attrib_identifiers): Likewise.
13544 (simple_cst_list_equal): Likewise.
13545 (omp_declare_simd_clauses_equal): Likewise.
13546 (attribute_value_equal): Likewise.
13547 (comp_type_attributes): Likewise.
13548 (build_type_attribute_variant): Likewise.
13549 (lookup_ident_attribute): Likewise.
13550 (remove_attribute): Likewise.
13551 (merge_attributes): Likewise.
13552 (merge_type_attributes): Likewise.
13553 (merge_decl_attributes): Likewise.
13554 (merge_dllimport_decl_attributes): Likewise.
13555 (handle_dll_attribute): Likewise.
13556 (attribute_list_equal): Likewise.
13557 (attribute_list_contained): Likewise.
13558 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13559 (lookup_attribute_by_prefix): Likewise.
13560 * bb-reorder.c: Include header files.
13561 * builtins.c: Likewise.
13562 * calls.c: Likewise.
13563 * cfgexpand.c: Likewise.
13564 * cgraph.c: Likewise.
13565 * cgraphunit.c: Likewise.
13566 * convert.c: Likewise.
13567 * dwarf2out.c: Likewise.
13568 * final.c: Likewise.
13569 * fold-const.c: Likewise.
13570 * function.c: Likewise.
13571 * gimple-expr.c: Likewise.
13572 * gimple-fold.c: Likewise.
13573 * gimple-pretty-print.c: Likewise.
13574 * gimple.c: Likewise.
13575 * gimplify.c: Likewise.
13576 * hsa-common.c: Likewise.
13577 * hsa-gen.c: Likewise.
13578 * internal-fn.c: Likewise.
13579 * ipa-chkp.c: Likewise.
13580 * ipa-cp.c: Likewise.
13581 * ipa-devirt.c: Likewise.
13582 * ipa-fnsummary.c: Likewise.
13583 * ipa-inline.c: Likewise.
13584 * ipa-visibility.c: Likewise.
13586 * lto-cgraph.c: Likewise.
13587 * omp-expand.c: Likewise.
13588 * omp-general.c: Likewise.
13589 * omp-low.c: Likewise.
13590 * omp-offload.c: Likewise.
13591 * omp-simd-clone.c: Likewise.
13592 * opts-global.c: Likewise.
13593 * passes.c: Likewise.
13594 * predict.c: Likewise.
13595 * sancov.c: Likewise.
13596 * sanopt.c: Likewise.
13597 * symtab.c: Likewise.
13598 * toplev.c: Likewise.
13599 * trans-mem.c: Likewise.
13600 * tree-chkp.c: Likewise.
13601 * tree-eh.c: Likewise.
13602 * tree-into-ssa.c: Likewise.
13603 * tree-object-size.c: Likewise.
13604 * tree-parloops.c: Likewise.
13605 * tree-profile.c: Likewise.
13606 * tree-ssa-ccp.c: Likewise.
13607 * tree-ssa-live.c: Likewise.
13608 * tree-ssa-loop.c: Likewise.
13609 * tree-ssa-sccvn.c: Likewise.
13610 * tree-ssa-structalias.c: Likewise.
13611 * tree-ssa.c: Likewise.
13612 * tree-streamer-in.c: Likewise.
13613 * tree-vectorizer.c: Likewise.
13614 * tree-vrp.c: Likewise.
13615 * tsan.c: Likewise.
13616 * ubsan.c: Likewise.
13617 * varasm.c: Likewise.
13618 * varpool.c: Likewise.
13619 * tree.c: Remove functions moved to attribs.[ch].
13620 * tree.h: Likewise.
13621 * config/aarch64/aarch64.c: Add attrs.h header file.
13622 * config/alpha/alpha.c: Likewise.
13623 * config/arc/arc.c: Likewise.
13624 * config/arm/arm.c: Likewise.
13625 * config/avr/avr.c: Likewise.
13626 * config/bfin/bfin.c: Likewise.
13627 * config/c6x/c6x.c: Likewise.
13628 * config/cr16/cr16.c: Likewise.
13629 * config/cris/cris.c: Likewise.
13630 * config/darwin.c: Likewise.
13631 * config/epiphany/epiphany.c: Likewise.
13632 * config/fr30/fr30.c: Likewise.
13633 * config/frv/frv.c: Likewise.
13634 * config/ft32/ft32.c: Likewise.
13635 * config/h8300/h8300.c: Likewise.
13636 * config/i386/winnt.c: Likewise.
13637 * config/ia64/ia64.c: Likewise.
13638 * config/iq2000/iq2000.c: Likewise.
13639 * config/lm32/lm32.c: Likewise.
13640 * config/m32c/m32c.c: Likewise.
13641 * config/m32r/m32r.c: Likewise.
13642 * config/m68k/m68k.c: Likewise.
13643 * config/mcore/mcore.c: Likewise.
13644 * config/microblaze/microblaze.c: Likewise.
13645 * config/mips/mips.c: Likewise.
13646 * config/mmix/mmix.c: Likewise.
13647 * config/mn10300/mn10300.c: Likewise.
13648 * config/moxie/moxie.c: Likewise.
13649 * config/msp430/msp430.c: Likewise.
13650 * config/nds32/nds32-isr.c: Likewise.
13651 * config/nds32/nds32.c: Likewise.
13652 * config/nios2/nios2.c: Likewise.
13653 * config/nvptx/nvptx.c: Likewise.
13654 * config/pa/pa.c: Likewise.
13655 * config/pdp11/pdp11.c: Likewise.
13656 * config/powerpcspe/powerpcspe.c: Likewise.
13657 * config/riscv/riscv.c: Likewise.
13658 * config/rl78/rl78.c: Likewise.
13659 * config/rx/rx.c: Likewise.
13660 * config/s390/s390.c: Likewise.
13661 * config/sh/sh.c: Likewise.
13662 * config/sol2.c: Likewise.
13663 * config/sparc/sparc.c: Likewise.
13664 * config/spu/spu.c: Likewise.
13665 * config/stormy16/stormy16.c: Likewise.
13666 * config/tilegx/tilegx.c: Likewise.
13667 * config/tilepro/tilepro.c: Likewise.
13668 * config/v850/v850.c: Likewise.
13669 * config/vax/vax.c: Likewise.
13670 * config/visium/visium.c: Likewise.
13671 * config/xtensa/xtensa.c: Likewise.
13673 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13676 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13677 constraints since the -mupper-regs-* switches have been
13679 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13680 into a vector from a double word element that was extracted from
13681 another vector, and eliminate extra XXPERMDI instructions.
13682 (vsx_concat_<mode>_2): Likewise.
13683 (vsx_concat_<mode>_3): Likewise.
13684 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13685 concat to allow optimizing inserts from previous extracts.
13687 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13689 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13690 memory reference to a SSP offset in TLS address space.
13691 (ix86_print_operand) <case '@'>: Remove.
13692 (ix86_print_operand_punct_valid_p): Remove '@' code.
13693 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13694 UNSPEC_SP_TLS_TEST.
13695 (stack_tls_protect_set_<mode>): Remove.
13696 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13697 (stack_tls_protect_test_<mode>): Remove.
13698 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13700 2017-08-07 Olivier Hainque <hainque@adacore.com>
13703 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13705 2017-08-07 Douglas Rupp <rupp@adacore.com>
13707 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13708 variable was referenced as multidir in command.
13710 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13713 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13716 2017-08-07 Martin Liska <mliska@suse.cz>
13718 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13719 * config/rl78/rl78.c: Add include of attribs.h.
13720 * config/sh/sh.c: Likewise.
13721 * config/v850/v850.c: Likewise.
13723 2017-08-07 Tom de Vries <tom@codesourcery.com>
13725 PR middle-end/78266
13726 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13728 2017-08-07 Martin Liska <mliska@suse.cz>
13730 * config/mips/mips.c: Include attribs.h.
13732 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13735 * doc/invoke.texi: Document change in behvaior for -Ofast for
13738 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13740 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13742 (aarch64_pop_regs): Likewise.
13743 (aarch64_gen_load_pair): Likewise.
13744 (aarch64_save_callee_saves): Likewise.
13745 (aarch64_restore_callee_saves): Likewise.
13747 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13749 * config/i386/i386.c: Revert the last change.
13751 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13754 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13756 (ix86_finalize_stack_frame_flags): This. Also clear
13757 frame_pointer_needed if -fno-omit-frame-pointer is used without
13759 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13760 with ix86_finalize_stack_frame_flags.
13761 (ix86_expand_epilogue): Likewise.
13762 (ix86_expand_split_stack_prologue): Likewise.
13764 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13767 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13768 to P_AES for Westmere.
13770 2017-08-07 Jonathan Yong <10walls@gmail.com>
13772 * config/i386/mingw.opt (fset-stack-executable): Removed.
13773 * config/i386/cygming.opt (fset-stack-executable): Moved
13775 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13777 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13779 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13781 2017-08-07 Marek Polacek <polacek@redhat.com>
13783 PR middle-end/81737
13784 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13786 2017-08-07 Martin Liska <mliska@suse.cz>
13788 * attribs.h (canonicalize_attr_name): New function.
13789 (cmp_attribs): Move from c-format.c and adjusted.
13790 (is_attribute_p): Moved from tree.h.
13791 * tree-inline.c: Add new includes.
13792 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13793 (private_is_attribute_p): Remove.
13794 (private_lookup_attribute): Likewise.
13795 (private_lookup_attribute_by_prefix): Simplify.
13796 (remove_attribute): Use is_attribute_p.
13797 * tree.h: Remove removed declarations.
13799 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13801 PR middle-end/81698
13802 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13803 instead of computing it in the function. Formatting fix.
13804 (expand_case): Don't rely on default_edge being the first edge,
13805 clear it if removing it, pass default_edge to
13806 emit_case_dispatch_table.
13807 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13810 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13812 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13813 insn in the function, emit NOP after the insn.
13815 2017-08-06 Tom de Vries <tom@codesourcery.com>
13817 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13820 2017-08-06 Tom de Vries <tom@codesourcery.com>
13822 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13825 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13827 PR tree-optimization/57371
13828 * match.pd: New pattern.
13830 2017-08-04 Marek Polacek <polacek@redhat.com>
13832 PR middle-end/81695
13833 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13834 perform the computation in offset_int.
13836 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13838 PR tree-optimization/81136
13839 * tree-vectorizer.h: Include tree-hash-traits.h.
13840 (vec_base_alignments): New typedef.
13841 (vec_info): Add a base_alignments field.
13842 (vect_record_base_alignments): Declare.
13843 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13845 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13846 (create_data_ref): Add an is_conditional_in_stmt argument.
13847 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13848 the is_conditional_in_stmt field.
13849 (data_ref_loc): Add an is_conditional_in_stmt field.
13850 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13851 (find_data_references_in_stmt): Update call to create_data_ref.
13852 (graphite_find_data_references_in_stmt): Likewise.
13853 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13854 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13855 (vect_record_base_alignment): New function.
13856 (vect_record_base_alignments): Likewise.
13857 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13858 for nested statements even if we fail to compute a misalignment.
13859 Use pooled base alignments for unconditional references.
13860 (vect_find_same_alignment_drs): Compare base addresses instead
13862 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13863 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13865 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13867 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13868 Add an explicit name for the enum. Use auto_vec for slp_instances
13869 and grouped_stores.
13870 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13872 (_bb_vec_info): Add a constructor and destructor.
13873 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13874 (destroy_loop_vec_info): Delete.
13875 (vect_destroy_datarefs): Likewise.
13876 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13877 (vec_info::vec_info): New function.
13878 (vec_info::~vec_info): Likewise.
13879 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13880 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13881 destroy_loop_vec_info.
13882 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13883 (_loop_vec_info::_loop_vec_info): ...this.
13884 (destroy_loop_vec_info): Replace with...
13885 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13886 the stmt_vec_infos. Leave handling of vec_info information to its
13887 destructor. Remove explicit vector releases.
13888 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13889 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13890 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13891 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13892 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13893 (destroy_bb_vec_info): Replace with...
13894 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13895 information to its destructor.
13896 (vect_slp_analyze_bb_1): Use new and delete instead of
13897 new_bb_vec_info and destroy_bb_vec_info.
13898 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13901 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13903 * tree-data-ref.h (subscript): Add access_fn field.
13904 (data_dependence_relation): Add could_be_independent_p.
13905 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13906 (same_access_functions): Move to tree-data-ref.c.
13907 * tree-data-ref.c (ref_contains_union_access_p): New function.
13908 (access_fn_component_p): Likewise.
13909 (access_fn_components_comparable_p): Likewise.
13910 (dr_analyze_indices): Add a reference to access_fn_component_p.
13911 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13913 (constant_access_functions): Likewise.
13914 (add_other_self_distances): Likewise.
13915 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13916 (initialize_data_dependence_relation): Use XCNEW and remove
13917 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13918 of access functions that have the same type. Allow the
13919 subsequence to end with different bases in some circumstances.
13920 Record the chosen access functions in SUB_ACCESS_FN.
13921 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13922 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13923 (subscript_dependence_tester_1): Likewise dra and drb.
13924 (build_classic_dist_vector): Update calls accordingly.
13925 (subscript_dependence_tester): Likewise.
13926 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13927 DDR_COULD_BE_INDEPENDENT_P.
13928 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13929 comp_alias_ddrs instead of may_alias_ddrs.
13930 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13932 (vect_analyze_data_ref_dependence): Use it if
13933 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13934 distance vectors if that fails.
13935 (dependence_distance_ge_vf): New function.
13936 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13937 LOOP_VINFO_MAY_ALIAS_DDRS.
13939 2017-08-04 Richard Biener <rguenther@suse.de>
13941 PR middle-end/81705
13942 * fold-const.c (fold_binary_loc): Properly restrict
13943 minus_var0 && minus_var1 case when associating undefined overflow
13946 2017-08-04 Tom de Vries <tom@codesourcery.com>
13948 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13950 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13952 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13953 Don't start diagnostic messages with a capital letter.
13954 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13956 (rs6000_invalid_builtin): Likewise.
13957 (rs6000_trampoline_init): Likewise.
13959 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13962 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13963 after setting changeable df flags.
13965 2017-08-03 Richard Biener <rguenther@suse.de>
13967 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13968 up if the use is in USE - X.
13970 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13972 * toplev.c (dumpfile.h): New include.
13973 (internal_error_reentered): New static function. Use it...
13974 (internal_error_function): ...here to handle reentered internal_error.
13976 2017-08-03 Richard Biener <rguenther@suse.de>
13978 PR middle-end/81148
13979 * fold-const.c (split_tree): Add minus_var and minus_con
13980 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13981 here but always use minus_*.
13982 (associate_trees): Assert we never associate with MINUS_EXPR
13983 and NULL first operand. Do not recurse for PLUS_EXPR operands
13984 when associating as MINUS_EXPR either.
13985 (fold_binary_loc): Track minus_var and minus_con.
13987 2017-08-03 Tom de Vries <tom@codesourcery.com>
13990 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13991 ACCEL_COMPILER, apply finish_options on
13992 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13994 2017-08-03 Tom de Vries <tom@codesourcery.com>
13997 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13998 function_entry_patch_area_size > 0.
14000 2017-08-03 Jakub Jelinek <jakub@redhat.com>
14003 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
14004 instead of 10??LU, perform unit multiplication in wide_int,
14005 don't change alloc_object_size_limit if the limit is larger
14008 PR tree-optimization/81655
14009 PR tree-optimization/81588
14010 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
14011 the case when ranges[i].low and high are 1 for unsigned type with
14014 PR middle-end/81052
14015 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
14016 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
14018 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
14020 * tree-vrp.h: Add include guard.
14022 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
14025 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14026 (ud2): New insn pattern.
14027 * config/i386/i386.c (ix86_expand_epilogue):
14028 For naked functions, generate ud2 instead of trap insn.
14030 2017-08-02 Marek Polacek <polacek@redhat.com>
14033 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14035 2017-08-02 Tom de Vries <tom@codesourcery.com>
14036 Cesar Philippidis <cesar@codesourcery.com>
14038 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14039 Add missing edge probabilities.
14041 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14043 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14044 Correct endianness.
14046 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14048 PR middle-end/79499
14049 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14050 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14051 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14052 prologue_seq sequences - if any.
14054 2017-08-02 Richard Biener <rguenther@suse.de>
14056 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14057 via vectors if supported, integer extracts via punning if supported
14058 or otherwise vector extracts.
14060 2017-08-02 Richard Biener <rguenther@suse.de>
14062 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14064 (bitmap_insert_into_set): ... this.
14066 2017-08-02 Richard Biener <rguenther@suse.de>
14068 PR tree-optimization/81633
14070 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14072 PR tree-optimization/71752
14073 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14075 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14077 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14078 (machine_function::call_ms2sysv_pad_out): Remove field.
14079 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14080 (ix86_compute_frame_layout): Likewise.
14082 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14085 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14086 attribute with interrupt attribute.
14088 2017-08-01 Andrew Pinski <apinski@cavium.com>
14090 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14091 BIT_INSERT_EXPR's operand 1
14092 to see if the types precision matches.
14094 2017-08-01 Martin Liska <mliska@suse.cz>
14096 PR middle-end/70140
14097 * builtins.c (expand_builtin_memcpy_args): Remove.
14098 (expand_builtin_memcpy): Call newly added function
14099 expand_builtin_memory_copy_args.
14100 (expand_builtin_memcpy_with_bounds): Likewise.
14101 (expand_builtin_mempcpy): Remove last argument.
14102 (expand_builtin_mempcpy_with_bounds): Likewise.
14103 (expand_builtin_memory_copy_args): New function created from
14104 expand_builtin_mempcpy_args with small modifications.
14105 (expand_builtin_mempcpy_args): Remove.
14106 (expand_builtin_stpcpy): Remove unused argument.
14107 (expand_builtin): Likewise.
14108 (expand_builtin_with_bounds): Likewise.
14110 2017-08-01 Martin Liska <mliska@suse.cz>
14113 Make mempcpy more optimal (PR middle-end/70140).
14115 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14118 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14119 __builtin_vec_cmpne verify both arguments are compatible vectors
14120 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14121 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14122 move computation of aligned to after checking the argument types.
14126 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14127 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14130 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14131 Doug Rupp <rupp@adacore.com>
14132 Olivier Hainque <hainque@adacore.com>
14134 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14135 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14137 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14138 for TARGET_OS_CPP_BUILTIN.
14139 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14140 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14142 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14143 passing required abi options to the assembler for EABI configurations.
14144 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14145 of .text.hot and .text.unlikely sections for kernel modules when
14146 using ARM style exceptions.
14147 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14148 options. Add EXTRA_CC1_SPEC.
14149 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14151 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14153 (ARM_TARGET2_DWARF_FORMAT): Define.
14154 * config/arm/t-vxworks: Adjust multilib control to removal of the
14155 Diab command line options.
14157 2017-08-01 Martin Liska <mliska@suse.cz>
14159 PR gcov-profile/81561
14160 * gcov.c (unblock): Make unblocking safe as we need to preserve
14161 index correspondence of blocks and block_lists.
14163 2017-08-01 Richard Biener <rguenther@suse.de>
14165 PR tree-optimization/81181
14166 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14167 (compute_antic): ... end of iteration here.
14169 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14171 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14172 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14173 (ftree-slp-vectorize): Likewise.
14174 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14175 can no longer be set independent of flag_tree_loop_vectorize.
14176 * omp-general.c (emp_max_vf): Likewise.
14177 * opts.c (enable_fdo_optimizations): Remove references to
14178 flag_tree_vectorize, these are now implicit.
14179 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14180 and leave it for the options machinery.
14182 2017-08-01 Martin Liska <mliska@suse.cz>
14184 PR middle-end/70140
14185 * builtins.c (expand_builtin_memcpy_args): Remove.
14186 (expand_builtin_memcpy): Call newly added function
14187 expand_builtin_memory_copy_args.
14188 (expand_builtin_memcpy_with_bounds): Likewise.
14189 (expand_builtin_mempcpy): Remove last argument.
14190 (expand_builtin_mempcpy_with_bounds): Likewise.
14191 (expand_builtin_memory_copy_args): New function created from
14192 expand_builtin_mempcpy_args with small modifications.
14193 (expand_builtin_mempcpy_args): Remove.
14194 (expand_builtin_stpcpy): Remove unused argument.
14195 (expand_builtin): Likewise.
14196 (expand_builtin_with_bounds): Likewise.
14198 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14201 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14202 print "ds:" only for immediates in generic address space.
14204 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14207 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14208 (ix86_function_ok_for_sibcall): Return false for naked functions.
14210 2017-08-01 Richard Biener <rguenther@suse.de>
14212 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14213 (compute_antic): Seed worklist with exit block predecessors.
14214 * cfganal.c (dfs_find_deadend): For a cycle return the source
14215 of the edge closing it.
14217 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14219 * config/aarch64/aarch64.c
14220 (aarch64_can_const_movi_rtx_p): Move 0 check.
14222 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14224 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14225 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14227 * match.pd: Ditto in address comparison pattern.
14229 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14231 PR tree-optimization/81627
14232 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14233 closed ssa form for store-store chain.
14235 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14237 PR tree-optimization/81620
14238 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14239 for store-store chain.
14241 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14243 PR tree-optimization/81588
14244 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14245 ranges[i].in_p, invert comparison code ccode. For >/>=,
14246 swap rhs1 and rhs2 and comparison code unconditionally,
14247 for </<= don't do that. Don't swap rhs1/rhs2 again if
14248 ranges[i].in_p, instead invert comparison code ccode if
14249 opcode or oe->rank is BIT_IOR_EXPR.
14252 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14253 a direct optab to conversion optab.
14254 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14255 with GET_MODE_INNER as last argument instead of optab_handler.
14256 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14257 vector extraction if possible and optab is available.
14258 * expr.c (store_constructor): Use convert_optab_handler instead
14259 of optab_handler. Use vector initialization from smaller
14260 vectors if possible and optab is available.
14261 * tree-vect-stmts.c (vectorizable_load): Likewise.
14262 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14263 now have two modes.
14264 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14265 of vec_init from half-sized vectors with the same element mode.
14266 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14267 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14268 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14269 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14270 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14271 after mode in gen_vec_extract* calls.
14272 (vec_extract<mode>): Renamed to ...
14273 (vec_extract<mode><ssescalarmodelower>): ... this.
14274 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14275 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14276 element mode after mode in gen_vec_init* calls.
14277 (VEC_INIT_HALF_MODE): New mode iterator.
14278 (vec_init<mode>): Renamed to ...
14279 (vec_init<mode><ssescalarmodelower>): ... this.
14280 (vec_init<mode><ssehalfvecmodelower>): New expander.
14281 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14282 (vec_extractv2sfsf): ... this.
14283 (vec_initv2sf): Renamed to ...
14284 (vec_initv2sfsf): ... this.
14285 (vec_extractv2si): Renamed to ...
14286 (vec_extractv2sisi): ... this.
14287 (vec_initv2si): Renamed to ...
14288 (vec_initv2sisi): ... this.
14289 (vec_extractv4hi): Renamed to ...
14290 (vec_extractv4hihi): ... this.
14291 (vec_initv4hi): Renamed to ...
14292 (vec_initv4hihi): ... this.
14293 (vec_extractv8qi): Renamed to ...
14294 (vec_extractv8qiqi): ... this.
14295 (vec_initv8qi): Renamed to ...
14296 (vec_initv8qiqi): ... this.
14297 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14298 (vec_init<mode>): Renamed to ...
14299 (vec_init<mode><VEC_base_l>): ... this.
14300 (vec_extract<mode>): Renamed to ...
14301 (vec_extract<mode><VEC_base_l>): ... this.
14302 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14303 (vec_initv2sfsf): ... this.
14304 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14305 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14306 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14307 element mode after mode in gen_vec_init* calls.
14308 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14309 (vec_init<mode><Vel>): ... this.
14310 (vec_extract<mode>): Renamed to ...
14311 (vec_extract<mode><Vel>): ... this.
14312 * config/aarch64/iterators.md (Vel): New mode attribute.
14313 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14314 Add element mode after mode in gen_vec_extract* calls.
14315 * config/s390/vector.md (non_vec_l): New mode attribute.
14316 (vec_extract<mode>): Renamed to ...
14317 (vec_extract<mode><non_vec_l>): ... this.
14318 (vec_init<mode>): Renamed to ...
14319 (vec_init<mode><non_vec_l>): ... this.
14320 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14321 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14323 * config/arm/iterators.md (V_elem_l): New mode attribute.
14324 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14325 (vec_extract<mode><V_elem_l>): ... this.
14326 (vec_extractv2di): Renamed to ...
14327 (vec_extractv2didi): ... this.
14328 (vec_init<mode>): Renamed to ...
14329 (vec_init<mode><V_elem_l>): ... this.
14330 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14331 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14332 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14333 Add element mode after gen_vec_extract* calls.
14334 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14335 (vec_init<mode><unitmode>): ... this.
14336 (vec_extract<mode>): Renamed to ...
14337 (vec_extract<mode><unitmode>): ... this.
14338 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14339 (vec_init<mode><unitmode>): ... this.
14340 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14341 (vec_initv2sfsf): ... this.
14342 (vec_extractv2sf): Renamed to ...
14343 (vec_extractv2sfsf): ... this.
14344 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14345 Add element mode after gen_vec_extract* calls.
14346 * config/mips/mips.md (unitmode): New mode iterator.
14347 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14348 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14349 * config/spu/spu.md (inner_l): New mode attribute.
14350 (vec_init<mode>): Renamed to ...
14351 (vec_init<mode><inner_l>): ... this.
14352 (vec_extract<mode>): Renamed to ...
14353 (vec_extract<mode><inner_l>): ... this.
14354 * config/sparc/sparc.md (veltmode): New mode iterator.
14355 (vec_init<VMALL:mode>): Renamed to ...
14356 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14357 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14358 (vec_initv2sisi): ... this.
14359 (vec_initv2sf): Renamed to ...
14360 (vec_initv2sfsf): ... this.
14361 (vec_extractv2sf): Renamed to ...
14362 (vec_extractv2sfsf): ... this.
14363 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14364 (vec_init<mode>): Renamed to ...
14365 (vec_init<mode><VEC_base_l>): ... this.
14366 (vec_extract<mode>): Renamed to ...
14367 (vec_extract<mode><VEC_base_l>): ... this.
14368 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14369 (vec_initv2sfsf): ... this.
14370 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14371 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14372 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14373 gen_vec_init* calls.
14375 2017-08-01 Richard Biener <rguenther@suse.de>
14377 PR tree-optimization/81297
14378 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14379 TREE_OVERFLOW from INTEGER_CSTs.
14381 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14383 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14385 2017-07-31 Carl Love <cel@us.ibm.com>
14387 * config/rs6000/rs6000-c: Add support for built-in functions
14388 vector signed char vec_xl_be (signed long long, signed char *);
14389 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14390 vector signed int vec_xl_be (signed long long, signed int *);
14391 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14392 vector signed long long vec_xl_be (signed long long, signed long long *);
14393 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14394 vector signed short vec_xl_be (signed long long, signed short *);
14395 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14396 vector double vec_xl_be (signed long long, double *);
14397 vector float vec_xl_be (signed long long, float *);
14398 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14399 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14400 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14402 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14403 (altivec_expand_builtin): Add switch statement to call
14404 altivec_expand_xl_be for each builtin.
14405 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14406 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14407 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14408 __builtin_vsx_le_be_v16qi.
14409 * doc/extend.texi: Update the built-in documentation file for the
14410 new built-in functions.
14412 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14415 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14417 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14419 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14421 * config.gcc: Add z14.
14422 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14423 CPU model numbers for z13s and z14.
14424 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14426 * config/s390/s390-opts.h (enum processor_type): Rename
14427 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14428 * config/s390/s390.c (processor_table): Add field for CPU name to
14429 be passed to Binutils.
14430 (s390_asm_output_machine_for_arch): Use the new field in
14431 processor_table for Binutils.
14432 (s390_expand_builtin): Replace arch12 with z14.
14433 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14434 (s390_get_sched_attrmask): Likewise.
14435 (s390_get_unit_mask): Likewise.
14436 * config/s390/s390.opt: Add z14 to processor_type enum.
14438 2017-07-31 Martin Jambor <mjambor@suse.cz>
14441 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14442 regardless of optimization level.
14444 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14445 Martin Liska <mliska@suse.cz>
14447 * predict.def: Remove old comment and adjust probability.
14448 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14449 PREDICT statements.
14451 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14454 * config/i386/i386.c (ix86_function_naked): New function.
14455 (ix86_can_use_return_insn_p): Return false for naked functions.
14456 (ix86_expand_prologue): Skip prologue for naked functions.
14457 (ix86_expand_epilogue): Skip epilogue for naked functions
14458 and emit trap instruction.
14459 (ix86_warn_func_return): New function.
14460 (ix86_attribute_table): Add "naked" attribute specification.
14461 (TARGET_WARN_FUNC_RETURN): Define.
14462 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14464 2017-07-31 Martin Liska <mliska@suse.cz>
14466 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14467 (dump_gimple_bb_header): Always dump BB info.
14468 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14470 2017-07-31 Martin Liska <mliska@suse.cz>
14473 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14474 also with current_function_decl non-null equality.
14476 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14479 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14480 change type to the element type, instead add eltype variable and
14481 use it where we are interested in the element type.
14483 PR tree-optimization/81603
14484 * ipa-polymorphic-call.c
14485 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14486 offset arithmetic in offset_int, bail out if the resulting bit offset
14487 doesn't fit into shwi.
14489 2017-07-31 Martin Liska <mliska@suse.cz>
14491 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14492 (gimplify_save_expr): Fix comment.
14494 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14497 * config/i386/i386.c (ix86_function_arg): Update arguments for
14499 (ix86_compute_frame_layout): Set the initial stack offset to
14500 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14501 INCOMING_FRAME_SP_OFFSET.
14502 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14503 stack before exception handler returns.
14504 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14505 the 'ERROR_CODE' for exception handler.
14507 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14509 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14510 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14511 (ASM_OUTPUT_REG_POP): Ditto.
14512 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14513 instead of asm_fprintf to output pure string.
14515 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14517 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14518 to imported_module_or_decl hook.
14519 (debug_nothing_tree_tree_tree_bool): Remove.
14520 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14521 * debug.c (do_nothing_debug_hooks): Use
14522 debug_nothing_tree_tree_tree_bool_bool instead of
14523 debug_nothing_tree_tree_tree_bool.
14524 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14525 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14526 * sdbout.c (sdb_debug_hooks): Likewise.
14527 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14528 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14530 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14531 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14532 attribute, don't add anything.
14534 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14536 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14537 (fold_build2_stat_loc): Likewise.
14538 (fold_build3_stat_loc): Likewise.
14539 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14540 (fold_build1_loc): Remove macro.
14541 (fold_build2_loc): Likewise.
14542 (fold_build3_loc): Likewise.
14544 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14546 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14547 (gimple_build_debug_bind_source_stat): Likewise.
14548 * gimple.h (gimple_build_debug_bind): Remove macro.
14549 (gimple_build_debug_bind_source): Likewise.
14551 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14553 * bitmap.c (bitmap_alloc): Adjust.
14554 (bitmap_gc_alloc): Likewise.
14555 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14557 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14559 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14560 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14561 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14562 (bitmap_gc_alloc_stat): Likewise.
14563 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14565 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14567 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14568 * rtl.h (shallow_copy_rtx): Remove macro.
14570 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14572 * emit-rtl.c (gen_raw_REG): Adjust.
14573 * gengenrtl.c (gendef): Likewise.
14574 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14575 * rtl.h (rtx_alloc): Remove macro.
14577 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14579 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14580 (build_tree_list_stat): Likewise.
14581 * tree.h (build_tree_list): Remove macro.
14582 (build_tree_list_vec): Likewise.
14584 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14586 * tree.c (make_vector_stat): Remove _stat from name.
14587 (build_vector_stat): Likewise.
14588 * tree.h (make_vector_stat): Remove macro.
14589 (build_vector_stat): Likewise.
14591 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14593 * tree.h (build_var_debug_value): Remove prototype.
14595 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14597 * tree.c (tree_cons_stat): Remove _stat from name.
14598 * tree.h (tree_cons): Remove macro.
14600 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14602 * tree.c (build_vl_exp_stat): Remove _stat from name.
14603 * tree.h (build_vl_exp): Remove macro.
14605 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14607 * tree.c (build_decl_stat): Remove _stat from name.
14608 * tree.h (build_decl): Remove macro.
14610 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14612 * gimple.c (gimple_build_with_ops_stat): Adjust.
14613 (gimple_alloc_stat): Remove _stat from name.
14614 * gimple.h (gimple_alloc): Remove macro.
14616 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14618 * tree.c (make_tree_vec_stat): Remove _stat from name.
14619 (grow_tree_vec_stat): Likewise.
14620 * tree.h (make_tree_vec_stat): Adjust prototype.
14621 (grow_tree_vec_stat): Likewise.
14622 (make_tree_vec): Remove macro.
14623 (grow_tree_vec): Likewise.
14625 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14627 * fold-const.c (fold_build1_stat_loc): Adjust.
14628 (fold_build2_stat_loc): Likewise.
14629 (fold_build3_stat_loc): Likewise.
14630 * tree.c (build0_stat): Remove _stat from name.
14631 (build1_stat): Likewise.
14632 (build2_stat): Likewise.
14633 (build3_stat): Likewise.
14634 (build4_stat): Likewise.
14635 (build5_stat): Likewise.
14636 * tree.h (build1_loc): Remove macro, and rename _stat function
14638 (build2_loc): Likewise.
14639 (build3_loc): Likewise.
14640 (build4_loc): Likewise.
14641 (build5_loc): Likewise.
14643 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14645 * tree.c (make_int_cst_stat): Remove _stat from name.
14646 * tree.h (make_int_cst_stat): Adjust prototype.
14647 (make_int_cst): Remove macro.
14649 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14651 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14652 * tree.h (make_tree_binfo_stat): Adjust prototype.
14653 (make_tree_binfo): Remove.
14655 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14657 * tree.c (copy_node_stat): Rename to copy_node.
14658 (build_distinct_type_copy): Adjust.
14659 * tree.h (copy_node_stat): Adjust prototype.
14660 (copy_node): Remove macro.
14662 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14664 * tree.c (make_node_stat): rename to make_node.
14665 (build_tree_list_stat): Adjust.
14666 (build0_stat): Likewise.
14667 (build2_stat): Likewise.
14668 (build3_stat): Likewise.
14669 (build4_stat): Likewise.
14670 (build5_stat): Likewise.
14671 (build_decl_stat): Likewise.
14672 * tree.h (make_node_stat): Adjust prototype.
14673 (make_node): remove macro.
14675 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14677 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14678 (PPC_FEATURE2_SCV): Likewise.
14679 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14681 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14683 * config/aarch64/aarch64.c
14684 (aarch64_internal_mov_immediate): Add new special pattern.
14685 * config/aarch64/aarch64.md (*movdi_aarch64):
14686 Add reg/32bit const mov case.
14688 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14689 Richard Sandiford <richard.sandiford@linaro.org>
14691 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14692 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14693 Add integer and movi cases.
14694 (movi-split-hf-df-sf split, fp16): New.
14695 (enabled): Added TARGET_FP_F16INST.
14696 * config/aarch64/iterators.md (GPF_HF): New.
14697 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14699 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14701 * config/aarch64/aarch64.c
14702 (aarch64_simd_container_mode): Add prototype.
14703 (aarch64_expand_mov_immediate): Add HI support.
14704 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14705 (aarch64_can_const_movi_rtx_p): New.
14706 (aarch64_preferred_reload_class):
14707 Remove restrictions of using FP registers for certain SIMD operations.
14708 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14709 (aarch64_valid_floating_const): Add integer move validation.
14710 (aarch64_simd_imm_scalar_p): Remove.
14711 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14712 (aarch64_legitimate_constant_p): Expand list of supported cases.
14713 * config/aarch64/aarch64-protos.h
14714 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14715 (aarch64_reinterpret_float_as_int): New.
14716 (aarch64_simd_imm_scalar_p): Remove.
14717 * config/aarch64/constraints.md (Uvi): New.
14718 (Dd): Split into Ds and new Dd.
14719 * config/aarch64/aarch64.md (*movsi_aarch64):
14721 (*movdi_aarch64): Add SIMD mov case.
14723 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14725 * tree-predcom.c: (struct chain): Handle store-store chain in which
14726 stores for elimination only store loop invariant values.
14727 (execute_pred_commoning_chain): Ditto.
14728 (prepare_initializers_chain_store_elim): Ditto.
14729 (prepare_finalizers): Ditto.
14730 (is_inv_store_elimination_chain): New function.
14731 (initialize_root_vars_store_elim_1): New function.
14733 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14735 * tree-predcom.c: Revise general description of the pass.
14736 (enum chain_type): New enum type for store elimination.
14737 (struct chain): New field supporting store elimination.
14738 (struct component): Ditto.
14739 (dump_chain): Dump store-stores chain.
14740 (release_chain): Release resources.
14741 (split_data_refs_to_components): Compute and create component
14742 contains only stores for elimination.
14743 (get_chain_last_ref_at): New function.
14744 (make_invariant_chain): Initialization.
14745 (make_rooted_chain): Specify chain type in parameter and record it.
14746 (add_looparound_copies): Skip for store-stores chain.
14747 (determine_roots_comp): Compute type of chain and pass it to
14749 (initialize_root_vars_store_elim_2): New function.
14750 (finalize_eliminated_stores): New function.
14751 (remove_stmt): Handle store for elimination.
14752 (execute_pred_commoning_chain): Execute predictive commoning on
14753 store-store chains.
14754 (determine_unroll_factor): Skip unroll for store-stores chain.
14755 (prepare_initializers_chain_store_elim): New function.
14756 (prepare_initializers_chain): Hanlde store-store chain.
14757 (prepare_finalizers_chain, prepare_finalizers): New function.
14758 (tree_predictive_commoning_loop): Return integer value indicating
14759 if loop is unrolled or lcssa form is corrupted.
14760 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14762 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14764 * tree-predcom.c (initialize_root): Delete.
14765 (execute_pred_commoning_chain): Initialize root vars and replace
14766 reference of non-combined chain directly, rather than call above
14769 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14771 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14772 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14774 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14776 * tree-predcom.c (struct chain): New field init_seq.
14777 (release_chain): Release init_seq.
14778 (prepare_initializers_chain): Record intialization stmts in above
14780 (insert_init_seqs): New function.
14781 (tree_predictive_commoning_loop): Call insert_init_seqs.
14783 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14785 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14787 2017-07-28 Richard Biener <rguenther@suse.de>
14789 * match.pd: Remove superfluous :c.
14790 * genmatch.c (simplify::id): Add member.
14791 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14793 (current_id): New global.
14794 (dt_node::parent): Move from ...
14795 (dt_operand::parent): ... here. Add for_id member.
14796 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14797 (decision_tree::find_node): Relax order requirement when
14798 merging DT_TRUE nodes to ones inbetween the current simplify
14799 and the one we try to merge with. Add diagnostic whenever
14800 we need to enforce pattern order by not merging.
14801 (decision_tree::insert): Set current_id.
14802 (decision_tree::print_node): Dump parent node and for_id.
14803 (parser::last_id): Add member.
14804 (parser::push_simplify): Assign unique id.
14805 (parser::parser): Initialize last_id.
14807 2017-07-28 Martin Liska <mliska@suse.cz>
14810 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14811 gimple_build_debug_bind.
14813 2017-07-28 Richard Biener <rguenther@suse.de>
14815 PR tree-optimization/81502
14816 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14818 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14820 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14821 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14822 for BIT_FIELD_REF args.
14823 * fold-const.c (make_bit_field_ref): Likewise.
14824 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14826 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14829 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14830 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14831 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14832 Or it into SANITIZER_UNDEFINED.
14833 * ubsan.c: Include gimple-fold.h and varasm.h.
14834 (ubsan_expand_ptr_ifn): New function.
14835 (instrument_pointer_overflow): New function.
14836 (maybe_instrument_pointer_overflow): New function.
14837 (instrument_object_size): Formatting fix.
14838 (pass_ubsan::execute): Call instrument_pointer_overflow
14839 and maybe_instrument_pointer_overflow.
14840 * internal-fn.c (expand_UBSAN_PTR): New function.
14841 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14842 * sanitizer.def (__ubsan_handle_pointer_overflow,
14843 __ubsan_handle_pointer_overflow_abort): New builtins.
14844 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14845 * internal-fn.def (UBSAN_PTR): New internal function.
14846 * opts.c (sanitizer_opts): Add pointer-overflow.
14847 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14848 * fold-const.c (build_range_check): Compute pointer range check in
14849 integral type if pointer arithmetics would be needed. Formatting
14852 2017-07-28 Martin Liska <mliska@suse.cz>
14855 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14856 parameters that are of a variable-length.
14858 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14860 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14862 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14863 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14864 (CRT_CALL_STATIC_FUNCTION): Likewise.
14865 (ASM_DEFAULT_SPEC): New define.
14866 (ASM_SPEC32): Likewise.
14867 (ASM_SPEC64): Likewise.
14868 (ASM_SPEC_COMMON): Likewise.
14869 (ASM_SPEC): Likewise.
14870 (INVALID_64BIT): Likewise.
14871 (LINK_OS_DEFAULT_SPEC): Likewise.
14872 (LINK_OS_SPEC32): Likewise.
14873 (LINK_OS_SPEC64): Likewise.
14874 (POWERPC_LINUX): Likewise.
14875 (PTRDIFF_TYPE): Likewise.
14876 (RESTORE_FP_PREFIX): Likewise.
14877 (RESTORE_FP_SUFFIX): Likewise.
14878 (SAVE_FP_PREFIX): Likewise.
14879 (SAVE_FP_SUFFIX): Likewise.
14880 (SIZE_TYPE): Likewise.
14881 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14882 (TARGET_64BIT): Likewise.
14883 (TARGET_64BIT): Likewise.
14884 (TARGET_AIX): Likewise.
14885 (WCHAR_TYPE_SIZE): Likewise.
14886 (WCHAR_TYPE): Undefine.
14887 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14888 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14889 (CPP_OS_RTEMS_SPEC): Delete.
14890 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14891 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14893 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14895 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14897 PR tree-optimization/81578
14898 * tree-parloops.c (build_new_reduction): Bail out if
14899 reduction_code isn't one of the standard OpenMP reductions.
14900 Move the details printing after that decision.
14902 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14904 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14905 related to reload_in_progress.
14906 (splat_input_operand): Likewise.
14907 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14909 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14911 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14912 (TARGET_INSTANTIATE_DECLS): Likewise.
14913 (legitimate_indexed_address_p): Delete reload_in_progress code.
14914 (rs6000_debug_legitimate_address_p): Likewise.
14915 (rs6000_eliminate_indexed_memrefs): Likewise.
14916 (rs6000_emit_le_vsx_store): Likewise.
14917 (rs6000_emit_move_si_sf_subreg): Likewise.
14918 (rs6000_emit_move): Likewise.
14919 (register_to_reg_type): Likewise.
14920 (rs6000_pre_atomic_barrier): Likewise.
14921 (rs6000_machopic_legitimize_pic_address): Likewise.
14922 (rs6000_allocate_stack_temp): Likewise.
14923 (rs6000_address_for_fpconvert): Likewise.
14924 (rs6000_address_for_altivec): Likewise.
14925 (rs6000_secondary_memory_needed_rtx): Delete function.
14926 (rs6000_check_sdmode): Likewise.
14927 (rs6000_alloc_sdmode_stack_slot): Likewise.
14928 (rs6000_instantiate_decls): Likewise.
14929 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14930 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14931 Delete reload_in_progress.
14932 (*vec_reload_and_plus_<mptrsize>): Likewise.
14933 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14934 (vsx_div_v2di): Likewise.
14935 (vsx_udiv_v2di): Likewise.
14937 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14939 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14940 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14941 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14942 (rs6000_debug_reg_global): Delete print of LRA status.
14943 (rs6000_option_override_internal): Delete dead LRA related code.
14944 (rs6000_lra_p): Delete function.
14945 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14947 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14949 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14950 * config/riscv/rtems.h: New file.
14952 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14953 Sudakshina Das <sudi.das@arm.com>
14955 * config/aarch64/aarch64.md
14956 (define_split for and<mode>3nr_compare): Move
14957 non aarch64_logical_operand to a register.
14958 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14959 register immediate operand to a register.
14960 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14962 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14964 PR middle-end/81564
14965 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14967 2017-07-27 Richard Biener <rguenther@suse.de>
14969 PR tree-optimization/81573
14970 PR tree-optimization/81494
14971 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14972 multi defuse cycle case.
14974 2017-07-27 Richard Biener <rguenther@suse.de>
14976 PR tree-optimization/81571
14977 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14980 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14982 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14983 earlier and only if MASK_FPU is set. Adjust formatting.
14985 2017-07-27 Martin Liska <mliska@suse.cz>
14987 * opt-functions.awk: Add validation of value of Init.
14988 * optc-gen.awk: Pass new argument.
14990 2017-07-27 Martin Liska <mliska@suse.cz>
14992 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14993 Fix wrong condition.
14995 2017-07-27 Martin Liska <mliska@suse.cz>
14997 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14998 BBs and edges seen by autoFDO.
15000 2017-07-27 Richard Biener <rguenther@suse.de>
15002 PR tree-optimization/81502
15003 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
15004 with incompatible but same sized type.
15005 (execute_update_addresses_taken): Likewise.
15007 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
15009 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
15010 flag_tree_loop_vectorize rather than flag_tree_vectorize.
15012 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15015 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
15016 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
15017 Change s_operand to memory_operand.
15019 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
15021 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
15022 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
15023 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
15024 Emit instructions rather than returning an expression. Handle TFmode
15025 and KFmode by casting to TImode.
15026 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15027 (rs6000_emit_le_vsx_store): Likewise.
15028 * config/rs6000/vsx.md (VSX_TI): New iterator.
15029 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15030 (*vsx_le_undo_permute_<mode>): Likewise.
15031 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15032 emit the split sequence.
15033 (*vsx_le_perm_store_<mode>): Likewise.
15035 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15037 PR tree-optimization/81555
15038 PR tree-optimization/81556
15039 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15040 if true, force CHANGED for the recursive invocation.
15041 (reassociate_bb): Remember original length of ops array, pass
15042 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15044 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15045 attributes for noipa attribute. For naked attribute use
15046 lookup_attribute first before lookup_attribute_spec.
15047 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15049 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15050 for functions with noipa attribute.
15051 (cgraph_externally_visible_p): Return true for functions with noipa
15053 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15054 for functions with noipa attribute.
15055 * doc/extend.texi: Document noipa function attribute.
15056 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15057 also for functions with noipa attribute.
15058 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15060 2017-07-26 Andrew Pinski <apinski@cavium.com>
15062 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15063 vec_unalign_load_cost and vec_unalign_store_cost.
15065 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15067 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15068 -mvsx-small-integer option.
15069 (ISA_3_0_MASKS_IEEE): Likewise.
15070 (OTHER_VSX_VECTOR_MASKS): Likewise.
15071 (POWERPC_MASKS): Likewise.
15072 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15073 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15074 code, only testing for DImode being allowed in non-VSX floating
15076 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15077 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15079 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15080 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15081 TARGET_P8_VECTOR test.
15082 (rs6000_secondary_reload_simple_move): Likewise.
15083 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15084 since TARGET_P9_VECTOR was already tested.
15085 (rs6000_opt_masks): Remove -mvsx-small-integer.
15086 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15087 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15089 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15090 test for TARGET_VEXTRACTUB was used, and that uses
15092 (p9 extract splitter): Likewise.
15093 (vsx_extract_<mode>_di_p9): Likewise.
15094 (vsx_extract_<mode>_store_p9): Likewise.
15095 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15096 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15097 the elimination of TARGET_VSX_SMALL_INTEGER.
15098 (vsx_extract_<mode>_p8): Likewise.
15099 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15100 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15101 (vsx_set_<mode>_p9): Likewise.
15102 (vsx_set_v4sf_p9): Likewise.
15103 (vsx_set_v4sf_p9_zero): Likewise.
15104 (vsx_insert_extract_v4sf_p9): Likewise.
15105 (vsx_insert_extract_v4sf_p9_2): Likewise.
15106 * config/rs6000/rs6000.md (sign extend splitter): Change
15107 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15108 (floatsi<mode>2_lfiwax_mem): Likewise.
15109 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15110 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15111 since a test for TARGET_P9_VECTOR was used.
15112 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15113 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15114 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15115 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15116 TARGET_P8_VECTOR test.
15117 (fix_trunc<mode>si2_stfiwx): Likewise.
15118 (fix_trunc<mode>si2_internal): Likewise.
15119 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15120 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15122 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15123 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15124 TARGET_P8_VECTOR test.
15125 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15126 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15127 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15129 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15130 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15131 since a test for TARGET_P9_VECTOR was used.
15132 (splitter for loading small constants): Likewise.
15134 2017-07-26 Andrew Pinski <apinski@cavium.com>
15136 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15139 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15142 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15143 (fp_valid_at): Likewise.
15145 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15147 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15148 (qdf24xx_addrcost_table): Likewise.
15149 (cortexa57_tunings): Update to use generic_branch_cost.
15150 (cortexa72_tunings): Likewise.
15151 (cortexa73_tunings): Likewise.
15152 (qdf24xx_tunings): Likewise.
15154 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15156 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15157 (thunderx2t99_branch_cost): Likewise.
15158 (cortexa35_tunings): Update to use generic_branch_cost.
15159 (cortexa53_tunings): Likewise.
15160 (cortexa57_tunings): Likewise.
15161 (cortexa72_tunings): Likewise.
15162 (cortexa73_tunings): Likewise.
15163 (thunderx2t99_tunings): Likewise.
15165 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15167 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15168 (sparc_option_override): Honour MASK_FSMULD.
15169 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15170 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15171 * config/sparc/sparc.opt (mfsmuld): New option.
15172 * doc/invoke.texi (mfsmuld): Document option.
15174 2017-07-26 Marek Polacek <polacek@redhat.com>
15176 PR middle-end/70992
15177 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15179 2017-07-26 Richard Biener <rguenther@suse.de>
15181 * gimple-match-head.c (do_valueize): Return OP if valueize
15183 (get_def): New helper to get at the def stmt of a SSA name
15184 if valueize allows.
15185 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15186 do_valueize to get at the def stmt.
15187 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15189 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15191 PR middle-end/46932
15192 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15194 2017-07-26 Martin Liska <mliska@suse.cz>
15197 * function.c (expand_function_start): Make expansion of
15198 nonlocal_goto_save_area after parm_birth_insn.
15200 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15202 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15203 from all CPU target flags enable members.
15205 2017-07-26 Richard Biener <rguenther@suse.de>
15207 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15208 (decision_tree::gen): Make 'type' const.
15209 (write_predicate): Likewise.
15211 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15213 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15214 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15215 (rs6000_option_override_internal): Likewise.
15216 (rs6000_expand_vector_set): Likewise.
15217 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15218 (TARGET_UPPER_REGS_SF): Likewise.
15219 (TARGET_UPPER_REGS_DI): Likewise.
15220 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15221 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15222 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15223 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15224 (Splitters for DI constants in Altivec registers): Likewise.
15225 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15226 (vsx_set_v4sf_p9): Likewise.
15227 (vsx_set_v4sf_p9_zero): Likewise.
15228 (vsx_insert_extract_v4sf_p9): Likewise.
15229 (vsx_insert_extract_v4sf_p9_2): Likewise.
15231 2017-07-25 Carl Love <cel@us.ibm.com>
15233 * doc/extend.texi: Update the built-in documentation file for the
15234 existing built-in functions
15235 vector signed char vec_cnttz (vector signed char);
15236 vector unsigned char vec_cnttz (vector unsigned char);
15237 vector signed short vec_cnttz (vector signed short);
15238 vector unsigned short vec_cnttz (vector unsigned short);
15239 vector signed int vec_cnttz (vector signed int);
15240 vector unsigned int vec_cnttz (vector unsigned int);
15241 vector signed long long vec_cnttz (vector signed long long);
15242 vector unsigned long long vec_cnttz (vector unsigned long long);
15244 2017-07-25 Andrew Pinski <apinski@cavium.com>
15246 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15247 accesses where the use is for the first operand of a BIT_INSERT.
15249 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15252 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15253 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15255 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15257 * config/i386/gstabs.h: Delete.
15258 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15260 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15262 * config/i386/i386.c (ix86_decompose_address): Do not check for
15263 register RTX when looking at index_reg or base_reg.
15264 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15266 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15268 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15269 to update EH info here.
15271 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15273 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15275 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15277 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15279 2017-07-25 Torsten Duwe <duwe@suse.de>
15281 * common.opt: Introduce -fpatchable-function-entry
15282 command line option, and its variables function_entry_patch_area_size
15283 and function_entry_patch_area_start.
15284 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15285 including a two-value parser.
15286 * target.def (print_patchable_function_entry): New target hook.
15287 * targhooks.h (default_print_patchable_function_entry): New function.
15288 * targhooks.c (default_print_patchable_function_entry): Likewise.
15289 * toplev.c (process_options): Switch off IPA-RA if
15290 patchable function entries are being generated.
15291 * varasm.c (assemble_start_function): Look at the
15292 patchable-function-entry command line switch and current
15293 function attributes and maybe generate NOP instructions by
15294 calling the print_patchable_function_entry hook.
15295 * doc/extend.texi: Document patchable_function_entry attribute.
15296 * doc/invoke.texi: Document -fpatchable_function_entry
15297 command line option.
15298 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15300 * doc/tm.texi: Re-generate.
15302 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15305 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15306 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15308 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15310 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15312 2017-07-25 Richard Biener <rguenther@suse.de>
15314 PR tree-optimization/81455
15315 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15316 not walk in cycles when looking for guards.
15318 2017-07-25 Richard Biener <rguenther@suse.de>
15320 PR tree-optimization/81529
15321 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15322 when optimizing backedge uses.
15324 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15326 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15328 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15329 to dl_section_ref. On AIX, append an expression to subtract
15330 the size of the section length to dl_section_ref.
15332 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15334 * configure.ac: If any of the config.* scripts fail, exit 1.
15335 * configure: Regenerate.
15337 2017-07-25 Richard Biener <rguenther@suse.de>
15339 PR middle-end/81546
15340 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15341 of immediate uses, be more verbose on errors.
15343 2017-07-25 Richard Biener <rguenther@suse.de>
15345 PR tree-optimization/81510
15346 * tree-vect-loop.c (vect_is_simple_reduction): When the
15347 reduction stmt is not inside the loop bail out.
15349 2017-07-25 Richard Biener <rguenther@suse.de>
15351 PR tree-optimization/81303
15352 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15353 profitability check against LOOP_VINFO_NITERSM1.
15355 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15357 * domwalk.c (cmp_bb_postorder): Simplify.
15358 (sort_bbs_postorder): New function. Use it...
15359 (dom_walker::walk): ...here to optimize common cases.
15361 2017-07-25 Martin Liska <mliska@suse.cz>
15364 * ipa-visibility.c (function_and_variable_visibility): Make the
15365 redirection just on target that supports aliasing.
15366 Fix GNU coding style.
15368 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15371 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15373 * config/arm/rtems.h: Likewise.
15374 * config/bfin/rtems.h: Likewise.
15375 * config/i386/rtemself.h: Likewise.
15376 * config/lm32/rtems.h: Likewise.
15377 * config/m32c/rtems.h: Likewise.
15378 * config/m68k/rtemself.h: Likewise.
15379 * config/microblaze/rtems.h: Likewise.
15380 * config/mips/rtems.h: Likewise.
15381 * config/moxie/rtems.h: Likewise.
15382 * config/nios2/rtems.h: Likewise.
15383 * config/powerpcspe/rtems.h: Likewise.
15384 * config/rs6000/rtems.h: Likewise.
15385 * config/rtems.h: Likewise.
15386 * config/sh/rtems.h: Likewise.
15387 * config/sh/rtemself.h: Likewise.
15388 * config/sparc/rtemself.h: Likewise.
15390 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15393 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15394 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15395 * tree-ssa-structalias.c (alias_get_name): Same.
15397 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15400 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15401 instructions if no du chain is found.
15403 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15405 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15407 2017-07-25 Richard Biener <rguenther@suse.de>
15409 PR middle-end/81505
15410 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15413 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15415 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15416 upper-regs options.
15417 (ISA_2_7_MASKS_SERVER): Likewise.
15418 (ISA_3_0_MASKS_IEEE): Likewise.
15419 (OTHER_P8_VECTOR_MASKS): Likewise.
15420 (OTHER_VSX_VECTOR_MASKS): Likewise.
15421 (POWERPC_MASKS): Likewise.
15422 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15423 duplicate list of options.
15424 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15425 explicit -mupper-regs options.
15426 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15427 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15428 alias for -mupper-regs-df.
15429 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15430 (rs6000_init_hard_regno_mode_ok): Likewise.
15431 (rs6000_option_override_internal): Likewise.
15432 (rs6000_opt_masks): Likewise.
15433 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15434 options in terms of whether -mvsx or -mpower8-vector was used.
15435 (TARGET_UPPER_REGS_DI): Likewise.
15436 (TARGET_UPPER_REGS_SF): Likewise.
15437 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15438 -mupper-regs-* options.
15440 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15442 * passes.c (emergency_dump_function): Print some empty lines and a
15443 header before the RTL dump.
15445 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15447 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15449 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15452 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15453 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15455 2017-07-24 Carl Love <cel@us.ibm.com>
15457 * config/rs6000/rs6000-c.c: Add support for built-in functions
15458 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15459 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15460 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15461 vec_extract_fp_from_shortl): Add defines for the two builtins.
15462 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15463 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15465 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15466 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15467 * doc/extend.texi: Update the built-in documentation file for the
15468 new built-in function.
15470 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15473 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15475 * doc/generic.texi: Likewise.
15476 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15477 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15479 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15481 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15482 (aarch64_mls_elt_merge<mode>): Likewise.
15484 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15486 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15487 having __cxa_atexit.
15489 2017-07-23 Michael Collison <michael.collison@arm.com>
15491 * config/arm/arm.c (arm_option_override): Deprecate
15492 use of -mstructure-size-boundary.
15493 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15494 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15496 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15499 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15500 Reduce cost estimate for direct moves.
15502 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15505 * config/i386/i386.c (ix86_option_override_internal): Disable
15506 BMI, BMI2 and TBM instructions for -m16.
15508 2017-07-21 Carl Love <cel@us.ibm.com>
15510 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15511 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15512 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15513 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15514 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15515 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15516 VMULOSW): New enum "unspec" values.
15517 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15518 altivec_vmulosw): New patterns.
15519 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15520 VMULOSW): Add definitions.
15522 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15524 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15525 (qdf24xx): Likewise.
15526 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15527 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15528 (AARCH64_FL_V8_1): Renumber.
15529 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15530 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15531 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15532 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15533 rdma to feature modifiers list.
15535 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15537 PR middle-end/56727
15538 * ipa-visibility (function_and_variable_visibility): Convert
15539 recursive PLT call to direct call if appropriate.
15541 2017-07-21 Andrew Pinski <apinski@cavium.com>
15543 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15544 operand 1 to see if the types precision matches.
15545 * fold-const.c (operand_equal_p): Likewise.
15547 2017-07-21 Richard Biener <rguenther@suse.de>
15549 PR tree-optimization/81303
15550 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15551 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15552 (vect_peeling_hash_get_lowest_cost): Adjust.
15553 (vect_enhance_data_refs_alignment): Likewise. Use
15554 vect_get_peeling_costs_all_drs to compute the penalty for no
15555 peeling to match up costs.
15557 2017-07-21 Richard Biener <rguenther@suse.de>
15559 PR tree-optimization/81500
15560 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15561 we didn't identify a reduction path.
15563 2017-07-21 Tom de Vries <tom@codesourcery.com>
15564 Cesar Philippidis <cesar@codesourcery.com>
15566 PR gcov-profile/81442
15567 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15570 2017-07-21 Tom de Vries <tom@codesourcery.com>
15573 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15575 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15576 nvptx_override_options_after_change.
15578 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15580 * dwarf2out.c (output_file_names): Avoid double testing for
15581 dwarf_version >= 5.
15583 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15585 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15587 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15589 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15591 (try_crossjump_to_edge): Do not punt on partitioned functions.
15593 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15595 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15596 Put all BBs reachable only via paths crossing cold region to cold
15598 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15600 2016-07-21 Richard Biener <rguenther@suse.de>
15602 PR tree-optimization/81303
15603 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15604 into account prologue and epilogue iterations when raising
15605 min_profitable_iters to sth at least covering one vector iteration.
15607 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15609 * config/arm/arm.c (arm_test_cpu_arch_dat):
15612 2017-07-20 Nathan Sidwell <nathan@acm.org>
15614 Remove TYPE_METHODS.
15615 * tree.h (TYPE_METHODS): Delete.
15616 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15617 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15618 (dbxout_type_methods): Scan TYPE_FIELDS.
15619 (dbxout_type): Don't check TYPE_METHODS here.
15620 * function.c (use_register_for_decl): Always ignore register for
15621 class types when not optimizing.
15622 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15623 * tree.c (free_lang_data_in_type): Stitch out member functions and
15624 templates from TYPE_FIELDS.
15625 (build_distinct_type_copy, verify_type_variant,
15626 verify_type): Member fns are on TYPE_FIELDS.
15627 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15628 * tree-pretty-print.c (dump_generic_node): Likewise.
15630 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15633 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15634 V2TImode and V4TImode.
15635 (ix86_expand_vector_extract): Likewise.
15636 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15637 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15638 (ssescalarmode): Handle V4TImode and V2TImode.
15639 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15640 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15641 (VEXTRACTI128_MODE): New mode iterator.
15642 (splitter for *vec_extractv?ti first element): New.
15643 (VEC_INIT_MODE): New mode iterator.
15644 (vec_init<mode>): Consolidate 3 expanders into one using
15645 VEC_INIT_MODE mode iterator.
15647 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15649 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15650 non_spilled_static_chain_regno_p.
15652 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15654 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15656 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15658 * bb-reorder.c (connect_traces): Allow copying of blocks within
15661 2017-07-20 Richard Biener <rguenther@suse.de>
15663 * gimple.h (gimple_phi_result): Add gphi * overload.
15664 (gimple_phi_result_ptr): Likewise.
15665 (gimple_phi_arg): Likewise. Adjust index assert to only
15666 allow actual argument accesses rather than all slots available
15668 (gimple_phi_arg_def): Add gphi * overload.
15669 * tree-phinodes.c (make_phi_node): Initialize only actual
15671 (resize_phi_node): Clear memory not covered by old node,
15672 do not initialize excess argument slots.
15673 (reserve_phi_args_for_new_edge): Initialize new argument slot
15676 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15678 PR tree-optimization/81388
15680 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15682 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15683 by removing computation of may_be_zero.
15685 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15686 Tom de Vries <tom@codesourcery.com>
15688 PR middle-end/81030
15689 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15690 when gimple level profile disagrees with what RTL expander did.
15692 2017-07-20 Richard Biener <rguenther@suse.de>
15694 PR tree-optimization/61171
15695 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15696 (vect_analyze_stmt): Add slp instance parameter.
15697 (vectorizable_reduction): Likewise.
15698 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15699 (vect_is_simple_reduction): Deal with chains not detected
15700 as SLP reduction chain, specifically not properly associated
15701 chains containing a mix of plus/minus.
15702 (get_reduction_op): Remove.
15703 (get_initial_defs_for_reduction): Simplify, pass in whether
15704 this is a reduction chain, pass in the SLP node for the PHIs.
15705 (vect_create_epilog_for_reduction): Get the SLP instance as
15707 (vectorizable_reduction): Get the SLP instance as arg.
15708 During analysis remember the SLP node with the PHIs in the
15709 instance. Simplify getting at the vectorized reduction PHIs.
15710 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15711 through SLP instance.
15712 (vect_slp_analyze_operations): Likewise.
15713 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15714 (vect_transform_stmt): Likewise.
15716 2017-07-20 Tom de Vries <tom@codesourcery.com>
15718 PR tree-optimization/81489
15719 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15720 read of phi arg location to before loop that modifies phi.
15722 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15724 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15727 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15729 PR middle-end/81331
15730 * except.c (execute): Fix ordering issue.
15732 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15734 PR rtl-optimization/81423
15735 * combine.c (make_compound_operation_int): Don't try to optimize
15736 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15738 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15740 PR rtl-optimization/81423
15741 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15742 with a constant that is -1 in the truncated to mode.
15744 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15746 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15747 (determine_unlikely_bbs): ... here.
15748 * predict.h (propagate_unlikely_bbs_forward): Declare.
15749 * cfgexpand.c (pass_expand::execute): Use it.
15750 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15752 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15753 propagate_unlikely_bbs_forward.
15755 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15757 PR middle-end/81331
15758 * except.c (maybe_add_nop_after_section_switch): New function.
15761 2017-07-19 Tom de Vries <tom@codesourcery.com>
15763 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15765 2017-07-19 Tom de Vries <tom@codesourcery.com>
15767 * gimple.h (gimple_phi_arg): Make assert more strict.
15769 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15771 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15772 * config/rs6000/mmintrin.h: New file.
15773 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15775 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15777 PR tree-optimization/81346
15778 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15780 2017-07-19 Tom de Vries <tom@codesourcery.com>
15782 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15784 2017-07-19 Tom de Vries <tom@codesourcery.com>
15786 * config/nvptx/nvptx-modes.def: Add V2DImode.
15787 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15788 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15789 (nvptx_output_mov_insn): Handle lack of mov.b128.
15790 (nvptx_print_operand): Handle 'H' and 'L' codes.
15791 (nvptx_vector_mode_supported): Allow V2DImode.
15792 (nvptx_preferred_simd_mode): New function.
15793 (nvptx_data_alignment): New function.
15794 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15795 nvptx_preferred_simd_mode.
15796 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15798 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15800 2017-07-19 Tom de Vries <tom@codesourcery.com>
15802 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15803 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15804 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15805 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15806 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15807 (mov<VECIM>_insn): New define_insn.
15808 (define_expand "mov<VECIM>): New define_expand.
15810 2017-07-19 Tom de Vries <tom@codesourcery.com>
15812 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15814 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15816 PR tree-optimization/81346
15817 * fold-const.h (fold_div_compare, range_check_type): Declare.
15818 * fold-const.c (range_check_type): New function.
15819 (build_range_check): Use range_check_type.
15820 (fold_div_compare): No longer static, rewritten into
15821 a match.pd helper function.
15822 (fold_comparison): Don't call fold_div_compare here.
15823 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15824 as helper function.
15826 2017-07-19 Nathan Sidwell <nathan@acm.org>
15828 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15829 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15830 * tree.c (find_decls_types_r, verify_type): Use
15831 TYPE_{MIN,MAX}_VALUE_RAW.
15832 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15833 (hash_tree): Likewise.
15834 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15836 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15839 2017-07-18 Tom de Vries <tom@codesourcery.com>
15841 PR middle-end/81464
15842 * omp-expand.c (expand_omp_for_static_chunk): Handle
15843 equal-argument loop exit phi.
15845 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15848 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15849 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15850 operand 2 predicate.
15851 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15852 operand 2 predicate.
15853 (ror,rol -> rorx splitters): Use const_int_operand as
15854 operand 2 predicate.
15856 2017-06-18 Richard Biener <rguenther@suse.de>
15858 PR tree-optimization/81410
15859 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15860 the gap in the ! slp_perm SLP case after each group.
15862 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15864 PR middle-end/81463
15865 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15868 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15870 PR middle-end/81462
15871 * predict.c (set_even_probabilities): Cleanup; do not affect
15872 probabilities that are already known.
15873 (combine_predictions_for_bb): Call even when count is set.
15875 2017-07-18 Nathan Sidwell <nathan@acm.org>
15877 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15880 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15883 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15884 optimization for loop niter analysis.
15886 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15889 * config/avr/avr.c (avr_optimize_casesi): Don't use
15890 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15892 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15894 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15895 body_cost_vec from _vect_peel_extended_info.
15896 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15897 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15900 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15902 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15904 2017-07-18 Richard Biener <rguenther@suse.de>
15906 PR tree-optimization/80620
15907 PR tree-optimization/81403
15908 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15909 info when re-using a VN table entry.
15911 2017-07-18 Richard Biener <rguenther@suse.de>
15913 PR tree-optimization/81418
15914 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15915 vectype_in. Verify that with lane-reducing reduction operations
15916 we have a single def-use cycle.
15918 2017-07-17 Carl Love <cel@us.ibm.com>
15920 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15922 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15923 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15924 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15925 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15926 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15927 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15928 VMULOSW): New enum "unspec" values.
15929 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15930 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15931 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15932 altivec_vmulosw): New patterns.
15933 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15934 VMULOSW): Add definitions.
15935 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15937 * config/alpha/alpha.c: Include predict.h.
15939 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15941 * tree-vrp.c (compare_assert_loc): Fix comparison function
15942 to return predictable results.
15944 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15946 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15948 (subdi3): Likewise.
15949 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15950 * doc/invoke.texi (mexpand-adddi): Update text.
15952 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15954 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15955 that also clobbers the CC register. The old expand code is moved
15957 (*arc_clzsi2): ... here.
15958 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15959 the CC register. The old expand code is moved to ...
15960 (arc_ctzsi2): ... here.
15962 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15964 * config/arc/arc.opt (mindexed-loads): Use initial value
15965 TARGET_INDEXED_LOADS_DEFAULT.
15966 (mauto-modify-reg): Use initial value
15967 TARGET_AUTO_MODIFY_REG_DEFAULT.
15968 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15969 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15970 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15971 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15973 2017-07-17 Martin Liska <mliska@suse.cz>
15976 * opts.c (finish_options): Do not allow -fgnu-tm
15977 w/ -fsanitize={kernel-,}address. Say sorry.
15979 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15982 * tree-loop-distribution.c (classify_partition): Only assert on
15983 numer of iterations.
15984 (merge_dep_scc_partitions): Delete prameter. Update function call.
15985 (distribute_loop): Remove code handling loop with unknown niters.
15986 (pass_loop_distribution::execute): Skip loop with unknown niters.
15988 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15991 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15992 function sort_partitions_by_post_order.
15994 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15996 PR tree-optimization/81374
15997 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15998 the max index of basic blocks, rather than number of basic blocks.
16000 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16002 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
16004 (arc_legitimate_pic_operand_p): Likewise.
16005 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
16007 (arc_needs_pcl_p): Likewise.
16008 (arc_legitimate_pc_offset_p): Likewise.
16009 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
16010 function is also used in constrains.md.
16011 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
16012 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
16013 PLUS. Only return true/false in known cases, otherwise assert.
16014 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
16015 is already called in arc_legitimate_constant_p.
16016 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
16018 (LEGITIMATE_PIC_OPERAND_P): Use
16019 arc_raw_symbolic_reference_mentioned_p function.
16020 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
16025 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16026 Andrew Burgess <andrew.burgess@embecosm.com>
16028 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16029 (arc_return_address_register): New function.
16030 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16031 (arc_handle_fndecl_attribute): Add naked attribute.
16032 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16033 (TARGET_WARN_FUNC_RETURN): Likewise.
16034 (arc_allocate_stack_slots_for_args): New function.
16035 (arc_warn_func_return): Likewise.
16036 (machine_function): Change type fn_type.
16037 (arc_compute_function_type): Consider new naked function type,
16038 change function return type.
16039 (arc_must_save_register): Adapt to handle new
16040 arc_compute_function_type's return type.
16041 (arc_expand_prologue): Likewise.
16042 (arc_expand_epilogue): Likewise.
16043 (arc_return_address_regs): Delete.
16044 (arc_return_address_register): New function.
16045 (arc_epilogue_uses): Use above function.
16046 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16047 (arc_function_type): Change encoding, add naked type.
16048 (ARC_INTERRUPT_P): Change to handle the new encoding.
16049 (ARC_FAST_INTERRUPT_P): Likewise.
16050 (ARC_NORMAL_P): Define.
16051 (ARC_NAKED_P): Likewise.
16052 (arc_compute_function_type): Delete prototype.
16053 * config/arc/arc.md (in_ret_delay_slot): Use
16054 arc_return_address_register function.
16055 (simple_return): Likewise.
16056 (p_return_i): Likewise.
16058 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16060 PR tree-optimization/81428
16061 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16062 can't be built for those types.
16064 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16066 Remove stuff dead since r239246.
16068 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16069 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16070 (avr_inform_devices): Remove dead stuff.
16072 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16074 * config/arm/arm_neon.h: Fix softp typo.
16076 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16078 PR tree-optimization/81365
16079 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16080 aggregate moves onto bb predecessor edges, make sure there are no
16081 loads that could alias the lhs in between the start of bb and the
16084 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16087 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16088 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16089 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16091 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16093 PR tree-optimization/81396
16094 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16095 (init_symbolic_number): Initialize it to 1.
16096 (perform_symbolic_merge): Add n_ops from both operands into the new
16098 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16099 without base_addr as useless if they need more than one operation.
16100 (bswap_replace): Handle !bswap case for NULL base_addr.
16102 2017-07-17 Tom de Vries <tom@codesourcery.com>
16105 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16108 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16110 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16111 conditional builtin define __FIX_LEON3FT_B2BST.
16113 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16115 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16116 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16119 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16121 PR rtl-optimization/81424
16122 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16123 to remove potential trapping from operands if -fnon-call-exceptions.
16125 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16127 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16128 profile_proability for scalling.
16129 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16131 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16133 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16135 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16137 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16138 fixpoint arithmetics.
16140 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16142 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16143 fixpoint arithmetics.
16145 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16147 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16148 fixpoint arithmetics.
16150 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16152 * profile-count.h (profile_probability::from_reg_br_prob_note,
16153 profile_probability::to_reg_br_prob_note): New functions.
16154 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16155 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16156 * predict.c (probability_reliable_p): Update.
16157 (edge_probability_reliable_p): Update.
16158 (br_prob_note_reliable_p): Update.
16159 (invert_br_probabilities): Update.
16160 (add_reg_br_prob_note): New function.
16161 (combine_predictions_for_insn): Update.
16162 * asan.c (asan_clear_shadow): Update.
16163 * cfgbuild.c (compute_outgoing_frequencies): Update.
16164 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16165 (update_br_prob_note): Update.
16166 (rtl_verify_edges): Update.
16167 (purge_dead_edges): Update.
16168 (fixup_reorder_chain): Update.
16169 * emit-rtl.c (try_split): Update.
16170 * ifcvt.c (cond_exec_process_insns): Update.
16171 (cond_exec_process_if_block): Update.
16172 (dead_or_predicable): Update.
16173 * internal-fn.c (expand_addsub_overflow): Update.
16174 (expand_neg_overflow): Update.
16175 (expand_mul_overflow): Update.
16176 * loop-doloop.c (doloop_modify): Update.
16177 * loop-unroll.c (compare_and_jump_seq): Update.
16178 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16179 * predict.h: Update.
16180 * reorg.c (mostly_true_jump): Update.
16182 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16183 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16184 * config/arc/arc.c: (emit_unlikely_jump): Update.
16185 * config/arm/arm.c: (emit_unlikely_jump): Update.
16186 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16187 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16188 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16189 (ix86_print_operand): Update.
16190 (ix86_split_fp_branch): Update.
16191 (predict_jump): Update.
16192 * config/ia64/ia64.c (ia64_print_operand): Update.
16193 * config/mmix/mmix.c (mmix_print_operand): Update.
16194 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16195 (rs6000_expand_split_stack_prologue): Update.
16196 * config/rs6000/rs6000.c: Update.
16197 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16198 (s390_expand_vec_movstr): Update.
16199 (s390_expand_cs_tdsi): Update.
16200 (s390_expand_split_stack_prologue): Update.
16201 * config/sh/sh.c (sh_print_operand): Update.
16202 (expand_cbranchsi4): Update.
16203 (expand_cbranchdi4): Update.
16204 * config/sparc/sparc.c (output_v9branch): Update.
16205 * config/spu/spu.c (get_branch_target): Update.
16206 (ea_load_store_inline): Update.
16207 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16208 * config/tilepro/tilepro.c: Update.
16210 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16212 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16213 (gimplify_save_expr): Likewise.
16215 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16217 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16219 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16221 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16223 * timevar.def (TV_IPA_FNSUMMARY): Define.
16225 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16227 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16228 to back store errata sensitive sequence from being generated.
16229 (sqrtdf2_fix): Likewise.
16231 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16233 * tree-ssa-threadupdate.c (compute_path_counts,
16234 update_joiner_offpath_counts): Use profile_probability.
16236 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16239 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16241 * config/arm/arm-c.c (arm_cpu_builtins): Define
16242 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16244 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16246 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16247 array entries to represent __ieee128 versions of the
16248 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16249 scalar_extract_sig, and scalar_insert_exp built-in functions.
16250 (altivec_resolve_overloaded_builtin): Add special case handling
16251 for the __builtin_scalar_insert_exp function, as represented by
16252 the P9V_BUILTIN_VEC_VSIEDP constant.
16253 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16254 exponent support for __ieee128 argument.
16255 (VSESQP): Add scalar extract signature support for __ieee128
16257 (VSTDCNQP): Add scalar test negative support for __ieee128
16259 (VSIEQP): Add scalar insert exponent support for __int128 argument
16260 with __ieee128 result.
16261 (VSIEQPF): Add scalar insert exponent support for __ieee128
16262 argument with __ieee128 result.
16263 (VSTDCQP): Add scalar test data class support for __ieee128
16265 (VSTDCNQP): Add overload support for scalar test negative with
16266 __ieee128 argument.
16267 (VSTDCQP): Add overload support for scalar test data class
16268 __ieee128 argument.
16269 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16270 UNSPEC_VSX_SXSIGDP.
16271 (UNSPEC_VSX_SIEXPQP): New constant.
16272 (xsxexpqp): New insn for VSX scalar extract exponent quad
16274 (xsxsigqp): New insn for VSX scalar extract significand quad
16276 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16277 precision with floating point argument.
16278 (xststdcqp): New expand for VSX scalar test data class quad
16280 (xststdcnegqp): New expand for VSX scalar test negative quad
16282 (xststdcqp): New insn to match expansions for VSX scalar test data
16283 class quad precision and VSX scalar test negative quad precision.
16284 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16285 special case operand checking to enforce that second operand of
16286 VSX scalar test data class with quad precision argument is a 7-bit
16288 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16289 prototypes and descriptions of __ieee128 versions of
16290 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16291 scalar_test_data_class, and scalar_test_neg built-in functions.
16293 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16295 PR tree-optimization/81162
16296 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16297 replace a negate with an add.
16299 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16301 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16303 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16305 * config/arm/arm-c.c (arm_cpu_builtins): Define
16306 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16308 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16310 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16311 (armv8-r): Set ARM Cortex-R52 as default CPU.
16312 * config/arm/arm-tables.opt: Regenerate.
16313 * config/arm/arm-tune.md: Regenerate.
16314 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16316 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16317 extension for -mcpu=cortex-r52.
16319 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16321 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16322 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16323 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16324 (fp-armv8): Define it as FP_ARMv8 only.
16325 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16326 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16328 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16329 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16330 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16331 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16332 than TARGET_FPU_ARMV8.
16333 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16334 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16335 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16337 * config/arm/neon.md (neon_vrint): Likewise.
16338 (neon_vcvt): Likewise.
16339 (neon_<fmaxmin_op><mode>): Likewise.
16340 (<fmaxmin><mode>3): Likewise.
16341 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16342 * config/arm/predicates.md (arm_cond_move_operator): Check against
16343 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16345 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16347 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16348 to top of function.
16350 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16352 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16353 loop in comment with memset.
16355 2017-07-14 Martin Liska <mliska@suse.cz>
16357 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16358 * dwarf2out.c (is_java): Remove the function.
16359 (output_pubname): Remove usage of the function.
16360 (lower_bound_default): Remove usage of DW_LANG_Java.
16361 (gen_compile_unit_die): Likewise.
16362 * gcc.c: Remove compiler defaults for .java and .zip files.
16363 * gimple-expr.c (remove_suffix): Change as there's no longer
16364 extension than 4-letter one.
16365 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16366 (gimplify_save_expr): Likewise.
16367 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16368 as it's possible even for other languages than Java.
16369 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16370 * lto-opts.c (lto_write_options): Remove reference to Java.
16371 * opts.c (strip_off_ending): Update file extension handling.
16372 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16373 * tree-eh.c (lower_resx): Likewise.
16374 * tree.c (free_lang_data_in_type): Remove dead code.
16375 (find_decls_types_r): Likewise.
16376 (build_common_builtin_nodes): Remove Java from a comment.
16377 (verify_type): Remove dead code.
16378 * varasm.c (assemble_external): Remove Java from a comment.
16380 2017-07-14 Martin Liska <mliska@suse.cz>
16382 * opts.c (finish_options): Add quotes.
16383 (common_handle_option): Likewise.
16385 2017-07-14 Martin Liska <mliska@suse.cz>
16387 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16388 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16389 Remove N_SO_PASCAL.
16390 * dwarf2out.c (lower_bound_default): Do not handle
16392 (gen_compile_unit_die): Likewise.
16393 * gcc.c: Remove default extension binding for GNU Pascal.
16394 * stmt.c: Remove Pascal language from a comment.
16395 * xcoffout.c: Likewise.
16397 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16400 * diagnostic-show-locus.c (fixit_cmp): New function.
16401 (layout::layout): Sort m_fixit_hints.
16402 (column_range::column_range): Assert that the values are valid.
16403 (struct char_span): New struct.
16404 (correction::overwrite): New method.
16405 (struct source_line): New struct.
16406 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16407 calls in terms of classes source_line and char_span, and
16408 correction::overwrite.
16409 (selftest::test_overlapped_fixit_printing_2): New function.
16410 (selftest::diagnostic_show_locus_c_tests): Call it.
16412 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16414 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16415 early if there is no lhs.
16417 2017-07-13 Martin Liska <mliska@suse.cz>
16419 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16420 (gen_reference_type_die): Likewise.
16421 * stor-layout.c: Remove Pascal-related comment.
16423 2017-07-13 Martin Liska <mliska@suse.cz>
16425 * opts.c (finish_options): Add quotes to error messages.
16426 (parse_sanitizer_options): Likewise.
16428 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16430 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16432 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16434 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16436 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16438 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16440 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16442 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16445 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16446 provides the hardware capability bits, define the macro
16447 __BUILTIN_CPU_SUPPORTS__.
16448 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16449 if GLIBC does not provide the hardware capability bits. Add a
16450 gcc_unreachable call if the built-in cpu function is neither
16451 __builtin_cpu_is nor __builtin_cpu_supports.
16452 (rs6000_get_function_versions_dispatcher): Change the warning
16453 that an old GLIBC is used which does not export the capability
16454 bits to be an error.
16455 * doc/extend.texi (target_clones attribute): Document the
16456 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16457 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16458 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16459 the macros defined by GCC if the newer GLIBC is available.
16461 2017-07-12 Jeff Law <law@redhat.com>
16463 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16464 remaining includes slightly.
16465 * config/riscv/riscv-builtins.c: Include profile-count.h.
16467 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16470 * config/avr/avr.c (avr_set_current_function): In diagnostic
16471 messages: Quote keywords and (parts of) identifiers.
16472 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16475 2017-07-12 Carl Love <cel@us.ibm.com>
16477 * config/rs6000/rs6000-c.c: Add support for built-in functions
16478 vector bool char vec_revb (vector bool char);
16479 vector bool short vec_revb (vector short char);
16480 vector bool int vec_revb (vector bool int);
16481 vector bool long long vec_revb (vector bool long long);
16482 * doc/extend.texi: Update the built-in documentation file for the
16483 new built-in functions.
16485 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16487 * config/s390/s390.md: Remove movcc splitter.
16489 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16491 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16492 load/store on condition.
16494 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16497 * config/avr/avr.c (avr_encode_section_info)
16498 [progmem && !TREE_READONLY]: Error if progmem object needs
16501 2017-07-11 Michael Collison <michael.collison@arm.com>
16503 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16506 2017-07-11 Carl Love <cel@us.ibm.com>
16508 * config/rs6000/rs6000-c.c: Add support for builtins
16509 vector unsigned int vec_parity_lsbb (vector signed int);
16510 vector unsigned int vec_parity_lsbb (vector unsigned int);
16511 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16512 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16513 vector unsigned long long vec_parity_lsbb (vector signed long long);
16514 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16515 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16516 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16517 * doc/extend.texi: Update the built-in documentation file for the
16518 new built-in functions.
16520 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16522 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16523 (layout::m_primary_loc): New field.
16524 (layout::layout): Initialize new field. Move location filtering
16525 logic from here to...
16526 (layout::maybe_add_location_range): ...this new method. Add
16527 support for filtering to just the lines already specified by other
16529 (layout::will_show_line_p): New method.
16530 (gcc_rich_location::add_location_if_nearby): New method.
16531 (selftest::test_add_location_if_nearby): New test function.
16532 (selftest::diagnostic_show_locus_c_tests): Call it.
16533 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16536 2017-07-11 Tom de Vries <tom@codesourcery.com>
16538 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16539 (bb_first_real_insn): New function.
16540 (nvptx_single): Add extra initialization of broadcasted condition
16543 2017-07-11 Nathan Sidwell <nathan@acm.org>
16545 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16547 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16549 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16550 Binutils doc as TEXI will mess them up.
16551 * doc/invoke.texi (AVR Options): Same here.
16553 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16555 * config/sparc/sparc.opt (mfix-ut700): New option.
16556 (mfix-gr712rc): Likewise.
16557 (sparc_fix_b2bst): New variable.
16558 * doc/invoke.texi (SPARC options): Document them.
16559 (ARM options): Fix warnings.
16560 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16561 instructions to prevent sequences that can trigger the store-store
16562 errata for certain LEON3FT processors.
16563 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16564 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16565 * config/sparc/sparc.md (fix_b2bst): New attribute.
16566 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16568 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16571 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16573 (*rsqrtsf2_sse): Ditto.
16575 (div<mode>3): Macroize insn from divdf3 and divsf3
16576 using MODEF mode iterator.
16578 2017-07-10 Martin Sebor <msebor@redhat.com>
16580 PR tree-optimization/80397
16581 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16582 instead of testing for equality to INTEGER_TYPE.
16584 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16586 * config.gcc: Remove uclibc from arc target spec.
16588 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16590 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16592 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16595 * lto-wrapper.c (remove_option): New function.
16596 (merge_and_complain): Merge PIC/PIE options more realistically.
16598 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16600 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16604 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16605 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16606 * config.in: Regenerate.
16607 * configure: Regenerate.
16608 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16609 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16610 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16611 (TARGET_GASISR_PROLOGUES): ...target mask.
16612 * common/config/avr/avr-common.c
16613 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16614 Set -mgas-isr-prologues.
16615 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16616 INSERT_PASS_BEFORE for it.
16617 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16618 * config/avr/avr.c (avr_option_override)
16619 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16620 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16621 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16622 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16623 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16625 (make_avr_pass_pre_proep): New function.
16626 (emit_push_sfr) <treg>: Add argument to function and use it
16627 instead of TMP_REG.
16628 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16629 and set machine->gasisr.yes.
16630 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16631 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16632 __gcc_isr.n_pushed to .L__stack_usage.
16633 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16634 (avr_asm_final_postscan_insn): ...this new static function.
16635 * config/avr/avr.h (machine_function)
16636 <is_no_gccisr, use_L__stack_usage>: New fields.
16637 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16638 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16639 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16640 (gasisr, *gasisr): New expander and insn.
16641 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16642 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16643 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16645 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16647 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16650 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16652 Move jump-tables out of .text again.
16655 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16656 (ASM_OUTPUT_ADDR_VEC): New function.
16657 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16658 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16659 INSN_ADDRESSes as asm comment.
16660 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16661 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16662 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16663 * config/avr/avr.md (*tablejump): Adjust comment.
16664 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16665 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16667 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16668 (avr_output_addr_vec): New proto.
16669 (avr_log_t) <insn_addresses>: New field.
16671 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16674 * config/i386/i386.c (ix86_function_arg_advance): Set
16675 outgoing_args_on_stack to true if there are outgoing arguments
16677 (ix86_function_arg): Likewise.
16678 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16679 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16680 * config/i386/i386.h (machine_function): Add
16681 outgoing_args_on_stack.
16683 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16685 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16686 supporting pthreds.
16687 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16689 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16691 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16692 (REAL_H): Remove $(MACHMODE_H).
16693 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16695 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16696 $(MACHMODE_H) and double-int.h.
16697 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16699 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16702 2017-07-07 Andrew Pinski <apinski@cavium.com>
16704 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16705 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16707 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16709 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16710 Add warning if GCC was not configured to link against a GLIBC that
16711 exports the hardware capability bits.
16712 (make_resolver_func): Make resolver function private and not a
16713 COMDAT function. Create the name with clone_function_name instead
16714 of make_unique_name.
16717 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16718 correct operand in doing the split.
16720 2017-07-07 Carl Love <cel@us.ibm.com>
16722 * config/rs6000/rs6000-c: Add support for built-in function
16723 vector unsigned short vec_pack_to_short_fp32 (vector float,
16725 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16726 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16727 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16728 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16729 (convert_4f32_8i16): Add define_expand.
16730 * doc/extend.texi: Update the built-in documentation file for the
16731 new built-in function.
16733 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16735 * config/sparc/m8.md: New file.
16736 * config/sparc/sparc.md: Include m8.md.
16738 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16740 * config/sparc/sparc.opt: New option -mvis4b.
16741 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16742 (sparc_option_override): Handle VIS4B.
16743 (enum sparc_builtins): Define
16744 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16745 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16746 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16747 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16748 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16749 (check_constant_argument): New function.
16750 (sparc_vis_init_builtins): Define builtins
16751 __builtin_vis_dictunpack{8,16,32},
16752 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16753 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16754 __builtin_vis_fpcmpde{8,16,32}shl and
16755 __builtin_vis_fpcmpur{8,16,32}shl.
16756 (sparc_expand_builtin): Check that the constant operands to
16757 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16758 constant and in range.
16759 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16761 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16762 (SPARC_IMM5_P): Likewise.
16763 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16764 (enabled): Handle vis4b.
16765 (UNSPEC_DICTUNPACK): New unspec.
16766 (UNSPEC_FPCMPSHL): Likewise.
16767 (UNSPEC_FPUCMPSHL): Likewise.
16768 (UNSPEC_FPCMPDESHL): Likewise.
16769 (UNSPEC_FPCMPURSHL): Likewise.
16770 (cpu_feature): New CPU feature `vis4b'.
16771 (dictunpack{8,16,32}): New insns.
16772 (FPCSMODE): New mode iterator.
16773 (fpcscond): New code iterator.
16774 (fpcsucond): Likewise.
16775 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16776 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16777 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16778 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16779 * config/sparc/constraints.md: Define constraints `q' for unsigned
16780 2-bit integer constants and `t' for unsigned 5-bit integer
16782 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16784 (imm5_operand_dictunpack16): Likewise.
16785 (imm5_operand_dictunpack32): Likewise.
16786 (imm2_operand): Likewise.
16787 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16788 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16789 ditunpack* and fpcmp*shl builtins.
16791 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16793 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16794 * config.in: Add HAVE_AS_SPARC6 define.
16795 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16797 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16799 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16800 (CPP_CPU_SPEC): Handle m8.
16801 (ASM_CPU_SPEC): Likewise.
16802 * config/sparc/sparc-opts.h (enum processor_type): Add
16804 * config/sparc/sparc.c (m8_costs): New struct.
16805 (sparc_option_override): Handle TARGET_CPU_m8.
16806 (sparc32_initialize_trampoline): Likewise.
16807 (sparc64_initialize_trampoline): Likewise.
16808 (sparc_issue_rate): Likewise.
16809 (sparc_register_move_cost): Likewise.
16810 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16811 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16812 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16813 (CPP_CPU_SPEC): Handle M8.
16814 (ASM_CPU_SPEC): Likewise.
16815 (AS_M8_FLAG): Define.
16816 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16817 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16818 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16820 * configure: Regenerate.
16821 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16824 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16826 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16828 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16829 ("*movdi_insn_sp32"): Do not set v3pipe.
16830 ("*movsi_insn"): Likewise.
16831 ("*movdi_insn_sp64"): Likewise.
16832 ("*movsf_insn"): Likewise.
16833 ("*movdf_insn_sp32"): Likewise.
16834 ("*movdf_insn_sp64"): Likewise.
16835 ("*zero_extendsidi2_insn_sp64"): Likewise.
16836 ("*sign_extendsidi2_insn"): Likewise.
16837 ("*mov<VM32:mode>_insn"): Likewise.
16838 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16839 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16840 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16841 ("<vlop:code><VL:mode>3"): Likewise.
16842 ("*not_<vlop:code><VL:mode>3"): Likewise.
16843 ("*nand<VL:mode>_vis"): Likewise.
16844 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16845 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16846 ("one_cmpl<VL:mode>2"): Likewise.
16847 ("faligndata<VM64:mode>_vis"): Likewise.
16848 ("alignaddrsi_vis"): Likewise.
16849 ("alignaddrdi_vis"): Likweise.
16850 ("alignaddrlsi_vis"): Likewise.
16851 ("alignaddrldi_vis"): Likewise.
16852 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16853 ("bmaskdi_vis"): Likewise.
16854 ("bmasksi_vis"): Likewise.
16855 ("bshuffle<VM64:mode>_vis"): Likewise.
16856 ("cmask8<P:mode>_vis"): Likewise.
16857 ("cmask16<P:mode>_vis"): Likewise.
16858 ("cmask32<P:mode>_vis"): Likewise.
16859 ("pdistn<P:mode>_vis"): Likewise.
16860 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16862 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16864 * config/sparc/sparc.md ("subtype"): New insn attribute.
16865 ("*wrgsr_sp64"): Set insn subtype.
16866 ("*rdgsr_sp64"): Likewise.
16867 ("alignaddrsi_vis"): Likewise.
16868 ("alignaddrdi_vis"): Likewise.
16869 ("alignaddrlsi_vis"): Likewise.
16870 ("alignaddrldi_vis"): Likewise.
16871 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16872 ("fexpand_vis"): Likewise.
16873 ("fpmerge_vis"): Likewise.
16874 ("faligndata<VM64:mode>_vis"): Likewise.
16875 ("bshuffle<VM64:mode>_vis"): Likewise.
16876 ("cmask8<P:mode>_vis"): Likewise.
16877 ("cmask16<P:mode>_vis"): Likewise.
16878 ("cmask32<P:mode>_vis"): Likewise.
16879 ("fchksm16_vis"): Likewise.
16880 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16881 ("fmean16_vis"): Likewise.
16882 ("fp<plusminus_insn>64_vis"): Likewise.
16883 ("<plusminus_insn>v8qi3"): Likewise.
16884 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16885 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16886 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16887 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16888 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16889 ("*movqi_insn"): Likewise.
16890 ("*movhi_insn"): Likewise.
16891 ("*movsi_insn"): Likewise.
16892 ("movsi_pic_gotdata_op"): Likewise.
16893 ("*movdi_insn_sp32"): Likewise.
16894 ("*movdi_insn_sp64"): Likewise.
16895 ("movdi_pic_gotdata_op"): Likewise.
16896 ("*movsf_insn"): Likewise.
16897 ("*movdf_insn_sp32"): Likewise.
16898 ("*movdf_insn_sp64"): Likewise.
16899 ("*zero_extendhisi2_insn"): Likewise.
16900 ("*zero_extendqihi2_insn"): Likewise.
16901 ("*zero_extendqisi2_insn"): Likewise.
16902 ("*zero_extendqidi2_insn"): Likewise.
16903 ("*zero_extendhidi2_insn"): Likewise.
16904 ("*zero_extendsidi2_insn_sp64"): Likewise.
16905 ("ldfsr"): Likewise.
16906 ("prefetch_64"): Likewise.
16907 ("prefetch_32"): Likewise.
16908 ("tie_ld32"): Likewise.
16909 ("tie_ld64"): Likewise.
16910 ("*tldo_ldub_sp32"): Likewise.
16911 ("*tldo_ldub1_sp32"): Likewise.
16912 ("*tldo_ldub2_sp32"): Likewise.
16913 ("*tldo_ldub_sp64"): Likewise.
16914 ("*tldo_ldub1_sp64"): Likewise.
16915 ("*tldo_ldub2_sp64"): Likewise.
16916 ("*tldo_ldub3_sp64"): Likewise.
16917 ("*tldo_lduh_sp32"): Likewise.
16918 ("*tldo_lduh1_sp32"): Likewise.
16919 ("*tldo_lduh_sp64"): Likewise.
16920 ("*tldo_lduh1_sp64"): Likewise.
16921 ("*tldo_lduh2_sp64"): Likewise.
16922 ("*tldo_lduw_sp32"): Likewise.
16923 ("*tldo_lduw_sp64"): Likewise.
16924 ("*tldo_lduw1_sp64"): Likewise.
16925 ("*tldo_ldx_sp64"): Likewise.
16926 ("*mov<VM32:mode>_insn"): Likewise.
16927 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16928 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16930 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16932 * config/sparc/sparc.md ("type"): New insn type viscmp.
16933 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16935 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16936 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16937 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16938 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16940 ("n7_vis_logical_11cycle"): Likewise.
16941 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16942 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16943 * config/sparc/niagara.md ("niag_vis"): Likewise.
16944 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16945 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16947 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16949 * config/sparc/sparc.md: New instruction type `bmask'.
16950 (bmaskdi_vis): Use the `bmask' type.
16951 (bmasksi_vis): Likewise.
16952 * config/sparc/ultra3.md (us3_array): Likewise.
16953 * config/sparc/niagara7.md (n7_array): Likewise.
16954 * config/sparc/niagara4.md (n4_array): Likewise.
16955 * config/sparc/niagara2.md (niag2_vis): Likewise.
16956 (niag3_vis): Likewise.
16957 * config/sparc/niagara.md (niag_vis): Likewise.
16959 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16961 * ipa-comdats.c: Remove optimize check from gate.
16962 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16963 for functions not optimized.
16964 (ipa_fn_summary_read): Skip optimize check.
16965 (ipa_fn_summary_write): Likewise.
16966 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16968 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16970 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16971 (check_callers): Give up on caller which is not optimized.
16972 (inline_small_functions): Likewise.
16973 (ipa_inline): Do not give up when not optimizing.
16974 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16975 away unoptimizes cdtors.
16976 (whole_program_function_and_variable_visibility): Do
16977 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16978 * ipa.c (process_references): Do not check optimize.
16979 (symbol_table::remove_unreachable_nodes): Update optimize check.
16980 (set_writeonly_bit): Update optimize check.
16981 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16982 (pass_ipa_single_use::gate): Remove.
16984 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16986 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16987 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16988 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16989 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16990 permute_load, permute_store, adjust_extract, adjust_splat,
16991 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16992 replace_swap_with_copy, dump_swap_insn_table,
16993 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16994 recombine_lvx_pattern, recombine_stvx_pattern,
16995 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16996 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16997 to file rs6000-p8swap.c.
16998 * config/rs6000/rs6000-p8swap.c: New file.
16999 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
17000 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
17001 and rs6000*-*-* targets.
17003 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17005 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
17007 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17009 * lto-wrapper.c (merge_and_complain): Do not merge
17010 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
17011 fsigned_zeros, ftrapping_math, fwrapv.
17012 (append_compiler_options): Do not track these options.
17013 (append_linker_options): Likewie
17015 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17017 * cgraphunit.c (cgraph_node::finalize_function): When
17018 !flag_toplevel_reorde set no_reorder flag.
17019 (varpool_node::finalize_decl): Likewise.
17020 (symbol_table::compile): Drop no toplevel reorder path.
17022 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17024 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
17025 edges; zero probability is not better than uninitialized.
17027 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17029 * asan.h (asan_sanitize_allocas_p): Declare.
17030 * asan.c (asan_sanitize_allocas_p): New function.
17031 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17032 (handle_builtin_alloca): Likewise.
17033 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17034 if !asan_sanitize_allocas_p.
17035 * params.def (asan-instrument-allocas): Add new option.
17036 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17037 * opts.c (common_handle_option): Disable allocas sanitization for
17040 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17042 * asan.c: Include gimple-fold.h.
17043 (get_last_alloca_addr): New function.
17044 (handle_builtin_stackrestore): Likewise.
17045 (handle_builtin_alloca): Likewise.
17046 (asan_emit_allocas_unpoison): Likewise.
17047 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17048 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17049 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17050 (instrument_builtin_call): Pass gimple iterator to
17051 get_mem_refs_of_builtin_call.
17052 (last_alloca_addr): New global.
17053 * asan.h (asan_emit_allocas_unpoison): Declare.
17054 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17055 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17056 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17057 if function calls alloca.
17058 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17059 * gimple-fold.h (replace_call_with_value): Declare.
17060 * internal-fn.c: Include asan.h.
17061 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17062 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17064 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17066 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17067 (C_SELFTEST_FLAGS): New.
17068 (CPP_SELFTEST_FLAGS): New.
17069 (SELFTEST_DEPS): New, from deps of s-selftest.
17070 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17071 (CPP_SELFTEST_DEPS): New.
17072 (selftest): Add dependency on s-selftest-c++.
17073 (s-selftest): Rename to...
17074 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17075 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17076 than SELFTEST_FLAGS.
17077 (selftest-gdb): Rename to...
17078 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17080 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17081 (selftest-valgrind): Rename to...
17082 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17084 (selftest-valgrind): Reintroduce as an alias for
17085 selftest-c-valgrind.
17086 (s-selftest-c++): New.
17087 (selftest-c++-gdb): New.
17088 (selftest-c++-valgrind): New.
17090 2017-07-06 Olivier Hainque <hainque@adacore.com>
17092 * gcc.c (process_command): When deciding if undefined variables
17093 should be ignored when processing specs, accept "gcc -v" as well.
17095 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17097 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17098 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17100 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17102 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17103 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17104 * config/arm/arm-tables.opt: Regenerate.
17105 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17107 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17109 2017-07-06 Carl Love <cel@us.ibm.com>
17111 * ChangeLog: Clean up from mid air collision
17113 2017-07-06 Carl Love <cel@us.ibm.com>
17115 * config/rs6000/rs6000-c.c: Add support for built-in functions
17116 vector signed int vec_subc (vector signed int, vector signed int);
17117 vector signed __int128 vec_subc (vector signed __int128,
17118 vector signed __int128);
17119 vector unsigned __int128 vec_subc (vector unsigned __int128,
17120 vector unsigned __int128);
17121 vector signed int vec_sube (vector signed int, vector signed int,
17122 vector signed int);
17123 vector unsigned int vec_sube (vector unsigned int,
17124 vector unsigned int,
17125 vector unsigned int);
17126 vector signed __int128 vec_sube (vector signed __int128,
17127 vector signed __int128,
17128 vector signed__int128);
17129 vector unsigned __int128 vec_sube (vector unsigned __int128,
17130 vector unsigned __int128,
17131 vector unsigned __int128);
17132 vector signed int vec_subec (vector signed int, vector signed int,
17133 vector signed int);
17134 vector unsigned int vec_subec (vector unsigned int,
17135 vector unsigned int,
17136 vector unsigned int);
17137 vector signed __int128 vec_subec (vector signed __int128,
17138 vector signed __int128,
17139 vector signed__int128);
17140 vector unsigned __int128 vec_subec (vector unsigned __int128,
17141 vector unsigned __int128,
17142 vector unsigned __int128);
17143 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17144 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17145 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17146 BU_ALTIVEC_OVERLOAD_X definitions.
17147 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17148 * doc/extend.texi: Update the built-in documentation file for the new
17149 built-in functions.
17151 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17154 * diagnostic-show-locus.c (layout::layout): Use start and finish
17155 spelling location for the start and finish of each range.
17156 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17157 Add unused aspect param.
17158 * input.c (expand_location_1): Add "aspect" param, and use it
17159 to access the correct part of the location.
17160 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17162 (expand_location_to_spelling_point): Likewise.
17163 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17164 param, and pass it to expand_location_1.
17166 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17168 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17169 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17170 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17171 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17172 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17173 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17174 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17175 _mm_maskz_getmant_ss): New intrinsics.
17176 (__builtin_ia32_getexpss128_mask): Changed to ...
17177 __builtin_ia32_getexpss128_round ... this.
17178 (__builtin_ia32_getexpsd128_mask): Changed to ...
17179 __builtin_ia32_getexpsd128_round ... this.
17180 * config/i386/i386-builtin-types.def
17181 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17182 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17183 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17184 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17185 __builtin_ia32_getmantss_mask_round): New builtins.
17186 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17187 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17188 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17189 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17190 * config/i386/sse.md
17191 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17192 avx512f_sgetexp<mode><mask_scalar_name>
17193 <round_saeonly_scalar_name> ... this.
17194 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17195 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17196 vgetexp<ssescalarmodesuffix>
17197 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17198 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17199 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17200 avx512f_vgetmant<mode><mask_scalar_name>
17201 <round_saeonly_scalar_name> ... this.
17202 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17203 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17204 vgetmant<ssescalarmodesuffix>
17205 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17206 %0<mask_scalar_operand4>, %1, %2
17207 <round_saeonly_scalar_mask_op4>, %3} ... this.
17208 * config/i386/subst.md (mask_scalar_operand4,
17209 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17210 round_saeonly_scalar_nimm_predicate): New subst attributes.
17212 2017-07-06 Julia Koval <julia.koval@intel.com>
17214 * config/i386/i386.c (ix86_erase_embedded_rounding):
17215 Remove code for old rounding pattern.
17217 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17219 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17221 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17223 * doc/sourcebuild.texi (Test Directives, Variants of
17224 dg-require-support): Add documentation for dg-require-stack-check.
17226 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17228 * config/i386/subst.md (mask_scalar, round_scalar,
17229 round_saeonly_scalar): New meta-templates.
17230 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17231 round_scalar_mask_operand3, round_scalar_mask_op3,
17232 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17233 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17234 round_saeonly_scalar_constraint,
17235 round_saeonly_scalar_prefix): New subst attribute.
17236 * config/i386/sse.md
17237 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17238 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17239 <round_scalar_name> ... this.
17240 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17241 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17242 <round_scalar_name> ... this.
17243 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17244 <sse>_vm<code><mode>3<mask_scalar_name>
17245 <round_saeonly_scalar_name> ... this.
17246 (v<plusminus_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<plusminus_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<multdiv_mnemonic><ssescalarmodesuffix>
17253 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17254 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17255 v<multdiv_mnemonic><ssescalarmodesuffix>
17256 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17257 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17258 (v<maxmin_float><ssescalarmodesuffix>
17259 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17260 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17261 v<maxmin_float><ssescalarmodesuffix>
17262 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17263 %0<mask_scalar_operand3>, %1, %<iptr>2
17264 <round_saeonly_scalar_mask_op3>} ... this.
17266 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17268 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17269 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17271 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17272 Alan Hayward <alan.hayward@arm.com>
17273 David Sherwood <david.sherwood@arm.com>
17275 * combine.c (simplify_if_then_else): Remove "enum" before
17277 * compare-elim.c (can_eliminate_compare): Likewise.
17278 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17280 (aarch64_lookup_simd_builtin_type): Likewise.
17281 (aarch64_simd_builtin_type): Likewise.
17282 (aarch64_init_simd_builtin_types): Likewise.
17283 (aarch64_simd_expand_args): Likewise.
17284 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17286 (aarch64_reverse_mask): Likewise.
17287 (aarch64_simd_emit_reg_reg_move): Likewise.
17288 (aarch64_gen_adjusted_ldpstp): Likewise.
17289 (aarch64_ccmp_mode_to_code): Likewise.
17290 (aarch64_operands_ok_for_ldpstp): Likewise.
17291 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17292 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17294 (aarch64_min_divisions_for_recip_mul): Likewise.
17295 (aarch64_reassociation_width): Likewise.
17296 (aarch64_get_condition_code_1): Likewise.
17297 (aarch64_simd_emit_reg_reg_move): Likewise.
17298 (aarch64_simd_attr_length_rglist): Likewise.
17299 (aarch64_reverse_mask): Likewise.
17300 (aarch64_operands_ok_for_ldpstp): Likewise.
17301 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17302 (aarch64_gen_adjusted_ldpstp): Likewise.
17303 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17305 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17306 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17307 (arm_lookup_simd_builtin_type): Likewise.
17308 (arm_simd_builtin_type): Likewise.
17309 (arm_init_simd_builtin_types): Likewise.
17310 (arm_expand_builtin_args): Likewise.
17311 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17312 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17313 (ft32_setup_incoming_varargs): Likewise.
17314 (ft32_function_arg): Likewise.
17315 (ft32_function_arg_advance): Likewise.
17316 (ft32_pass_by_reference): Likewise.
17317 (ft32_arg_partial_bytes): Likewise.
17318 (ft32_valid_pointer_mode): Likewise.
17319 (ft32_addr_space_pointer_mode): Likewise.
17320 (ft32_addr_space_legitimate_address_p): Likewise.
17321 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17323 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17324 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17325 (iamcu_alignment): Likewise.
17326 (canonicalize_vector_int_perm): Likewise.
17327 (ix86_noce_conversion_profitable_p): Likewise.
17328 (ix86_mpx_bound_mode): Likewise.
17329 (ix86_operands_ok_for_move_multiple): Likewise.
17330 * config/microblaze/microblaze-protos.h
17331 (microblaze_expand_conditional_branch_reg): Likewise.
17332 * config/microblaze/microblaze.c
17333 (microblaze_expand_conditional_branch_reg): Likewise.
17334 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17336 (rs6000_reassociation_width): Likewise.
17337 (rs6000_invalid_binary_op): Likewise.
17338 (fusion_p9_p): Likewise.
17339 (emit_fusion_p9_load): Likewise.
17340 (emit_fusion_p9_store): Likewise.
17341 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17343 (riscv_hard_regno_mode_ok_p): Likewise.
17344 (riscv_address_insns): Likewise.
17345 (riscv_split_symbol): Likewise.
17346 (riscv_legitimize_move): Likewise.
17347 (riscv_function_value): Likewise.
17348 (riscv_hard_regno_nregs): Likewise.
17349 (riscv_expand_builtin): Likewise.
17350 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17351 (riscv_build_integer): Likewise.
17352 (riscv_split_integer): Likewise.
17353 (riscv_legitimate_constant_p): Likewise.
17354 (riscv_cannot_force_const_mem): Likewise.
17355 (riscv_regno_mode_ok_for_base_p): Likewise.
17356 (riscv_valid_base_register_p): Likewise.
17357 (riscv_valid_offset_p): Likewise.
17358 (riscv_valid_lo_sum_p): Likewise.
17359 (riscv_classify_address): Likewise.
17360 (riscv_legitimate_address_p): Likewise.
17361 (riscv_address_insns): Likewise.
17362 (riscv_load_store_insns): Likewise.
17363 (riscv_force_binary): Likewise.
17364 (riscv_split_symbol): Likewise.
17365 (riscv_force_address): Likewise.
17366 (riscv_legitimize_address): Likewise.
17367 (riscv_move_integer): Likewise.
17368 (riscv_legitimize_const_move): Likewise.
17369 (riscv_legitimize_move): Likewise.
17370 (riscv_address_cost): Likewise.
17371 (riscv_subword): Likewise.
17372 (riscv_output_move): Likewise.
17373 (riscv_canonicalize_int_order_test): Likewise.
17374 (riscv_emit_int_order_test): Likewise.
17375 (riscv_function_arg_boundary): Likewise.
17376 (riscv_pass_mode_in_fpr_p): Likewise.
17377 (riscv_pass_fpr_single): Likewise.
17378 (riscv_pass_fpr_pair): Likewise.
17379 (riscv_get_arg_info): Likewise.
17380 (riscv_function_arg): Likewise.
17381 (riscv_function_arg_advance): Likewise.
17382 (riscv_arg_partial_bytes): Likewise.
17383 (riscv_function_value): Likewise.
17384 (riscv_pass_by_reference): Likewise.
17385 (riscv_setup_incoming_varargs): Likewise.
17386 (riscv_print_operand): Likewise.
17387 (riscv_elf_select_rtx_section): Likewise.
17388 (riscv_save_restore_reg): Likewise.
17389 (riscv_for_each_saved_reg): Likewise.
17390 (riscv_register_move_cost): Likewise.
17391 (riscv_hard_regno_mode_ok_p): Likewise.
17392 (riscv_hard_regno_nregs): Likewise.
17393 (riscv_class_max_nregs): Likewise.
17394 (riscv_memory_move_cost): Likewise.
17395 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17396 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17397 (rl78_addr_space_address_mode): Likewise.
17398 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17400 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17401 (rs6000_reassociation_width): Likewise.
17402 (rs6000_invalid_binary_op): Likewise.
17403 (fusion_p9_p): Likewise.
17404 (emit_fusion_p9_load): Likewise.
17405 (emit_fusion_p9_store): Likewise.
17406 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17407 (ok_for_simple_move_operands): Likewise.
17408 (ok_for_simple_move_strict_operands): Likewise.
17409 (ok_for_simple_arith_logic_operands): Likewise.
17410 (visium_legitimize_reload_address): Likewise.
17411 (visium_select_cc_mode): Likewise.
17412 (output_cbranch): Likewise.
17413 (visium_split_double_move): Likewise.
17414 (visium_expand_copysign): Likewise.
17415 (visium_expand_int_cstore): Likewise.
17416 (visium_expand_fp_cstore): Likewise.
17417 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17418 (visium_function_arg): Likewise.
17419 (visium_function_arg_advance): Likewise.
17420 (visium_libcall_value): Likewise.
17421 (visium_setup_incoming_varargs): Likewise.
17422 (visium_legitimate_constant_p): Likewise.
17423 (visium_legitimate_address_p): Likewise.
17424 (visium_legitimize_address): Likewise.
17425 (visium_secondary_reload): Likewise.
17426 (visium_register_move_cost): Likewise.
17427 (visium_memory_move_cost): Likewise.
17428 (prepare_move_operands): Likewise.
17429 (ok_for_simple_move_operands): Likewise.
17430 (ok_for_simple_move_strict_operands): Likewise.
17431 (ok_for_simple_arith_logic_operands): Likewise.
17432 (visium_function_value_1): Likewise.
17433 (rtx_ok_for_offset_p): Likewise.
17434 (visium_legitimize_reload_address): Likewise.
17435 (visium_split_double_move): Likewise.
17436 (visium_expand_copysign): Likewise.
17437 (visium_expand_int_cstore): Likewise.
17438 (visium_expand_fp_cstore): Likewise.
17439 (visium_split_cstore): Likewise.
17440 (visium_select_cc_mode): Likewise.
17441 (visium_split_cbranch): Likewise.
17442 (output_cbranch): Likewise.
17443 (visium_print_operand_address): Likewise.
17444 * expmed.c (flip_storage_order): Likewise.
17445 * expmed.h (emit_cstore): Likewise.
17446 (flip_storage_order): Likewise.
17447 * genrecog.c (validate_pattern): Likewise.
17448 * hsa-gen.c (gen_hsa_addr): Likewise.
17449 * internal-fn.c (expand_arith_overflow): Likewise.
17450 * ira-color.c (allocno_copy_cost_saving): Likewise.
17451 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17452 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17453 (process_invariant_for_inheritance): Likewise.
17454 * lra-eliminations.c (move_plus_up): Likewise.
17455 * omp-low.c (lower_oacc_reductions): Likewise.
17456 * simplify-rtx.c (simplify_subreg): Likewise.
17457 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17458 (TARGET_CHKP_BOUND_MODE): Likewise..
17459 * targhooks.c (default_chkp_bound_mode): Likewise.
17460 (default_setup_incoming_vararg_bounds): Likewise.
17461 * targhooks.h (default_chkp_bound_mode): Likewise.
17462 (default_setup_incoming_vararg_bounds): Likewise.
17463 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17464 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17465 (have_whole_vector_shift): Likewise.
17466 * tree-vect-stmts.c (vectorizable_load): Likewise.
17467 * doc/tm.texi: Regenerate.
17469 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17471 Graceful degrade if Binutils PR21472 is not available.
17474 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17475 .rodata in flash test fails.
17476 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17477 * confgure: Regenerate.
17478 * config.in: Regenerate.
17479 * config/avr/avr.c (avr_asm_named_section)
17480 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17481 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17482 (avr_asm_init_sections): Same.
17484 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17486 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17487 (fma<VH:mode>4_intrinsic): Likewise.
17488 (*fmsub<VCVTF:mode>4): Likewise.
17489 (*fmsub<VH:mode>4_intrinsic): Likewise.
17491 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17494 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17495 Don't depend on "optimize > 0".
17496 (out_movhi_r_mr, out_movqi_mr_r): Same.
17497 (out_movhi_mr_r, out_movqi_r_mr): Same.
17498 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17499 io_address_operand on "optimize > 0".
17501 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17503 * tree-loop-distribution.c: Add general explanantion on the pass.
17504 (generate_loops_for_partition): Mark distributed loop.
17505 (pg_add_dependence_edges): New parameter. Handle alias data
17506 dependence specially and record it in the parameter if asked.
17507 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17508 (init_partition_graph_vertices, add_partition_graph_edge): New.
17509 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17510 (free_partition_graph_vdata, build_partition_graph): New.
17511 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17512 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17513 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17514 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17515 (version_for_distribution_p, finalize_partitions): New.
17516 (distribute_loop): Handle alias data dependence specially. Factor
17517 out loop fusion code as functions and call these functions.
17519 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17521 * tree-loop-distribution.c (classify_partition): New parameter and
17522 better handle reduction statement.
17523 (rdg_build_partitions): Revise comment.
17524 (distribute_loop): Compute statements in all partitions and pass it
17525 to classify_partition.
17527 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17529 * tree-loop-distribution.c (enum partition_type): New.
17530 (struct partition): New field type.
17531 (partition_merge_into): Add parameter. Update partition type.
17532 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17533 (build_rdg_partition_for_vertex): Compute partition type.
17534 (rdg_build_partitions): Dump partition type.
17535 (distribute_loop): Update calls to partition_merge_into.
17537 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17539 * tree-loop-distribution.c (struct ddr_hasher): New.
17540 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17542 (classify_partition): Call get_data_dependence.
17543 (pg_add_dependence_edges): Ditto.
17544 (distribute_loop): Release data dependence hash table.
17546 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17548 * tree-loop-distribution.c (ref_base_address): Delete.
17549 (similar_memory_accesses): Rename ...
17550 (share_memory_accesses): ... to this. Check if partitions access
17551 the same memory reference.
17552 (distribute_loop): Call share_memory_accesses.
17554 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17556 * tree-loop-distribution.c (struct partition): New field recording
17557 its data reference.
17558 (partition_alloc, partition_free): Init and release data refs.
17559 (partition_merge_into): Merge data refs.
17560 (build_rdg_partition_for_vertex): Collect data refs for partition.
17561 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17563 (distribute_loop): Remve data refs from vertice data of partition
17566 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17568 * tree-loop-distribution.c (params.h): Include header file.
17569 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17570 (datarefs_vec): New global var.
17571 (create_rdg_vertices): Use datarefs_vec directly.
17572 (free_rdg): Don't free data references.
17573 (build_rdg): Update use. Don't free data references.
17574 (distribute_loop): Compute global variable for data references.
17575 Bail out if there are too many data references.
17577 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17579 * tree-loop-distribution.c (loop_nest): New global var.
17580 (build_rdg): Use loop directly, rather than loop nest.
17581 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17583 (distribute_loop): Compute global variable loop nest. Update use.
17585 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17587 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17588 (partition_merge_into): New parameter. Dump reason for fusion.
17589 (distribute_loop): Update use of partition_merge_into.
17591 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17593 * tree-loop-distribution.c (bb_top_order_index): New.
17594 (bb_top_order_index_size, bb_top_order_cmp): New.
17595 (stmts_from_loop): Use topological order.
17596 (pass_loop_distribution::execute): Compute and release topological
17597 order for basic blocks.
17599 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17601 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17604 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17606 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17607 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17608 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17609 * internal-fn.def (LOOP_DIST_ALIAS): New.
17610 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17611 (fold_loop_internal_call): ... this.
17612 (vect_loop_dist_alias_call): New function.
17613 (set_uid_loop_bbs): Call fold_loop_internal_call.
17614 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17617 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17620 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17621 Require dead FLAGS_REG at the beginning of a peephole.
17623 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17626 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17627 arguments in the call to __builtin_ia32_sbb_u32.
17628 (_subborrow_u64): Swap _X and _Y arguments in the call to
17629 __builtin_ia32_sbb_u64.
17631 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17634 * tree-vrp.c (compare_assert_loc): Turn into a function template
17635 with stable template parameter. Only test if a->e is NULL,
17636 !a->e == !b->e has been verified already. Use e == NULL or
17637 e != NULL instead of e or ! e tests. If stable is true, don't use
17638 iterative_hash_expr, on the other side allow a or b or both NULL
17639 and sort the NULLs last.
17640 (process_assert_insertions): Sort using compare_assert_loc<false>
17641 instead of compare_assert_loc, later sort using
17642 compare_assert_loc<true> before calling process_assert_insertions_for
17643 in a loop. Use break instead of continue once seen NULL pointer.
17645 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17647 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17648 Cortex-R7 and Cortex-R8 processors.
17650 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17652 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17653 uninitialized while src is not.
17655 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17657 * common/config/arm/arm-common.c: Adjust include path for
17659 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17660 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17661 (arm-cpu-data.h): Likewise.
17662 (arm-cpu-cdata.h): Likewise.
17663 * config/arm/arm-cpu.h: Delete.
17664 * config/arm/arm-cpu-cdata.h: Delete.
17665 * config/arm/arm-cpu-data.h: Delete.
17667 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17669 * config/arm/arm-cpus.in (cortex-a55): New.
17670 (cortex-a75): Likewise.
17671 (cortex-a75.cortex-a55): Likewise.
17672 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17674 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17675 * config/arm/arm-cpu-cdata.h: Regenerate.
17676 * config/arm/arm-cpu-data.h: Regenerate.
17677 * config/arm/arm-cpu.h: Regenerate.
17678 * config/arm/arm-tables.opt: Regenerate.
17679 * config/arm/arm-tune.md: Regenerate.
17681 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17683 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17685 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17687 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17690 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17692 PR tree-optimization/81292
17693 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17694 full_string_p, also call adjust_related_strinfos if the adjustment
17695 is simple, otherwise invalidate related strinfos.
17697 2017-07-04 Martin Liska <mliska@suse.cz>
17700 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17701 newly created variable as DECL_IGNORED_P.
17703 2017-07-04 Martin Liska <mliska@suse.cz>
17706 * ipa-inline.c (inline_small_functions):
17707 Use xstrdup_for_dump.
17709 2017-07-04 Tom de Vries <tom@codesourcery.com>
17711 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17713 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17716 * config/darwin.c (darwin_function_switched_text_sections):
17719 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17721 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17723 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17725 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17727 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17729 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17730 min_profitable_iters, and th as inclusive lower bounds.
17731 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17732 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17733 for min_profitable_iters and min_profitable_estimate.
17734 (vect_transform_loop): Treat th as an inclusive lower bound.
17735 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17737 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17740 * config/darwin.c (darwin_function_switched_text_sections):
17741 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17742 in two pieces, and suppress the use of buf.
17744 2017-07-03 Nathan Sidwell <nathan@acm.org>
17746 * hash-table.h (hash_table_mod1): Fix indentation.
17748 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17750 PR middle-end/81290
17751 * predict.c (force_edge_cold): Be more careful about propagation
17753 * profile-count.h (profile_probability::guessed,
17754 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17756 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17758 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17760 * doc/invoke.texi (rcpc architecture extension): Document it.
17762 2017-07-03 Richard Biener <rguenther@suse.de>
17764 PR tree-optimization/60510
17765 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17766 the scalar reduction PHI and use it.
17767 (vectorizable_reduction): Properly guard the single_defuse_cycle
17768 path for non-SLP reduction chains where we cannot use it.
17769 Rework reduc_def/index and vector type deduction. Rework
17770 vector operand gathering during reduction op code-gen.
17771 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17772 chains dissolve the chain and leave it to non-SLP reduction
17775 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17777 * tree-data-ref.h (dr_alignment): Declare.
17778 * tree-data-ref.c (dr_alignment): New function.
17779 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17780 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17782 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17784 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17786 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17787 and base_misalignment fields.
17788 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17789 * tree-data-ref.c: Include builtins.h.
17790 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17791 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17792 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17793 * tree-vect-data-refs.c: Include tree-cfg.h.
17794 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17795 fields instead of calculating an alignment here.
17796 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17797 innermost_loop_behavior fields.
17799 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17801 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17803 (DR_STEP_ALIGNMENT): New macro.
17804 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17805 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17806 (create_data_ref): Print it.
17807 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17808 to tell whether the step preserves vector (mis)alignment.
17809 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17810 Move the check for an integer step and generalise to all INTEGER_CST.
17811 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17812 Print the outer step alignment.
17814 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17816 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17817 with offset_alignment.
17818 (DR_ALIGNED_TO): Delete.
17819 (DR_OFFSET_ALIGNMENT): New macro.
17820 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17821 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17822 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17823 (create_data_ref): Likewise.
17824 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17825 (vect_analyze_data_refs): Likewise.
17826 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17827 creating dummy innermost behavior.
17829 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17831 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17832 with a "innermost_loop_behavior *" and refeence tree.
17833 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17834 (create_data_ref): Update call accordingly.
17835 * tree-predcom.c (find_looparound_phi): Likewise.
17837 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17839 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17840 fields with dr_wrt_vec_loop.
17841 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17842 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17843 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17844 (vect_dr_behavior): New function.
17845 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17846 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17847 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17848 track whether the step preserves the misalignment.
17849 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17850 Use vect_dr_behavior.
17851 (vect_setup_realignment): Update call accordingly.
17852 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17853 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17854 call to vect_create_addr_base_for_vector_ref.
17855 (vect_create_cond_for_align_checks): Likewise.
17856 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17857 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17858 (vect_recog_mask_conversion_pattern): Likewise.
17859 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17860 (new_stmt_vec_info): Remove redundant zeroing.
17862 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17864 * common/config/arm/arm-common.c (arm_be8_option): New function.
17865 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17866 (ISA_ARMv6): Add isa_bit_be8.
17867 * config/arm/arm.h (arm_be8_option): Add prototype.
17868 (BE8_SPEC_FUNCTION): New define.
17869 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17870 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17871 (mlittle-endian): Similarly.
17872 (mbe8, mbe32): New options.
17873 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17874 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17876 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17878 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17880 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17882 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17883 (cleanup_tree_cfg_bb): Use it.
17884 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17886 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17888 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17891 * loop-doloop.c (add_test): Update profile.
17893 2017-07-03 Martin Liska <mliska@suse.cz>
17896 * sanopt.c (rewrite_usage_of_param): New function.
17897 (sanitize_rewrite_addressable_params): Likewise.
17898 (pass_sanopt::execute): Call rewrite_usage_of_param.
17900 2017-07-03 Richard Biener <rguenther@suse.de>
17902 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17903 back to using VIEW_CONVERT_EXPR.
17905 2017-07-03 Martin Liska <mliska@suse.cz>
17908 * doc/extend.texi: Document when a resolver function is
17909 generated for target_clones.
17911 2017-07-03 Martin Liska <mliska@suse.cz>
17913 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17914 and shadow memory of auto variables which are subject of
17915 use-after-scope sanitization.
17916 (asan_expand_mark_ifn): Add do set only when is_poison.
17918 2016-07-03 Richard Biener <rguenther@suse.de>
17920 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17922 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17923 (vectorizable_reduction): Perform reduction PHI creation when
17924 visiting a reduction PHI and adjust and simplify code generation
17925 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17926 (vect_transform_loop): Visit reduction PHIs.
17927 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17928 defs into the SLP tree.
17929 (vect_build_slp_tree): Reduction defs terminate the recursion.
17930 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17932 (vect_get_vec_defs_for_stmt_copy): Export.
17933 (vect_get_vec_defs): Likewise.
17934 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17936 (vect_get_vec_defs_for_stmt_copy): Declare.
17937 (vect_get_vec_defs): Likewise.
17939 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17941 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17942 parameter with a "loop" parameter and use it instead of the
17943 loop containing DR_STMT. Don't check simple_iv when doing
17944 BB analysis. Describe the two analysis modes in the comment.
17946 2017-07-03 Tom de Vries <tom@codesourcery.com>
17948 PR tree-optimization/69468
17949 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17950 (find_same_succ_bb): Handle ignore_edge_flags.
17952 2017-07-03 Tom de Vries <tom@codesourcery.com>
17954 PR tree-optimization/81192
17955 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17957 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17959 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17961 2017-07-03 Tom de Vries <tom@codesourcery.com>
17963 PR tree-optimization/81192
17964 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17965 BB_SAME_SUCC (bb) == NULL.
17967 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17969 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17972 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17974 * dumpfile.c: Include profile-count.h
17975 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17977 (insert_cond_bb): Update profile.
17978 * tree-cfg.h (insert_cond_bb): Update prototype.
17979 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17980 * tree-dump.c: Do not include tree-cfg.
17982 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17984 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17986 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17988 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17991 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17993 * tree-complex.c (expand_complex_div_wide): update profile.
17995 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17996 Alan Hayward <alan.hayward@arm.com>
17997 David Sherwood <david.sherwood@arm.com>
17999 * Makefile.in (MACHMODE_H): Remove insn-modes.h
18000 (CORETYPES_H): New define.
18001 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
18002 (insn-modes-inline.h, s-modes-inline-h): New rules.
18003 (generated_files): Add insn-modes-inline.h.
18004 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
18005 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
18006 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
18007 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
18008 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
18009 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
18010 (build/gencodes.o, build/genconditions.o): Likewise.
18011 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
18012 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
18013 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
18014 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
18015 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
18016 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
18017 * coretypes.h: Include everything up to real.h for generators.
18018 Include insn-modes.h first. Include wide-int-print.h after
18019 wide-int.h. Include insn-modes-inline.h and then machmode.h.
18020 * machmode.h: Don't include insn-modes.h here.
18021 * function-tests.c: Remove includes of signop.h, machmode.h,
18022 double-int.h and wide-int.h.
18024 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18026 * optc-save-gen.awk: Likewise.
18027 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18028 * godump.c: Remove include of wide-int-print.h.
18029 * pretty-print.h: Likewise.
18030 * wide-int-print.cc: Likewise.
18031 * wide-int.cc: Likewise.
18032 * hash-map-tests.c: Remove include of signop.h.
18033 * hash-set-tests.c: Likewise.
18034 * rtl-tests.c: Likewise.
18035 * mkconfig.sh: Remove include of machmode.h.
18036 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18038 (emit_insn_modes_inline_h): ...this new function. Emit the code
18039 into an insn-modes-inline.h header file, adding appropriate
18040 include guards and end comments.
18041 (emit_insn_modes_c_header): Remove include of machmode.h.
18042 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18044 (main): Handle -i flag and call emit_insn_modes_inline_h when
18047 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18049 * tree-ssa-strlen.c (strinfo): Rename the length field to
18050 nonzero_chars. Add a full_string_p field.
18051 (compare_nonzero_chars, zero_length_string_p): New functions.
18052 (get_addr_stridx): Add an offset_out parameter.
18053 Use compare_nonzero_chars.
18054 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18055 (new_strinfo): Update after above changes to strinfo.
18056 (set_endptr_and_length): Set full_string_p.
18057 (get_string_length): Update after above changes to strinfo.
18058 (unshare_strinfo): Update call to new_strinfo.
18059 (maybe_invalidate): Likewise.
18060 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18061 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18062 as a uhwi instead of an shwi. Update after above changes to
18063 strinfo and new_strinfo.
18064 (zero_length_string): Assert that chainsi contains full strings.
18065 Use zero_length_string_p. Update call to new_strinfo.
18066 (adjust_related_strinfos): Update after above changes to strinfo.
18067 Copy full_string_p from origsi.
18068 (adjust_last_stmt): Use zero_length_string_p.
18069 (handle_builtin_strlen): Update after above changes to strinfo and
18070 new_strinfo. Install the lhs as the string length if the previous
18071 entry didn't describe a full string.
18072 (handle_builtin_strchr): Update after above changes to strinfo
18074 (handle_builtin_strcpy): Likewise.
18075 (handle_builtin_strcat): Likewise.
18076 (handle_builtin_malloc): Likewise.
18077 (handle_pointer_plus): Likewise.
18078 (handle_builtin_memcpy): Likewise. Track nonzero characters
18079 that aren't necessarily followed by a nul terminator.
18080 (handle_char_store): Likewise.
18082 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18084 PR tree-optimization/80769
18085 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18086 for malloc and calloc. Document the new invariant that all related
18087 strinfos have delayed lengths or none do.
18088 (verify_related_strinfos): Move earlier in file.
18089 (set_endptr_and_length): New function, split out from...
18090 (get_string_length): ...here. Also set the lengths of related
18092 (zero_length_string): Assert that chainsi has known (rather than
18094 (adjust_related_strinfos): Likewise.
18096 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18098 PR tree-optimization/81136
18099 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18100 assert that two references with the same misalignment have the same
18101 compile-time misalignment if those compile-time misalignments
18104 2017-07-01 Andi Kleen <ak@linux.intel.com>
18106 * print-tree.c (print_node): Print all attributes.
18108 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18110 * cfg.c (scale_bbs_frequencies): New function.
18111 * cfg.h (scale_bbs_frequencies): Declare it.
18112 * cfgloopanal.c (single_likely_exit): Cleanup.
18113 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18115 (scale_loop_profile): Likewise.
18116 (loop_version): Likewise.
18117 (create_empty_loop_on_edge): Update.
18118 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18119 scale_loop_frequencies, scale_loop_profile, loopify,
18120 loop_version): Update prototypes.
18121 * modulo-sched.c (sms_schedule): Update.
18122 * predict.c (unlikely_executed_edge_p): Also check probability.
18123 (probably_never_executed_edge_p): Fix typo.
18124 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18125 * tree-parloops.c (gen_parallel_loop): Update.
18126 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18127 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18128 * tree-ssa-loop-split.c (split_loop): Update.
18129 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18130 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18131 (vect_loop_versioning): Update.
18132 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18134 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18136 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18138 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18140 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18141 to keep profile consistent.
18143 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18145 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18146 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18147 * profile-count.h (max_safe_multiplier): Make unsigned.
18148 (profile_count::guessed_zero): New.
18150 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18152 * bb-reorder.c (fix_up_crossing_landing_pad,
18153 fix_crossing_conditional_branches): Use make_single_succ_edge
18154 to keep profile consistent.
18156 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18158 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18161 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18164 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18165 the right scopes, make sure cond_jump isn't preserved between multiple
18166 iterations. Search for fallthru edge whenever there are 3+ edges and
18167 use find_fallthru_edge for it.
18169 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18171 Patch by Alexander Monakov <amonakov@ispras.ru>
18172 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18173 probabilities consistently.
18175 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18177 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18178 emit_cmp_and_jump_insns.
18180 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18183 * tree-inline.c (expand_call_inline): Combine profile statuses.
18185 2017-06-30 Andrew Pinski <apinski@cavium.com>
18187 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18188 fold_stmt returned true.
18190 2017-06-30 Nathan Sidwell <nathan@acm.org>
18192 * ggc.h (empty_string): Delete.
18193 * cfgexpand.c (expand_asm_stmt): Use plain "".
18194 * optabs.c (expand_asm_memory_barrier): Likewise.
18195 * stringpool.c (empty_string): Delete.
18196 (digit_vector, digit_string): Delete.
18197 (ggc_alloc_string): Use plain "", don't optimize single digit
18198 strings. Use ggc_alloc_atomic.
18200 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18202 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18203 comparison set and one other set, use the cost of the non-comparison
18206 2017-06-30 Nathan Sidwell <nathan@acm.org>
18208 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18211 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18213 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18214 loops. Remove now unneeded calls to gimple_switch_set_label() that
18215 just set removed labels to NULL_TREE.
18217 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18219 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18220 (set_range_info): ...here. Only call set_range_info_raw if domain
18222 (set_nonzero_bits): Call set_range_info_raw.
18223 * tree-ssanames.h (set_range_info_raw): New.
18225 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18228 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18229 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18230 of nonimmediate_operand and <store_mask_constraint> instead of m
18231 for the input operand. For V8FI iterator, always split if input
18232 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18233 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18234 <store_mask_predicate> instead of register_operand and
18235 <store_mask_constraint> instead of v for the input operand. Make
18236 sure both operands aren't MEMs for if not <mask_applied>.
18238 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18240 * lto-wrapper.c (copy_file) Close both file descriptors before
18243 2017-06-30 Martin Liska <mliska@suse.cz>
18246 * multiple_target.c (create_dispatcher_calls): Make ifunc
18247 also for function that don't have calls or are not referenced.
18249 2017-06-30 Richard Biener <rguenther@suse.de>
18251 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18252 analyze the first scalar stmt. Move vector type computation
18253 for the BB case here from ...
18254 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18255 live operation processing in the SLP case properly.
18257 2017-06-30 Richard Biener <rguenther@suse.de>
18259 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18261 2017-06-30 Martin Liska <mliska@suse.cz>
18264 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18265 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18267 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18269 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18272 2017-06-30 Richard Biener <rguenther@suse.de>
18274 PR tree-optimization/81249
18275 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18276 condition reduction result to original scalar type.
18278 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18280 * profile-count.h (enum profile_quality): Fix typos and whitespace
18283 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18285 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18286 type for branch probabilities.
18288 2017-06-29 Julian Brown <julian@codesourcery.com>
18289 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18291 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18292 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18293 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18294 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18296 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18298 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18299 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18300 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18301 CC usage from generic code to here.
18302 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18303 CC usage into the target macros.
18305 2017-06-29 Maya Rashish <coypu@sdf.org>
18307 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18310 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18312 * arm/arm-builtins.c: Include profile-count.h
18313 * except.c (sjlj_emit_function_enter): Use
18314 profile_probability::unlikely.
18316 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18318 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18319 and tocrel_offset be pointer args rather than implicitly using
18321 (legitimate_constant_pool_address_p, rs6000_emit_move,
18322 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18323 tocrel_offset and use in toc_relative_expr_p call.
18324 (print_operand, print_operand_address): Use static tocrel_base_oac
18325 and tocrel_offset_oac.
18326 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18329 2017-06-29 Maya Rashish <coypu@sdf.org>
18331 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18333 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18335 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18336 objects, take into account only the alignment of 'op0' and 'mode1' if
18339 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18341 * ccmp.c (ccmp_tree_comparison_p): New function.
18342 (ccmp_candidate_p): Update to use above function.
18343 (get_compare_parts): New function.
18344 (expand_ccmp_next): Update to use new functions.
18345 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18347 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18348 take mode as argument.
18349 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18350 * expr.c (expand_expr_real_1): Pass mode as argument.
18352 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18354 * combine.c (combine_instructions): Print insns to dump_file, together
18357 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18359 * asan.c (asan_emit_stack_protection): Update.
18360 (create_cond_insert_point): Update.
18361 * auto-profile.c (afdo_propagate_circuit): Update.
18362 * basic-block.h (struct edge_def): Turn probability to
18363 profile_probability.
18364 (EDGE_FREQUENCY): Update.
18365 * bb-reorder.c (find_traces_1_round): Update.
18366 (better_edge_p): Update.
18367 (sanitize_hot_paths): Update.
18368 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18369 (make_single_succ_edge): Update.
18370 (check_bb_profile): Update.
18371 (dump_edge_info): Update.
18372 (update_bb_profile_for_threading): Update.
18373 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18374 probabilitycount to 0.
18375 * cfgbuild.c (compute_outgoing_frequencies): Update.
18376 * cfgcleanup.c (try_forward_edges): Update.
18377 (outgoing_edges_match): Update.
18378 (try_crossjump_to_edge): Update.
18379 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18380 (expand_gimple_tailcall): Update.
18381 (construct_init_block): Use make_single_succ_edge.
18382 (construct_exit_block): Use make_single_succ_edge.
18383 * cfghooks.c (verify_flow_info): Update.
18384 (redirect_edge_succ_nodup): Update.
18385 (split_edge): Update.
18386 (account_profile_record): Update.
18387 * cfgloopanal.c (single_likely_exit): Update.
18388 * cfgloopmanip.c (scale_loop_profile): Update.
18389 (set_zero_probability): Remove.
18390 (duplicate_loop_to_header_edge): Update.
18391 * cfgloopmanip.h (loop_version): Update prototype.
18392 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18393 (force_nonfallthru_and_redirect): Update.
18394 (update_br_prob_note): Update.
18395 (rtl_verify_edges): Update.
18396 (purge_dead_edges): Update.
18397 (rtl_lv_add_condition_to_bb): Update.
18398 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18399 * cgraphunit.c (init_lowered_empty_function): Update.
18400 (cgraph_node::expand_thunk): Update.
18401 * cilk-common.c: Include profile-count.h
18402 * dojump.c (inv): Remove.
18403 (jumpifnot): Update.
18404 (jumpifnot_1): Update.
18405 (do_jump_1): Update.
18407 (do_jump_by_parts_greater_rtx): Update.
18408 (do_compare_rtx_and_jump): Update.
18409 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18410 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18411 * dwarf2cfi.c: Include profile-count.h
18412 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18413 (sjlj_emit_dispatch_table): Likewise.
18414 * explow.c: Include profile-count.h
18415 * expmed.c (emit_store_flag_force): Update.
18416 (do_cmp_and_jump): Update.
18417 * expr.c (compare_by_pieces_d::generate): Update.
18418 (compare_by_pieces_d::finish_mode): Update.
18419 (emit_block_move_via_loop): Update.
18420 (store_expr_with_bounds): Update.
18421 (store_constructor): Update.
18422 (expand_expr_real_2): Update.
18423 (expand_expr_real_1): Update.
18424 * expr.h (try_casesi, try_tablejump): Update prototypes.
18425 * gimple-pretty-print.c (dump_probability): Update.
18426 (dump_profile): New.
18427 (dump_gimple_label): Update.
18428 (dump_gimple_bb_header): Update.
18429 * graph.c (draw_cfg_node_succ_edges): Update.
18430 * hsa-gen.c (convert_switch_statements): Update.
18431 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18432 (find_if_case_1): Update.
18433 (find_if_case_2): Update.
18434 * internal-fn.c (expand_arith_overflow_result_store): Update.
18435 (expand_addsub_overflow): Update.
18436 (expand_neg_overflow): Update.
18437 (expand_mul_overflow): Update.
18438 (expand_vector_ubsan_overflow): Update.
18439 * ipa-cp.c (good_cloning_opportunity_p): Update.
18440 * ipa-split.c (split_function): Use make_single_succ_edge.
18441 * ipa-utils.c (ipa_merge_profiles): Update.
18442 * loop-doloop.c (add_test): Update.
18443 (doloop_modify): Update.
18444 * loop-unroll.c (compare_and_jump_seq): Update.
18445 (unroll_loop_runtime_iterations): Update.
18446 * lra-constraints.c (lra_inheritance): Update.
18447 * lto-streamer-in.c (input_cfg): Update.
18448 * lto-streamer-out.c (output_cfg): Update.
18449 * mcf.c (adjust_cfg_counts): Update.
18450 * modulo-sched.c (sms_schedule): Update.
18451 * omp-expand.c (expand_omp_for_init_counts): Update.
18452 (extract_omp_for_update_vars): Update.
18453 (expand_omp_ordered_sink): Update.
18454 (expand_omp_for_ordered_loops): Update.
18455 (expand_omp_for_generic): Update.
18456 (expand_omp_for_static_nochunk): Update.
18457 (expand_omp_for_static_chunk): Update.
18458 (expand_cilk_for): Update.
18459 (expand_omp_simd): Update.
18460 (expand_omp_taskloop_for_outer): Update.
18461 (expand_omp_taskloop_for_inner): Update.
18462 * omp-simd-clone.c (simd_clone_adjust): Update.
18463 * optabs.c (expand_doubleword_shift): Update.
18464 (expand_abs): Update.
18465 (emit_cmp_and_jump_insn_1): Update.
18466 (expand_compare_and_swap_loop): Update.
18467 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18468 * predict.c (predictable_edge_p): Update.
18469 (edge_probability_reliable_p): Update.
18470 (set_even_probabilities): Update.
18471 (combine_predictions_for_insn): Update.
18472 (combine_predictions_for_bb): Update.
18473 (propagate_freq): Update.
18474 (estimate_bb_frequencies): Update.
18475 (force_edge_cold): Update.
18476 * profile-count.c (profile_count::dump): Add missing space into dump.
18477 (profile_count::debug): Add newline.
18478 (profile_count::differs_from_p): Explicitly convert to unsigned.
18479 (profile_count::stream_in): Update.
18480 (profile_probability::dump): New member function.
18481 (profile_probability::debug): New member function.
18482 (profile_probability::differs_from_p): New member function.
18483 (profile_probability::differs_lot_from_p): New member function.
18484 (profile_probability::stream_in): New member function.
18485 (profile_probability::stream_out): New member function.
18486 * profile-count.h (profile_count_quality): Rename to ...
18487 (profile_quality): ... this one.
18488 (profile_probability): New.
18489 (profile_count): Update.
18490 * profile.c (compute_branch_probabilities): Update.
18491 * recog.c (peep2_attempt): Update.
18492 * sched-ebb.c (schedule_ebbs): Update.
18493 * sched-rgn.c (find_single_block_region): Update.
18494 (compute_dom_prob_ps): Update.
18495 (schedule_region): Update.
18496 * sel-sched-ir.c (compute_succs_info): Update.
18497 * stmt.c (struct case_node): Update.
18498 (do_jump_if_equal): Update.
18499 (get_outgoing_edge_probs): Update.
18500 (conditional_probability): Update.
18501 (emit_case_dispatch_table): Update.
18502 (expand_case): Update.
18503 (expand_sjlj_dispatch_table): Update.
18504 (emit_case_nodes): Update.
18505 * targhooks.c: Update.
18506 * tracer.c (better_p): Update.
18507 (find_best_successor): Update.
18508 * trans-mem.c (expand_transaction): Update.
18509 * tree-call-cdce.c: Update.
18510 * tree-cfg.c (gimple_split_edge): Upate.
18511 (move_sese_region_to_fn): Upate.
18512 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18513 * tree-eh.c (lower_resx): Upate.
18514 (cleanup_empty_eh_move_lp): Upate.
18515 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18516 * tree-inline.c (copy_edges_for_bb): Update.
18517 (copy_cfg_body): Update.
18518 * tree-parloops.c (gen_parallel_loop): Update.
18519 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18520 (gimple_gen_time_profiler): Update.
18521 * tree-ssa-dce.c (remove_dead_stmt): Update.
18522 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18523 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18524 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18525 (unloop_loops): Update.
18526 (try_peel_loop): Update.
18527 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18528 * tree-ssa-loop-split.c (connect_loops): Update.
18529 (split_loop): Update.
18530 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18531 (hoist_guard): Update.
18532 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18533 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18534 (value_replacement): Update.
18535 * tree-ssa-reassoc.c (branch_fixup): Update.
18536 * tree-ssa-tail-merge.c (replace_block_by): Update.
18537 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18538 (create_edge_and_update_destination_phis): Update.
18539 (compute_path_counts): Update.
18540 (recompute_probabilities): Update.
18541 (update_joiner_offpath_counts): Update.
18542 (freqs_to_counts_path): Update.
18543 (duplicate_thread_path): Update.
18544 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18545 (struct switch_conv_info): Update.
18546 (gen_inbound_check): Update.
18547 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18548 (vect_do_peeling): Update.
18549 (vect_loop_versioning): Update.
18550 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18551 (optimize_mask_stores): Update.
18552 * ubsan.c (ubsan_expand_null_ifn): Update.
18553 * value-prof.c (gimple_divmod_fixed_value): Update.
18554 (gimple_divmod_fixed_value_transform): Update.
18555 (gimple_mod_pow2): Update.
18556 (gimple_mod_pow2_value_transform): Update.
18557 (gimple_mod_subtract): Update.
18558 (gimple_mod_subtract_transform): Update.
18559 (gimple_ic): Update.
18560 (gimple_stringop_fixed_value): Update.
18561 (gimple_stringops_transform): Update.
18562 * value-prof.h: Update.
18564 2017-06-29 Carl Love <cel@us.ibm.com>
18566 * config/rs6000/rs6000-c.c: Add support for built-in functions
18567 vector signed int vec_signed (vector float);
18568 vector signed long long vec_signed (vector double);
18569 vector signed int vec_signed2 (vector double, vector double);
18570 vector signed int vec_signede (vector double);
18571 vector signed int vec_signedo (vector double);
18572 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18573 instruction generator.
18574 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18575 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18576 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18578 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18579 vunsignede_v2df): Add define_expands.
18580 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18581 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18582 VEC_UNSIGNEDO): Add definitions.
18583 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18584 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18585 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18586 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18587 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18588 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18589 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18590 vec_unsignede, vec_unsignedo): Add builtin defines.
18591 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18593 * doc/extend.texi: Update the built-in documentation file for the
18594 new built-in functions.
18596 2017-06-29 Richard Biener <rguenther@suse.de>
18598 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18599 reduction chains to LOOP_VINFO_REDUCTIONS.
18600 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18601 SLP reductions after processing reduction chains.
18603 2017-06-29 Nathan Sidwell <nathan@acm.org>
18605 * builtins.c (fold_builtin_FUNCTION): Use
18606 lang_hooks.decl_printable_name.
18608 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18610 PR middle-end/81194
18611 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18612 with only one label.
18613 * stmt.c (expand_case): Assert NCASES is greater than one.
18615 2017-06-29 Richard Biener <rguenther@suse.de>
18617 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18619 (group_case_labels): Likewise.
18620 (find_taken_edge): Push sanity checking on val to workers...
18621 (find_taken_edge_cond_expr): ... here
18622 (find_taken_edge_switch_expr): ... and here, handle cases
18623 with just a default label.
18624 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18625 (group_case_labels): Likewise.
18626 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18627 group_case_labels does anything cleanup the CFG again.
18629 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18631 PR tree-optimization/81196
18632 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18633 exit condition comparing two IVs.
18635 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18637 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18638 profile to the dummy entry at the end of the list of architectures.
18639 * config/arm/arm-cpu-cdata.h: Regenerated.
18641 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18642 Michael Collison <michael.collison@arm.com>
18645 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18647 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18648 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18649 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18650 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18651 of shift when the shift amount is masked with constant equal to
18652 the size of the mode.
18653 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18656 2017-06-29 Martin Liska <mliska@suse.cz>
18658 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18660 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18663 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18666 (edge_growth_cache_entry::edge_growth_cache_entry): New
18668 (reset_edge_growth_cache): Update to use constructor.
18670 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18672 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18673 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18674 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18676 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18678 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18679 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18681 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18683 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18684 (*-linux-uclibc*): Add t-uclibc tmake_file.
18685 * config/t-musl: New.
18686 * config/t-uclibc: New.
18688 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18690 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18692 (gen_comm_data): Emit architectural setting of arch_prof.
18693 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18695 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18696 (armv8-m.base, armv8-m.main): Likewise.
18697 * arm-protos.h (arm_build_target): Add profile field.
18698 (arch_option): Likewise.
18699 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18701 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18702 arm_active_target.profile.
18704 2017-06-28 Richard Biener <rguenther@suse.de>
18706 PR middle-end/81227
18707 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18708 TYPE_OVERFLOW_WRAPS.
18709 * match.pd (negate_expr_p): Likewise.
18710 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18711 fold_build2, not fold_binary.
18713 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18715 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18716 Convert memory address to Pmode.
18717 (aarch64_print_operand): Assert MEM operands are always Pmode.
18719 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18722 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18723 Remove redundant if.
18724 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18725 * config/arm/aarch-common-protos.h
18726 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18727 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18729 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18732 * multiple_target.c (create_dispatcher_calls): Set the default
18733 clone to be static, not public.
18735 2017-06-28 Richard Biener <rguenther@suse.de>
18737 * tree-vect-loop.c (vectorizable_reduction): Move special
18738 cond reduction IV var creation ...
18739 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18740 parameter. Use STMT_VINFO_VECTYPE.
18741 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18744 2017-06-28 Martin Liska <mliska@suse.cz>
18747 * ipa-visibility.c (non_local_p): Handle visibility.
18749 2017-06-28 Martin Liska <mliska@suse.cz>
18752 * common.opt: Add IntegerRange to various options.
18753 * opt-functions.awk (integer_range_info): New function.
18754 * optc-gen.awk: Add integer_range_info to cl_options struct.
18755 * opts-common.c (decode_cmdline_option): Handle
18756 CL_ERR_INT_RANGE_ARG.
18757 (cmdline_handle_error): Likewise.
18758 * opts.c (print_filtered_help): Show valid interval in
18759 when --help is provided.
18760 * opts.h (struct cl_option): Add range_min and range_max fields.
18761 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18763 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18765 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18766 (x * C EQ/NE y * C): New transformation.
18768 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18770 * genmultilib (combination_space): Accept '+' in option names.
18772 2017-06-28 Martin Liska <mliska@suse.cz>
18775 * asan.c (instrument_derefs): Bail out inner references
18776 that are hard register variables.
18778 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18781 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18782 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18784 2017-06-28 Richard Biener <rguenther@suse.de>
18786 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18787 (vect_get_slp_defs): Adjust.
18788 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18790 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18792 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18793 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18794 (vectorizable_reduction): Adjust.
18795 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18797 (vect_get_slp_defs): Likewise.
18798 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18799 (vectorizable_bswap): Adjust.
18800 (vectorizable_call): Likewise.
18801 (vectorizable_conversion): Likewise.
18802 (vectorizable_assignment): Likewise.
18803 (vectorizable_shift): Likewise.
18804 (vectorizable_operation): Likewise.
18805 (vectorizable_store): Likewise.
18806 (vectorizable_condition): Likewise.
18807 (vectorizable_comparison): Likewise.
18809 2017-06-28 Michael Collison <michael.collison@arm.com>
18812 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18814 (arm_gen_movmemqi): Removed unused variable 'i'.
18815 Convert 'for' loop into 'while' loop.
18816 (arm_expand_prologue): Remove last unnecessary set of insn.
18817 (thumb_pop): Remove unused variable 'pushed_words'.
18818 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18820 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18822 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18823 * config/s390/s390-protos.h: Add prototype of
18824 s390_rel_address_ok_p.
18825 * config/s390/s390.c (s390_got_symbol): New function.
18826 (s390_rel_address_ok_p): New function.
18827 (legitimize_pic_address): Use s390_rel_address_ok_p.
18828 (s390_load_got): Use s390_got_symbol.
18829 (s390_option_override): Issue error if
18830 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18831 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18833 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18835 2017-06-27 Andrew Pinski <apinski@cavium.com>
18837 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18838 (X * copysign (1.0, X)): New pattern.
18839 (X * copysign (1.0, -X)): New pattern.
18840 (copysign (-1.0, CST)): New pattern.
18842 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18844 * genmultilib (combination_space): Remove variable.
18845 Validate reuse rules against regular expression for any sequence
18846 of multilib options in any order.
18848 2017-06-27 Michael Collison <michael.collison@arm.com>
18850 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18851 call aarch64_split_simd_combine.
18852 * (aarch64_combine_internal<mode>): Delete pattern.
18853 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18854 Allow register and subreg operands.
18856 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18858 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18859 specific need, just fallback on defaults.
18860 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18862 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18863 Olivier Hainque <hainque@adacore.com>
18865 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18867 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18868 targets. Pick a default if no particular attempt applied.
18869 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18872 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18874 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18875 (x86_64-wrs-vxworks7): Likewise.
18877 2017-06-27 Marek Polacek <polacek@redhat.com>
18880 * ubsan.c (instrument_null): Check get_base_address's result for null.
18882 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18884 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18886 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18888 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18889 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18890 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18891 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18892 New function types.
18893 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18894 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18895 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18896 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18897 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18898 BUILT_IN_FEUPDATEENV): New builtins.
18899 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18900 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18901 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18902 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18904 (builtin_structptr_types): Adjust size.
18905 * tree.c (builtin_structptr_types): Add four entries.
18907 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18908 Olivier Hainque <hainque@adacore.com>
18910 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18911 (TLS_SYM): New local macro, forcing reference to __tls__ on
18912 link command lines for VxWorks 7 RTPs, triggering initialization
18914 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18915 OS features TLS support, true for RTPs on VxWorks 7.
18916 * config/vxworks.c (vxworks_override_options): Setup emutls
18919 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18921 * predict.c (test_prediction_value_range): Use -1U instead of -1
18922 to avoid narrowing conversion warning.
18923 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18924 to avoid narrowing conversion warning.
18925 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18927 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18929 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18931 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18932 64bit configurations.
18933 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18934 (SIZE_TYPE): Likewise.
18935 * config/vxworks.c (vxworks_emutls_var_fields): Use
18936 long_unsigned_type_node instead of unsigned_type_node as the offset
18937 field type, which is "pointer" mode in emutls.c.
18939 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18942 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18944 PR middle-end/81207
18945 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18946 gimple_vuse copying separately from gimple_vdef copying.
18948 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18950 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18953 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18954 Olivier Hainque <hainque@adacore.com>
18956 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18957 for all vxworks7 targets.
18958 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18959 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18960 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18961 variations for VX6/VX7 and 32/64bits later on in ...
18962 (VXWORKS_LIB_SPEC): Leverage new macros.
18963 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18964 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18966 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18968 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18969 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18971 2017-06-26 Carl Love <cel@us.ibm.com>
18973 * config/rs6000/rs6000-c.c: Add support for built-in functions
18974 vector bool char vec_reve (vector bool char);
18975 vector signed char vec_reve (vector signed char);
18976 vector unsigned char vec_reve (vector unsigned char);
18977 vector bool int vec_reve (vector bool int);
18978 vector signed int vec_reve (vector signed int);
18979 vector unsigned int vec_reve (vector unsigned int);
18980 vector bool long long vec_reve (vector bool long long);
18981 vector signed long long vec_reve (vector signed long long);
18982 vector unsigned long long vec_reve (vector unsigned long long);
18983 vector bool short vec_reve (vector bool short);
18984 vector signed short vec_reve (vector signed short);
18985 vector double vec_reve (vector double);
18986 vector float vec_reve (vector float);
18987 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18988 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18989 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18990 (altivec_vreve): New pattern.
18991 * config/rs6000/altivec.h (vec_reve): New define.
18992 * doc/extend.texi (vec_rev): Update the built-in documentation file
18993 for the new built-in functions.
18995 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18997 PR tree-optimization/71815
18998 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
19000 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
19002 (slsr_process_phi): Likewise.
19003 (replace_uncond_cands_and_profitable_phis): Don't replace a
19004 multiply candidate with a stride of 1 (copy or cast).
19005 (phi_incr_cost): Call uses_consumed_by_stmt rather than
19007 (lowest_cost_path): Likewise.
19008 (total_savings): Likewise.
19010 2017-06-26 Richard Biener <rguenther@suse.de>
19013 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
19014 Use def_builtin_pure for all gather builtins.
19016 2017-06-26 Richard Biener <rguenther@suse.de>
19018 PR tree-optimization/81203
19019 * tree-tailcall.c (find_tail_calls): Do not move stmts into
19020 non-dominating BBs.
19022 2017-06-26 Marek Polacek <polacek@redhat.com>
19025 * doc/invoke.texi: Document -Wmultistatement-macros.
19027 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19029 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19030 arm_neon_ok_no_float_abi effective target.
19032 2017-06-26 Richard Biener <rguenther@suse.de>
19034 PR tree-optimization/80928
19035 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19036 (copy_bbs): Set BB_DUPLICATED flag early.
19037 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19039 (execute_on_shrinking_pred): Likewise.
19040 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19041 BB_DUPLICATED blocks.
19042 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19043 iterate over all PHIs considering removal of *gsi.
19045 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19047 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19050 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19052 * config/rs6000/rs6000-string.c: (expand_block_clear,
19053 do_load_for_compare, select_block_compare_mode,
19054 compute_current_alignment, expand_block_compare,
19055 expand_strncmp_align_check, expand_strn_compare,
19056 expand_block_move, rs6000_output_load_multiple)
19057 Move functions related to string/block move/compare
19058 to a separate file.
19059 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19060 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19061 for this function which is now used in two files.
19062 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19063 * config.gcc: Add rs6000-string.o to extra_objs for
19064 targets powerpc*-*-* and rs6000*-*-*.
19066 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19069 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19070 32-bit, since indexed is not valid for DImode.
19071 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19072 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19073 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19074 (define_peephole2 for Altivec d-form store): Likewise.
19077 * multiple_target.c (create_dispatcher_calls): Only create the
19078 dispatcher call if the function is the default clone of a
19079 versioned function.
19081 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19083 PR middle-end/80902
19084 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19085 a call, force the call to not be a tail call.
19087 2017-06-23 Jeff Law <law@redhat.com>
19089 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19092 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19094 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19095 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19096 handling for early expansion of vector shifts (sl,sr,sra,rl).
19097 (builtin_function_type): Add vector shift right instructions
19098 to the unsigned argument list.
19100 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19102 rtl-optimizatoin/79286
19103 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19104 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19105 trap. PIC register plus a const unspec without offset can never trap.
19107 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19109 * tree.h (builtin_structptr_type): New type.
19110 (builtin_structptr_types): Declare new array.
19111 * tree.c (builtin_structptr_types): New array.
19112 (free_lang_data, build_common_tree_nodes): Use it.
19114 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19117 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19120 2017-06-22 Matt Turner <mattst88@gmail.com>
19122 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19123 Lake models to skylake case. Assume skylake for unknown
19124 models with clflushopt.
19126 2017-06-22 Jeff Law <law@redhat.com>
19128 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19129 frame sizes that do not satisfy aarch64_uimm12_shift.
19131 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19133 * profile-count.h (apply_probability,
19134 apply_scale, probability_in): Fix checks for zero.
19136 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19138 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19139 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19141 2016-06-22 Richard Biener <rguenther@suse.de>
19143 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19144 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19145 REDUC_MAX_EXPR support.
19146 (vectorizable_reduction): Likewise.
19147 (vect_create_epilog_for_reduction): Likewise.
19149 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19151 * match.pd (A / (1 << B) -> A >> B): New.
19152 * generic-match-head.c: Include optabs-tree.h.
19153 * gimple-match-head.c: Likewise.
19154 * optabs-tree.h (target_supports_op_p): New.
19155 * optabs-tree.c (target_supports_op_p): New.
19157 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19159 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19160 $gcc_cv_ld --help output.
19161 (gcc_cv_ld_demangle): Likewise.
19162 (gcc_cv_ld_eh_frame_hdr): Likewise.
19163 (gcc_cv_ld_pie): Likewise.
19164 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19165 (gcc_cv_ld_buildid): Likewise.
19166 (gcc_cv_ld_sysroot): Likewise.
19167 (ld_bndplt_support): Likewise.
19168 (ld_pushpopstate_support): Likewise.
19169 * configure: Regenerate.
19170 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19172 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19175 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19176 operands indexes to avoid gap between operands and match_dups.
19178 2017-06-21 Andrew Pinski <apinski@cavium.com>
19180 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19181 Increment Arith_shift and Arith_shift_reg by 1.
19182 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19184 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19185 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19186 (aarch64_strip_extend): Add new argument and test for it.
19187 (aarch64_cheap_mult_shift_p): New function.
19188 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19189 add a cost if it is true.
19190 Update calls to aarch64_strip_extend.
19191 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19193 2017-06-21 Andrew Pinski <apinski@cavium.com>
19195 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19197 (thunderxt88): Likewise.
19198 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19199 (thunderx_prefetch_tune): New variable.
19200 (thunderx2t99_prefetch_tune): Update for the correct values.
19201 (thunderxt88_tunings): New variable.
19202 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19203 generic_prefetch_tune.
19204 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19206 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19208 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19209 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19210 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19211 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19212 (aarch64_atomic_cas<mode>, GPI): Likewise.
19214 2017-06-21 Martin Liska <mliska@suse.cz>
19216 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19217 statements on cold and hot labels.
19218 * predict.c (tree_estimate_probability_bb): Remove the
19219 prediction from this place.
19221 2017-06-21 Martin Liska <mliska@suse.cz>
19223 PR tree-optimization/79489
19224 * gimplify.c (maybe_add_early_return_predict_stmt): New
19226 (gimplify_return_expr): Call the function.
19227 * predict.c (tree_estimate_probability_bb): Remove handling
19229 * predict.def: Update comment about early return predictor.
19230 * gimple-predict.h (is_gimple_predict): New function.
19231 * predict.def: Change default value of early return to 66.
19232 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19234 * passes.def: Put pass_strip_predict_hints to the beginning of
19237 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19239 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19240 FUNCTION_DECL declarations.
19241 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19243 (dwaf2out_decl): Likewise.
19244 * godump.c (go_early_global_decl): Skip call to the real debug hook
19245 for FUNCTION_DECL declarations.
19246 * passes.c (rest_of_decl_compilation): Skip call to the
19247 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19248 -fdump-go-spec is passed.
19250 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19252 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19253 Reorder for compactness.
19254 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19255 (def_builtin_pure, def_builtin_pure2): New functions.
19256 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19258 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19260 * match.pd (nop_convert): New predicate.
19261 ((A +- CST1) +- CST2): Allow some NOP conversions.
19263 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19266 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19267 with ctors/dtors if GOVD_SHARED is set.
19269 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19271 * config/aarch64/aarch64.md (movti_aarch64):
19272 Emit mov rather than orr.
19273 (movtf_aarch64): Likewise.
19274 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19275 Emit mov rather than orr.
19277 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19279 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19280 Swap alternatives, make integer dup more expensive.
19282 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19284 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19285 Return true for non-tls symbols.
19287 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19289 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19290 (cortex-a75): Likewise.
19291 (cortex-a75.cortex-a55): Likewise.
19292 * config/aarch64/aarch64-tune.md: Regenerate.
19293 * doc/invoke.texi (-mtune): Document new values for -mtune.
19295 2017-06-21 Tom de Vries <tom@codesourcery.com>
19297 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19298 stack_size feature.
19299 (Effective-Target Keywords, Other attributes): Suggest using
19300 dg-add-options stack_size feature to get stack limit in stack_size
19301 effective target documentation.
19303 2017-06-21 Julian Brown <julian@codesourcery.com>
19304 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19306 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19307 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19308 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19310 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19311 attribute type list for neon_multiply.
19312 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19313 attribute type list for neon_multiply.
19314 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19315 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19316 attribute type list for neon_multiply.
19317 * config/arm/types.md (crypto_pmull): Add.
19318 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19319 attribute type list.
19321 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19323 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19326 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19328 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19329 to make sure not to dereference a NULL cost_classes_ptr pointer.
19331 2017-06-20 Carl Love <cel@us.ibm.com>
19333 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19334 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19335 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19336 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19337 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19338 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19339 VMULOSW): New enum "unspec" values.
19340 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19341 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19342 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19343 altivec_vmulosw): New patterns.
19344 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19345 VMULOSW): Add definitions.
19347 2017-06-20 Julia Koval <julia.koval@intel.com>
19349 * config/i386/i386.c: Fix rounding expand for new pattern.
19350 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19352 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19354 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19355 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19357 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19359 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19362 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19364 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19365 architecture, then by implementer ID.
19366 * config/aarch64/aarch64-tune.md: Regenerate.
19368 2017-06-20 Richard Biener <rguenther@suse.de>
19370 PR middle-end/81097
19371 * fold-const.c (split_tree): Fold to type before negating.
19373 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19375 * diagnostic-show-locus.c
19376 (selftest::test_fixit_deletion_affecting_newline): New function.
19377 (selftest::diagnostic_show_locus_c_tests): Call it.
19379 2017-06-20 Andreas Schwab <schwab@suse.de>
19382 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19385 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19387 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19388 __ARM_FEATURE_COPROC according to support.
19390 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19392 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19393 Rewritten to avoid overflow for > 32-bit pointers.
19396 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19397 by removing enum keyword.
19398 (ubsan_type_descriptor): Likewise. Formatting fix.
19401 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19402 splitter): Require TARGET_SSE2 in the condition.
19404 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19407 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19408 for doing vector set of SFmode on ISA 3.0.
19409 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19410 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19412 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19413 SFmode value into a V4SF variable that was extracted from another
19414 V4SF variable without converting the element to double precision
19415 and back to single precision vector format.
19416 (vsx_insert_extract_v4sf_p9_2): Likewise.
19418 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19420 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19421 in UWHI to avoid undefined overflow.
19424 * ubsan.h (enum ubsan_encode_value_phase): New.
19425 (ubsan_encode_value): Change second argument to
19426 enum ubsan_encode_value_phase with default value of
19427 UBSAN_ENCODE_VALUE_GENERIC.
19428 * ubsan.c (ubsan_encode_value): Change second argument to
19429 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19430 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19431 create_tmp_var_raw instead of create_tmp_var and use a
19433 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19434 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19435 ubsan_encode_value callers.
19438 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19439 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19440 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19442 2017-06-19 Richard Biener <rguenther@suse.de>
19444 PR middle-end/81118
19445 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19446 estimates if we changed anything.
19448 2017-06-19 Richard Biener <rguenther@suse.de>
19450 PR tree-optimization/80887
19451 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19452 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19453 simplified lookups, then reset mprts_hook.
19454 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19456 (try_to_simplify): Likewise.
19458 2017-06-19 Martin Liska <mliska@suse.cz>
19461 * gimplify.c (gimplify_switch_expr):
19462 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19464 2017-06-19 Martin Liska <mliska@suse.cz>
19466 * doc/install.texi: Document that PGO runs in 4 stages.
19468 2017-06-19 Martin Liska <mliska@suse.cz>
19471 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19472 to dispatcher function name.
19473 * multiple_target.c (replace_function_decl): New function.
19474 (create_dispatcher_calls): Redirect both edges and references.
19476 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19478 * profile-count.c (profile_count::dump): Dump quality.
19479 (profile_count::differs_from_p): Update for unsigned val.
19480 * profile-count.h (profile_count_quality): New enum.
19481 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19483 2017-06-19 Richard Biener <rguenther@suse.de>
19485 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19486 struct function as arg.
19487 (estimate_numbers_of_iterations): Export overload with loop arg.
19488 (free_numbers_of_iterations_estimates_loop): Use an overload of
19489 free_numbers_of_iterations_estimates instead.
19490 * tree-cfg.c (remove_bb): Adjust.
19491 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19492 * tree-parloops.c (gen_parallel_loop): Likewise.
19493 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19495 (tree_unroll_loops_completely): Likewise.
19496 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19497 Use an overload instead and export.
19498 (estimated_loop_iterations): Adjust.
19499 (max_loop_iterations): Likewise.
19500 (likely_max_loop_iterations): Likewise.
19501 (estimate_numbers_of_iterations): Take struct function as arg
19503 (loop_exits_before_overflow): Adjust.
19504 (free_numbers_of_iterations_estimates_loop): Use an overload.
19505 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19506 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19508 2017-06-19 Richard Biener <rguenther@suse.de>
19511 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19512 RANGE_EXPR conservatively.
19514 2017-06-16 Carl Love <cel@us.ibm.com>
19516 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19517 definitions for vec_float, vec_float2, vec_floato,
19518 vec_floate built-ins.
19519 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19520 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19522 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19523 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19524 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19525 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19526 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19527 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19528 vec_floato): Add builtin defines.
19529 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19530 Update the built-in documentation file for the new built-in
19533 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19535 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19536 (mthumb): Mark as the negative of -marm.
19538 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19540 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19542 (ARM Options, -mtune): Document that this accepts the same
19543 extension options as -mcpu.
19544 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19546 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19548 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19549 permitted extensions.
19551 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19553 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19554 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19555 (armv8-m.main): Add option +nodsp.
19556 * config/arm/arm-cpu-cdata.h: Regenerated.
19558 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19560 * config/arm/t-fuchsia: New file.
19561 * config.gcc (arm*-*-fuchsia*): Use it.
19563 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19565 * config/arm/t-symbian: Rewrite for new option infrastructure.
19567 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19569 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19570 (MULTILIB_REQUIRED): Likewise.
19572 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19574 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19575 (MULTILIB_RESUE): Likewise.
19576 (MULTILIB_MATCHES): Likewise.
19577 (MULTLIB_REQUIRED): Likewise.
19579 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19581 * config/arm/t-rtems: Rewrite for new option framework.
19583 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19585 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19586 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19587 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19588 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19589 * config/arm/t-multilib: ... here.
19590 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19591 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19592 armv7-a and armv8*-a when A-profile libraries have not been built.
19593 * config/arm/t-rmprofile: Rewrite.
19595 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19597 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19598 with a backslash. Remove the backslash after substituting unescaped
19600 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19602 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19604 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19605 the driver to pass through the default setting of -mfloat-abi.
19606 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19608 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19609 (all_feat_combs): New rule.
19610 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19612 * config/arm/t-aprofile: Rewrite.
19614 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19616 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19617 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19618 fpu is not specified by the user/command-line.
19619 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19620 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19621 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19622 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19623 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19624 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19626 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19628 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19629 * config/arm/t-arm-elf: Rewritten.
19631 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19633 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19634 have some floating-point instructions.
19635 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19636 (TARGET_MAYBE_HARD_FLOAT): New macro.
19637 * config/arm/arm-builtins.c (arm_init_builtins): Use
19638 TARGET_MAYBE_HARD_FLOAT.
19639 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19641 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19643 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19644 (configargs.h): Include it.
19645 (arm_print_hint_for_fpu_option): New function.
19646 (arm_parse_fpu_option): New function.
19647 (candidate_extension): New class.
19648 (arm_canon_for_multilib): New function.
19649 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19650 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19651 (ARCH_CANONICAL_SPECS): New macro.
19652 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19654 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19656 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19657 are set after handling multilib fragments. Set target_cpu_default2
19660 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19662 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19664 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19665 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19667 (check_arch): Likewise.
19668 * config/arm/arm.c (arm_configure_build_target): Handle
19669 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19670 options in the default.
19672 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19674 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19675 when an option is an alias of another.
19676 * config/arm/parsecpu.awk (optalias): New parser token.
19677 (gen_comm_data): Mark non-alias options as such. Emit entries
19678 for extension aliases.
19679 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19680 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19681 (armv6kz, armv6zk, armv6t2): Likewise.
19682 (armv7): Make vfpv3-d16 an alias.
19683 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19685 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19686 Sort in canonical order.
19687 (armv8-a): Sort in canonical order.
19688 (armv8.1-a, armv8.2-a): Likewise.
19689 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19691 (cortex-a9): Sort in canonical order.
19692 * config/arm/arm.c (selftests.h): Include it.
19693 (arm_test_cpu_arch_data): New function.
19694 (arm_run_self_tests): New function.
19695 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19696 (targetm): Move declaration to the end of the file.
19697 * arm-cpu-cdata.h: Regenerated.
19699 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19701 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19702 call to target_mode_check describing the type of option passed.
19703 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19704 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19705 arm_parse_cpu_option_name to match parameters against list of
19707 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19708 arm_arch_core_flags data structure.
19709 * config/arm/arm-cpu_cdata.h: Regenerated.
19711 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19713 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19715 (arm_print_hint_for_cpu_option): Likewise.
19716 (arm_print_hint_for_arch_option): Likewise.
19717 (arm_parse_cpu_option_name): Likewise.
19718 (arm_parse_arch_option_name): Likewise.
19719 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19720 of entries in the all_fpus list.
19721 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19722 (arm_parse_cpu_option_name): Declare.
19723 (arm_parse_arch_option_name): Declare.
19724 (arm_parse_option_features): Declare.
19725 (arm_intialize_isa): Declare.
19726 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19728 (gen_comm_data): ... here. Make definitions non-static.
19729 * config/arm/arm-cpu-data.h: Regenerated.
19730 * config/arm/arm-cpu-cdata.h: Regenerated.
19732 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19734 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19735 (cpu_arch_extension): New structure.
19736 (cpu_arch_option, arch_option, cpu_option): New structures.
19737 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19738 architecture types.
19739 (gen_data): Generate new format data tables.
19740 * config/arm/arm.c (cpu_tune): New structure.
19741 (cpu_option, processors): Delete.
19742 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19743 (arm_print_hint_for_cpu_option): ... this and ...
19744 (arm_print_hint_for_arch_option): ... this.
19745 (arm_parse_arch_cpu_name): Delete. Replace with ...
19746 (arm_parse_cpu_option_name): ... this and ...
19747 (arm_parse_arch_option_name): ... this.
19748 (arm_unrecognized_feature): Change type of target parameter to
19750 (arm_parse_arch_cpu_features): Delete. Replace with ...
19751 (arm_parse_option_features): ... this.
19752 (arm_configure_build_target): Rework to use new configuration data
19754 (arm_print_tune_info): Rework for new configuration data tables.
19755 * config/arm/arm-cpu-data.h: Regenerated.
19756 * config/arm/arm-cpu.h: Regenerated.
19758 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19760 * Makefile.in (OBJS): Move sbitmap.o from here ...
19761 (OBJS-libcommon): ... to here.
19763 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19765 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19766 (ISA_ALL_CRYPTO): New macro.
19767 (ISA_ALL_SIMD): New macro
19768 (ISA_ALL_FP): New macro.
19769 * config/arm/arm.c (fpu_bitlist): Update initializer.
19770 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19772 (arm9e): Add fpu. Add option for nofp
19773 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19774 (arm926ej-s, arm1026ej-s): Likewise.
19775 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19776 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19777 neon-fp16, neon-vfpv4, nofp and nosimd.
19778 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19779 (cortex-a8): Add fpu. Add option for nofp.
19780 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19781 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19782 (cortex-r4f): Add fpu.
19783 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19784 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19786 (cortex-r8): Likewise.
19787 (cortex-m4): Add fpu. Add option for nofp.
19788 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19789 (cortex-a17.cortex-a7): Likewise.
19790 (cortex-a32): Add fpu. Add options for crypto and nofp.
19791 (cortex-a35, cortex-a53): Likewise.
19792 (cortex-a57): Add fpu. Add option for crypto.
19793 (cortex-a72, cortex-a73): Likewise.
19794 (exynos-m1): Likewise.
19795 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19796 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19797 (cortex-m33): Add fpu. Add option for nofp.
19798 * config/arm/arm-cpu-cdata.h: Regenerated
19799 * config/arm/arm-cpu-data.h: Regenerated.
19801 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19803 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19804 (armv5te, armv5tej): Likewise.
19805 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19806 (armv7): Add options fp and vfpv3-d16.
19807 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19808 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19810 (armv7ve): Likewise.
19811 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19812 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19813 (armv8-a): Add nocrypto option.
19814 (armv8.1-a, armv8.2-a): Likewise.
19815 (armv8-m.main): add options fp, fp.dp and nofp.
19817 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19819 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19821 (armv8-a+crc): Delete.
19822 (armv8.1-a): Add options simd, crypto and nofp.
19823 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19824 (armv8.2-a+fp16): Delete.
19825 (armv8-m.main): Add option dsp.
19826 (armv8-m.main+dsp): Delete.
19827 (cortex-a8): Add fpu. Add nofp option.
19828 (cortex-a9): Add fpu. Add nofp and nosimd options.
19829 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19830 link to main cpu and architecture data structures.
19831 (gen_comm_data): Only put isa attributes from the main architecture
19833 (option): New statement for architecture and CPU entries.
19834 * arm.c (struct cpu_option): New structure.
19835 (struct processors): Add entry for options.
19836 (arm_unrecognized_feature): New function.
19837 (arm_parse_arch_cpu_name): Ignore any characters after the first
19839 (arm_parse_arch_cpu_feature): New function.
19840 (arm_configure_build_target): Separate out any CPU and architecture
19841 features and parse separately. Don't error out if -mfpu=auto is
19842 used with only an architecture string.
19843 (arm_print_asm_arch_directives): New function.
19844 (arm_file_start): Call it.
19845 * config/arm/arm-cpu-cdata.h: Regenerated.
19846 * config/arm/arm-cpu-data.h: Likewise.
19847 * config/arm/arm-tables.opt: Likewise.
19849 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19851 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19852 assembler when it is not -mfpu=auto.
19854 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19856 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19857 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19858 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19859 (ASM_CPU_SPEC): Rewrite.
19860 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19861 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19862 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19863 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19864 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19865 copied string is NUL-terminated. Also strip any characters prefixed
19867 (arm_rewrite_selected_arch): New function.
19868 (arm_rewrite_march): New function.
19870 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19872 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19873 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19874 (march, mcpu, mtune): Convert to string-based options.
19875 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19876 (arm_parse_arch_cpu_name): New function.
19877 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19878 identify selected architecture or CPU.
19879 (arm_option_save): New function.
19880 (TARGET_OPTION_SAVE): Redefine.
19881 (arm_option_restore): Restore string options.
19882 (arm_option_print): Print string options.
19884 2017-06-16 Martin Sebor <msebor@redhat.com>
19886 PR tree-optimization/80933
19887 PR tree-optimization/80934
19888 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19889 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19890 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19891 (gimple_fold_builtin): Call them.
19893 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19895 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19896 as unlikely; update profile.
19898 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19900 * predict.c (force_edge_cold): Handle declaring edges impossible
19903 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19905 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19907 (try_unroll_loop_completely): Fix reporting.
19909 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19911 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19913 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19916 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19917 if given a non-constant argument for an intrinsic which requires a
19920 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19922 * profile.c (compare_freqs): New function.
19923 (branch_prob): Sort edge list.
19924 (find_spanning_tree): Assume that the list is priority sorted.
19926 2017-06-16 Richard Biener <rguenther@suse.de>
19928 PR tree-optimization/81090
19929 * passes.def (pass_record_bounds): Remove.
19930 * tree-pass.h (make_pass_record_bounds): Likewise.
19931 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19932 make_pass_record_bounds): Likewise.
19933 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19934 not free niter estimates at the beginning but at the end.
19935 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19937 2017-06-16 Richard Biener <rguenther@suse.de>
19939 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19940 initializer to workaround ICE in host GCC 4.8.
19942 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19944 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19946 (clone_inlined_nodes): Update.
19948 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19950 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19951 prefetch settings, and enable prefetching by default at -O3.
19953 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19955 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19956 Set flag_prefetch_loop_arrays according to tuning data.
19958 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19960 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19961 New tune structure.
19962 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19963 [Unrelated to main purpose of the patch] Place the pointer field last
19964 to enable type checking errors when tune structure are wrongly merged.
19965 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19966 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19967 (thunderx2t99_prefetch_tune): New tune constants.
19968 (tune_params *_tunings): Update all tunings (no functional change).
19969 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19970 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19971 from tunings structures.
19973 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19976 * ubsan.c (instrument_null): Add T argument, use it instead
19977 of computing it based on IS_LHS.
19978 (instrument_object_size): Likewise.
19979 (pass_ubsan::execute): Adjust instrument_null and
19980 instrument_object_size callers to pass gimple_get_lhs or
19981 gimple_assign_rhs1 result to it. Use instrument_null instead of
19982 calling get_base_address and instrument_mem_ref. Handle
19983 aggregate call arguments for object-size sanitization.
19985 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19987 PR tree-optimization/81089
19988 * tree-vrp.c (is_masked_range_test): Validate operands of
19991 2017-06-15 Martin Sebor <msebor@redhat.com>
19994 * dumpfile.c (dump_register): Avoid calling memset to initialize
19995 a class with a default ctor.
19996 * gcc.c (struct compiler): Remove const qualification.
19997 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19998 * hash-table.h: Ditto.
19999 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
20001 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
20002 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
20004 * params.h (struct param_info): Make struct members non-const.
20005 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
20006 with default initialization.
20007 * vec.h (vec_copy_construct, vec_default_construct): New helper
20009 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
20010 with vec_copy_construct.
20011 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
20012 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
20013 * doc/invoke.texi (-Wclass-memaccess): Document.
20015 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
20017 * emit-rtl.h (is_leaf): Update comment about local
20018 register allocator.
20020 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20023 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
20024 for a variable to have a section before checking if the section has a
20026 Set section to.persistent if persistent attribute is set.
20027 Warn if .persistent attribute is used on an automatic variable.
20029 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20031 PR rtl-optimization/80474
20032 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20034 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20036 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20039 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20041 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20042 (casesi): Emit bounds checking as RTL.
20043 (casesi_internal_mips16_<mode>): Remove bounds checking.
20045 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20047 * config/xtensa/xtensa.c (xtensa_option_override): Append
20048 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20049 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20050 xtensa_doloop_hooks): Define unconditionally.
20051 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20053 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20054 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20055 for it in xtensa_option_override.
20056 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20057 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20059 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20061 * doc/cppopts.texi: Document '-' special value to -MF.
20063 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20065 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20066 (cortex_a53_fconst): Likewise.
20067 (cortex_a53_fpmul): Likewise.
20068 (cortex_a53_f_load_64): Likewise.
20069 (cortex_a53_f_load_many): Likewise.
20070 (cortex_a53_advsimd_alu): Likewise.
20071 (cortex_a53_advsimd_alu_q): Likewise.
20072 (cortex_a53_advsimd_mul): Likewise.
20073 (cortex_a53_advsimd_mul_q): Likewise.
20074 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20075 Add missing fmul, r2f_cvt and fconst cases.
20077 2017-06-14 Richard Biener <rguenther@suse.de>
20079 PR middle-end/81088
20080 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20082 (fold_binary_loc): When associating do not treat pre-existing
20083 TREE_OVERFLOW on literal constants as a reason to allow
20084 TREE_OVERFLOW on associated literal constants.
20086 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20088 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20089 (MASK_FEATURES): New macro.
20090 * config/sparc/sparc.c (sparc_option_override): Remove the special
20091 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20093 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20095 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20096 a division of 0 if non-call exceptions are enabled.
20098 2017-06-14 Andrew Pinski <apinski@cavium.com>
20099 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20102 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20103 Improve vector initialization code gen for only variable case.
20105 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20107 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20109 2017-06-14 Richard Biener <rguenther@suse.de>
20111 PR tree-optimization/81083
20112 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20115 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20117 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20118 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20119 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20120 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20121 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20122 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20124 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20126 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20127 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20129 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20131 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20133 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20135 * config/rs6000/t-rtems: Don't handle SPE.
20137 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20139 * config/rs6000/t-linux: Don't handle SPE.
20141 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20143 * config/rs6000/eabispe.h: Delete file.
20145 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20147 * config/rs6000/t-spe: Delete file.
20149 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20151 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20152 (rs6000_legitimate_offset_address_p): Return false for anything in
20153 V2SImode or V2SFmode.
20155 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20157 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20158 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20159 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20161 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20162 (rs6000_legitimate_offset_address_p): Ditto.
20163 (rs6000_emit_move): Ditto.
20164 (rs6000_init_builtins): Remove V4HI_type_node.
20166 2017-06-13 Martin Liska <mliska@suse.cz>
20169 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20170 (gate_asan): Likewise.
20171 * asan.h (asan_no_sanitize_address_p): Remove the function.
20172 (sanitize_flags_p): New function.
20173 * builtins.def: Fix coding style.
20174 * common.opt: Use renamed enum value.
20175 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20176 * doc/extend.texi: Document no_sanitize attribute.
20177 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20178 to SANITIZE_UNDEFINED_NONDEFAULT.
20179 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20180 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20181 Use sanitize_flags_p.
20182 * gimplify.c (gimplify_function_tree): Likewise.
20183 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20184 * opts.c (parse_no_sanitize_attribute): New function.
20185 (common_handle_option): Use renamed enum value.
20186 * opts.h (parse_no_sanitize_attribute): Declare.
20187 * tree.c (sanitize_flags_p): New function.
20188 * tree.h: Declared here.
20189 * tsan.c: Use sanitize_flags_p.
20190 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20191 (instrument_mem_ref): Likewise.
20192 (instrument_bool_enum_load): Likewise.
20193 (do_ubsan_in_current_function): Remove the function.
20194 (pass_ubsan::execute): Use sanitize_flags_p.
20195 * ubsan.h: Remove do_ubsan_in_current_function
20196 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20197 (dump_function_to_file): Use it here.
20199 2017-06-13 Martin Jambor <mjambor@suse.cz>
20201 PR tree-optimization/80803
20202 PR tree-optimization/81063
20203 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20204 (propagate_subaccesses_across_link): Enqueue subtree whenever
20205 necessary instead of relying on the caller.
20207 2017-06-13 Martin Jambor <mjambor@suse.cz>
20209 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20210 that have a first_link.
20211 (sort_and_splice_var_accesses): Do not check first_link before
20213 (subtree_mark_written_and_enqueue): Likewise.
20214 (propagate_all_subaccesses): Likewise and do not stop at first
20215 parent with a first_link.
20217 2017-06-13 Martin Jambor <mjambor@suse.cz>
20219 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20222 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20224 * match.pd: New pattern.
20226 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20228 * tree-vrp.c (is_masked_range_test): New function.
20229 (register_edge_assert_for): Determine ranges for
20232 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20234 PR tree-optimization/67328
20235 * fold-const.c (maskable_range_p): New function.
20236 (build_range_check): Generate bittests if possible.
20238 2017-06-13 Martin Liska <mliska@suse.cz>
20240 * gimple-pretty-print.c (dump_probability): Add new argument.
20241 (dump_edge_probability): Dump both probability and count.
20242 (dump_gimple_label): Likewise.
20243 (dump_gimple_bb_header): Likewise.
20245 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20248 * config/avr/avr-devices.c: Fix indentation.
20249 * config/avr/gen-avr-mmcu-specs.c: Dito.
20251 2017-06-13 Richard Biener <rguenther@suse.de>
20253 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20254 instead get vector type from stmt_info.
20255 (vectorizable_reduction): Adjust. Remove dead code.
20257 2017-06-13 Richard Biener <rguenther@suse.de>
20259 PR middle-end/81065
20260 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20261 case of C * (x * C2 + C3).
20262 (fold_addr_of_array_ref_difference): Properly fold index difference.
20264 2017-06-12 David S. Miller <davem@davemloft.net>
20267 * config/sparc/sparc.md (return expander): Emit frame blockage if
20268 function uses alloca.
20270 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20272 * combine.c (make_field_assignment): Check len rather than the mode
20273 precision when calling force_to_mode.
20275 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20277 Support multilibs and devices that see flash in RAM address range.
20280 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20281 (avr_mcu_t) <flash_pm_offset>: New field.
20282 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20283 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20284 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20285 (AVR_TINY_PM_OFFSET): Remove macro.
20286 * config/avr/avr.opt (-mshort-calls): New option.
20287 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20288 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20289 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20290 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20291 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20292 instead of avr_arch->have_jmp_call.
20293 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20294 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20295 avr_arch->flash_pm_offset to define.
20296 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20297 new field flash_pm_offset. Add entry for avrxmega3.
20298 (avr_texinfo): Add entry for avrxmega3.
20299 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20300 attiny212, attiny214,
20301 attiny412, attiny414, attiny416, attiny417,
20302 attiny814, attiny816, attiny817,
20303 attiny1614, attiny1616, attiny1617,
20304 attiny3214, attiny3216, attiny3217.
20305 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20306 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20307 (avr_print_operand_address) [AVR_TINY]: Same.
20308 (avr_asm_init_sections) <readonly_data_section>: Only patch
20309 callback if avr_arch->flash_pm_offset = 0.
20310 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20311 for rodata if avr_arch->flash_pm_offset != 0.
20312 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20313 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20314 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20315 (m_options): Append opt_rcall.
20316 (m_dirnames): Append dir_rcall.
20317 * config/avr/t-multilib: Regenerate.
20319 * configure.ac [target=avr]: Check whether avrxmega3 default
20320 linker description file works as needed.
20321 * configure: Regenerate.
20322 * doc/avr-mmcu.texi: Regenerate.
20323 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20324 <__AVR_ARCH__>: Document avrxmega3 and 103.
20325 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20326 <__AVR_SHORT_CALLS__>: Document it.
20327 <__AVR_PM_BASE_ADDRESS__>: Document it.
20328 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20329 (AVR Variable Attributes) <progmem>: Document this is
20330 not needed for avrxmega3.
20331 (AVR Named Address Spaces) <__flash>: Dito.
20333 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20335 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20337 2017-06-12 Doug Rupp <rupp@adacore.com>
20339 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20340 Append vxworks-stdint.h to the tm_file list.
20341 * config/vxworks-stdint.h: New file.
20343 2017-06-12 Martin Liska <mliska@suse.cz>
20345 PR tree-optimization/81041
20346 * tree-profile.c (gimple_gen_ic_func_profiler):
20347 Create an extra BB in profile-generate
20348 (gimple_gen_time_profiler): Likewise.
20350 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20352 PR tree-optimization/81003
20353 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20354 (update_range_test): Use it instead of force_gimple_operand_gsi.
20356 2017-06-12 Richard Biener <rguenther@suse.de>
20358 PR tree-optimization/81053
20359 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20360 with backedge value not defined in loop. Simplify def stmt
20363 2017-06-11 Tom de Vries <tom@codesourcery.com>
20366 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20368 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20369 nvptx_cannot_force_const_mem.
20371 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20373 * opts.c (finish_options): Move test for flag_split_stack after
20374 it has been initialized.
20376 2017-06-11 Jason Merrill <jason@redhat.com>
20378 * tree.h (id_equal): New.
20379 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20380 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20381 instead of strcmp of IDENTIFIER_POINTER.
20383 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20385 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20386 (mark_all_inlined_calls_cdtor): Fix formating.
20387 (inline_transform): Rescale profile before inlining.
20389 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20391 * cgraph.h (cgraph_edge::clone): Update prototype.
20392 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20393 (cgraph_node::create_clone): Update.
20394 (cgraph_node::create_version_clone): Update.
20395 * tree-inline.c (copy_bb): Update.
20396 (expand_call_inline): Update.
20398 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20400 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20401 factored out from ...
20402 (rs6000_emit_prologue): ... here.
20404 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20406 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20407 factored out from ...
20408 (rs6000_emit_prologue): ... here.
20410 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20412 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20414 (handle_missing_profiles): Fix computation of tp_first_run.
20415 (counts_to_freqs): Do not touch freqs when count is 0.
20417 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20419 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20422 2017-06-10 Tom de Vries <tom@codesourcery.com>
20424 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20425 attributes): Document signal effective target.
20427 2017-06-10 Tom de Vries <tom@codesourcery.com>
20429 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20430 Document effective target stack_size.
20432 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20434 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20435 to the edit_context if they can be auto-applied.
20437 2017-06-9 Ian Lance Taylor <iant@golang.org>
20439 * opts.c (finish_options): If -fsplit-stack, disable implicit
20440 -forder-blocks-and-partition.
20441 * doc/invoke.texi (Optimize Options): Document that when using
20442 -fsplit-stack -forder-blocks-and-partition is not implicitly
20445 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20447 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20448 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20449 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20450 * builtins.def (abort, trap, unreachable): Declare cold.
20451 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20452 * tree-core.h (ECF_COLD): New.
20453 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20454 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20456 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20458 * predict.c (unlikely_executed_stmt_p): Cleanup.
20460 2017-06-09 Richard Biener <rguenther@suse.de>
20462 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20463 model if the ref is always written to.
20465 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20467 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20469 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20471 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20474 2017-06-09 Tom de Vries <tom@codesourcery.com>
20477 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20478 "target cannot support label values" when encountering LABEL_REF.
20480 2017-06-09 Martin Liska <mliska@suse.cz>
20482 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20483 (gimple_gen_ic_func_profiler): Emit direct comparison
20484 of __gcov_indirect_call_callee with NULL.
20485 (gimple_gen_time_profiler): Change probability from
20486 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20488 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20490 * profile.c (edge_gcov_counts): Turn to pointer.
20491 (compute_branch_probabilities, compute_branch_probabilities): Update.
20492 (branch_prob): Do not clear edge_gcov_count.
20493 * profile.h (edge_gcov_counts): Turn to pointer.
20494 (edge_gcov_count): Update.
20496 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20498 * gimple.h (gimple_check_failed): Mark cold.
20500 2017-06-09 Richard Biener <rguenther@suse.de>
20502 PR tree-optimization/66623
20503 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20504 refactor check_reduction into two parts, properly computing
20505 whether we have to check reduction validity for outer loop
20508 2017-06-09 Richard Biener <rguenther@suse.de>
20510 PR tree-optimization/79483
20511 * graphite-scop-detection.c (order): New global.
20512 (get_order): Compute bb to order mapping that satisfies code
20513 generation constraints.
20514 (cmp_pbbs): New helper.
20515 (build_scops): Start domwalk at entry block, sort generated
20518 2017-06-09 Richard Biener <rguenther@suse.de>
20520 PR middle-end/81007
20521 * ipa-polymorphic-call.c
20522 (ipa_polymorphic_call_context::restrict_to_inner_class):
20523 Skip FIELD_DECLs with error_mark_node type.
20524 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20527 2017-06-09 Martin Liska <mliska@suse.cz>
20529 * predict.c (struct branch_predictor): New struct.
20530 (test_prediction_value_range): New test.
20531 (predict_c_tests): New function.
20532 * selftest-run-tests.c (selftest::run_tests): Run the function.
20533 * selftest.h: Declare new tests.
20535 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20538 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20539 gen_add3_insn did not fail.
20540 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20541 r0, construct that number in a temporary reg and add that reg to r0.
20542 If asked to put the result in r0 as well, fail.
20544 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20546 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20547 for early expansion of vec_eqv.
20549 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20551 PR middle-end/81005
20552 * ubsan.c (instrument_null): Avoid pointless code temporary.
20553 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20555 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20559 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20562 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20565 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20567 * predict.c (unlikely_executed_edge_p): Move ahead.
20568 (probably_never_executed_edge_p): Use it.
20570 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20572 PR middle-end/79988
20573 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20574 gimple_call_builtin_p call.
20576 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20578 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20579 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20580 rtl_check_failed_type2, rtl_check_failed_code1,
20581 rtl_check_failed_code2, rtl_check_failed_code_mode,
20582 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20583 rtvec_check_failed_bounds, rtl_check_failed_flag,
20584 _fatal_insn_not_found, _fatal_insn): Likewise.
20585 * tree.h (tree_contains_struct_check_failed,
20586 tree_check_failed, tree_not_check_failed,
20587 tree_class_check_failed, tree_range_check_failed,
20588 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20589 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20590 tree_operand_check_failed, omp_clause_check_failed,
20591 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20594 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20596 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20597 flag_branch_probabilities.
20598 * ipa-inline.c (edge_badness): Likewise.
20599 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20600 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20601 * predict.c (maybe_hot_frequency_p): Likewise.
20602 (probably_never_executed): Likewise.
20603 * sched-ebb.c (schedule_ebbs): Likewise.
20604 * sched-rgn.c (find_single_block_region): Likewise.
20605 * tracer.c (tail_duplicate): Likewise.
20607 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20609 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20610 longer requires x_flag_profile_use.
20612 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20614 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20615 instead of flag_reorder_blocks_and_partition.
20616 * dbxout.c (dbxout_function_end): Likewise.
20617 * dwarf2out.c (gen_subprogram_die): Likewise.
20618 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20619 * hw-doloop.c (reorg_loops): Likewise.
20620 * varasm.c (assemble_start_function,
20621 assemble_end_function): Likewise.
20622 (decide_function_section): Do not check for
20623 flag_reorder_blocks_and_partition.
20625 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20627 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20629 (chkp_get_hard_register_fake_addr_expr): Ditto.
20630 (chkp_build_addr_expr): Add check for hard reg case.
20631 (chkp_parse_array_and_component_ref): Ditto.
20632 (chkp_find_bounds_1): Ditto.
20633 (chkp_process_stmt): Don't generate bounds store for
20636 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20638 * predict.c (maybe_hot_bb_p): Do not check profile status.
20639 (maybe_hot_edge_p): Likewise.
20640 (probably_never_executed): Check for zero counts even if profile
20642 (unlikely_executed_edge_p): New function.
20643 (unlikely_executed_stmt_p): New function.
20644 (unlikely_executed_bb_p): New function.
20645 (set_even_probabilities): Use unlikely predicates.
20646 (combine_predictions_for_bb): Likewise.
20647 (predict_paths_for_bb): Likewise.
20648 (predict_paths_leading_to_edge): Likewise.
20649 (determine_unlikely_bbs): New function.
20650 (estimate_bb_frequencies): Use it.
20651 (compute_function_frequency): Use zero counts even if profile is
20653 * profile-count.h: Fix typo.
20655 2017-08-08 Julia Koval <julia.koval@intel.com>
20657 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20658 _mm512_mask_cvtsepi16_storeu_epi8,
20659 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20660 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20661 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20662 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20663 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20664 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20665 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20666 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20667 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20668 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20669 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20670 __builtin_ia32_pmovuswb256mem_mask,
20671 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20672 __builtin_ia32_pmovwb512mem_mask): New builtins.
20674 2017-08-08 Julia Koval <julia.koval@intel.com>
20676 PR target/73350,80862
20677 * config/i386/subst.md (round): Fix round pattern.
20678 * config/i386/i386.c (ix86_erase_embedded_rounding):
20679 Fix erasing rounding for the fixed pattern.
20681 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20683 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20685 2017-06-08 Martin Liska <mliska@suse.cz>
20687 PR gcov-profile/80911
20688 * gcov.c (block_info::block_info): New constructor.
20690 2017-06-07 Carl Love <cel@us.ibm.com>
20692 * config/rs6000/rs6000-c: The return type of the following
20693 built-in functions was implemented as int not long long. Fix sign
20694 of return value for the unsigned version of vec_mulo and vec_mule.
20695 vector unsigned long long vec_bperm (vector unsigned long long,
20696 vector unsigned char)
20697 vector signed long long vec_mule (vector signed int,
20699 vector unsigned long long vec_mule (vector unsigned int,
20700 vector unsigned int)
20701 vector signed long long vec_mulo (vector signed int,
20703 vector unsigned long long vec_mulo (vector unsigned int,
20704 vector unsigned int)
20705 * doc/extend.texi: Fix the documentation for the built-in
20708 2017-06-07 Carl Love <cel@us.ibm.com>
20711 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20714 2017-06-07 Carl Love <cel@us.ibm.com>
20716 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20717 support, Generate doublehv for signed int/float for BE case only.
20719 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20721 * doc/invoke.texi (mcx16): Rewrite.
20723 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20725 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20726 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20727 *mov<mode>_softfloat, and an anonymous splitter): Use
20728 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20730 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20732 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20734 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20735 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20736 (rs6000_debug_reg_global): Adjust.
20737 (rs6000_init_hard_regno_mode_ok): Adjust.
20738 (rs6000_dbx_register_number): Adjust.
20739 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20740 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20741 Remove SPE_ACC and SPEFSCR.
20742 (REG_ALLOC_ORDER): Ditto.
20743 (FRAME_POINTER_REGNUM): Change to 111.
20744 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20745 (REG_CLASS_NAMES): Ditto.
20746 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20747 (REGISTER_NAMES): Ditto.
20748 (ADDITIONAL_REG_NAMES): Ditto.
20749 (rs6000_reg_names): Ditto.
20750 * config/rs6000/rs6000.md: Renumber some register number
20753 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20755 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20757 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20758 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20760 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20761 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20762 Delete the SPE high registers.
20763 (REG_ALLOC_ORDER): Ditto.
20764 (enum reg_class): Remove SPE_HIGH_REGS.
20765 (REG_CLASS_NAMES): Ditto.
20766 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20767 (REGISTER_NAMES): Ditto.
20768 (rs6000_reg_names): Ditto.
20769 * doc/tm.texi.in: Remove SPE as example.
20770 * doc/tm.texi: Regenerate.
20772 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20774 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20775 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20776 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20777 * config/rs6000/rs6000.md (type): Remove "brinc".
20779 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20781 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20782 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20783 * config/rs6000/linuxspe.h: Delete file.
20784 * config/rs6000/rs6000.md: Don't include spe.md.
20785 * config/rs6000/spe.h: Delete file.
20786 * config/rs6000/spe.md: Delete file.
20787 * config/rs6000/t-rs6000: Remove spe.md.
20789 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20791 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20792 (reg_or_none500mem_operand): Delete.
20793 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20794 instead of reg_or_none500mem_operand.
20796 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20798 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20799 handling of SPE flags.
20800 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20802 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20804 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20806 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20807 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20808 paired_divv2sf3): Similar.
20809 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20810 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20811 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20813 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20814 Rename the paired_* instruction patterns.
20815 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20817 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20818 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20819 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20820 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20821 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20822 PAIRED_VECTOR_MODE.
20823 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20824 (spe_func_has_64bit_regs_p): Delete.
20825 (spe_expand_predicate_builtin): Delete.
20826 (spe_expand_evsel_builtin): Delete.
20827 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20828 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20829 (invalid_e500_subreg): Delete.
20830 (rs6000_legitimize_address): Always force_reg op2 as well, for
20831 paired single memory accesses.
20832 (rs6000_member_type_forces_blk): Delete.
20833 (rs6000_spe_function_arg): Delete.
20834 (rs6000_expand_unop_builtin): Delete SPE handling.
20835 (rs6000_expand_binop_builtin): Ditto.
20836 (spe_expand_stv_builtin): Delete.
20837 (bdesc_2arg_spe): Delete.
20838 (spe_expand_builtin): Delete.
20839 (spe_expand_predicate_builtin): Delete.
20840 (spe_expand_evsel_builtin): Delete.
20841 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20842 (spe_init_builtins): Delete.
20843 (spe_func_has_64bit_regs_p): Delete.
20844 (savres_routine_name): Delete "info" parameter. Adjust callers.
20845 (rs6000_emit_stack_reset): Ditto.
20846 (rs6000_dwarf_register_span): Delete.
20847 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20848 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20849 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20851 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20853 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20854 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20855 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20856 mulv2sf3, divv2sf3): Delete expanders.
20858 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20860 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20862 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20864 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20865 * config/rs6000/rs6000.c: Ditto.
20867 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20869 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20870 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20871 comparison_operator.
20873 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20875 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20876 * config/rs6000/rs6000.opt: Ditto.
20877 * config/rs6000/t-rtems: Ditto.
20879 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20881 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20882 TARGET_E500_SINGLE by 0, simplify.
20883 * config/rs6000/rs6000.c: Ditto.
20884 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20885 (spe_build_register_parallel): Delete.
20886 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20887 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20888 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20889 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20890 (E500_CONVERT): Delete.
20891 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20893 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20895 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20896 * config/rs6000/dfp.md: Ditto.
20897 (negdd2, *negdd2_fpr): Merge.
20898 (absdd2, *absdd2_fpr): Merge.
20899 (negtd2, *negtd2_fpr): Merge.
20900 (abstd2, *abstd2_fpr): Merge.
20901 * config/rs6000/e500.h: Delete file.
20902 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20903 TARGET_FPRS by 1 and simplify.
20904 * config/rs6000/rs6000-c.c: Ditto.
20905 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20906 TARGET_DF_SPE by 0.
20907 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20909 * config/rs6000/rs6000.md: Ditto.
20910 (floatdidf2, *floatdidf2_fpr): Merge.
20911 (move_from_CR_gt_bit): Delete.
20912 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20913 (E500_CR_IOR_COMPARE): Delete.
20914 (All patterns that require !TARGET_FPRS): Delete.
20915 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20917 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20919 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20921 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20923 * graphds.c (add_edge): Intitialize edge's attached data.
20924 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20925 pointer parameter. Call pointed function on each edge during
20926 graph traversing. Skip traversing the edge when the function
20928 (graphds_dfs, graphds_scc): Ditto.
20929 (for_each_edge): New parameter. Pass the new parameter to callback
20931 * graphds.h (skip_edge_callback): New function pointer type.
20932 (graphds_dfs, graphds_scc): New function pointer parameter.
20933 (graphds_edge_callback, for_each_edge): New parameter.
20935 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20937 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20938 out code checking if runtime alias check is possible to below ...
20939 Call the new function.
20940 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20941 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20943 2017-06-07 Marek Polacek <polacek@redhat.com>
20946 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20947 TYPE_OVERFLOW_WRAPS check.
20949 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20951 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20952 if versioning is required.
20953 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20954 peeling with the check for versioning.
20956 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20958 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20959 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20960 Set true to new parameter if new ssa variable is defined.
20961 (vect_gen_vector_loop_niters): Refactor. Set range information
20962 for the new vector loop bound variable.
20963 (vect_do_peeling): Ditto.
20965 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20967 * tree-affine.c (ssa.h): Include header file.
20968 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20969 has wrapping overflow behavior.
20971 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20973 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20975 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20977 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20978 (tree_to_aff_combination): ... here.
20980 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20982 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20983 reg_pressure model function.
20984 (ivopts_global_cost_for_size): Delete.
20985 (determine_set_costs, iv_ca_recount_cost): Call new model function
20986 ivopts_estimate_reg_pressure.
20988 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20990 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20991 expensive than udiv. Remove floating point cases from mod.
20993 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20995 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20996 Increase idiv cost.
20998 2017-06-07 Tamar Christina <tamar.christina@arm.com>
21000 * config/aarch64/aarch64.md
21001 (copysignsf3): Fix mask generation.
21003 2017-06-07 Jakub Jelinek <jakub@redhat.com>
21005 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
21007 (class dump_manager): Add register_dumps method.
21008 * dumpfile.c: Include langhooks.h.
21009 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
21010 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
21011 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
21012 (dump_manager::dump_register): Start with 512 entries instead of 32.
21013 (dump_manager::register_dumps): New method.
21014 * toplev.c (general_init): Instead of invoking register_dumps
21015 langhook, invoke register_dumps method on the dump manager.
21016 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
21019 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
21021 * doc/md.texi: Clarify the restrictions on a define_insn condition.
21022 Say that # requires an associated define_split to exist, and that
21023 the define_split must be suitable for use after register allocation.
21025 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21027 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21028 (compute_outgoing_frequencies): Also initialize zero counts.
21029 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21030 around loops; preserve more of profile when nothing changes.
21032 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21034 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21036 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21037 * config/arm/arm-cpu-cdata.h: Regenerate.
21038 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21039 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21040 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21041 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21042 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21044 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21046 * config/arm/t-rmprofile: Likewise.
21047 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21049 2017-06-06 David S. Miller <davem@davemloft.net>
21052 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21053 blockage if function uses alloca.
21055 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21057 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21058 New "uid" fields to hold pretty-print IDs of group and ref.
21059 Memory references are now identified as <group_id>:<ref_id>
21060 instead of using [random] addresses.
21061 (dump_mem_details): Simplify, no functional change.
21062 (dump_mem_ref): Simplify and make output more concise.
21063 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21064 (find_or_create_group): Initialize group uid.
21065 (record_ref): Initialize ref uid. Improve debug output.
21066 (prune_group_by_reuse, should_issue_prefetch_p,)
21067 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21068 (mark_nontemporal_store, determine_loop_nest_reuse):
21069 Improve debug output.
21071 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21073 * dbgcnt.def (prefetch): New debug counter.
21074 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21075 (schedule_prefetches): Stop issueing prefetches if debug counter
21078 2017-06-06 Tom de Vries <tom@codesourcery.com>
21080 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21081 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21083 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21085 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21086 Use aarch64_reg_or_zero predicate for operand 4.
21087 (aarch64_compare_and_swap<mode> define_insn_and_split):
21088 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21089 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21091 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21093 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21094 (arm_compute_save_core_reg_mask): This.
21095 (thumb1_compute_save_reg_mask): Rename into ...
21096 (thumb1_compute_save_core_reg_mask): This.
21097 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21098 (arm_compute_frame_layout): Likewise.
21100 2017-06-06 Richard Biener <rguenther@suse.de>
21102 PR tree-optimization/80974
21103 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21104 keep or clear leaders SSA info.
21106 2017-06-06 Tom de Vries <tom@codesourcery.com>
21108 * config/nvptx/nvptx.c (split_mode_p): New function.
21109 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21111 2017-06-06 Tom de Vries <tom@codesourcery.com>
21113 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21115 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21118 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21121 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21123 * shrink-wrap.c (handle_simple_exit): Update profile.
21124 (try_shrink_wrapping): Upate profile.
21126 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21128 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21129 (tree_guess_outgoing_edge_probabilities): New.
21130 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21131 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21133 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21135 * ipa-split.c (split_function): Initialize return bb profile.
21137 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21139 * profile.c (compute_branch_probabilities): Also initialize
21140 EXIT_BLOCK profile.
21142 2017-06-06 Richard Biener <rguenther@suse.de>
21144 PR tree-optimization/80928
21145 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21146 (vect_analyze_loop_operations): Properly guard analysis for
21148 (vect_transform_loop): Likewise.
21149 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21150 (vect_model_induction_cost): Do not cost for pure SLP.
21151 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21152 of induction in inner loop vectorization.
21153 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21154 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21155 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21157 (vect_analyze_slp_cost_1): Cost induction.
21158 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21159 (vect_get_slp_vect_defs): Likewise.
21160 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21161 (vect_transform_stmt): Handle SLP reductions.
21162 * tree-vectorizer.h (vectorizable_induction): Adjust.
21164 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21166 * config/rs6000/rs6000.c (make_resolver_func): Update
21167 init_lowered_empty_function call.
21169 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21171 * doc/invoke.texi: Document the -fprofile-abs-path option.
21172 * common.opt (fprofile-abs-path): New option.
21173 * gcov-io.h (gcov_write_filename): Declare.
21174 * gcov-io.c (gcov_write_filename): New function.
21175 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21176 * profile.c (output_location): Likewise.
21178 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21180 * shring-wrap.c: Revert accidental commit.
21182 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21184 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21186 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21188 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21190 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21191 profile in callgraph edge.
21192 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21193 (apply_scale): Likewise.
21194 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21195 Also scale profile when inlining function with zero profile.
21196 (initialize_cfun): Update exit block profile even when it is zero.
21197 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21198 when profile is read.
21200 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21202 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21203 (CLONE_*): New constants to define the processors we can generate
21204 code for with the target_clone attribute.
21205 (rs6000_clone_map): New array to identify which clone processors
21206 the current program is running on.
21207 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21208 target_clone attribute.
21209 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21210 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21211 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21212 (cpu_expand_builtin): Add support for target_clone attribute.
21213 (rs6000_valid_attribute_p): Allow "default" attribute.
21214 (get_decl_name): New debug function to simplify printing the
21215 current function name in debugging statements.
21216 (rs6000_clone_priority): New functions to support the target_clone
21217 attribute, and be able to generate code to switch between ISA 2.05
21218 through ISA 3.0 (power6 through power9).
21219 (rs6000_compare_version_priority): Likewise.
21220 (rs6000_get_function_versions_dispatcher): Likewise.
21221 (make_resolver_func): Likewise.
21222 (add_condition_to_bb): Likewise.
21223 (dispatch_function_versions): Likewise.
21224 (rs6000_generate_version_dispatcher_body): Likewise.
21225 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21226 (fusion_gpr_load_p): Fix a spacing issue.
21227 * doc/extend.texi (Common Function Attributes): Document that the
21228 PowerPC supports the target_clone attribute.
21230 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21232 * config/arm/arm.h: explain F symbol found in description of ARM
21233 register allocation in its legend.
21235 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21237 * config/mips/frame-header-opt.c: Include profile-count.h.
21238 * config/riscv/riscv.c: Include profile-count.h
21240 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21242 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21244 (sm_set_flag_if_changed): Add bbs field.
21245 (execute_sm_if_changed_flag_set): Pass BBS.
21246 (execute_sm): Update.
21248 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21250 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21253 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21255 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21256 (peephole2): New peephole2 to emit the above.
21257 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21259 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21261 * config/aarch64/aarch64.c (define_peephole2 above
21262 *sub_<shift>_<mode>): New peephole.
21264 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21266 * config/i386/i386.c (make_resolver_func): Update.
21267 * Makefile.in: Add profile-count.h and profile-count.o
21268 * auto-profile.c (afdo_indirect_call): Update to new API.
21269 (afdo_set_bb_count): Update.
21270 (afdo_propagate_edge): Update.
21271 (afdo_propagate_circuit): Update.
21272 (afdo_calculate_branch_prob): Update.
21273 (afdo_annotate_cfg): Update.
21274 * basic-block.h: Include profile-count.h
21275 (struct edge_def): Turn count to profile_count.
21276 (struct basic_block_def): Likewie.
21277 (REG_BR_PROB_BASE): Move to profile-count.h
21278 (RDIV): Move to profile-count.h
21279 * bb-reorder.c (max_entry_count): Turn to profile_count.
21280 (find_traces): Update.
21281 (rotate_loop):Update.
21282 (connect_traces):Update.
21283 (sanitize_hot_paths):Update.
21284 * bt-load.c (migrate_btr_defs): Update.
21285 * cfg.c (RDIV): Remove.
21286 (init_flow): Use alloc_block.
21287 (alloc_block): Uninitialize count.
21288 (unchecked_make_edge): Uninitialize count.
21289 (check_bb_profile): Update.
21290 (dump_edge_info): Update.
21291 (dump_bb_info): Update.
21292 (update_bb_profile_for_threading): Update.
21293 (scale_bbs_frequencies_int): Update.
21294 (scale_bbs_frequencies_gcov_type): Update.
21295 (scale_bbs_frequencies_profile_count): New.
21296 * cfg.h (update_bb_profile_for_threading): Update.
21297 (scale_bbs_frequencies_profile_count): Declare.
21298 * cfgbuild.c (compute_outgoing_frequencies): Update.
21299 (find_many_sub_basic_blocks): Update.
21300 * cfgcleanup.c (try_forward_edges): Update.
21301 (try_crossjump_to_edge): Update.
21302 * cfgexpand.c (expand_gimple_tailcall): Update.
21303 (construct_exit_block): Update.
21304 * cfghooks.c (verify_flow_info): Update.
21305 (dump_bb_for_graph): Update.
21306 (split_edge): Update.
21307 (make_forwarder_block): Update.
21308 (duplicate_block): Update.
21309 (account_profile_record): Update.
21310 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21311 (get_estimated_loop_iterations): Update.
21312 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21313 (single_likely_exit): Update.
21314 * cfgloopmanip.c (scale_loop_profile): Update.
21316 (set_zero_probability): Update.
21317 (lv_adjust_loop_entry_edge): Update.
21318 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21319 (purge_dead_edges): Update.
21320 (rtl_account_profile_record): Update.
21321 * cgraph.c (cgraph_node::create): Uninitialize count.
21322 (symbol_table::create_edge): Uninitialize count.
21323 (cgraph_update_edges_for_call_stmt_node): Update.
21324 (cgraph_edge::dump_edge_flags): Update.
21325 (cgraph_node::dump): Update.
21326 (cgraph_edge::maybe_hot_p): Update.
21327 * cgraph.h: Include profile-count.h
21328 (create_clone), create_edge, create_indirect_edge): Update.
21329 (cgraph_node): Turn count to profile_count.
21330 (cgraph_edge0: Likewise.
21331 (make_speculative, clone): Update.
21332 (create_edge): Update.
21333 (init_lowered_empty_function): Update.
21334 * cgraphclones.c (cgraph_edge::clone): Update.
21335 (duplicate_thunk_for_node): Update.
21336 (cgraph_node::create_clone): Update.
21337 * cgraphunit.c (cgraph_node::analyze): Update.
21338 (cgraph_node::expand_thunk): Update.
21339 * final.c (dump_basic_block_info): Update.
21340 * gimple-streamer-in.c (input_bb): Update.
21341 * gimple-streamer-out.c (output_bb): Update.
21342 * graphite.c (print_global_statistics): Update.
21343 (print_graphite_scop_statistics): Update.
21344 * hsa-brig.c: Include basic-block.h.
21345 * hsa-dump.c: Include basic-block.h.
21346 * hsa-gen.c (T sum_slice): Update.
21347 (convert_switch_statements):Update.
21348 * hsa-regalloc.c: Include basic-block.h.
21349 * ipa-chkp.c (chkp_produce_thunks): Update.
21350 * ipa-cp.c (struct caller_statistics): Update.
21351 (init_caller_stats): Update.
21352 (gather_caller_stats): Update.
21353 (ipcp_cloning_candidate_p): Update.
21354 (good_cloning_opportunity_p): Update.
21355 (get_info_about_necessary_edges): Update.
21356 (dump_profile_updates): Update.
21357 (update_profiling_info): Update.
21358 (update_specialized_profile): Update.
21359 (perhaps_add_new_callers): Update.
21360 (decide_about_value): Update.
21361 (ipa_cp_c_finalize): Update.
21362 * ipa-devirt.c (struct odr_type_warn_count): Update.
21363 (struct decl_warn_count): Update.
21364 (struct final_warning_record): Update.
21365 (possible_polymorphic_call_targets): Update.
21366 (ipa_devirt): Update.
21367 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21368 * ipa-icf.c (sem_function::merge): Update.
21369 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21370 * ipa-inline.c (compute_uninlined_call_time): Update.
21371 (compute_inlined_call_time): Update.
21372 (want_inline_small_function_p): Update.
21373 (want_inline_self_recursive_call_p): Update.
21374 (edge_badness): Update.
21375 (lookup_recursive_calls): Update.
21376 (recursive_inlining): Update.
21377 (inline_small_functions): Update.
21378 (dump_overall_stats): Update.
21379 (dump_inline_stats): Update.
21380 * ipa-profile.c (ipa_profile_generate_summary): Update.
21381 (ipa_propagate_frequency): Update.
21382 (ipa_profile): Update.
21383 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21384 * ipa-utils.c (ipa_merge_profiles): Update.
21385 * loop-doloop.c (doloop_modify): Update.
21386 * loop-unroll.c (report_unroll): Update.
21387 (unroll_loop_runtime_iterations): Update.
21388 * lto-cgraph.c (lto_output_edge): Update.
21389 (lto_output_node): Update.
21390 (input_node): Update.
21391 (input_edge): Update.
21392 (merge_profile_summaries): Update.
21393 * lto-streamer-in.c (input_cfg): Update.
21394 * lto-streamer-out.c (output_cfg): Update.
21395 * mcf.c (create_fixup_graph): Update.
21396 (adjust_cfg_counts): Update.
21397 (sum_edge_counts): Update.
21398 * modulo-sched.c (sms_schedule): Update.
21399 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21400 * predict.c (maybe_hot_count_p): Update.
21401 (probably_never_executed): Update.
21402 (dump_prediction): Update.
21403 (combine_predictions_for_bb): Update.
21404 (propagate_freq): Update.
21405 (handle_missing_profiles): Update.
21406 (counts_to_freqs): Update.
21407 (rebuild_frequencies): Update.
21408 (force_edge_cold): Update.
21409 * predict.h: Include profile-count.h
21410 (maybe_hot_count_p, counts_to_freqs): UPdate.
21411 * print-rtl-function.c: Do not include cfg.h
21412 * print-rtl.c: Include basic-block.h
21413 * profile-count.c: New file.
21414 * profile-count.h: New file.
21415 * profile.c (is_edge_inconsistent): Update.
21416 (correct_negative_edge_counts): Update.
21417 (is_inconsistent): Update.
21418 (set_bb_counts): Update.
21419 (read_profile_edge_counts): Update.
21420 (compute_frequency_overlap): Update.
21421 (compute_branch_probabilities): Update; Initialize and deinitialize
21423 (branch_prob): Update.
21424 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21425 (edge_gcov_count): New.
21426 (bb_gcov_count): New.
21427 * shrink-wrap.c (try_shrink_wrapping): Update.
21428 * tracer.c (better_p): Update.
21429 * trans-mem.c (expand_transaction): Update.
21430 (ipa_tm_insert_irr_call): Update.
21431 (ipa_tm_insert_gettmclone_call): Update.
21432 * tree-call-cdce.c: Update.
21433 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21434 (gimple_duplicate_sese_tail): Update.
21435 (gimple_account_profile_record): Update.
21436 (execute_fixup_cfg): Update.
21437 * tree-inline.c (copy_bb): Update.
21438 (copy_edges_for_bb): Update.
21439 (initialize_cfun): Update.
21440 (freqs_to_counts): Update.
21441 (copy_cfg_body): Update.
21442 (expand_call_inline): Update.
21443 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21444 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21445 (try_unroll_loop_completely): Update.
21446 (try_peel_loop): Update.
21447 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21448 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21449 * tree-ssa-loop-split.c (connect_loops): Update.
21450 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21451 * tree-ssa-reassoc.c (branch_fixup): Update.
21452 * tree-ssa-tail-merge.c (replace_block_by): Update.
21453 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21454 (compute_path_counts): Update.
21455 (update_profile): Update.
21456 (recompute_probabilities): Update.
21457 (update_joiner_offpath_counts): Update.
21458 (estimated_freqs_path): Update.
21459 (freqs_to_counts_path): Update.
21460 (clear_counts_path): Update.
21461 (ssa_fix_duplicate_block_edges): Update.
21462 (duplicate_thread_path): Update.
21463 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21464 (struct switch_conv_info): Update.
21465 * tree-tailcall.c (decrease_profile): Update.
21466 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21467 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21468 * value-prof.c (check_counter): Update.
21469 (gimple_divmod_fixed_value): Update.
21470 (gimple_mod_pow2): Update.
21471 (gimple_mod_subtract): Update.
21472 (gimple_ic_transform): Update.
21473 (gimple_stringop_fixed_value): Update.
21474 * value-prof.h (gimple_ic): Update.
21476 2017-06-02 Carl Love <cel@us.ibm.com>
21478 * config/rs6000/rs6000-c: Add support for built-in functions
21479 vector double vec_doublee (vector signed int);
21480 vector double vec_doublee (vector unsigned int);
21481 vector double vec_doublee (vector float);
21482 vector double vec_doubleh (vector signed int);
21483 vector double vec_doubleh (vector unsigned int);
21484 vector double vec_doubleh (vector float);
21485 vector double vec_doublel (vector signed int);
21486 vector double vec_doublel (vector unsigned int);
21487 vector double vec_doublel (vector float);
21488 vector double vec_doubleo (vector signed int);
21489 vector double vec_doubleo (vector unsigned int);
21490 vector double vec_doubleo (vector float);.
21491 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21492 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21494 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21495 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21496 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21498 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21499 vec_doublel, vec_doubleh.
21500 * doc/extend.texi: Update the built-in documentation file for the
21501 new built-in functions.
21503 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21506 * ipa-inline-analysis.c (free_growth_caches): Set
21507 edge_removal_hook_holder to NULL after removing it.
21509 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21511 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21512 comparision with zero.
21514 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21515 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21516 for early expansion of vec_min and vec_max builtins.
21517 (builtin_function_type): Add min/max unsigned variants to those
21518 identified as having unsigned arguments.
21520 2017-06-02 Olivier Hainque <hainque@adacore.com>
21522 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21524 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21526 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21527 Use VALL_F16 iterator rather than VALL.
21529 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21531 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21532 Emit CBNZ inside loop when doing a strong exchange and comparing
21533 against zero. Generate the CC flags after the loop.
21535 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21537 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21538 (dl_section_ref): New.
21539 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21540 On AIX, append an expression to subtract the size of the
21541 section length to dl_section_ref.
21543 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21545 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21546 for early expansion of vector absolute builtins.
21548 2017-06-02 Richard Biener <rguenther@suse.de>
21550 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21551 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21553 2017-06-02 Richard Biener <rguenther@suse.de>
21555 PR tree-optimization/80948
21556 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21557 stmt order as well.
21559 2017-06-02 Richard Biener <rguenther@suse.de>
21561 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21563 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21564 for inductions as relevant.
21566 2017-06-02 Richard Biener <rguenther@suse.de>
21568 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21569 (vectorizable_induction): ... this. Remove dead code.
21571 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21573 * builtins. (expand_builtin_alloca): Remove second parameter and
21574 infer its value from the first parameter instead.
21575 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21577 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21579 PR rtl-optimization/80903
21580 * loop-doloop.c (add_test): Unshare sequence.
21582 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21584 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21586 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21588 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21590 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21591 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21592 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21593 (xlogue_layout::get_stub_rtx): Make static.
21594 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21595 (xlogue_layout::compute_stub_managed_regs): Rename to...
21596 (xlogue_layout::count_stub_managed_regs): ...this.
21597 (xlogue_layout::is_stub_managed_reg): New function.
21598 (xlogue_layout::m_stub_names): Rename to...
21599 (xlogue_layout::s_stub_names): ...this, make static.
21600 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21601 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21602 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21603 xlogue_layout::s_stub_names): Instantiate statics.
21604 (stub_managed_regs): Remove.
21605 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21606 (disable_call_ms2sysv_xlogues): Rename to...
21607 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21608 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21610 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21611 change after reload_completed.
21612 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21614 (ix86_expand_prologue): Likewise.
21615 (ix86_expand_epilogue): Likewise.
21616 (ix86_expand_split_stack_prologue): Likewise.
21617 (ix86_compute_frame_layout): Remove frame parameter ...
21618 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21619 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21621 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21622 (ix86_frame): Move from here ...
21623 * config/i386/i386.h (ix86_frame): ... to here.
21624 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21625 complete ix86_frame data structure instead. Remove some_ld_name.
21627 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21629 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21630 symbols that hold a DECL_VALUE_EXPR.
21632 2017-06-01 Martin Jambor <mjambor@suse.cz>
21634 PR tree-optimization/80898
21635 * tree-sra.c (process_subtree_disqualification): Removed.
21636 (disqualify_candidate): Do not acll
21637 process_subtree_disqualification.
21638 (subtree_mark_written_and_enqueue): New function.
21639 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21640 RHS has been disqualified and re-queue LHS if necessary. Apart
21641 from that, ignore disqualified RHS.
21643 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21645 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21646 address fetch for z10 or later.
21648 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21650 * config/arc/arc.md (tst_movb): Add guard when splitting.
21652 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21654 * config/arc/arc.c (arc_can_eliminate): Test against
21655 arc_frame_pointer_needed.
21657 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21659 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21660 to prevent store reordering.
21661 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21662 (type): Add block type.
21663 (stack_tie): Define special instruction to be used in
21666 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21668 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21669 constraint. It is not valid for the pattern.
21670 (noncommutative_binary_comparison): Likewise.
21672 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21674 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21677 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21679 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21680 be used by the reg-alloc.
21682 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21684 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21685 reg-alloc when having mul64 or mul32x16 instructions.
21686 (mulsidi3): Likewise.
21687 (umulsidi3): Likewise.
21688 (mulsi32x16): New pattern.
21689 (mulsi64): Likewise.
21690 (mulsidi64): Likewise.
21691 (umulsidi64): Likewise.
21692 (MUL32x16_REG): Define.
21693 (mul64_600): Use MUL32x16_REG.
21694 (mac64_600): Likewise.
21695 (umul64_600): Likewise.
21696 (umac64_600): Likewise.
21698 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21700 * config/arc/arc.md (mulsi3_700): Make it commutative.
21702 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21704 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21706 (*sign_extendsidi2_insn): Likewise for movstosw.
21708 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21710 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21711 the type of the input discriminant value. Convert the
21712 discriminant value of signedness vary.
21714 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21716 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21717 Add to -Wall section.
21719 2017-06-01 Richard Biener <rguenther@suse.de>
21721 PR middle-end/66313
21722 * fold-const.c (fold_plusminus_mult_expr): If the factored
21723 factor may be zero use a wrapping type for the inner operation.
21724 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21725 and handle moved defs.
21726 (process_assignment): Properly guard the unary op case. Return a
21727 tri-state indicating that moving the stmt before the call may allow
21728 to continue. Pass through to_move.
21729 (find_tail_calls): Handle moving unrelated defs before
21732 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21735 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21736 splitter result in the canonical way.
21738 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21740 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21741 also for 32bit target. Update insn attributes.
21742 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21744 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21746 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21747 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21748 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21749 (_mm_maskz_min_ss): New intrinsics.
21751 2017-05-31 Martin Liska <mliska@suse.cz>
21753 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21754 Change comment style to one we normally use.
21755 (vectorizable_reduction): Likewise.
21756 (vectorizable_induction): Likewise.
21757 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21758 (vectorizable_call): Likewise.
21759 (vectorizable_simd_clone_call): Likewise.
21760 (vectorizable_conversion): Likewise.
21761 (vectorizable_assignment): Likewise.
21762 (vectorizable_shift): Likewise.
21763 (vectorizable_operation): Likewise.
21764 (vectorizable_store): Likewise.
21765 (vectorizable_load): Likewise.
21766 * tree-vectorizer.h: Likewise.
21768 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21770 * passes.c (emergency_dump_function): New.
21771 * tree-pass.h (emergency_dump_function): Declare.
21772 * plugin.c (plugins_internal_error_function): Remove.
21773 * plugin.h (plugins_internal_error_function): Remove declaration.
21774 * toplev.c (internal_error_function): New static function. Use it...
21775 (general_init): ...here.
21777 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21779 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21780 (arc_rtx_costs): Add costs for new patterns.
21781 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21782 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21784 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21786 * tree-ssa-strlen.c (get_next_strinfo): New function.
21787 (get_stridx_plus_constant): Use it.
21788 (zero_length_string): Likewise.
21789 (adjust_related_strinfos): Likewise.
21790 (adjust_last_stmt): Likewise.
21792 2017-05-31 Richard Biener <rguenther@suse.de>
21795 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21796 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21798 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21800 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21801 loop_vinfo argument and use of dependence distance vectors.
21802 Check instead whether the two references differ only in their
21803 initial value and assume that they have the same alignment if the
21804 difference is a multiple of the vector alignment.
21805 (vect_analyze_data_refs_alignment): Update call accordingly.
21807 2017-05-31 Martin Liska <mliska@suse.cz>
21810 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21812 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21814 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21815 (create_intersect_range_checks): Move from ...
21816 * tree-data-ref.c (create_intersect_range_checks_index)
21817 (create_intersect_range_checks): ... to here.
21818 (create_runtime_alias_checks): New function factored from ...
21819 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21820 here. Call above function.
21821 * tree-data-ref.h (create_runtime_alias_checks): New function.
21823 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21825 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21826 segment length for dr_b and compute it in wide_int.
21828 2017-05-31 Richard Biener <rguenther@suse.de>
21830 PR tree-optimization/80906
21831 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21832 and pass through iv_map.
21833 (copy_bb_and_scalar_dependences): Adjust.
21834 (translate_pending_phi_nodes): Likewise.
21835 (copy_loop_close_phi_args): Handle code-generating IVs instead
21838 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21840 * diagnostic-color.c (color_dict): Add "type-diff".
21841 (parse_gcc_colors): Update comment.
21842 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21843 -fdiagnostics-show-template-tree and -fno-elide-type.
21844 (GCC_COLORS): Add type-diff to example.
21846 (-fdiagnostics-show-template-tree): New.
21847 (-fno-elide-type): New.
21848 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21849 the pp_format_decoder callback. Call any m_format_postprocessor's
21851 (pretty_printer::pretty_printer): Initialize
21852 m_format_postprocessor.
21853 (pretty_printer::~pretty_printer): Delete any
21854 m_format_postprocessor.
21855 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21856 (class format_postprocessor): New class.
21857 (struct pretty_printer::format_decoder): Document the new parameters.
21858 (struct pretty_printer::m_format_postprocessor): New field.
21859 * tree-diagnostic.c (default_tree_printer): Update for new
21860 bool and const char ** params.
21861 * tree-diagnostic.h (default_tree_printer): Likewise.
21863 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21865 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21866 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21867 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21868 rs6000_gen_cell_microcode code.
21869 (rs6000_final_prescan_insn): Delete.
21870 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21871 "warn-cell-microcode" entries.
21872 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21873 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21874 throughout. Change cc_reg_not_micro_cr0_operand to
21875 cc_reg_not_cr0_operand throughout.
21876 (*extendhi<mode>2_noload): Delete.
21877 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21878 (mwarn-cell-microcode): Delete.
21879 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21880 -mgen-cell-microcode and -mwarn-cell-microcode.
21882 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21885 * config/i386/constraints.md (Yd): New constraint.
21887 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21888 and (?Yd, r) alternatives. Update insn attributes.
21889 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21890 and (?*Yd, r) alternatives. Update insn attributes.
21891 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21893 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21895 * gimplify.c (gimplify_modify_expr): Don't create a
21896 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21899 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21901 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21903 2017-05-30 Richard Biener <rguenther@suse.de>
21905 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21906 and reduc_def fields.
21907 (STMT_VINFO_REDUC_TYPE): New define.
21908 (STMT_VINFO_REDUC_DEF): Likewise.
21909 (vect_force_simple_reduction): Adjust prototype.
21910 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21911 (vect_is_simple_reduction): Remove check_reduction argument.
21912 (vect_force_simple_reduction): Adjust and set
21913 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21914 (vectorizable_reduction): Do not re-do reduction analysis
21915 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21916 * tree-parloops.c (gather_scalar_reductions): Adjust.
21918 2017-05-30 Richard Biener <rguenther@suse.de>
21920 PR middle-end/80901
21921 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21924 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21926 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21927 Introduce unknown_misalignment parameter and remove vf.
21928 (vect_peeling_hash_get_lowest_cost):
21929 Pass unknown_misalignment parameter.
21930 (vect_enhance_data_refs_alignment):
21931 Fix unsupportable data ref treatment.
21933 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21935 * tree-vect-data-refs.c (vect_get_data_access_cost):
21936 Workaround for SLP handling.
21937 (vect_enhance_data_refs_alignment):
21938 Compute costs for doing no peeling at all, compare to the best
21939 peeling costs so far and avoid peeling if cheaper.
21941 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21943 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21944 Return peeling info and set costs to zero for unlimited cost
21946 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21947 with unknown misalignment. Compute and costs for unknown
21948 misalignment, compare them to the costs for known misalignment
21949 and choose the cheapest for peeling.
21951 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21953 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21954 (vect_get_peeling_costs_all_drs): Create function.
21955 (vect_peeling_hash_get_lowest_cost):
21956 Use vect_get_peeling_costs_all_drs.
21957 (vect_peeling_supportable): Create function.
21958 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21960 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21962 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21963 DR_HAS_NEGATIVE_STEP.
21964 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21965 (vect_enhance_data_refs_alignment): Use.
21966 (vect_duplicate_ssa_name_ptr_info): Use.
21967 * tree-vectorizer.h (dr_misalignment): Use.
21968 (known_alignment_for_access_p): Use.
21970 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21973 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21975 (has_section_name): New function.
21977 2017-05-30 Martin Liska <mliska@suse.cz>
21980 * auto-profile.c (get_function_decl_from_block): Fix
21983 2017-05-30 Richard Biener <rguenther@suse.de>
21985 PR middle-end/80876
21986 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21988 2017-05-30 Martin Liska <mliska@suse.cz>
21990 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21991 * dumpfile.h (struct dump_file_info): Remove ctors.
21993 2017-05-30 Martin Liska <mliska@suse.cz>
21995 * predict.def: Fix GNU coding style.
21997 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21999 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22000 Mark 'to' argument with ATTRIBUTE_UNUSED.
22002 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
22004 * config/xtensa/xtensa.c (xtensa_emit_call): Use
22005 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
22006 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
22009 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
22011 * doc/install.texi (Options specification): Restore entry of
22012 --enable-sjlj-exceptions.
22014 2017-05-27 Michael Eager <eager@eagercon.com>
22017 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
22019 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
22021 * config/microblaze/microblaze.h
22022 (FIXED_REGISTERS): Update in macro.
22023 (CALL_USED_REGISTERS): Update in macro.
22025 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22027 * doc/install.texi: Add links to macOS binary distributions.
22029 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22033 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22035 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22037 2017-05-26 Martin Liska <mliska@suse.cz>
22039 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22041 2017-05-26 Martin Liska <mliska@suse.cz>
22043 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22044 always leading ';; '.
22045 (dump_bb_info): Likewise.
22046 (brief_dump_cfg): Likewise.
22047 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22048 * dumpfile.c: Remove usage of TDF_VERBOSE.
22049 * dumpfile.h (enum dump_kind): Likewise.
22050 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22051 * print-tree.c (debug_verbose): Remove.
22052 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22053 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22054 * tree-diagnostic.c (default_tree_printer): Replace
22055 TDF_DIAGNOSTIC with TDF_SLIM.
22057 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22059 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22060 in parameter loop, rather than loop_vinfo.
22061 (create_intersect_range_checks): Ditto.
22062 (vect_create_cond_for_alias_checks): Update call to above functions.
22064 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22066 PR tree-optimization/80815
22067 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22068 for merging runtime alias checks. Handle negative DR_STEPs.
22070 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22072 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22074 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22075 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22076 out code pruning runtime alias checks.
22077 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22078 factored out from above.
22079 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22081 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22083 (prune_runtime_alias_test_list): New decalaration.
22085 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22087 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22088 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22089 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22090 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22091 (operator==, comp_dr_with_seg_len_pair): Ditto.
22092 (vect_prune_runtime_alias_test_list): Ditto.
22094 2017-05-26 Martin Liska <mliska@suse.cz>
22097 * params.def: Bound partial-inlining-entry-probability param.
22099 2017-05-26 Marek Polacek <polacek@redhat.com>
22102 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22105 2017-05-26 Richard Biener <rguenther@suse.de>
22107 PR tree-optimization/80842
22108 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22111 2017-05-26 Richard Biener <rguenther@suse.de>
22113 PR tree-optimization/80844
22114 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22116 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22118 * doc/md.texi (Machine Constraints): Update x86 family
22119 machine constraints section to match 'config/i386/constraints.md'.
22121 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22123 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22125 2017-05-25 Nathan Sidwell <nathan@acm.org>
22127 * doc/invoke.texi (--enable-languages): Update documentation.
22129 2017-05-25 Martin Liska <mliska@suse.cz>
22131 * dumpfile.c: Add TDF_FOLDING.
22132 * dumpfile.h (enum dump_kind): Likewise.
22133 * genmatch.c (dt_simplify::gen_1): Use it.
22135 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22137 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22139 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22141 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22142 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22144 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22146 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22147 * match.pd (X == C): Rewrite it here.
22148 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22149 with_certain_nonzero_bits2): New predicates.
22150 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22152 2017-05-24 Nathan Sidwell <nathan@acm.org>
22154 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22157 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22160 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22162 * config/powerpcspe: New port. Files are copied from the rs6000
22163 port, with "rs6000" in filenames replaced by "powerpcspe".
22165 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22167 PR rtl-optimization/80754
22168 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22170 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22172 * config/sparc/sparc.md (length): Return the correct value for -mflat
22173 sibcalls to match output_sibcall.
22175 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22179 * config/rs6000/rs6000.c (struct machine_function): Add new field
22181 (rs6000_get_separate_components): Init that field, use it.
22182 (rs6000_components_for_bb): Use the field.
22184 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22186 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22188 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22190 PR middle-end/80823
22191 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22193 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22196 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22197 address against address_operand predicate.
22198 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22200 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22202 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22203 parameters passed indirectly.
22205 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22207 * config/i386/i386.md (*movdi_internal): Remove SSE4
22208 alternative 18 (?r, *v). Update insn attributes.
22209 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22210 Update insn attributes.
22211 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22212 Update insn attributes.
22213 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22214 alternative 1 (r, v). Remove isa attribute.
22215 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22216 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22217 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22219 2017-05-23 Tom de Vries <tom@codesourcery.com>
22221 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22224 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22226 * cgraphunit.c (symbol_table::process_new_functions): Update.
22227 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22228 (inline_generate_summary): Rename to ...
22229 (ipa_fn_summary_generate): ... this one.
22230 (inline_read_summary): Rename to ...
22231 (ipa_fn_summary_read): ... this one.
22232 (inline_write_summary): Rename to ...
22233 (ipa_fn_summary_write): ... this one.
22234 (inline_free_summary): Rename to ...
22235 (ipa_free_fn_summary): ... this one.
22236 (pass_data_local_fn_summary, pass_local_fn_summary,
22237 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22238 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22239 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22240 make_pass_ipa_fn_summary): New.
22241 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22242 inline_write_summary, inline_free_summary): Remove.
22243 (ipa_free_fn_summary) : New.
22244 * ipa-inline.c (ipa_inline): Update.
22245 (pass_ipa_inline): Do not generate summaries.
22246 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22248 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22249 and add pass_ipa_fn_summary.
22250 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22252 (make_pass_inline_parameters): Remove.
22254 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22256 * omp-low.c (struct omp_context): Remove "default_kind" member.
22259 * omp-offload.c (execute_oacc_device_lower): Remove the
22260 parallelism dimensions function attributes for unparallelized
22261 OpenACC kernels constructs.
22263 2017-05-23 Martin Liska <mliska@suse.cz>
22265 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22267 (cgraph_edge::make_speculative): Likewise.
22268 (cgraph_edge::resolve_speculation): Likewise.
22269 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22270 (cgraph_node::dump): Likewise.
22271 * cgraph.h: Likewise.
22272 * cgraphunit.c (analyze_functions): Likewise.
22273 (symbol_table::compile): Likewise.
22274 * ipa-cp.c (print_all_lattices): Likewise.
22275 (determine_versionability): Likewise.
22276 (initialize_node_lattices): Likewise.
22277 (ipcp_verify_propagated_values): Likewise.
22278 (estimate_local_effects): Likewise.
22279 (update_profiling_info): Likewise.
22280 (create_specialized_node): Likewise.
22281 (perhaps_add_new_callers): Likewise.
22282 (decide_about_value): Likewise.
22283 (decide_whether_version_node): Likewise.
22284 (identify_dead_nodes): Likewise.
22285 (ipcp_store_bits_results): Likewise.
22286 * ipa-devirt.c (dump_targets): Likewise.
22287 (ipa_devirt): Likewise.
22288 * ipa-icf.c (sem_item::dump): Likewise.
22289 (sem_function::equals): Likewise.
22290 (sem_variable::equals): Likewise.
22291 (sem_item_optimizer::read_section): Likewise.
22292 (sem_item_optimizer::execute): Likewise.
22293 (congruence_class::dump): Likewise.
22294 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22295 (dump_inline_summary): Likewise.
22296 (estimate_node_size_and_time): Likewise.
22297 (inline_analyze_function): Likewise.
22298 * ipa-inline-transform.c (inline_call): Likewise.
22299 * ipa-inline.c (report_inline_failed_reason): Likewise.
22300 (want_early_inline_function_p): Likewise.
22301 (edge_badness): Likewise.
22302 (update_edge_key): Likewise.
22303 (inline_small_functions): Likewise.
22304 * ipa-profile.c (ipa_profile): Likewise.
22305 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22306 (ipa_make_edge_direct_to_target): Likewise.
22307 (remove_described_reference): Likewise.
22308 (ipa_impossible_devirt_target): Likewise.
22309 (propagate_controlled_uses): Likewise.
22310 (ipa_print_node_params): Likewise.
22311 (ipcp_transform_function): Likewise.
22312 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22313 (propagate_pure_const): Likewise.
22314 * ipa-reference.c (generate_summary): Likewise.
22315 (read_write_all_from_decl): Likewise.
22316 (propagate): Likewise.
22317 (ipa_reference_read_optimization_summary): Likewise.
22318 * ipa-utils.c (ipa_merge_profiles): Likewise.
22319 * ipa.c (walk_polymorphic_call_targets): Likewise.
22320 (symbol_table::remove_unreachable_nodes): Likewise.
22321 (ipa_single_use): Likewise.
22322 * passes.c (execute_todo): Likewise.
22323 * predict.c (drop_profile): Likewise.
22324 * symtab.c (symtab_node::get_dump_name): New function.
22325 (symtab_node::dump_name): Likewise.
22326 (symtab_node::dump_asm_name): Likewise.
22327 (symtab_node::dump_references): Likewise.
22328 (symtab_node::dump_referring): Likewise.
22329 (symtab_node::dump_base): Likewise.
22330 (symtab_node::debug_symtab): Likewise.
22331 * tree-sra.c (convert_callers_for_node): Likewise.
22332 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22333 * value-prof.c (init_node_map): Likewise.
22335 2017-05-23 Martin Liska <mliska@suse.cz>
22337 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22338 and symtab_node::debug_symtab to symbol_table::debug.
22339 * cgraphunit.c (analyze_functions): Use the renamed function.
22340 (symbol_table::compile): Likewise.
22341 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22342 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22343 * passes.c (execute_todo): Likewise.
22344 * symtab.c (symbol_table::dump): New function.
22345 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22347 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22349 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22350 that nonconst implies exec.
22352 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22354 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22355 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22356 (inline_edge_summary_vec): Turn into ...
22357 (ipa_call_summaries): ... this one.
22358 (redirect_to_unreachable, edge_set_predicate,
22359 evaluate_properties_for_edge, inline_summary_alloc,
22360 reset_ipa_call_summary, reset_inline_summary,
22361 inline_summary_t::duplicate): Update.
22362 (inline_edge_duplication_hook): Turn to ...
22363 (ipa_call_summary_t::duplicate): ... this one.
22364 (inline_edge_removal_hook): Turn to ...
22365 (ipa_call_summary_t::remove): ... this one.
22366 (dump_inline_edge_summary): Turn to ...
22367 (dump_ipa_call_summary): ... this one.
22368 (estimate_function_body_sizes): Update.
22369 (inline_update_callee_summaries): Update.
22370 (remap_edge_change_prob): Update.
22371 (remap_edge_summaries): Update.
22372 (inline_merge_summary): Update.
22373 (do_estimate_edge_time): Update.
22374 (inline_generate_summary): Update.
22375 (inline_read_section): Update.
22376 (inline_read_summary): Update.
22377 (inline_free_summary): Update.
22378 * ipa-inline.c (can_inline_edge_p): Update.
22379 (compute_inlined_call_time): Update.
22380 (want_inline_small_function_p): Update.
22381 (edge_badness): Update.
22382 (early_inliner): Update.
22383 * ipa-inline.h (inline_edge_summary): Turn to ...
22384 (ipa_call_summary): ... this one.
22385 (ipa_call_summary_t): New class.
22386 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22387 (ipa_call_summaries): New.
22388 (inline_edge_summary): Remove.
22389 (estimate_edge_growth): Update.
22390 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22391 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22392 * ipa-split.c (execute_split_functions): Update.
22393 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22395 2017-05-23 Tom de Vries <tom@codesourcery.com>
22397 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22398 attributes): Document rdrand effective target.
22400 2017-05-23 Tom de Vries <tom@codesourcery.com>
22402 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22403 attributes): Sort alphabetically.
22405 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22407 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22409 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22412 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22413 V2DF/V2DI splat into two separate patterns, one that handles
22414 registers, and the other that only handles memory. Drop support
22415 for splatting from a GPR on ISA 2.07 and then splitting the
22416 splat into direct move and splat.
22417 (vsx_splat_<mode>_reg): Likewise.
22418 (vsx_splat_<mode>_mem): Likewise.
22420 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22422 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22424 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22426 PR middle-end/80809
22427 * omp-low.c (finish_taskreg_remap): New function.
22428 (finish_taskreg_scan): If unit size of ctx->record_type
22429 is non-constant, unshare the size expression and replace
22430 decls in it with possible outer var refs.
22432 PR middle-end/80809
22433 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22434 GOVD_SHARED rather than GOVD_PRIVATE with it.
22435 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22436 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22438 PR middle-end/80853
22439 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22440 as last argument to build_outer_var_ref for pointer bases of array
22441 section reductions.
22443 2017-05-19 Martin Sebor <msebor@redhat.com>
22445 * print-tree.c (print_node): Print DECL_READ_P flag.
22447 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22449 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22450 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22451 * cgraph.c: Likewise.
22452 * cgraphunit.c: Likewise.
22453 * gengtype.c: Likewise.
22454 * ipa-cp.c: Likewise.
22455 * ipa-devirt.c: Likewise.
22456 * ipa-icf.c: Likewise.
22457 * ipa-predicate.c: Likewise.
22458 * ipa-profile.c: Likewise.
22459 * ipa-prop.c: Likewise.
22460 * ipa-split.c: Likewise.
22462 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22463 edge_predicate_pool, dump_inline_hints,
22464 inline_summary::account_size_time, redirect_to_unreachable,
22465 edge_set_predicate, set_hint_predicate,
22466 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22467 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22468 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22469 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22470 ipa_call_summary_t::remove, initialize_growth_caches,
22471 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22472 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22473 mark_modified, unmodified_parm_1, unmodified_parm,
22474 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22475 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22476 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22477 will_be_nonconstant_predicate, record_modified_bb_info,
22478 get_minimal_bb, record_modified, param_change_prob,
22479 phi_result_unknown_predicate, predicate_for_phi_result,
22480 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22481 estimate_function_body_sizes, compute_inline_parameters,
22482 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22483 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22484 inline_update_callee_summaries, remap_edge_change_prob,
22485 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22486 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22487 inline_analyze_function, inline_summary_t::insert,
22488 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22489 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22490 inline_free_summary): Move to ipa-fnsummary.h
22491 (predicate_t): Remove.
22492 * ipa-fnsummary.c: New file.
22493 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22494 (enum inline_hints_vals, inline_hints, agg_position_info,
22495 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22496 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22497 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22498 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22499 inline_read_summary, inline_write_summary, inline_free_summary,
22500 inline_analyze_function, initialize_inline_failed,
22501 inline_merge_summary, inline_update_overall_summary,
22502 compute_inline_parameters): Move to ipa-fnsummary.h
22503 * ipa-fnsummary.h: New file.
22504 * ipa-inline-transform.h: Include ipa-inline.h.
22505 * ipa-inline.c: LIkewise.
22507 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22509 * ipa-inline.c (edge_badness): Use inlined_time instead of
22510 inline_summaries->get.
22512 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22514 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22516 2017-05-22 Nathan Sidwell <nathan@acm.org>
22518 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22519 (fdump-lang): Document 'raw' option.
22520 * dumpfile.h (TDI_tu): Delete.
22521 * dumpfile.c (dump_files): Remove translation-unit.
22522 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22524 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22526 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22527 command option from $(AWK) call.
22528 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22529 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22530 [FORMAT]: Remove handling of variable.
22531 * config/avr/t-multilib: Regenerate.
22533 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22535 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22537 (dump_inline_summary): Do not print self_time.
22538 (estimate_function_body_sizes): Do not set self_time.
22539 (compute_inline_parameters): Likewise.
22540 (inline_read_section, inline_write_summary): Do not stream self_time.
22541 * ipa-inline.h (inline_summary): Drop self_time.
22543 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22545 * ipa-inline-analysis.c (account_size_time): Rename to ...
22546 (inline_summary::account_size_time): ... this one.
22547 (reset_ipa_call_summary): Turn to ...
22548 (ipa_call_summary::reset): ... this one.
22549 (reset_inline_summary): Turn to ...
22550 (inline_summary::reset): ... this one.
22551 (inline_summary_t::remove): Update.
22552 (inline_summary_t::duplicate): Update.
22553 (ipa_call_summary_t::remove): Update.
22554 (dump_inline_summary): Update.
22555 (estimate_function_body_sizes): Update.
22556 (compute_inline_parameters): Update.
22557 (estimate_node_size_and_time): Update.
22558 (inline_merge_summary): Update.
22559 (inline_update_overall_summary): Update.
22560 (inline_read_section): Update.
22561 (inline_write_summary): Update.
22562 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22563 add account_size_time and reset member functions.
22564 (ipa_call_summary): Add reset function.
22565 * ipa-predicate.h (predicate::operator &): Constify.
22567 2017-05-22 Richard Biener <rguenther@suse.de>
22569 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22571 2017-05-19 Jason Merrill <jason@redhat.com>
22573 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22575 2017-05-19 Marek Polacek <polacek@redhat.com>
22578 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22579 TYPE_OVERFLOW_WRAPS checks.
22581 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22583 * tree-core.h (enum omp_clause_default_kind): Add
22584 "OMP_CLAUSE_DEFAULT_PRESENT".
22585 * tree-pretty-print.c (dump_omp_clause): Handle it.
22586 * gimplify.c (enum gimplify_omp_var_data): Add
22587 "GOVD_MAP_FORCE_PRESENT".
22588 (gimplify_adjust_omp_clauses_1): Map it to
22589 "GOMP_MAP_FORCE_PRESENT".
22590 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22592 * gimplify.c (oacc_default_clause): Clarify.
22594 2017-05-19 Nathan Sidwell <nathan@acm.org>
22596 LANG_HOOK_REGISTER_DUMPS
22597 * toplev.c (general_init): Call register dump lang hook.
22598 * doc/invoke.texi: Document -fdump-lang option family.
22599 * dumpfile.c (dump_files): Remove class dump here.
22600 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22601 * dumpfile.h (tree_dump_index): Remove TDI_class.
22602 * langhooks-def.h (lhd_register_dumps): Declare.
22603 (LANG_HOOKS_REGISTER_DUMPS): Define.
22604 (LANG_HOOKS_INITIALIZER): Add it.
22605 * langhooks.c (lhd_register_dumps): Define.
22606 * langhooks.h (struct lang_hooks): Add register_dumps.
22608 2017-05-19 Nathan Sidwell <nathan@acm.org>
22610 * context.h (context::set_passes): New.
22611 * context.c (context::context): Do not create pass manager.
22612 * toplev.c (general_init): Create pass manager here.
22614 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22616 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22617 use this splitter if two add or or instructions would also work for
22618 the constant we want to generate.
22620 2017-05-19 Richard Biener <rguenther@suse.de>
22623 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22624 predicate evaluation.
22626 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22628 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22630 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22633 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22635 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22636 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22637 (inline_edge_summary_vec): Turn into ...
22638 (ipa_call_summaries): ... this one.
22639 (redirect_to_unreachable, edge_set_predicate,
22640 evaluate_properties_for_edge, inline_summary_alloc,
22641 reset_ipa_call_summary, reset_inline_summary,
22642 inline_summary_t::duplicate): Update.
22643 (inline_edge_duplication_hook): Turn to ...
22644 (ipa_call_summary_t::duplicate): ... this one.
22645 (inline_edge_removal_hook): Turn to ...
22646 (ipa_call_summary_t::remove): ... this one.
22647 (dump_inline_edge_summary): Turn to ...
22648 (dump_ipa_call_summary): ... this one.
22649 (estimate_function_body_sizes): Update.
22650 (inline_update_callee_summaries): Update.
22651 (remap_edge_change_prob): Update.
22652 (remap_edge_summaries): Update.
22653 (inline_merge_summary): Update.
22654 (do_estimate_edge_time): Update.
22655 (inline_generate_summary): Update.
22656 (inline_read_section): Update.
22657 (inline_read_summary): Update.
22658 (inline_free_summary): Update.
22659 * ipa-inline.c (can_inline_edge_p): Update.
22660 (compute_inlined_call_time): Update.
22661 (want_inline_small_function_p): Update.
22662 (edge_badness): Update.
22663 (early_inliner): Update.
22664 * ipa-inline.h (inline_edge_summary): Turn to ...
22665 (ipa_call_summary): ... this one.
22666 (ipa_call_summary_t): New class.
22667 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22668 (ipa_call_summaries): New.
22669 (inline_edge_summary): Remove.
22670 (estimate_edge_growth): Update.
22671 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22672 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22673 * ipa-split.c (execute_split_functions): Update.
22674 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22676 2017-05-19 Richard Biener <rguenther@suse.de>
22678 PR middle-end/80764
22679 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22681 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22683 * config/rs6000/rs6000.c (struct machine_function): Add field
22684 fpr_is_wrapped_separately.
22685 (rs6000_get_separate_components): Use 64 components. Handle the
22686 new FPR components.
22687 (rs6000_components_for_bb): Handle the FPR components.
22688 (rs6000_emit_prologue_components): Handle the FPR components.
22689 (rs6000_emit_epilogue_components): Handle the FPR components.
22690 (rs6000_set_handled_components): Handle the FPR components.
22691 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22692 that are already separately shrink-wrapped.
22693 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22694 that are already separately shrink-wrapped.
22696 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22699 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22702 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22703 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22704 cases where the register allocator uses a move and an offsettable
22705 memory operation to/from a FPR register on ISA 2.06/2.07.
22706 (define_peephole2 for Altivec d-form store): Likewise.
22708 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22711 * config/i386/mmx.md (*mov<mode>_internal): Enable
22712 alternatives 11, 12, 13 and 14 also for 32bit targets.
22713 Remove alternatives 15, 16, 17 and 18.
22714 * config/i386/sse.md (vec_concatv2di): Change
22715 alternative (!x, *y) to (x, ?!*Yn).
22717 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22719 * dumpfile.h (enum dump_kind): Remove stray comma.
22721 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22723 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22724 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22725 predicate::num_conditions
22726 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22727 (CHANGED): turn into predicate::changed.
22728 (agg_position_info): Move to ipa-predicate.h
22729 (add_condition, predicate::add_clause, predicate::operator &=,
22730 predicate::or_with, predicate::evaluate, predicate::probability,
22731 dump_condition, dump_clause, predicate::dump,
22732 predicate::remap_after_duplication, predicate::remap_after_inlining,
22733 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22734 (evaluate_conditions_for_known_args): Update.
22735 (set_cond_stmt_execution_predicate): Update.
22736 * ipa-inline.h: Include ipa-predicate.h
22737 (condition, inline_param_summary, conditions, agg_position_info,
22738 predicate): Move to ipa-predicate.h
22739 * ipa-predicate.c: New file.
22740 * ipa-predicate.h: New file.
22742 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22744 * final.c (leaf_function_p): Check we are not in a sequence.
22746 2017-05-18 Martin Liska <mliska@suse.cz>
22748 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22749 * dumpfile.c (dump_register): Use new enum dump_kind.
22750 (get_dump_file_name): Likewise.
22751 (dump_enable_all): Likewise.
22752 (dump_switch_p_1): Likewise.
22753 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22754 * dumpfile.h (enum dump_kind): New enum type.
22755 (struct dump_file_info): Create constructor and
22756 format fields and comments.
22757 * passes.c (pass_manager::register_one_dump_file):
22759 * statistics.c (statistics_early_init): Likewise.
22760 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22761 TDF_TREE with TDF_SLIM.
22762 (gather_memory_references_ref): Likewise.
22764 2017-05-18 Martin Liska <mliska@suse.cz>
22766 * vec.h (struct vnull): Use it.
22768 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22770 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22771 (true_predicate, false_predicate, true_predicate_p,
22772 false_predicate_p): Remove.
22773 (single_cond_predicate, not_inlined_predicate): Turn to member function
22775 (add_condition): Update.
22776 (add_clause): Turn to...
22777 (predicate::add_clause): ... this one; update; allow passing NULL
22779 (and_predicates): Turn to ...
22780 (predicate::operator &=): ... this one.
22781 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22782 (or_predicates): Turn to ...
22783 (predicate::or_with): ... this one.
22784 (evaluate_predicate): Turn to ...
22785 (predicate::evaluate): ... this one.
22786 (predicate_probability): Turn to ...
22787 (predicate::probability): ... this one.
22788 (dump_condition): Update.
22789 (dump_predicate): Turn to ...
22790 (predicate::dump): ... this one.
22791 (account_size_time): Update.
22792 (edge_set_predicate): Update.
22793 (set_hint_predicate): UPdate.
22794 (evaluate_conditions_for_known_args): Update.
22795 (evaluate_properties_for_edge): Update.
22796 (remap_predicate_after_duplication): Turn to...
22797 (predicate::remap_after_duplication): ... this one.
22798 (remap_hint_predicate_after_duplication): Update.
22799 (inline_summary_t::duplicate): UPdate.
22800 (dump_inline_edge_summary): Update.
22801 (dump_inline_summary): Update.
22802 (set_cond_stmt_execution_predicate): Update.
22803 (set_switch_stmt_execution_predicate): Update.
22804 (compute_bb_predicates): Update.
22805 (will_be_nonconstant_expr_predicate): Update.
22806 (will_be_nonconstant_predicate): Update.
22807 (phi_result_unknown_predicate): Update.
22808 (predicate_for_phi_result): Update.
22809 (array_index_predicate): Update.
22810 (estimate_function_body_sizes): Update.
22811 (estimate_node_size_and_time): Update.
22812 (estimate_ipcp_clone_size_and_time): Update.
22813 (remap_predicate): Rename to ...
22814 (predicate::remap_after_inlining): ... this one.
22815 (remap_hint_predicate): Update.
22816 (inline_merge_summary): Update.
22817 (inline_update_overall_summary): Update.
22818 (estimate_size_after_inlining): Update.
22819 (read_predicate): Rename to ...
22820 (predicate::stream_in): ... this one.
22821 (read_inline_edge_summary): Update.
22822 (write_predicate): Rename to ...
22823 (predicate::stream_out): ... this one.
22824 (write_inline_edge_summary): Update.
22825 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22826 (clause_t): Turn to uint32_t
22827 (predicate): Turn to class; implement constructor and operators
22829 (size_time_entry): Update.
22830 (inline_summary): Update.
22831 (inline_edge_summary): Update.
22833 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22835 * fold-const.c (fold_binary_loc): Move transformation...
22836 * match.pd (C - X CMP X): ... here.
22838 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22840 * config/sparc/sparc.c (sparc_option_override): Set function
22841 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22842 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22844 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22846 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22848 2017-05-18 Marek Polacek <polacek@redhat.com>
22851 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22852 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22853 gimple_assign_load_p.
22855 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22857 PR middle-end/80692
22858 * real.c (do_compare): Give decimal_do_compare preference over
22859 comparing just the signs.
22861 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22863 * doc/md.texi (Canonicalization of Instructions): Describe the
22864 canonical form of instructions that inherently set a condition
22867 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22869 PR middle-end/80775
22870 * tree-cfg.c: Move deletion of unreachable case statements to after
22871 the merging of consecutive case labels.
22873 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22875 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22876 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22877 restoring of callee-saved registers.
22879 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22881 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22882 * config/visium/visium.c (single_set_and_flags): Likewise.
22883 * config/visium/visium.md (Substitutions): Likewise.
22885 2017-05-17 Martin Liska <mliska@suse.cz>
22887 * cfg.c: Introduce dump_flags_t type and
22888 use it instead of int type.
22890 * cfghooks.c: Likewise.
22891 * cfghooks.h (struct cfg_hooks): Likewise.
22892 * cfgrtl.c: Likewise.
22893 * cfgrtl.h: Likewise.
22894 * cgraph.c (cgraph_node::get_body): Likewise.
22895 * coretypes.h: Likewise.
22896 * domwalk.c: Likewise.
22897 * domwalk.h: Likewise.
22898 * dumpfile.c (struct dump_option_value_info): Likewise.
22899 (dump_enable_all): Likewise.
22900 (dump_switch_p_1): Likewise.
22901 (opt_info_switch_p): Likewise.
22902 * dumpfile.h (enum tree_dump_index): Likewise.
22903 (struct dump_file_info): Likewise.
22904 * genemit.c: Likewise.
22905 * generic-match-head.c: Likewise.
22906 * gengtype.c (open_base_files): Likewise.
22907 * gimple-pretty-print.c: Likewise.
22908 * gimple-pretty-print.h: Likewise.
22909 * graph.c (print_graph_cfg): Likewise.
22910 * graphite-scop-detection.c (dot_all_sese): Likewise.
22911 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22912 * loop-unroll.c (report_unroll): Likewise.
22913 * passes.c (pass_manager::register_one_dump_file): Likewise.
22914 * print-tree.c: Likewise.
22915 * statistics.c: Likewise.
22916 * tree-cfg.c: Likewise.
22917 * tree-cfg.h: Likewise.
22918 * tree-dfa.c: Likewise.
22919 * tree-dfa.h: Likewise.
22920 * tree-dump.c (dump_function): Likewise.
22921 * tree-dump.h (struct dump_info): Likewise.
22922 * tree-pretty-print.c: Likewise.
22923 * tree-pretty-print.h: Likewise.
22924 * tree-ssa-live.c: Likewise.
22925 * tree-ssa-live.h: Likewise.
22926 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22927 * tree-vect-loop.c: Likewise.
22928 * tree-vect-slp.c: Likewise.
22930 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22931 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22933 PR tree-optimization/80457
22934 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22935 of all arguments to a statement as scalar_to_vec operations.
22936 (vectorizable_call): Adjust call to vect_model_simple_cost for
22938 (vectorizable_conversion): Likewise.
22939 (vectorizable_assignment): Likewise.
22940 (vectorizable_shift): Likewise.
22941 (vectorizable_operation): Likewise.
22942 (vectorizable_comparison): Likewise.
22943 (vect_is_simple_cond): Record the def types for operands.
22944 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22945 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22946 for statement argument count.
22948 2017-05-16 Carl Love <cel@us.ibm.com>
22950 * config/rs6000/rs6000-c: Add support for built-in functions
22951 vector unsigned long long vec_bperm (vector unsigned long long,
22952 vector unsigned char)
22953 vector signed long long vec_mule (vector signed int,
22955 vector unsigned long long vec_mule (vector unsigned int,
22956 vector unsigned int)
22957 vector signed long long vec_mulo (vector signed int,
22959 vector unsigned long long vec_mulo (vector unsigned int,
22960 vector unsigned int)
22961 vector signed char vec_sldw (vector signed char,
22962 vector signed char,
22964 vector unsigned char vec_sldw (vector unsigned char,
22965 vector unsigned char,
22967 vector signed short vec_sldw (vector signed short,
22968 vector signed short,
22970 vector unsigned short vec_sldw (vector unsigned short,
22971 vector unsigned short,
22973 vector signed int vec_sldw (vector signed int,
22976 vector unsigned int vec_sldw (vector unsigned int,
22977 vector unsigned int,
22979 vector signed long long vec_sldw (vector signed long long,
22980 vector signed long long,
22982 vector unsigned long long vec_sldw (vector unsigned long long,
22983 vector unsigned long long,
22985 * config/rs6000/rs6000-c: Add support for built-in functions
22986 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22987 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22988 * doc/extend.texi: Update the built-in documentation for the
22989 new built-in functions.
22991 2017-05-16 Marek Polacek <polacek@redhat.com>
22995 * tree.c (save_expr): Don't fold the expression.
22997 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22999 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
23000 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
23001 and (?*y,m). Update insn attributes.
23003 2017-05-16 Martin Liska <mliska@suse.cz>
23005 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
23006 flags argument of print_gimple_stmt, print_gimple_expr,
23007 print_generic_stmt and print_generic_expr.
23008 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
23009 * coretypes.h: Likewise.
23010 * except.c (dump_eh_tree): Likewise.
23011 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
23012 * gimple-pretty-print.h: Likewise.
23013 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
23014 (backprop::push_to_worklist): Likewise.
23015 (backprop::pop_from_worklist): Likewise.
23016 (backprop::process_use): Likewise.
23017 (backprop::intersect_uses): Likewise.
23018 (note_replacement): Likewise.
23019 * gimple-ssa-store-merging.c
23020 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
23021 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
23022 (pass_store_merging::execute): Likewise.
23023 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
23024 (ssa_base_cand_dump_callback): Likewise.
23025 (dump_incr_vec): Likewise.
23026 (replace_refs): Likewise.
23027 (replace_mult_candidate): Likewise.
23028 (create_add_on_incoming_edge): Likewise.
23029 (create_phi_basis): Likewise.
23030 (insert_initializers): Likewise.
23031 (all_phi_incrs_profitable): Likewise.
23032 (introduce_cast_before_cand): Likewise.
23033 (replace_one_candidate): Likewise.
23034 * gimplify.c (gimplify_expr): Likewise.
23035 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23036 (set_rename): Likewise.
23037 (rename_uses): Likewise.
23038 (copy_loop_phi_nodes): Likewise.
23039 (add_close_phis_to_merge_points): Likewise.
23040 (copy_loop_close_phi_args): Likewise.
23041 (copy_cond_phi_args): Likewise.
23042 (graphite_copy_stmts_from_block): Likewise.
23043 (translate_pending_phi_nodes): Likewise.
23044 * graphite-poly.c (print_pdr): Likewise.
23045 (dump_gbb_cases): Likewise.
23046 (dump_gbb_conditions): Likewise.
23047 (print_scop_params): Likewise.
23048 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23049 (build_cross_bb_scalars_use): Likewise.
23050 (gather_bbs::before_dom_children): Likewise.
23051 * hsa-dump.c (dump_hsa_immed): Likewise.
23052 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23053 (get_replacement_map): Likewise.
23054 * ipa-inline-analysis.c (dump_condition): Likewise.
23055 (estimate_function_body_sizes): Likewise.
23056 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23057 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23058 * ipa-prop.c (ipa_dump_param): Likewise.
23059 (ipa_print_node_jump_functions_for_edge): Likewise.
23060 (ipa_modify_call_arguments): Likewise.
23061 (ipa_modify_expr): Likewise.
23062 (ipa_dump_param_adjustments): Likewise.
23063 (ipa_dump_agg_replacement_values): Likewise.
23064 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23065 * ipa-pure-const.c (check_stmt): Likewise.
23066 (pass_nothrow::execute): Likewise.
23067 * ipa-split.c (execute_split_functions): Likewise.
23068 * omp-offload.c (dump_oacc_loop_part): Likewise.
23069 (dump_oacc_loop): Likewise.
23070 * trans-mem.c (tm_log_emit): Likewise.
23071 (tm_memopt_accumulate_memops): Likewise.
23072 (dump_tm_memopt_set): Likewise.
23073 (dump_tm_memopt_transform): Likewise.
23074 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23075 (print_loop): Likewise.
23076 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23077 (chkp_gather_checks_info): Likewise.
23078 (chkp_get_check_result): Likewise.
23079 (chkp_remove_check_if_pass): Likewise.
23080 (chkp_use_outer_bounds_if_possible): Likewise.
23081 (chkp_reduce_bounds_lifetime): Likewise.
23082 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23083 (chkp_mark_completed_bounds): Likewise.
23084 (chkp_register_incomplete_bounds): Likewise.
23085 (chkp_mark_invalid_bounds): Likewise.
23086 (chkp_maybe_copy_and_register_bounds): Likewise.
23087 (chkp_build_returned_bound): Likewise.
23088 (chkp_get_bound_for_parm): Likewise.
23089 (chkp_build_bndldx): Likewise.
23090 (chkp_get_bounds_by_definition): Likewise.
23091 (chkp_generate_extern_var_bounds): Likewise.
23092 (chkp_get_bounds_for_decl_addr): Likewise.
23093 * tree-chrec.c (chrec_apply): Likewise.
23094 * tree-data-ref.c (dump_data_reference): Likewise.
23095 (dump_subscript): Likewise.
23096 (dump_data_dependence_relation): Likewise.
23097 (analyze_overlapping_iterations): Likewise.
23098 * tree-inline.c (expand_call_inline): Likewise.
23099 (tree_function_versioning): Likewise.
23100 * tree-into-ssa.c (dump_defs_stack): Likewise.
23101 (dump_currdefs): Likewise.
23102 (dump_names_replaced_by): Likewise.
23103 (dump_update_ssa): Likewise.
23104 (update_ssa): Likewise.
23105 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23106 * tree-parloops.c (build_new_reduction): Likewise.
23107 (try_create_reduction_list): Likewise.
23108 (ref_conflicts_with_region): Likewise.
23109 (oacc_entry_exit_ok_1): Likewise.
23110 (oacc_entry_exit_single_gang): Likewise.
23111 * tree-pretty-print.h: Likewise.
23112 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23113 (get_scalar_evolution): Likewise.
23114 (add_to_evolution): Likewise.
23115 (get_loop_exit_condition): Likewise.
23116 (analyze_evolution_in_loop): Likewise.
23117 (analyze_initial_condition): Likewise.
23118 (analyze_scalar_evolution): Likewise.
23119 (instantiate_scev): Likewise.
23120 (number_of_latch_executions): Likewise.
23121 (gather_chrec_stats): Likewise.
23122 (final_value_replacement_loop): Likewise.
23123 (scev_const_prop): Likewise.
23124 * tree-sra.c (dump_access): Likewise.
23125 (disqualify_candidate): Likewise.
23126 (create_access): Likewise.
23127 (reject): Likewise.
23128 (maybe_add_sra_candidate): Likewise.
23129 (create_access_replacement): Likewise.
23130 (analyze_access_subtree): Likewise.
23131 (analyze_all_variable_accesses): Likewise.
23132 (sra_modify_assign): Likewise.
23133 (initialize_constant_pool_replacements): Likewise.
23134 (find_param_candidates): Likewise.
23135 (decide_one_param_reduction): Likewise.
23136 (replace_removed_params_ssa_names): Likewise.
23137 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23138 * tree-ssa-copy.c (dump_copy_of): Likewise.
23139 (copy_prop_visit_cond_stmt): Likewise.
23140 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23141 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23142 (record_equivalences_from_stmt): Likewise.
23143 * tree-ssa-dse.c (compute_trims): Likewise.
23144 (delete_dead_call): Likewise.
23145 (delete_dead_assignment): Likewise.
23146 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23147 (forward_propagate_into_cond): Likewise.
23148 (pass_forwprop::execute): Likewise.
23149 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23150 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23152 (move_computations_worker): Likewise.
23153 (execute_sm): Likewise.
23154 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23155 (remove_exits_and_undefined_stmts): Likewise.
23156 (remove_redundant_iv_tests): Likewise.
23157 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23158 (adjust_iv_update_pos): Likewise.
23159 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23160 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23161 (value_replacement): Likewise.
23162 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23163 * tree-ssa-pre.c (print_pre_expr): Likewise.
23164 (get_representative_for): Likewise.
23165 (create_expression_by_pieces): Likewise.
23166 (insert_into_preds_of_block): Likewise.
23167 (eliminate_insert): Likewise.
23168 (eliminate_dom_walker::before_dom_children): Likewise.
23169 (eliminate): Likewise.
23170 (remove_dead_inserted_code): Likewise.
23171 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23172 * tree-ssa-reassoc.c (get_rank): Likewise.
23173 (eliminate_duplicate_pair): Likewise.
23174 (eliminate_plus_minus_pair): Likewise.
23175 (eliminate_not_pairs): Likewise.
23176 (undistribute_ops_list): Likewise.
23177 (eliminate_redundant_comparison): Likewise.
23178 (update_range_test): Likewise.
23179 (optimize_range_tests_var_bound): Likewise.
23180 (optimize_vec_cond_expr): Likewise.
23181 (rewrite_expr_tree): Likewise.
23182 (rewrite_expr_tree_parallel): Likewise.
23183 (linearize_expr): Likewise.
23184 (break_up_subtract): Likewise.
23185 (linearize_expr_tree): Likewise.
23186 (attempt_builtin_powi): Likewise.
23187 (attempt_builtin_copysign): Likewise.
23188 (transform_stmt_to_copy): Likewise.
23189 (transform_stmt_to_multiply): Likewise.
23190 (dump_ops_vector): Likewise.
23191 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23192 (print_scc): Likewise.
23193 (set_ssa_val_to): Likewise.
23194 (visit_reference_op_store): Likewise.
23195 (visit_use): Likewise.
23196 (sccvn_dom_walker::before_dom_children): Likewise.
23197 (run_scc_vn): Likewise.
23198 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23200 (expr_hash_elt::print): Likewise.
23201 (const_and_copies::pop_to_marker): Likewise.
23202 (const_and_copies::record_const_or_copy_raw): Likewise.
23203 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23204 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23205 (dump_predicates): Likewise.
23206 (find_uninit_use): Likewise.
23207 (warn_uninitialized_phi): Likewise.
23208 (pass_late_warn_uninitialized::execute): Likewise.
23209 * tree-ssa.c (verify_vssa): Likewise.
23210 (verify_ssa): Likewise.
23211 (maybe_optimize_var): Likewise.
23212 * tree-vrp.c (dump_value_range): Likewise.
23213 (dump_all_value_ranges): Likewise.
23214 (dump_asserts_for): Likewise.
23215 (register_edge_assert_for_2): Likewise.
23216 (vrp_visit_cond_stmt): Likewise.
23217 (vrp_visit_switch_stmt): Likewise.
23218 (vrp_visit_stmt): Likewise.
23219 (vrp_visit_phi_node): Likewise.
23220 (simplify_cond_using_ranges_1): Likewise.
23221 (fold_predicate_in): Likewise.
23222 (evrp_dom_walker::before_dom_children): Likewise.
23223 (evrp_dom_walker::push_value_range): Likewise.
23224 (evrp_dom_walker::pop_value_range): Likewise.
23225 (execute_early_vrp): Likewise.
23227 2017-05-16 Richard Biener <rguenther@suse.de>
23229 * dwarf2out.c (loc_list_from_tree_1): Do not create
23230 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23232 2017-05-16 Richard Biener <rguenther@suse.de>
23234 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23236 (note_variable_value_in_expr): If we resolved the decl ref
23237 do not push to the stack.
23239 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23241 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23242 operations in fast-math mode.
23243 (vaddq_f16): Likewise.
23244 (vmul_f16): Likewise.
23245 (vmulq_f16): Likewise.
23246 (vsub_f16): Likewise.
23247 (vsubq_f16): Likewise.
23248 * config/arm/neon.md (add<mode>3): New.
23250 (fma:<VH:mode>3): New. Also remove outdated comment.
23253 2017-05-16 Martin Liska <mliska@suse.cz>
23257 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23258 (odr_types_equivalent_p): Likewise.
23260 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23262 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23264 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23267 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23268 non-interunit SSE move alternatives with '?'.
23269 (zero-extendsidi peephole2): New peephole to skip intermediate
23270 general register in SSE zero-extend sequence.
23272 2017-05-15 Jeff Law <law@redhat.com>
23274 * reorg.c (relax_delay_slots): Create a new variable to hold
23275 the temporary target rather than clobbering TARGET_LABEL.
23277 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23278 missing argument to extract_bit_field call.
23279 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23281 2017-05-15 Martin Liska <mliska@suse.cz>
23284 * gcc.c (process_command): Do not allow empty argument of -o option.
23286 2017-05-15 Renlin Li <renlin.li@arm.com>
23288 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23289 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23290 * config/aarch64/constraints.md (Usf): Add long call check.
23291 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23292 (call_value): Likewise.
23293 (sibcall): Likewise.
23294 (sibcall_value): Likewise.
23296 (call_value_insn): New.
23297 (sibcall_insn): Update rtx pattern.
23298 (sibcall_value_insn): Likewise.
23299 (call_internal): Remove.
23300 (call_value_internal): Likewise.
23301 (sibcall_internal): Likewise.
23302 (sibcall_value_internal): Likewise.
23303 (call_reg): Likewise.
23304 (call_symbol): Likewise.
23305 (call_value_reg): Likewise.
23306 (call_value_symbol): Likewise.
23308 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23311 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23313 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23315 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23316 compatible with CCGOCmode and with CCZmode.
23318 2017-05-14 Martin Sebor <msebor@redhat.com>
23320 PR middle-end/77671
23321 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23322 (gimple_fold_builtin_snprintf): Same.
23323 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23324 (gimple_fold_builtin_snprintf): Same.
23325 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23326 of character types.
23327 (is_call_safe): New function.
23328 (try_substitute_return_value): Call it.
23329 (try_simplify_call): New function.
23330 (pass_sprintf_length::handle_gimple_call): Call it.
23332 2017-05-14 Martin Sebor <msebor@redhat.com>
23334 PR middle-end/80669
23335 * builtins.c (expand_builtin_stpncpy): Simplify.
23337 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23339 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23340 * config/i386/i386.h
23341 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23342 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23343 (struct machine_function): Add new members call_ms2sysv,
23344 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23345 (struct machine_frame_state): New fields sp_realigned and
23346 sp_realigned_offset.
23347 * config/i386/i386.c
23348 (enum xlogue_stub): New enum.
23349 (enum xlogue_stub_sets): New enum.
23350 (class xlogue_layout): New class.
23351 (struct ix86_frame): New fields stack_realign_allocate_offset,
23352 stack_realign_offset and outlined_save_offset. Modify comments to
23353 detail stack layout when using out-of-line stubs.
23354 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23355 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23356 -mcall-ms2sysv-xlogues.
23357 (stub_managed_regs): New static variable.
23358 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23359 registers managed by out-of-line stub.
23360 (disable_call_ms2sysv_xlogues): New function.
23361 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23362 m->call_ms2sysv when appropriate and compute frame layout for
23364 (sp_valid_at, fp_valid_at): New inline functions.
23365 (choose_basereg): New function.
23366 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23368 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23369 Use align parameter of choose_baseaddr to generated aligned SSE movs
23371 (pro_epilogue_adjust_stack): Modify to track
23372 machine_frame_state::sp_realigned.
23373 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23374 (ix86_nsaved_sseregs): Likewise.
23375 (ix86_emit_save_regs): Likewise.
23376 (ix86_emit_save_regs_using_mov): Likewise.
23377 (ix86_emit_save_sse_regs_using_mov): Likewise.
23378 (get_scratch_register_on_entry): Likewise.
23379 (gen_frame_set): New function.
23380 (gen_frame_load): Likewise.
23381 (gen_frame_store): Likewise.
23382 (emit_outlined_ms2sysv_save): Likewise.
23383 (emit_outlined_ms2sysv_restore): Likewise.
23384 (ix86_expand_prologue): Modify stack re-alignment code and call
23385 emit_outlined_ms2sysv_save when appropriate.
23386 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23387 parameter rtx_insn *insn, which allows the function to be used to only
23388 generate the notes.
23389 (ix86_expand_epilogue): Modify validity checks of frame and stack
23390 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23391 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23392 * config/i386/predicates.md
23393 (save_multiple): New predicate.
23394 (restore_multiple): Likewise.
23395 * config/i386/sse.md
23396 (save_multiple<mode>): New pattern.
23397 (save_multiple_realign<mode>): Likewise.
23398 (restore_multiple<mode>): Likewise.
23399 (restore_multiple_and_return<mode>): Likewise.
23400 (restore_multiple_leave_return<mode>): Likewise.
23401 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23403 2017-05-14 Julia Koval <julia.koval@intel.com>
23405 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23406 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23407 (__builtin_ia32_xsetbv): New builtins.
23408 * config/i386/i386.c (ix86_expand_special_args_builtin):
23410 (ix86_expand_builtin): Special expand for new intrinsics.
23411 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23412 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23413 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23415 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23417 * cfganal.c (inverted_post_order_compute): Change argument type
23419 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23420 * df-core.c (rest_of_handle_df_initialize): Adjust.
23421 (rest_of_handle_df_finish): Likewise.
23422 (df_analyze_1): Likewise.
23423 (df_analyze): Likewise.
23424 (loop_inverted_post_order_compute): Change argument to be a vec *.
23425 (df_analyze_loop): Adjust.
23426 (df_get_n_blocks): Likewise.
23427 (df_get_postorder): Likewise.
23428 * df.h (struct df_d): Change field to be a vec.
23429 * lcm.c (compute_laterin): Adjust.
23430 (compute_available): Likewise.
23431 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23432 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23433 * tree-ssa-pre.c (compute_antic): Likewise.
23435 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23437 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23438 (depth_first_search::depth_first_search): Change structure init
23439 function to this constructor.
23440 (depth_first_search::add_bb): Rename function to this member.
23441 (depth_first_search::execute): Likewise.
23442 (flow_dfs_compute_reverse_finish): Adjust.
23444 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23446 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23447 (longest_simple_path): Likewise.
23448 * shrink-wrap.c (spread_components): Likewise.
23449 (disqualify_problematic_components): Likewise.
23450 (emit_common_heads_for_components): Likewise.
23451 (emit_common_tails_for_components): Likewise.
23452 (insert_prologue_epilogue_for_components): Likewise.
23454 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23456 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23459 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23461 * df-core.c (df_set_blocks): Start using auto_bitmap.
23462 (df_compact_blocks): Likewise.
23463 * df-problems.c (df_rd_confluence_n): Likewise.
23464 * df-scan.c (df_insn_rescan_all): Likewise.
23465 (df_process_deferred_rescans): Likewise.
23466 (df_update_entry_block_defs): Likewise.
23467 (df_update_exit_block_uses): Likewise.
23468 (df_entry_block_bitmap_verify): Likewise.
23469 (df_exit_block_bitmap_verify): Likewise.
23470 (df_scan_verify): Likewise.
23471 * lra-constraints.c (lra_constraints): Likewise.
23472 (undo_optional_reloads): Likewise.
23473 (lra_undo_inheritance): Likewise.
23474 * lra-remat.c (calculate_gen_cands): Likewise.
23475 (do_remat): Likewise.
23476 * lra-spills.c (assign_spill_hard_regs): Likewise.
23477 (spill_pseudos): Likewise.
23478 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23479 (bitmap_set_subtract_values): Likewise.
23481 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23483 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23484 management with auto_bitmap.
23485 (fix_inter_tick): Likewise.
23486 (fix_recovery_deps): Likewise.
23487 * ira.c (add_store_equivs): Likewise.
23488 (find_moveable_pseudos): Likewise.
23489 (split_live_ranges_for_shrink_wrap): Likewise.
23490 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23491 (rtx_reuse_manager::seen_def_p): Likewise.
23492 (rtx_reuse_manager::set_seen_def): Likewise.
23493 * print-rtl.h (class rtx_reuse_manager): Likewise.
23495 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23497 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23499 (migrate_btr_def): Likewise.
23500 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23501 * df-core.c (loop_post_order_compute): Likewise.
23502 (loop_inverted_post_order_compute): Likewise.
23503 * hsa-common.h: Likewise.
23504 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23505 * init-regs.c (initialize_uninitialized_regs): Likewise.
23506 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23507 (inline_small_functions): Likewise.
23508 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23509 * ira.c (combine_and_move_insns): Likewise.
23510 (build_insn_chain): Likewise.
23511 * loop-invariant.c (find_invariants): Likewise.
23512 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23513 * predict.c (tree_predict_by_opcode): Likewise.
23514 (predict_paths_leading_to): Likewise.
23515 (predict_paths_leading_to_edge): Likewise.
23516 (estimate_loops_at_level): Likewise.
23517 (estimate_loops): Likewise.
23518 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23519 (spread_components): Likewise.
23520 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23521 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23522 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23523 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23524 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23525 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23526 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23527 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23528 (mark_threaded_blocks): Likewise.
23529 (thread_through_all_blocks): Likewise.
23530 * tree-ssa.c (verify_ssa): Likewise.
23531 (execute_update_addresses_taken): Likewise.
23532 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23534 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23536 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23538 (post_order_compute): Likewise.
23539 (inverted_post_order_compute): Likewise.
23540 (pre_and_rev_post_order_compute_fn): Likewise.
23542 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23544 * genrecog.c (int_set::int_set): Explicitly construct our
23545 auto_vec base class.
23546 * vec.h (auto_vec::auto_vec): New constructor.
23548 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23550 * bitmap.h (class auto_bitmap): New constructor taking
23551 bitmap_obstack * argument.
23553 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23555 * bitmap.h (class auto_bitmap): Change type of m_bits to
23556 bitmap_head, and adjust ctor / dtor and member operators.
23558 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23560 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23561 when returned register mode doesn't match original mode.
23563 2017-05-12 Jeff Law <law@redhat.com>
23564 Jakub Jelinek <jakub@redhat.com>
23566 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23567 we look for cc setter after the compare-elim changes.
23568 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23569 within the vector to match what compare-elim now expects.
23570 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23571 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23573 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23574 after the compare-elim changes.
23575 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23576 the vector to match what compare-elim now expects.
23577 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23578 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23579 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23580 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23581 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23583 * config/visium/visium.c (single_set_and_flags): Fix where
23584 we look for cc setter after the compare-elim changes.
23585 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23586 with the vector to match what compare-elim now expects.
23587 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23588 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23589 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23590 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23591 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23592 (neg<mode>2_insn_set_overflow): Likewise.
23594 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23596 PR middle-end/79794
23597 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23598 maybe_expand_insn call, set ops[0].target. If still set after call,
23599 set alt_rtl. Add extra arg to recursive calls.
23600 (extract_bit_field): Add alt_rtl argument. Pass to
23602 * expmed.h (extract_bit_field): Fix prototype.
23603 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23604 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23605 to extract_bit_field_calls.
23606 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23607 Pass alt_rtl to extract_bit_field calls.
23608 * calls.c (store_unaligned_arguments_into_psuedos)
23609 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23610 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23612 * optabs.h (struct expand_operand): Add target bitfield.
23614 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23616 * compare-elim.c (try_eliminate_compare): Canonicalize
23617 operation with embedded compare to
23618 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23619 (set (reg) (operation)].
23621 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23623 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23626 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23627 cost of adding a carry flag for ADC instruction.
23628 [case MINUS]: Ignore the cost of subtracting a carry flag
23629 for SBB instruction.
23631 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23633 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23635 * config/rs6000/bmiintrin.h: New file.
23636 * config/rs6000/bmi2intrin.h: New file.
23637 * config/rs6000/x86intrin.h: New file.
23639 2017-05-12 Jeff Law <law@redhat.com>
23641 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23644 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23646 PR middle-end/80707
23647 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23649 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23651 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23652 early expansion of vector divide builtins.
23653 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23654 builtins identified as having unsigned arguments.
23656 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23658 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23659 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23660 expansion of vector logical operations (and, andc, or, xor,
23663 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23665 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23666 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23668 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23670 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23671 early GIMPLE expansion of vector multiplies.
23673 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23675 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23676 TARGET_HAVE_MOVT conditional.
23677 (movt splitter): Likewise.
23679 2017-05-12 Richard Biener <rguenther@suse.de>
23681 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23682 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23683 Fold all stmts not inplace.
23685 2017-05-12 Richard Biener <rguenther@suse.de>
23687 PR tree-optimization/80713
23688 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23689 inserted_exprs bit for not removed stmts.
23691 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23693 PR middle-end/69921
23694 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23695 parallelized" attribute for parallelized OpenACC kernels.
23696 * omp-offload.c (execute_oacc_device_lower): Use it.
23698 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23699 Set "oacc kernels" attribute.
23700 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23701 parameter. Adjust all users.
23702 (oacc_fn_attrib_kernels_p): Remove function.
23703 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23704 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23705 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23706 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23707 assert "oacc kernels" attribute is set.
23709 2017-05-11 Carl Love <cel@us.ibm.com>
23711 * config/rs6000/rs6000-c: Add support for built-in functions
23712 vector unsigned char vec_popcnt (vector signed char)
23713 vector unsigned char vec_popcnt (vector unsigned char)
23714 vector unsigned short vec_popcnt (vector signed short)
23715 vector unsigned short vec_popcnt (vector unsigned short)
23716 vector unsigned int vec_popcnt (vector signed int)
23717 vector unsigned int vec_popcnt (vector unsigned int)
23718 vector unsigned long long vec_popcnt (vector signed long long)
23719 vector unsigned long long vec_popcnt (vector unsigned long long)
23720 vector signed long long vec_slo (vector signed long long,
23721 vector signed char)
23722 vector signed long long vec_slo (vector signed long long,
23723 vector unsigned char)
23724 vector unsigned long long vec_slo (vector unsigned long long,
23725 vector signed char)
23726 vector unsigned long long vec_slo (vector unsigned long long,
23727 vector unsigned char)
23728 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23729 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23730 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23731 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23732 * doc/extend.texi: Update the built-in documentation file for the
23733 new built-in functions.
23735 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23737 * attribs.h (sorted_attr_string): Move machine independent
23738 functions for target clone support from the i386 port to common
23739 code. Rename ix86_function_versions to common_function_versions.
23740 Rename make_name to make_unique_name.
23741 (common_function_versions): Likewise.
23742 (make_unique_name): Likewise.
23743 (make_dispatcher_decl): Likewise.
23744 (is_function_default_version): Likewise.
23745 * attribs.c (attr_strcmp): Likewise.
23746 (sorted_attr_string): Likewise.
23747 (common_function_versions): Likewise.
23748 (make_unique_name): Likewise.
23749 (make_dispatcher_decl): Likewise.
23750 (is_function_default_version): Likewise.
23751 * config/i386/i386.c (attr_strcmp): Likewise.
23752 (sorted_attr_string): Likewise.
23753 (ix86_function_versions): Likewise.
23754 (make_name): Likewise.
23755 (make_dispatcher_decl): Likewise.
23756 (is_function_default_version): Likewise.
23757 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23759 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23762 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23763 Account for direct move costs for vec_construct of integer
23766 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23769 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23770 (UNSPEC_STX_ATOMIC): Ditto.
23771 (loaddi_via_sse): New insn.
23772 (storedi_via_sse): Ditto.
23773 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23774 Update corresponding peephole2 patterns.
23775 (atomic_storedi_fpu): Ditto.
23777 2017-05-11 Julia Koval <julia.koval@intel.com>
23779 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23780 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23782 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23783 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23784 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23786 2017-05-11 Nathan Sidwell <nathan@acm.org>
23788 * graphite-poly.c: Include dumpfile.h.
23790 * dumpfle.h (dump_function): Declare here ...
23791 * tree-dump.h (dump_function): ... not here.
23792 * dumpfile.c: #include tree-cfg.h.
23793 (dump_function): Move here from ...
23794 * tree-dump.c (dump_function): ... here.
23795 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23796 * graphite-poly.c: Don't include tree-dump.h.
23797 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23798 * print-tree.c: Likewise.
23799 * stor-layout.c: Likewise.
23800 * tree-nested.c: Likewise.
23802 * dumpfile.c (dump_start): Use TDF_FLAGS.
23803 (dump_enable_all): Fix TDF_KIND check thinko.
23805 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23807 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23808 array entries to represent two legal parameterizations of the
23809 overloaded __builtin_cmpb function, as represented by the
23810 P6_OV_BUILTIN_CMPB constant.
23811 (altivec_resolve_overloaded_builtin): Add special case handling
23812 for the __builtin_cmpb function, as represented by the
23813 P6_OV_BUILTIN_CMPB constant.
23814 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23815 (BU_P6_64BIT_2): New macro.
23816 (BU_P6_OVERLOAD_2): New macro
23817 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23818 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23819 (CMPB): Add overload support to represent both 32-bit and 64-bit
23820 compare-bytes function.
23821 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23822 support for TARGET_CMPB.
23823 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23824 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23825 documentation of the __builtin_cmpb overloaded built-in function.
23827 2017-05-11 Richard Biener <rguenther@suse.de>
23829 PR tree-optimization/80705
23830 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23831 bases are not vectorizable.
23833 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23835 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23836 when counting register pressure.
23838 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23840 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23841 (iv_ca_more_deps): Renamed to ...
23842 (iv_ca_compare_deps): ... this.
23843 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23845 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23847 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23849 (determine_group_iv_costs): ... here.
23850 (find_inv_vars_cb): Record inv var if it's not recorded before.
23852 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23854 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23855 (get_shiftadd_cost): Ditto.
23857 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23859 * tree-ssa-address.c: Include header file.
23860 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23862 (add_to_parts): Refactor.
23863 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23864 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23867 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23869 PR tree-optimization/53090
23870 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23872 (extract_cond_operands): Detect condition with IV on both sides
23873 and return COMP_IV_EXPR_2.
23874 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23875 (rewrite_use_compare): Simplify by removing call to function
23876 extract_cond_operands.
23878 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23880 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23881 (extract_cond_operands): Detect condition comparing against non-
23882 invariant bound and return appropriate enum value.
23883 (find_interesting_uses_cond): Update use of extract_cond_operands.
23884 Handle its return value accordingly.
23885 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23887 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23889 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23890 nonlinear iv_use computation in loop invariant sensitive way.
23892 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23894 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23895 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23897 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23899 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23900 (dump_cand): Support iv_cand.inv_exprs.
23901 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23903 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23906 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23908 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23910 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23911 as local function. Include necessary header files.
23912 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23914 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23916 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23918 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23920 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23921 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23922 RSHIFT_EXPR and BIT_NOT_EXPR.
23924 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23926 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23927 (adjust_setup_cost): New parameter supporting round up adjustment.
23928 (struct address_cost_data): Delete.
23929 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23930 (split_address_cost, ptr_difference_cost): Delete.
23931 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23932 (struct ainc_cost_data): New struct.
23933 (get_address_cost_ainc): New function.
23934 (get_address_cost, get_computation_cost): Reimplement.
23935 (determine_group_iv_cost_address): Record inv_expr for all uses of
23937 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23938 (iv_ca_has_deps): Reimplemented to ...
23939 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23941 (iv_ca_extend): Call iv_ca_more_deps.
23943 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23945 * tree-ssa-address.c (struct mem_address): Move to header file.
23946 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23947 * tree-ssa-address.h (struct mem_address): Move from C file.
23948 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23950 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23952 * tree-affine.h (aff_combination_type): New interface.
23953 (aff_combination_zero_p): Remove static.
23954 (aff_combination_const_p): New interface.
23955 (aff_combination_singleton_var_p): New interfaces.
23957 2017-05-11 Richard Biener <rguenther@suse.de>
23959 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23960 Skip unreachable blocks and destinations.
23961 (eliminate): Move stmt removal and fixup ...
23962 (fini_eliminate): ... here. Skip inserted exprs.
23963 (pass_pre::execute): Move fini_pre after fini_eliminate.
23964 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23965 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23966 PRE to get rid of dead code that has invalid SSA form and
23967 split critical edges again.
23969 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23971 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23973 2017-05-11 Richard Biener <rguenther@suse.de>
23975 * passes.c (execute_function_todo): Verify loops if they are
23976 said to be up-to-date.
23977 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23978 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23980 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23983 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23984 handle calling assemble_external ourself.
23987 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23988 modes with zero size. Enhance comment.
23990 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23992 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23993 built-ins for vec_xl and vec_xst with short and char pointer
23996 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23998 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23999 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
24000 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
24001 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
24002 (_mm_maskz_min_round_ss): New intrinsics.
24003 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
24004 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
24005 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
24006 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
24007 (__builtin_ia32_minss_mask_round): New builtins.
24008 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24009 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24010 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
24012 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
24013 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
24015 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
24018 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
24020 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
24021 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
24022 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
24023 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
24024 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
24025 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24026 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24027 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24028 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24029 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24030 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24031 (__builtin_ia32_mulss_mask_round): New builtins.
24032 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24033 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24034 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24036 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24037 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24039 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24042 2017-05-10 Julia Koval <julia.koval@intel.com>
24044 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24045 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24046 (_mm256_setr_m128i): New intrinsics.
24048 2017-05-10 Julia Koval <julia.koval@intel.com>
24050 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24051 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24052 (_mm_maskz_rcp14_ss): New intrinsics.
24053 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24054 (__builtin_ia32_rcp14ss_mask): New builtins.
24055 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24057 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24059 PR tree-optimization/51513
24060 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24061 (assert_unreachable_fallthru_edge_p): Use it.
24062 (group_case_labels_stmt): Likewise.
24063 * tree-cfg.h: Prototype it.
24064 * stmt.c: Include cfghooks.h and tree-cfg.h.
24065 (emit_case_dispatch_table) <gap_label>: New local variable.
24066 Use it to fill dispatch table gaps.
24067 Test for default_label before updating probabilities.
24068 (expand_case) <default_label>: Remove unneeded initialization.
24069 Test for unreachable default case statement and remove its edge.
24070 Set default_label accordingly.
24071 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24073 2017-05-10 Carl Love <cel@us.ibm.com>
24075 * config/rs6000/rs6000-c: Add support for built-in functions
24076 vector signed char vec_neg (vector signed char)
24077 vector signed short int vec_neg (vector short int)
24078 vector signed int vec_neg (vector signed int)
24079 vector signed long long vec_neg (vector signed long long)
24080 vector float vec_neg (vector float)
24081 vector double vec_neg (vector double)
24082 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24084 * config/rs6000/altivec.h: Add define for vec_neg
24085 * doc/extend.texi: Update the built-in documentation for the
24086 new built-in functions.
24088 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24090 PR tree-optimization/77644
24091 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24093 2017-05-10 Nathan Sidwell <nathan@acm.org>
24095 * dumpfile.h (TDI_lang_all): New.
24096 (TDF_KIND): New. Renumber others
24097 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24099 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24101 (get_dump_file_name): Adjust suffix generation.
24102 (dump_enable_all): Use TDF_KIND.
24103 * doc/invoke.texi (-fdump-lang-all): Document.
24105 * dumpfile.h: Tabify.
24107 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24110 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24111 Move member access before delete.
24113 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24115 * tree-inline.c (expand_call_inline): Split block at stmt
24118 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24121 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24122 are now unused after splitting mov{sf,sd}_hardfloat.
24123 (f32_lr2): Likewise.
24124 (f32_lm): Likewise.
24125 (f32_lm2): Likewise.
24126 (f32_li): Likewise.
24127 (f32_li2): Likewise.
24128 (f32_lv): Likewise.
24129 (f32_sr): Likewise.
24130 (f32_sr2): Likewise.
24131 (f32_sm): Likewise.
24132 (f32_sm2): Likewise.
24133 (f32_si): Likewise.
24134 (f32_si2): Likewise.
24135 (f32_sv): Likewise.
24136 (f32_dm): Likewise.
24137 (f32_vsx): Likewise.
24138 (f32_av): Likewise.
24139 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24140 For movsf, order stores so the VSX stores occur before the GPR
24141 store which encourages the register allocator to use a traditional
24142 FPR instead of a GPR. For movsd, order the stores so that the GPR
24143 store comes before the VSX stores to allow the power6 to work.
24144 This is due to the power6 not having a 32-bit integer store
24145 instruction from a FPR.
24146 (movsf_hardfloat): Likewise.
24147 (movsd_hardfloat): Likewise.
24149 2017-05-09 Martin Sebor <msebor@redhat.com>
24151 PR translation/80280
24152 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24155 PR translation/80280
24156 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24157 data member added in r247778.
24158 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24160 2017-05-09 Nathan Sidwell <nathan@acm.org>
24162 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24164 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24167 2017-05-09 Marek Polacek <polacek@redhat.com>
24169 * doc/invoke.texi: Fix typo.
24171 2017-05-09 Richard Biener <rguenther@suse.de>
24173 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24174 (vrp_val_is_min): Likewise.
24175 (set_value_range_to_value): Likewise.
24176 (set_value_range_to_nonnegative): Likewise.
24177 (gimple_assign_nonzero_p): Likewise.
24178 (gimple_stmt_nonzero_p): Likewise.
24179 (vrp_int_const_binop): Likewise. Remove unreachable case.
24180 (adjust_range_with_scev): Adjust comments.
24181 (compare_range_with_value): Likewise.
24182 (extract_range_from_phi_node): Likewise.
24183 (test_for_singularity): Likewise.
24185 2017-05-09 Richard Biener <rguenther@suse.de>
24187 * tree-vrp.c (get_single_symbol): Add assert that we don't
24188 get overflowed constants as invariant part.
24189 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24190 checks. Use wi::cmp instead of recursing for integer constants.
24191 (compare_values): Just ignore whether we assumed undefined
24192 overflow instead of failing the compare.
24193 (extract_range_for_var_from_comparison_expr): Add comment before the
24194 TREE_NO_WARNING sets.
24195 (test_for_singularity): Likewise.
24196 (extract_range_from_comparison): Do not disable optimization
24197 when we assumed undefined overflow.
24198 (extract_range_basic): Remove init of unused var.
24200 2017-05-09 Richard Biener <rguenther@suse.de>
24202 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24203 (extract_range_from_multiplicative_op_1): Adjust.
24204 (extract_range_from_binary_expr_1): Use int_const_binop.
24206 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24209 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24210 rs6000_store_data_bypass_p in seven define_bypass directives and
24211 in several comments.
24212 * config/rs6000/rs6000-protos.h: Add prototype for
24213 rs6000_store_data_bypass_p function.
24214 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24215 function implements slightly different (rs6000-specific) semantics
24216 than store_data_bypass_p, returning false rather than aborting
24217 with assertion error when arguments do not satisfy the
24218 requirements of store data bypass.
24219 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24220 rs6000_store_data_bypass_p.
24222 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24224 * config/xtensa/xtensa-protos.h
24225 (xtensa_initial_elimination_offset): New declaration.
24226 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24227 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24228 macro definition, add case for FRAME_POINTER_REGNUM when
24229 FRAME_GROWS_DOWNWARD.
24230 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24231 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24232 xtensa_initial_elimination_offset.
24234 2017-05-08 Nathan Sidwell <nathan@acm.org>
24236 * doc/invoke.texi: Alphabetize -fdump options.
24238 2017-05-08 Martin Sebor <msebor@redhat.com>
24240 PR translation/80280
24241 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24243 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24245 * target.def (compute_frame_layout): New optional target hook.
24246 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24247 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24248 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24250 * reload1.c (verify_initial_elim_offsets): Likewise.
24251 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24252 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24253 (arm_get_frame_offsets): Split up into this ...
24254 (arm_compute_frame_layout): ... and this function.
24256 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24258 * config/aarch64/constraints.md (Usa): New constraint.
24259 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24261 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24263 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24264 with_multilib_list after it has been checked.
24266 2017-05-08 Richard Biener <rguenther@suse.de>
24268 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24269 (bitmap_set_subtract_values): Likewise.
24271 2017-05-08 Richard Biener <rguenther@suse.de>
24273 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24274 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24276 (gimple_stmt_nonzero_warnv_p): Rename to ...
24277 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24279 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24280 (extract_range_basic): Adjust, do not disable propagation on
24281 strict overflow sensitive simplification.
24282 (vrp_visit_cond_stmt): Likewise.
24284 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24286 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24287 body size unconditionally.
24289 2017-05-07 Jeff Law <law@redhat.com>
24292 2017-05-06 Jeff Law <law@redhat.com>
24293 PR tree-optimization/78496
24294 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24297 PR tree-optimization/78496
24298 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24299 (simplify_stmt_using_ranges): Call it.
24300 (vrp_dom_walker::before_dom_children): Extract equivalences
24301 from an ASSERT_EXPR with an equality comparison against a
24304 2017-05-06 Jeff Law <law@redhat.com>
24306 PR tree-optimization/78496
24307 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24310 PR tree-optimization/78496
24311 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24312 (simplify_stmt_using_ranges): Call it.
24313 (vrp_dom_walker::before_dom_children): Extract equivalences
24314 from an ASSERT_EXPR with an equality comparison against a
24317 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24319 * lra-constraints.c (lra_copy_reg_equiv): New function.
24320 (split_reg): Use it to copy equivalence information from the
24321 original register to the spill register.
24323 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24325 PR rtl-optimization/75964
24326 * simplify-rtx.c (simplify_const_relational_operation): Remove
24327 invalid handling of comparisons of integer ABS.
24329 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24331 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24332 initialize to zero.
24333 (init_regs): Remove declaration.
24334 (function_arg_advance_32): Initialize error_p as boolean variable.
24336 2017-05-05 Nathan Sidwell <nathan@acm.org>
24338 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24339 lines. Use for (;;).
24341 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24343 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24344 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24345 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24346 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24347 VF=2 that require versioning.
24349 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24351 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24354 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24356 * diagnostic.h (diagnostic_override_option_index): Convert from
24357 macro to inline function.
24359 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24361 * diagnostic.c (last_module_changed_p): New function.
24362 (set_last_module): New function.
24363 (diagnostic_report_current_module): Convert macro usage to
24364 the above functions.
24365 * diagnostic.h (diagnostic_context::last_module): Strengthen
24366 from const line_map * to const line_map_ordinary *.
24367 (diagnostic_last_module_changed): Delete macro.
24368 (diagnostic_set_last_module): Delete macro.
24370 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24372 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24373 with diagnostic_report_diagnostic.
24374 (diagnostic_n_impl_richloc): Likewise.
24375 * diagnostic.h (report_diagnostic): Delete macro.
24376 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24377 with diagnostic_report_diagnostic.
24378 * substring-locations.c (format_warning_va): Likewise.
24380 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24382 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24383 save/restor of format_spec. Move option-printing code to...
24384 (print_option_information): ...this new function, and
24385 reimplement by simply printing to the pretty_printer,
24386 rather than appending to the format string.
24388 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24390 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24391 handling logic into...
24392 (update_effective_level_from_pragmas): ...this new function.
24394 2017-05-04 Andrew Waterman <andrew@sifive.com>
24396 * config/riscv/riscv.opt (mstrict-align): New option.
24397 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24398 (SLOW_UNALIGNED_ACCESS): Define.
24399 (riscv_slow_unaligned_access): Declare.
24400 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24402 (riscv_slow_unaligned_access): New variable.
24403 (rocket_tune_info): Set slow_unaligned_access to true.
24404 (optimize_size_tune_info): Set slow_unaligned_access to false.
24405 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24406 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24407 (riscv_option_override): Set riscv_slow_unaligned_access.
24408 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24410 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24412 * config/riscv/riscv.md: Unify indentation.
24414 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24419 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24421 (extendsi<mode>2): Add support for doing sign extension via
24422 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24423 don't have ISA 3.0 instructions.
24424 (extendsi<mode>2 splitter): Likewise.
24425 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24426 generate the normal insns since SImode can now go in vector
24427 registers. Disallow the special UNSPECs needed for previous
24428 machines to hide SImode being used. Add new insns
24429 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24430 (fix_trunc<mode>si2_stfiwx): Likewise.
24431 (fix_trunc<mode>si2_internal): Likewise.
24432 (fixuns_trunc<mode>si2): Likewise.
24433 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24434 (fctiw<u>z_<mode>_smallint): Likewise.
24435 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24436 of floating point to 32-bit integer from doing a direct move to
24437 the GPR registers to do a store.
24438 (fctiwz_<mode>): Break long line.
24440 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24442 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24443 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24444 (addr_list, addr_offset_valid_p): New.
24445 (split_address_groups): Check offset validity with above function.
24446 (gt-tree-ssa-loop-ivopts.h): Include header file.
24448 2017-05-05 Nathan Sidwell <nathan@acm.org>
24450 * config.gcc (arm*-*-*): Add missing 'fi'.
24452 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24454 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24455 included in -fopt-info does not matter.
24456 * doc/optinfo.texi (-fopt-info): Fix description of default
24457 behavour. Explicitly say order of options included in -fopt-info
24460 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24462 * config.gcc: Allow combinations of aprofile and rmprofile values for
24463 --with-multilib-list.
24464 * config/arm/t-multilib: New file.
24465 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24466 variables. Remove setting of ISA and floating-point ABI in
24467 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24468 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24469 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24470 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24472 * config/arm/t-rmprofile: Likewise except for the matches changes.
24473 * doc/install.texi (--with-multilib-list): Document the combination of
24474 aprofile and rmprofile values and warn about pitfalls in doing that.
24476 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24478 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24479 (movdi_aarch64): Likewise.
24481 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24483 PR tree-optimization/80632
24484 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24486 (build_arrays): Initialize it for virtual phis.
24487 (fix_phi_nodes): Use it for virtual phis.
24489 PR tree-optimization/80558
24490 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24491 [x, y] op z into [x op, y op z] for op & or | if conditions
24494 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24495 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24498 * config/arm/arm.md (use_literal_pool): Remove.
24499 (64-bit immediate split): No longer takes cost into consideration
24500 if arm_disable_literal_pool is enabled.
24501 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24502 used when arm_disable_literal_pool is enabled.
24503 (arm_max_const_double_inline_cost): Remove use of
24504 arm_disable_literal_pool.
24505 (push_minipool_fix): Add assert.
24506 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24507 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24508 (no_literal_pool_sf_immediate): New.
24510 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24512 PR tree-optimization/80613
24513 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24514 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24516 2017-05-05 Richard Biener <rguenther@suse.de>
24518 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24520 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24522 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24523 of this flag from insn conditions due to removal from r247495.
24525 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24527 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24529 (arm_early_store_addr_dep_ptr): Likewise.
24530 * config/arm/aarch-common-protos.h
24531 (arm_early_load_addr_dep_ptr): Add prototype.
24532 (arm_early_store_addr_dep_ptr): Likewise.
24533 * config/arm/cortex-a53.md: Add new bypasses.
24535 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24537 * tree.c (next_type_uid): Change type to unsigned.
24538 (type_hash_canon): Decrement back next_type_uid if
24539 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24540 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24543 2017-05-04 Martin Sebor <msebor@redhat.com>
24545 * builtins.c: Fix a trivial typo in a comment.
24547 PR middle-end/79234
24548 * builtins.c (check_sizes): Adjust to handle reading past the end.
24549 Avoid printing excessive upper bound of ranges. Use %E to print
24550 tree nodes instead of converting them to %wu.
24551 (expand_builtin_memchr): New function.
24552 (compute_dest_size): Rename...
24553 (compute_objsize): ...to this.
24554 (expand_builtin_memcpy): Adjust.
24555 (expand_builtin_mempcpy): Adjust.
24556 (expand_builtin_strcat): Adjust.
24557 (expand_builtin_strcpy): Adjust.
24558 (check_strncat_sizes): Adjust.
24559 (expand_builtin_strncat): Adjust.
24560 (expand_builtin_strncpy): Adjust and simplify.
24561 (expand_builtin_memset): Adjust.
24562 (expand_builtin_bzero): Adjust.
24563 (expand_builtin_memcmp): Adjust.
24564 (expand_builtin): Handle memcmp.
24565 (maybe_emit_chk_warning): Check strncat just once.
24567 2017-05-04 Martin Sebor <msebor@redhat.com>
24569 PR preprocessor/79214
24570 PR middle-end/79222
24571 PR middle-end/79223
24572 * builtins.c (check_sizes): Add inlining context and issue
24573 warnings even when -Wno-system-headers is set.
24574 (check_strncat_sizes): Same.
24575 (expand_builtin_strncat): Same.
24576 (expand_builtin_memmove): New function.
24577 (expand_builtin_stpncpy): Same.
24578 (expand_builtin): Handle memmove and stpncpy.
24580 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24582 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24583 which is not used any more.
24585 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24587 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24589 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24591 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24592 (cortexa53_tunings): Likewise.
24593 (cortexa57_tunings): Likewise.
24594 (cortexa72_tunings): Likewise.
24595 (cortexa73_tunings): Likewise.
24597 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24599 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24600 Set loop alignment to 8.
24602 2017-05-04 Martin Sebor <msebor@redhat.com>
24604 PR translation/80280
24605 * builtins.c (expand_builtin_object_size): Add missing quoting to
24606 %D and like directives.
24607 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24608 (hsa_type_for_tree_type): Same.
24609 (verify_function_arguments): Same.
24610 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24611 * varasm.c (get_section): Same.
24614 2017-05-04 Martin Sebor <msebor@redhat.com>
24616 PR translation/80280
24617 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24619 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24621 * config/aarch64/aarch64.c (generic_addrcost_table):
24622 Change HI/TI mode setting.
24624 2017-05-04 Martin Jambor <mjambor@suse.cz>
24626 PR tree-optimization/80622
24627 * tree-sra.c (comes_initialized_p): New function.
24628 (build_accesses_from_assign): Only set write lazily when
24629 comes_initialized_p is false.
24630 (analyze_access_subtree): Use comes_initialized_p.
24631 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24632 instead of testing for PARM_DECL.
24634 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24636 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24637 constraint on operand 0 to allow more general addressing modes.
24638 Adjust output template.
24639 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24641 * config/aarch64/aarch64-protos.h
24642 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24643 * config/aarch64/constraints.md (Dp): New address constraint.
24644 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24647 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24649 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24650 update use of estimate_ipcp_clone_size_and_time.
24651 (estimate_local_effects): Update use of
24652 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24653 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24654 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24655 Return nonspecialized time.
24657 2017-05-04 Richard Biener <rguenther@suse.de>
24659 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24660 for the last VUSE which def dominates the PHI. Directly call
24662 (get_continuation_for_phi_1): Remove.
24664 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24666 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24667 to explain the use of truncating division. Cap the number of
24668 iterations to the maximum given by nb_iterations_upper_bound,
24671 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24673 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24674 * configure: Regenerate.
24675 * config.in: Regenerate.
24676 * config/i386/driver-mingw32.c: new file.
24677 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24678 * config.host: Link driver-mingw32.o on MinGW host.
24679 * doc/install.texi: Document new --enable-mingw-wildcard configure
24682 2017-05-04 Marek Polacek <polacek@redhat.com>
24684 PR tree-optimization/80612
24685 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24687 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24688 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24690 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24691 (movt splitter): Likewise.
24692 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24693 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24694 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24695 block for Thumb-1 with MOVT.
24696 (thumb2_legitimate_address_p): Move code block ...
24697 (can_avoid_literal_pool_for_label_p): ... into this new function.
24698 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24700 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24701 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24702 "M-profile targets with the MOVT instruction".
24704 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24706 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24707 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24708 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24710 2017-05-04 Martin Liska <mliska@suse.cz>
24712 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24713 variable cond_code.
24715 2017-05-04 Richard Biener <rguenther@suse.de>
24717 * tree.c (array_at_struct_end_p): Handle arrays at struct
24718 end with flexarrays more conservatively. Refactor and treat
24719 arrays of arrays or aggregates more strict. Fix
24720 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24721 * tree.c (array_at_struct_end_p): Adjust prototype.
24722 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24723 * gimple-fold.c (get_range_strlen): Likewise.
24724 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24726 2017-05-04 Richard Biener <rguenther@suse.de>
24728 PR tree-optimization/31130
24729 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24731 (supports_overflow_infinity): Likewise.
24732 (is_negative_overflow_infinity): Likewise.
24733 (is_positive_overflow_infinity): Likewise.
24734 (is_overflow_infinity): Likewise.
24735 (stmt_overflow_infinity): Likewise.
24736 (overflow_infinity_range_p): Likewise.
24737 (usable_range_p): Remove as always returning true.
24738 (make_overflow_infinity): Remove.
24739 (negative_overflow_infinity): Likewise.
24740 (positive_overflow_infinity): Likewise.
24741 (avoid_overflow_infinity): Likewise.
24742 (set_value_range): Adjust accordingly.
24743 (set_value_range_to_nonnegative): Likewise, remove now unused
24744 overflow_infinity arg.
24745 (vrp_operand_equal_p): Adjust.
24746 (update_value_range): Likewise.
24747 (range_int_cst_singleton_p): Likewise.
24748 (operand_less_p): Likewise.
24749 (compare_values_warnv): Likewise.
24750 (extract_range_for_var_from_comparison_expr): Likewise.
24751 (vrp_int_const_binop): Likewise.
24752 (zero_nonzero_bits_from_vr): Likewise.
24753 (extract_range_from_multiplicative_op_1): Likewise.
24754 (extract_range_from_binary_expr_1): Likewise.
24755 (extract_range_from_unary_expr): Likewise.
24756 (extract_range_from_comparison): Likewise.
24757 (extract_range_basic): Likewise.
24758 (adjust_range_with_scev): Likewise.
24759 (compare_ranges): Likewise.
24760 (compare_range_with_value): Likewise.
24761 (dump_value_range): Likewise.
24762 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24764 (simplify_cond_using_ranges): Adjust.
24766 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24768 * brig-builtins.def: Added a builtin for class_f64.
24769 * builtin-types.def: Added a builtin type needed by class_f64.
24771 2017-05-03 Jason Merrill <jason@redhat.com>
24773 * timevar.def: Add TV_CONSTEXPR.
24775 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24777 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24779 2017-05-03 Martin Jambor <mjambor@suse.cz>
24781 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24782 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24783 * ipa-cp.c (ipcp_propagate_stage): Do not call
24784 ipa_update_after_lto_read.
24785 * ipa-inline.c (ipa_inline): Likewise.
24787 2017-05-03 Martin Jambor <mjambor@suse.cz>
24789 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24790 tag. Added a default constructor and a destructor.
24791 (ipa_edge_args_sum_t): New class;
24792 (ipa_edge_args_sum): Declare.
24793 (ipa_edge_args_vector): Remove declaration.
24794 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24795 (ipa_free_edge_args_substructures): Remove declaration.
24796 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24797 (ipa_edge_args_info_available_for_edge_p): Likewise.
24798 * ipa-prop.c (ipa_edge_args_vector): Removed.
24799 (edge_removal_hook_holder): Likewise.
24800 (edge_duplication_hook_holder): Likewise.
24801 (ipa_edge_args_sum): New variable.
24802 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24803 ipa_edge_args_vector.
24804 (ipa_free_edge_args_substructures): Likewise.
24805 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24806 ipa_edge_args_vector.
24807 (ipa_edge_removal_hook): Turned into method
24808 ipa_edge_args_sum_t::remove.
24809 (ipa_edge_duplication_hook): Turned into method
24810 ipa_edge_args_sum_t::duplicate.
24811 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24812 registering edge hooks.
24813 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24814 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24815 ipa_edge_args_sum instead of ipa_edge_args_vector.
24816 * ipa-profile.c (ipa_profile): Likewise.
24818 2017-05-03 Martin Jambor <mjambor@suse.cz>
24820 * symbol-summary.h (function_summary): New method exists.
24821 (function_summary::symtab_removal): Deallocate through release.
24822 (call_summary): New class.
24823 (gt_ggc_mx): New overload.
24824 (gt_pch_nx): Likewise.
24825 (gt_pch_nx): Likewise.
24827 2017-05-03 Jeff Law <law@redhat.com>
24829 PR tree-optimization/78496
24830 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24831 from simplify_cond_using_ranges. Split off code to walk
24832 backwards through casts into ...
24833 (simplify_cond_using_ranges_2): New function.
24834 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24835 (execute_vrp): After identifying jump threads, call
24836 simplify_cond_using_ranges_2.
24838 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24841 * ipa-inline.h (inline_summary): Add ctor.
24842 (create_ggc): Do not use ggc_cleared_alloc.
24844 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24845 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24847 * gcc.c (handle_braces): Support escaping in switch matching
24849 * doc/invoke.texi (Spec Files): Document it.
24850 Remove superfluous @code markup in items.
24852 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24854 * diagnostic-show-locus.c (struct column_range): New struct.
24855 (get_affected_columns): New function.
24856 (get_printed_columns): New function.
24857 (struct correction): New struct.
24858 (correction::ensure_capacity): New function.
24859 (correction::ensure_terminated): New function.
24860 (struct line_corrections): New struct.
24861 (line_corrections::~line_corrections): New dtor.
24862 (line_corrections::add_hint): New function.
24863 (layout::print_trailing_fixits): Reimplement in terms of the new
24865 (selftest::test_overlapped_fixit_printing): New function.
24866 (selftest::diagnostic_show_locus_c_tests): Call it.
24868 2017-05-03 Nathan Sidwell <nathan@acm.org>
24870 Canonicalize canonical type hashing
24871 * tree.h (type_hash_canon_hash): Declare.
24872 * tree.c (type_hash_list, attribute_hash_list): Move into
24873 type_hash_canon_hash.
24874 (build_type_attribute_qual_variant): Break out hash code calc into
24875 type_hash_canon_hash.
24876 (type_hash_canon_hash): New. Generic type hash computation.
24877 (build_range_type_1, build_array_type_1, build_function_type,
24878 build_method_type_directly, build_offset_type, build_complex_type,
24879 make_vector_type): Call it.
24881 2017-05-03 Richard Biener <rguenther@suse.de>
24883 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24884 When all DRs have unknown misaligned do not always peel
24885 when there is a store but apply the same costing model as if
24886 there were only loads.
24888 2017-05-03 Richard Biener <rguenther@suse.de>
24891 PR tree-optimization/80492
24892 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24893 compare_base_decls returning dont-know properly.
24895 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24897 * config/arm/iterators.md (CCSI): New mode iterator.
24898 (arch): New mode attribute.
24899 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24900 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24901 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24902 code iterator for success result mode.
24903 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24904 the corresponding new insn generators.
24906 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24909 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24910 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24912 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24914 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24915 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24916 (DDR_A): Wrap DDR argument in brackets.
24917 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24918 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24919 (DDR_REVERSED_P): Likewise.
24921 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24923 PR tree-optimization/79472
24924 * tree-switch-conversion.c (struct switch_conv_info): Add
24925 contiguous_range and default_case_nonstandard fields.
24926 (collect_switch_conv_info): Compute contiguous_range and
24927 default_case_nonstandard fields, don't clear final_bb if
24928 contiguous_range and only the default case doesn't have the required
24930 (check_all_empty_except_final): Set default_case_nonstandard instead
24931 of failing if contiguous_range and the default case doesn't have empty
24933 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24934 and only the default case doesn't have the required constants. Skip
24936 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24937 if default_case_nonstandard.
24938 (build_constructors): Build constant 1 just once. Assert that default
24939 values aren't inserted in between cases if contiguous_range. Skip
24941 (build_arrays): Skip virtual phis.
24942 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24943 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24944 Handle virtual phis.
24945 (gen_inbound_check): Handle default_case_nonstandard case.
24946 (process_switch): Adjust check_final_bb caller. Call
24947 gather_default_values with the first non-default case instead of
24948 default case if default_case_nonstandard.
24950 2017-05-02 Nathan Sidwell <nathan@acm.org>
24952 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24953 check. Fix formatting.
24955 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24957 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24958 errors when comparing specialized and unspecialized times.
24960 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24962 * diagnostic-show-locus.c
24963 (layout::should_print_annotation_line_p): Make private.
24964 (layout::print_annotation_line): Make private.
24965 (layout::annotation_line_showed_range_p): Make private.
24966 (layout::show_ruler): Make private.
24967 (layout::print_source_line): Make private. Pass in line and
24968 line_width, rather than calling location_get_source_line. Drop
24970 (layout::print_leading_fixits): New method.
24971 (layout::print_any_fixits): Rename to...
24972 (layout::print_trailing_fixits): ...this, and make private.
24973 Don't print newline fixits.
24974 (diagnostic_show_locus): Move logic for printing one row into...
24975 (layout::print_line): ...this new function. Move the
24976 location_get_source_line call and error-handling from
24977 print_source_line to here. Call print_leading_fixits, and rename
24978 print_any_fixits to print_trailing_fixits.
24979 (selftest::test_fixit_insert_containing_newline): Update now that
24980 newlines are partially supported.
24981 (selftest::test_fixit_insert_containing_newline_2): New test.
24982 (selftest::test_fixit_replace_containing_newline): Update comments.
24983 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24984 * edit-context.c (class added_line): New class.
24985 (class edited_line): Describe newline handling in comment.
24986 (edited_line::actually_edited_p): New method.
24987 (edited_line::print_content): Delete redundant decl.
24988 (edited_line::m_predecessors): New field.
24989 (edited_file::print_content): Call edited_line::print_content.
24990 (edited_file::print_diff): Update to support newlines.
24991 (edited_file::print_diff_hunk): Likewise.
24992 (edited_file::print_run_of_changed_lines): New function.
24993 (edited_file::print_diff_line): Convert to...
24994 (print_diff_line): ...this.
24995 (edited_file::get_effective_line_count): New function.
24996 (edited_line::edited_line): Initialize new field m_predecessors.
24997 (edited_line::~edited_line): Clean up m_predecessors.
24998 (edited_line::apply_fixit): Handle newlines.
24999 (edited_line::get_effective_line_count): New function.
25000 (edited_line::print_content): New function.
25001 (edited_line::print_diff_lines): New function.
25002 (selftest::test_applying_fixits_insert_containing_newline): New
25004 (selftest::test_applying_fixits_replace_containing_newline): New
25006 (selftest::insert_line): New function.
25007 (selftest::test_applying_fixits_multiple_lines): Add example of
25009 (selftest::edit_context_c_tests): Call the new tests.
25011 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25013 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
25014 parameter cand. Update dump information.
25015 (get_computation_cost): Update uses.
25017 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25019 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
25020 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
25021 (get_computation_at, rewrite_use_address): Update use of
25022 get_computation_aff.
25024 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25026 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25027 (get_computation): Delete.
25028 (get_computation_cost): Implement like get_computation_cost_at.
25029 Use get_computation_at.
25030 (get_computation_cost_at): Delete.
25031 (rewrite_use_nonlinear_expr): Use get_computation_at.
25032 (rewrite_use_compare, remove_unused_ivs): Ditto.
25034 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25036 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25038 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25040 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25041 (ivopts_global_cost_for_size): Rename parameter and update uses.
25042 (iv_ca_recount_cost): Update uses.
25043 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25044 candidates seperately in n_invs and n_cands.
25045 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25047 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25049 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25050 (find_inv_vars_cb): New.
25051 (find_depends): Renamed to ...
25052 (find_inv_vars): ... this.
25053 (add_candidate_1, force_var_cost): Call find_inv_vars.
25054 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25056 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25058 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25059 inv_vars. Add inv_exprs.
25060 (struct iv_cand): Rename depends_on to inv_vars.
25061 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25062 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25063 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25064 (dump_cand): Dump inv_vars.
25065 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25066 (record_invariant, find_depends, add_candidate_1): Ditto.
25067 (set_group_iv_cost, force_var_cost): Ditto.
25068 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25069 (get_computation_cost_at, get_computation_cost): Ditto.
25070 (determine_group_iv_cost_generic): Ditto.
25071 (determine_group_iv_cost_address): Ditto.
25072 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25073 (determine_group_iv_costs): Ditto.
25074 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25075 (iv_ca_set_remove_invariants): Renamed to ...
25076 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25077 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25078 (iv_ca_set_add_invariants): Renamed to ...
25079 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25080 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25081 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25082 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25083 (create_new_ivs): Remove useless dump.
25085 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25087 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25089 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25090 (iv_ca_set_no_cp, create_new_iv): Ditto.
25092 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25094 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25096 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25098 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25099 function tree_check2.
25101 2017-05-02 Martin Liska <mliska@suse.cz>
25103 * doc/gcov.texi: Add missing preposition.
25104 * gcov.c (function_info::function_info): Properly fill up
25105 all member variables.
25107 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25109 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25111 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25113 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25115 2017-05-02 Martin Liska <mliska@suse.cz>
25118 * lto-streamer-in.c (lto_read_tree_1): Remove
25119 LTO_STREAMER_DEBUG.
25120 * lto-streamer.c (struct tree_hash_entry): Likewise.
25121 (struct tree_entry_hasher): Likewise.
25122 (tree_entry_hasher::hash): Likewise.
25123 (tree_entry_hasher::equal): Likewise.
25124 (lto_streamer_init): Likewise.
25125 (lto_orig_address_map): Likewise.
25126 (lto_orig_address_get): Likewise.
25127 (lto_orig_address_remove): Likewise.
25128 * lto-streamer.h: Likewise.
25129 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25130 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25132 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25134 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25135 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25136 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25137 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25138 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25139 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25140 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25141 (mm_maskz_sub_ss): New intrinsics.
25142 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25143 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25144 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25145 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25146 (__builtin_ia32_subss_mask_round): New builtins.
25147 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25148 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25149 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25151 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25152 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25154 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25157 2017-05-02 Martin Jambor <mjambor@suse.cz>
25159 PR tree-optimization/78687
25160 * tree-sra.c (access): New field parent.
25161 (process_subtree_disqualification): New function.
25162 (disqualify_candidate): Call it.
25163 (build_accesses_from_assign): Reset write flag if creating an
25165 (build_access_subtree): Fill in parent field and also prpagate
25166 down grp_write flag.
25167 (create_artificial_child_access): New parameter set_grp_write, set
25168 grp_write to its value.
25169 (propagate_subaccesses_across_link): Also propagate grp_write flag
25171 (propagate_all_subaccesses): Push the closest parent back to work
25172 queue if add_access_to_work_queue returned true.
25174 2017-05-02 Richard Biener <rguenther@suse.de>
25176 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25177 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25178 -fstrict-overflow documentation.
25179 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25180 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25181 flag_strict_overflow.
25182 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25183 * lto-opts.c (lto_write_options): Do not stream it.
25184 * lto-wrapper.c (merge_and_complain): Do not handle it.
25185 * opts.c (default_options_table): Do not set -fstrict-overflow.
25186 (finish_options): Likewise do not clear it when sanitizing.
25187 * simplify-rtx.c (simplify_const_relational_operation): Do not
25188 test flag_strict_overflow.
25190 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25192 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25193 using enabled attribute.
25194 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25195 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25196 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25197 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25198 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25199 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25200 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25201 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25202 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25203 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25205 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25207 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25209 2017-05-02 Richard Biener <rguenther@suse.de>
25211 PR tree-optimization/80591
25213 2017-04-10 Richard Biener <rguenther@suse.de>
25215 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25218 2017-05-02 Richard Biener <rguenther@suse.de>
25220 PR tree-optimization/80549
25221 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25222 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25223 headers if they do not have a preheader.
25225 2017-05-02 Martin Liska <mliska@suse.cz>
25228 * common.opt: Fix typo.
25229 * doc/invoke.texi: Likewise.
25231 2017-05-01 Jan Beulich <jbeulich@suse.com>
25233 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25234 swapping, add (x,x,m,x,n) alternative.
25236 2017-05-01 Nathan Sidwell <nathan@acm.org>
25238 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25239 unnecessary unadjusted_alignment check.
25241 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25244 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25246 * gimplify.c (gimplify_cilk_detach): New function.
25247 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25248 * tree-core.h: Document EXPR_CILK_SPAWN.
25249 * tree.h (EXPR_CILK_SPAWN): Define.
25251 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25253 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25254 to use new fixit_hint representation, using the "replace" logic.
25255 (get_line_span_for_fixit_hint): Likewise.
25256 (layout::print_any_fixits): Likewise.
25257 (selftest::test_one_liner_many_fixits): Rename to...
25258 (selftest::test_one_liner_many_fixits_1): ...this, and update
25259 comment and expected output to reflect that the multiple fix-it
25260 hints are now consolidated into one insertion.
25261 (selftest::test_one_liner_many_fixits_2): New test.
25262 (selftest::test_diagnostic_show_locus_one_liner): Update for
25264 (selftest::test_fixit_consolidation): Update for fix-it API
25266 * diagnostic.c (print_parseable_fixits): Likewise.
25267 * edit-context.c (edited_line::m_line_events): Convert from
25268 auto_vec <line_event *> to auto_vec <line_event>.
25269 (class line_event): Convert from abstract base class to a concrete
25270 class, taking over the role of replace_event.
25271 (class insert_event): Delete.
25272 (class replace_event): Rename to class line_event. Convert to
25274 (edit_context::add_fixits): Reimplement.
25275 (edit_context::apply_insert): Delete.
25276 (edit_context::apply_replace): Rename to...
25277 (edit_context::apply_fixit): ...this. Convert to half-open range.
25278 (edited_file::apply_insert): Delete.
25279 (edited_file::apply_replace): Rename to...
25280 (edited_file::apply_fixit): ...this.
25281 (edited_line::~edited_line): Drop deletion of events.
25282 (edited_line::apply_insert): Delete.
25283 (edited_line::apply_replace): Rename to...
25284 (edited_line::apply_fixit): ...this. Convert to half-open range.
25285 Update for change to type of m_line_events.
25286 * edit-context.h (edit_context::apply_insert): Delete.
25287 (edit_context::apply_replace): Rename to...
25288 (edit_context::apply_fixit): ...this.
25290 2017-05-01 Martin Sebor <msebor@redhat.com>
25292 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25295 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25298 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25299 __get_cpuid_max returns 0.
25300 (__get_cpuid_count): Ditto.
25302 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25304 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25305 replacement expression is another instance of one of its arguments.
25307 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25310 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25311 check for stack push/pop autoinc.
25312 * config/i386/i386.c (ix86_agi_dependent): Return false
25313 if the only reason why modified_in_p returned true is that
25314 addr is SP based and set_insn is a push or pop.
25316 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25318 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25321 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25324 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25325 (account_size_time): Use two predicates - exec_pred and
25327 (evaluate_conditions_for_known_args): Compute both clause and
25329 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25330 (inline_summary_t::duplicate): Update.
25331 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25333 (compute_inline_parameters): Likewise.
25334 (estimate_edge_size_and_time): Update caluclation of time.
25335 (estimate_node_size_and_time): Compute both time and nonspecialized
25337 (estimate_ipcp_clone_size_and_time): Update.
25338 (inline_merge_summary): Update.
25339 (do_estimate_edge_time): Update.
25340 (do_estimate_edge_size): Update.
25341 (do_estimate_edge_hints): Update.
25342 (inline_read_section, inline_write_summary): Stream both new predicates.
25343 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25345 (compute_inlined_call_time): Cleanup.
25346 (big_speedup_p): Update.
25347 (edge_badness): Update.
25348 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25349 (size_time_entry): Replace predicate by exec_predicate and
25350 nonconst_predicate.
25351 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25352 (estimate_edge_time): Return also nonspec_time.
25353 (reset_edge_growth_cache): Update.
25355 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25357 PR rtl-optimization/80491
25358 * ifcvt.c (noce_process_if_block): When looking for x setter
25359 with missing else_bb, don't check only the insn right before
25360 cond_earliest, but look for the last insn that x is modified in
25361 within the same bb.
25363 PR rtl-optimization/80491
25364 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25366 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25368 PR tree-optimization/80487
25369 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25371 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25373 PR tree-optimization/79697
25374 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25375 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25376 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25378 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25379 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25381 2017-04-28 Martin Sebor <msebor@redhat.com>
25383 PR tree-optimization/80523
25384 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25385 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25387 (maybe_warn, format_directive, parse_directive): Use new functions.
25388 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25390 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25392 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25394 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25396 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25397 target_header_dir): Set correctly.
25398 * configure: Regenerated.
25399 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25400 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25401 instead of SYSTEM_HEADER_DIR.
25403 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25405 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25406 (estimate_local_effects): Likewise.
25407 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25408 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25409 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25410 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25411 do_estimate_edge_time, estimate_edge_time): Likewise.
25412 * ipa-inline-analysis.c (estimate_node_size_and_time,
25413 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25414 (estimate_time_after_inlining): Remove.
25416 2017-04-28 Martin Liska <mliska@suse.cz>
25418 * doc/gcov.texi: Enhance documentation of gcov.
25420 2017-04-28 Martin Liska <mliska@suse.cz>
25422 * doc/gcov.texi: Sort options in alphabetic order.
25423 * doc/gcov-dump.texi: Likewise.
25424 * doc/gcov-tool.texi: Likewise.
25425 * gcov.c (print_usage): Likewise.
25426 * gcov-dump.c (print_usage): Likewise.
25427 * gcov-tool.c (print_merge_usage_message): Likewise.
25428 (print_rewrite_usage_message): Likewise.
25429 (print_overlap_usage_message): Likewise.
25431 2017-04-28 Martin Liska <mliska@suse.cz>
25433 PR gcov-profile/53915
25434 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25436 2017-04-28 Martin Liska <mliska@suse.cz>
25438 PR gcov-profile/79891
25439 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25440 is marked by compiler as living on a line.
25441 (get_cycles_count): Remove usage of the union.
25442 (output_intermediate_file): Likewise.
25443 (find_source): Fix GNU coding style.
25444 (accumulate_line_counts): Remove old non-all block mode.
25445 (output_lines): Remove usage of the union.
25446 * profile.c (output_location): Include all BBs, even if
25447 belonging to a same line (and file) as a previous BB.
25449 2017-04-28 Martin Liska <mliska@suse.cz>
25451 * gcov.c (process_args): Handle new argument 'w'.
25452 (read_graph_file): Assign ID to BBs.
25453 (output_branch_count): Display BB # if verbose flag is set.
25454 (output_lines): Likewise for arcs.
25455 (print_usage): Add '--verbose' option help.
25456 * doc/gcov.texi: Document --verbose (-w) option.
25458 2017-04-28 Martin Liska <mliska@suse.cz>
25460 * gcov.c (struct block_location_info): New struct.
25461 (process_file): Fill up the new structure.
25462 (read_graph_file): Replace usage of encoding by the newly added
25464 (add_line_counts): Likewise.
25465 (accumulate_line_counts): Remove usage of the union.
25466 (function_info::function_info): New function.
25467 (function_info::~function_info): Likewise.
25468 (process_file): Call delete instead of release_function.
25469 (release_function): Release the function.
25470 (release_structures): Call delete instead of release_function.
25471 (solve_flow_graph): Replace usage of num_blocks.
25472 (find_exception_blocks): Likewise.
25473 (output_lines): Fix GNU coding style.
25475 2017-04-28 Martin Liska <mliska@suse.cz>
25478 * coverage.c (coverage_remove_note_file): New function.
25479 * coverage.h: Declare the function.
25480 * toplev.c (finalize): Clean if an error has been seen.
25482 2017-04-28 Martin Liska <mliska@suse.cz>
25484 PR gcov-profile/80031
25485 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25486 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25487 * gcov.c (read_graph_file): Read just number of blocks.
25488 * profile.c (branch_prob): Do not stream 0 flags per a basic
25491 2017-04-28 Martin Liska <mliska@suse.cz>
25493 * gcov-dump.c (tag_*): Add new argument to declarations.
25494 (dump_gcov_file): Likewise.
25495 (tag_blocks): Add and use new argument depth.
25496 (tag_arcs): Likewise.
25497 (tag_lines): Likewise.
25498 (tag_counters): Likewise.
25499 (tag_summary): Likewise.
25500 (dump_working_sets): Use depth to do a proper indentation.
25502 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25505 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25506 * symtab.c (symtab_node::debug_symtab): Move definition here.
25508 2017-04-28 Richard Biener <rguenther@suse.de>
25510 * lto-streamer.h (LTO_major_version): Bump to 7.
25512 2017-04-28 Richard Biener <rguenther@suse.de>
25514 * tree-vrp.c (assert_info): New struct.
25515 (add_assert_info): New helper.
25516 (register_edge_assert_for_2): Refactor to add asserts to a vector
25518 (register_edge_assert_for_1): Likewise.
25519 (register_edge_assert_for): Likewise.
25520 (finish_register_edge_assert_for): New helper actually registering
25521 asserts where live on edge.
25522 (find_conditional_asserts): Adjust.
25523 (find_switch_asserts): Likewise.
25524 (evrp_dom_walker::try_find_new_range): Generalize.
25525 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25527 2017-04-27 Marek Polacek <polacek@redhat.com>
25530 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25531 arg10 and arg11 to itype.
25533 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25535 * doc/extend.texi (Object Size Checking): Improve grammar.
25537 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25540 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25541 that the logic for permitting reciprocal estimates matches that
25544 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25547 * tree.c (type_cache_hasher::equal): Only compare
25548 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25549 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25550 non-aggregate element types.
25551 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25552 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25554 2017-04-27 Richard Biener <rguenther@suse.de>
25556 PR middle-end/80533
25557 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25558 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25559 keeping a reference to a trailing array.
25561 2017-04-27 Richard Biener <rguenther@suse.de>
25563 PR middle-end/80539
25564 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25565 being in loop-closed SSA form conservatively.
25566 (chrec_fold_multiply_poly_poly): Likewise.
25568 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25570 PR middle-end/79665
25571 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25572 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25574 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25577 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25578 (aarch64_function_arg_alignment): Return unsigned int again, but still
25579 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25580 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25581 Don't emit -Wpsabi note.
25582 (aarch64_function_arg_boundary): Likewise.
25583 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25586 2017-04-26 Nathan Sidwell <nathan@acm.org>
25588 * tree.h (crc32_unsigned_n): Declare.
25589 (crc32_unsigned, crc32_unsigned): Make inline.
25590 * tree.c (crc32_unsigned_bits): Replace with ...
25591 (crc32_unsigned_n): ... this.
25592 (crc32_unsigned, crc32_byte): Remove.
25593 (crc32_string): Remove unnecessary braces.
25595 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25597 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25598 * ipa-inline-analysis.c (MAX_TIME): Remove.
25599 (account_size_time): Use sreal for time.
25600 (dump_inline_summary): Update.
25601 (estimate_function_body_sizes): Update.
25602 (estimate_edge_size_and_time): Update.
25603 (estimate_calls_size_and_time): Update.
25604 (estimate_node_size_and_time): Update.
25605 (inline_merge_summary): Update.
25606 (inline_update_overall_summary): Update.
25607 (estimate_time_after_inlining): Update.
25608 (inline_read_section): Update.
25609 (inline_write_summary): Update.
25610 * ipa-inline.c (compute_uninlined_call_time): Update.
25611 (compute_inlined_call_time): Update.
25612 (recursive_inlining): Update.
25613 (inline_small_functions): Update.
25614 (dump_overall_stats): Update.
25615 * ipa-inline.h: Include sreal.h.
25616 (size_time_entry): Turn time to sreal.
25617 (inline_summary): Turn self_time nad time to sreal.
25619 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25621 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25623 (sreal::stream_out, sreal::stream_in): New.
25624 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25626 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25628 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25631 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25634 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25636 (dimode_scalar_chain::compute_convert_gain): Ditto.
25637 (dimode_scalar_chain::make_vector_copies): Ditto.
25638 (dimode_scalar_chain::convert_reg): Ditto.
25639 (dimode_scalar_chain::convert_insn): Ditto.
25640 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25641 (VI248_AVX512BW_1): New mode iterator.
25642 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25643 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25646 2017-04-25 Martin Sebor <msebor@redhat.com>
25648 PR tree-optimization/80497
25649 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25650 constants are representable in HOST_WIDE_INT.
25651 (parse_directive): Ditto.
25653 2017-04-25 Martin Sebor <msebor@redhat.com>
25656 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25657 (new_zero_array): Adjust signature.
25658 (dom_info::dom_init): Used unsigned rather that size_t.
25659 (dom_info::dom_info): Same.
25661 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25662 Jakub Jelinek <jakub@redhat.com>
25665 * config/arm/arm.c: Include gimple.h.
25666 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25667 returns negative, increment ncrn only if it returned positive.
25668 (arm_needs_doubleword_align): Return int instead of bool,
25669 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25670 members, but if there is any such non-FIELD_DECL
25671 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25672 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25673 returns negative, increment nregs only if it returned positive.
25674 (arm_setup_incoming_varargs): Likewise.
25675 (arm_function_arg_boundary): Emit -Wpsabi note if
25676 arm_needs_doubleword_align returns negative, return
25677 DOUBLEWORD_ALIGNMENT only if it returned positive.
25679 2017-04-25 Marek Polacek <polacek@redhat.com>
25682 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25683 first argument to type.
25685 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25688 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25689 type checks to test for compatibility instead of equality.
25691 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25692 Jakub Jelinek <jakub@redhat.com>
25695 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25697 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25698 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25699 the alignment computation, but return their maximum in warn_alignment.
25700 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25701 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25703 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25704 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25707 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25709 * config/arc/simdext.md (dmpyh): Fix typo.
25711 2017-04-25 Richard Biener <rguenther@suse.de>
25713 PR tree-optimization/80492
25714 * alias.c (compare_base_decls): Handle registers with asm
25715 specification conservatively.
25716 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25717 compare_base_decls returning dont-know properly.
25719 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25721 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25722 (legitimate_offset_address_p): New function.
25723 (arc_legitimate_address_p): Use above function.
25725 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25727 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25729 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25731 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25732 ACCH registers whenever they are available.
25734 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25736 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25737 double regs fix when not used.
25739 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25741 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25743 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25744 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25746 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25748 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25749 when emitting short ADD instructions.
25751 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25753 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25755 (cmpsi_cc_c_insn): Likewise.
25756 (cbranchsi4_scratch): Compute proper instruction length using
25757 compact_hreg_operand.
25758 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25760 2017-04-25 Richard Biener <rguenther@suse.de>
25762 PR middle-end/80509
25763 * passes.c (pass_manager::pass_manager): Initialize
25764 m_name_to_pass_map.
25766 2017-04-25 Richard Biener <rguenther@suse.de>
25768 PR tree-optimization/79201
25769 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25771 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25774 * config/s390/vector.md: Split MEM->GPR vector moves for
25775 non-s_operand addresses.
25777 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25780 * config/s390/predicates.md (reload_const_wide_int_operand): New
25782 * config/s390/s390.md ("movti"): Remove d/P alternative.
25783 ("movti_bigconst"): New pattern definition.
25785 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25788 * s390-protos.h (s390_expand_cs_hqi): Removed.
25789 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25790 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25791 modes as well as CCZ1mode and CCZmode.
25792 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25793 signature of s390_emit_compare_and_swap.
25794 (s390_expand_cs_hqi): Likewise, make static.
25795 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25796 compare-and-swap, in some cases.
25797 (s390_expand_cs): Wrapper function.
25798 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25800 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25801 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25802 patterns for small and large integers. Forbid symref memory operands.
25803 Move expander to s390.c. Require cc register.
25804 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25805 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25806 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25807 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25808 symref memory operands. Remove CC mode and call s390_match_ccmode
25810 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25812 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25814 * config/s390/s390.md (define_peephole2): New peephole to help
25815 combining the load-and-test pattern with volatile memory.
25817 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25819 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25820 with CCZmode for TARGET_Z196.
25822 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25824 PR rtl-optimization/80501
25825 * combine.c (make_compound_operation_int): Set subreg_code to SET
25826 even for AND with mask of the sign bit of mode.
25828 PR rtl-optimization/80500
25829 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25830 sum's initial value.
25832 2017-04-25 Julian Brown <julian@codesourcery.com>
25833 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25835 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25837 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25839 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25841 2017-04-25 Julian Brown <julian@codesourcery.com>
25842 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25844 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25845 (thunderx2t99_sha): New Reservation.
25847 2017-04-25 Julian Brown <julian@codesourcery.com>
25848 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25850 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25851 type for 1-element load.
25853 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25855 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25857 2017-04-24 Martin Jambor <mjambor@suse.cz>
25859 PR tree-optimization/80293
25860 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25861 char arrays not totally scalarizable if it is false.
25862 (analyze_all_variable_accesses): Pass correct value in the new
25863 parameter. Add a statistics counter.
25865 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25867 PR middle-end/79931
25868 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25870 2017-04-24 Richard Biener <rguenther@suse.de>
25872 PR tree-optimization/80494
25873 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25874 out for complex types.
25876 2017-04-24 Richard Biener <rguenther@suse.de>
25878 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25879 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25880 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25881 (DFS): Adjust and never fail.
25882 (sccvn_dom_walker::fail): Remove.
25883 (sccvn_dom_walker::before_dom_children): Adjust.
25884 (run_scc_vn): Likewise and never fail.
25885 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25886 (pass_fre::execute): Likewise.
25888 2017-04-24 Richard Biener <rguenther@suse.de>
25890 PR tree-optimization/79725
25891 * tree-ssa-sink.c (statement_sink_location): Return whether
25892 failure reason was zero uses. Move that check later.
25893 (sink_code_in_bb): Deal with zero uses by removing the stmt
25896 2017-04-24 Richard Biener <rguenther@suse.de>
25899 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25900 pointer-based references.
25902 2017-04-24 Richard Biener <rguenther@suse.de>
25905 * pass_manager.h (pass_manager::operator new): Remove.
25906 (pass_manager::operator delete): Likewise.
25907 * passes.c (pass_manager::operator new): Remove.
25908 (pass_manager::operator delete): Likewise.
25909 (pass_manager::pass_manager): Zero individual pass members.
25911 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25914 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25915 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25916 Check "XEXP (src, 1)" operand here.
25917 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25918 Check "XEXP (src, 1)" operand here.
25919 (dimode_scalar_chain::make_vector_copies): Detect count register
25920 of a shift instruction. Zero extend count register from QImode
25921 to DImode to satisfy vector shift pattern count operand predicate.
25922 Substitute vector shift count operand with a DImode copy.
25923 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25926 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25928 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25929 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25930 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25931 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25932 (UNSPEC_NOREX_MEM): Remove definition.
25934 2017-04-21 Richard Biener <rguenther@suse.de>
25936 PR tree-optimization/79547
25937 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25938 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25939 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25940 without any constraints.
25942 2017-04-21 Richard Biener <rguenther@suse.de>
25944 PR tree-optimization/78847
25945 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25947 2017-04-21 Richard Biener <rguenther@suse.de>
25949 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25950 (build_distinct_type_copy): Likewise.
25951 (build_variant_type_copy): Likewise.
25952 * tree.c (build_qualified_type): Pass down mem-stat info.
25953 (build_distinct_type_copy): Likewise.
25954 (build_variant_type_copy): Likewise.
25956 2017-04-21 Richard Biener <rguenther@suse.de>
25958 PR tree-optimization/80237
25959 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25961 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25962 for a simplified result.
25964 2016-04-21 Richard Biener <rguenther@suse.de>
25966 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25967 sth as strict as a simple_iv but a chrec without symbols and an
25968 operand defined in the loop we are peeling (and not some subloop).
25969 (propagate_constants_for_unrolling): Propagate all constants.
25971 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25974 * config/i386/i386.c (print_reg): Remove assert for disalowed
25975 regno values, call output_operand_lossage instead.
25977 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25980 * config/i386/constraints.md (Yc): New register constraint.
25981 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25982 Use Yc constraint for alternative 2 of operand 0. Remove
25983 preferred_for_speed attribute.
25985 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25987 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25988 lastprivate clauses in SIMT case.
25990 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25992 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25994 2017-04-20 Richard Biener <rguenther@suse.de>
25996 PR tree-optimization/57796
25997 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25998 as N scalar stores.
25999 (vect_model_load_cost): Cost gathers as N scalar loads.
26001 2017-04-20 Richard Biener <rguenther@suse.de>
26003 * ggc-page.c (ggc_allocated_p): Rename to ...
26004 (safe_lookup_page_table_entry): ... this and return the lookup
26006 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
26008 2017-04-20 Richard Biener <rguenther@suse.de>
26010 PR tree-optimization/80453
26011 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
26012 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
26013 from the conditions.
26014 (vn_phi_eq): Pass them down.
26015 (vn_phi_lookup): Record them.
26016 (vn_phi_insert): Likewise.
26018 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
26020 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
26021 uninitialized variable warning to avoid buffer overrun.
26023 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26026 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26027 is suppressed for '{ 0 }' in C.
26029 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26031 * BASE-VER: Set to 8.0.0.
26033 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26035 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26036 priority .init_array and .fini_array section with SECTION_NOTYPE
26039 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26041 PR middle-end/80423
26042 * tree.h (build_array_type): Add typeless_storage default argument.
26043 * tree.c (type_cache_hasher::equal): Also compare
26044 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26045 (build_array_type): Add typeless_storage argument, set
26046 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26048 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26049 (build_array_type): Likewise. Add typeless_storage argument.
26051 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26052 Jakub Jelinek <jakub@redhat.com>
26054 PR tree-optimization/80426
26055 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26056 operation on symbolic operands, also compute the overflow for the
26057 invariant part when the operation degenerates into a negation.
26059 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26062 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26063 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26066 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26068 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26071 * config/avr/avr.c (tree.h): Include it.
26072 (cgraph.h): Include it.
26073 (avr_encode_section_info): Don't warn for uninitialized progmem
26074 variable if it's just an alias.
26076 2017-04-19 Richard Biener <rguenther@suse.de>
26079 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26080 when needed by AutoPGO.
26082 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26085 * doc/lto.texi: Remove an extra 'that'.
26087 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26089 PR rtl-optimization/80429
26090 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26091 are only used in debug insns.
26093 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26094 Vladimir Makarov <vmakarov@redhat.com>
26096 * config/sparc/predicates.md (input_operand): Add comment. Return
26097 true for any memory operand when LRA is in progress.
26098 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26100 2017-04-18 Jeff Law <law@redhat.com>
26103 * mips.md ({return,simple_return}_internal): Do not overwrite
26106 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26108 PR tree-optimization/80443
26109 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26110 instead of adding 1, subtract -1 and similarly instead of subtracting
26113 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26115 PR rtl-optimization/80357
26116 * haifa-sched.c (tmp_bitmap): New variable.
26117 (model_recompute): Handle duplicate use records.
26118 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26119 (free_global_sched_pressure_data): Free it.
26121 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26124 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26125 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26126 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26127 instead of SYSTEM_HEADER_DIR.
26129 2017-04-18 Jeff Law <law@redhat.com>
26131 PR middle-end/80422
26132 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26133 predecessors after walking up the insn chain.
26135 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26138 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26139 sizetype type into debug info.
26141 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26144 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26145 unneeded test for TARGET_UPPER_REGS_SF.
26146 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26148 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26151 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26152 instead of gsi_after_labels.
26154 2017-04-18 Jeff Law <law@redhat.com>
26156 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26160 2017-04-13 Jeff Law <law@redhat.com>
26161 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26162 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26164 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26167 * config/avr/avr.c (intl.h): Include it.
26168 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26170 2017-04-18 Martin Liska <mliska@suse.cz>
26172 PR gcov-profile/78783
26173 * gcov-tool.c (gcov_output_files): Validate that destination
26174 file is either removed by the tool or by a user.
26176 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26177 Guy Benyei <guybe@mellanox.com>
26179 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26180 block, and do not negate it, the stored id is already negative.
26182 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26184 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26186 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26189 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26190 masks of options that should be turned off if the VSX vector
26191 options are turned off.
26192 (OTHER_P8_VECTOR_MASKS): Likewise.
26193 (OTHER_VSX_VECTOR_MASKS): Likewise.
26194 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26195 rs6000_disable_incompatible_switches to validate no type switches
26197 (rs6000_incompatible_switch): New function to disallow turning on
26198 other vector options if -mno-vsx, -mno-power8-vector, or
26199 -mno-power9-vector are specified.
26201 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26203 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26205 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26207 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26208 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26209 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26210 (ARG_POINTER_CFA_OFFSET): Likewise.
26212 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26214 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26215 conditions to take advantage of various optimizations.
26217 2017-04-13 Jeff Law <law@redhat.com>
26219 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26220 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26221 (zero_extendsidi2_dext): Likewise.
26223 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26226 * fold-const.c (fold_ternary_loc): Revert
26227 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26230 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26232 PR rtl-optimization/80343
26233 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26234 new scratch pseudo.
26236 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26239 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26240 to ubsan_encode_value.
26242 2017-04-13 Jeff Law <law@redhat.com>
26244 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26245 appearing in DEBUG_INSNs.
26247 2017-04-13 Martin Liska <mliska@suse.cz>
26249 PR gcov-profile/80413
26250 * gcov-io.c (gcov_write_string): Copy to buffer just when
26251 allocated size is greater than zero.
26253 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26256 * dwarf2out.c (decls_for_scope): Ignore declarations of
26257 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26259 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26262 * ipa-visibility.c (non_local_p): Fix typos.
26263 (localize_node): When localizing symbol in same comdat group,
26264 dissolve the group only when we know external symbols are going
26266 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26268 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26270 PR tree-optimization/79390
26271 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26272 order does not result in usable sequence, retry with reversed operand
26278 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26279 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26280 op0 instead of fold_convert_loc (loc, type, arg0).
26282 2017-04-12 Jeff Law <law@redhat.com>
26284 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26285 has a delay slot in the generated code.
26287 * config/cris/cris.md (cris_preferred_reload_class): Return
26288 GENNONACR_REGS rather than GENERAL_REGS.
26290 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26293 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26294 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26295 signedness of the result type.
26297 2017-04-12 Richard Biener <rguenther@suse.de>
26298 Jeff Law <law@redhat.com>
26300 PR tree-optimization/80359
26301 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26302 trim stores to TARGET_MEM_REFs.
26304 2017-04-12 Richard Biener <rguenther@suse.de>
26306 PR tree-optimization/79390
26307 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26308 threading case even more.
26310 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26313 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26314 for quad_address_p for TImode, instead of just not indexed_address.
26316 2017-04-12 Richard Biener <rguenther@suse.de>
26317 Bernd Edlinger <bernd.edlinger@hotmail.de>
26319 PR middle-end/79671
26320 * alias.c (component_uses_parent_alias_set_from): Handle
26321 TYPE_TYPELESS_STORAGE.
26322 (get_alias_set): Likewise.
26323 * tree-core.h (tree_type_common): Add typeless_storage flag.
26324 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26325 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26326 for types containing members with TYPE_TYPELESS_STORAGE.
26327 (place_field): Likewise.
26328 (layout_type): Likewise for ARRAY_TYPE.
26329 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26330 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26331 TYPE_TYPELESS_STORAGE.
26332 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26334 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26337 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26338 first argument to type.
26340 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26344 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26345 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26346 (rs6000_expand_binop_builtin): Likewise.
26347 (rs6000_expand_ternop_builtin): Likewise; also add missing
26348 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26349 vshasigma built-ins.
26350 * doc/extend.texi: Document that vec_xxpermdi's third argument
26351 must be a constant.
26353 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26355 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26356 Use shift_const cost parameter when calculating gain of STV shifts.
26358 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26360 PR rtl-optimization/70478
26361 * lra-constraints.c (process_alt_operands): Check memory for
26362 disfavoring memory insn operand.
26364 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26366 PR middle-end/80100
26367 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26368 left shift in unsigned HOST_WIDE_INT type.
26370 PR rtl-optimization/80385
26371 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26372 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26375 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26376 if they have any depend clauses.
26378 2017-04-11 Martin Liska <mliska@suse.cz>
26381 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26382 * ipa-split.c (split_function): Create a local comdat symbol
26383 if caller is in a comdat group.
26385 2017-04-11 Martin Liska <mliska@suse.cz>
26388 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26391 2017-04-11 Martin Sebor <msebor@redhat.com>
26393 PR middle-end/80364
26394 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26395 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26397 (directive::set_width, directive::set_precision, format_character):
26399 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26402 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26405 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26406 conflict, set target->arch_name instead of target->cpu_name.
26408 2017-04-11 Richard Biener <rguenther@suse.de>
26410 PR tree-optimization/80374
26411 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26412 build_zero_cst, remove fold_convertible_p check again.
26414 2017-04-11 Martin Liska <mliska@suse.cz>
26417 * ubsan.c (instrument_object_size): Do not instrument register
26420 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26423 * config/i386/i386-builtin-types.def
26424 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26425 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26426 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26427 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26428 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26429 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26430 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26431 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26432 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26433 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26434 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26435 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26436 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26437 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26438 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26439 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26440 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26441 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26442 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26443 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26444 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26445 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26446 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26447 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26448 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26449 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26450 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26451 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26452 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26453 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26454 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26455 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26456 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26457 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26458 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26459 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26460 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26461 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26462 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26463 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26464 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26465 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26466 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26467 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26468 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26469 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26471 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26472 flag to second_arg_count, handle 4 argument function type _COUNT
26473 aliases, handle second_arg_count on second argument rather than last.
26475 2017-04-10 Jeff Law <law@redhat.com>
26477 PR tree-optimization/80374
26478 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26479 record anything if we can not convert integer_zero_node to the
26482 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26485 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26486 Enhance special handling given to the TARGET_P9_MINMAX option in
26487 relation to certain other options.
26489 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26491 PR tree-optimization/80153
26492 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26493 remove POINTER_PLUS_EXPR's base part directly, rather than through
26496 2017-04-10 Richard Biener <rguenther@suse.de>
26497 Bin Cheng <bin.cheng@arm.com>
26499 PR tree-optimization/80153
26500 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26501 the first element of pointer type aff_tree. Build result expr in
26503 (add_elt_to_tree): Convert to type unconditionally. Remove other
26504 fold_convert calls.
26505 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26506 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26508 2017-04-10 Richard Biener <rguenther@suse.de>
26510 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26513 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26515 PR rtl-optimization/70478
26516 * lra-constraints.c (curr_small_class_check): New.
26517 (update_and_check_small_class_inputs): New.
26518 (process_alt_operands): Update curr_small_class_check. Disfavor
26519 alternative insn memory operands. Check available regs for small
26522 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26525 * config/mips/mips.opt (-mvirt): Update description.
26526 * doc/invoke.texi (-mvirt): Likewise.
26528 2017-04-10 Richard Biener <rguenther@suse.de>
26530 PR middle-end/80362
26531 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26532 looking for NEGATE_EXPR in -A / -B to A / B folding.
26534 2017-04-10 Martin Liska <mliska@suse.cz>
26536 PR gcov-profile/80224
26537 * gcov.c (print_usage): Fix usage string.
26538 (get_gcov_intermediate_filename): Remove.
26539 (output_gcov_file): Use both for normal and intermediate format.
26540 (generate_results): Do not initialize special file for
26541 intermediate format.
26543 2017-04-10 Richard Biener <rguenther@suse.de>
26545 PR tree-optimization/80304
26546 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26549 2017-04-10 Nathan Sidwell <nathan@acm.org>
26552 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26553 (rs6000_init_builtins): Use it.
26555 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26557 * config/arm/arm.md (<mrc>): Add mode to SET source.
26558 (<mrrc>): Likewise.
26560 2017-04-10 Richard Biener <rguenther@suse.de>
26562 PR middle-end/80344
26563 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26565 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26568 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26569 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26570 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26571 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26572 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26573 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26574 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26575 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26576 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26577 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26578 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26579 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26580 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26581 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26582 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26583 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26584 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26585 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26586 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26587 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26588 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26589 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26590 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26592 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26594 PR rtl-optimization/70478
26595 * lra-constraints.c: Reverse the last patch.
26597 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26599 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26600 Add comment for WCHAR_T.
26602 2017-04-08 Martin Liska <mliska@suse.cz>
26605 2017-04-07 Martin Liska <mliska@suse.cz>
26608 * ipa-split.c (split_function): Add function part to a same comdat
26611 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26614 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26616 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26618 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26619 (vec_pairing): ...to this.
26620 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26621 (rs6000_sched_init): Adjust for name change.
26622 (struct rs6000_sched_context): Likewise.
26623 (rs6000_init_sched_context): Likewise.
26624 (rs6000_set_sched_context): Likewise.
26626 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26632 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26634 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26635 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26636 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26638 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26640 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26642 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26644 PR rtl-optimization/70703
26645 * ira-color.c (update_conflict_hard_regno_costs): Use
26646 int64_t instead of HOST_WIDE_INT.
26648 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26650 PR rtl-optimization/70478
26651 * lra-constraints.c (process_alt_operands): Disfavor alternative
26652 insn memory operands.
26654 2017-04-07 Jeff Law <law@redhat.com>
26656 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26657 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26659 2017-04-07 Martin Liska <mliska@suse.cz>
26662 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26663 Show error message instead of an ICE.
26665 2017-04-07 Martin Liska <mliska@suse.cz>
26668 * ipa-split.c (split_function): Add function part to a same comdat
26671 2017-04-07 Richard Biener <rguenther@suse.de>
26673 PR middle-end/80341
26674 * tree.c (get_unwidened): Also handle ! for_type case for
26676 * convert.c (do_narrow): Split out from ...
26677 (convert_to_integer_1): ... here. Do not pass final truncation
26678 type to get_unwidened for TRUNC_DIV_EXPR.
26680 2017-04-07 Richard Biener <rguenther@suse.de>
26682 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26684 (aff_combination_const): Adjust.
26685 (aff_combination_scale): Likewise.
26686 (aff_combination_add_elt): Likewise.
26687 (aff_combination_add_cst): Likewise.
26688 (aff_combination_convert): Likewise.
26689 (add_elt_to_tree): Likewise. Remove unused argument.
26690 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26692 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26694 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26696 * config/arm/arm.c (arm_default_short_enums): Use
26697 ARM_DEFAULT_SHORT_ENUMS.
26698 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26700 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26703 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26704 members with redundant out-of-class redeclaration.
26706 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26709 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26710 * config/i386/i386.md (*zero_extendsidi2):
26711 Add (?*x,*x) and (?*v,*v) alternatives.
26713 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26716 * config/i386/i386.c (ix86_expand_builtin)
26717 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26718 mode from insn data. Convert operands to insn operand mode.
26719 Copy operands that don't satisfy insn predicate to a register.
26721 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26723 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26726 2017-04-06 Richard Biener <rguenther@suse.de>
26728 PR tree-optimization/80334
26729 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26730 preserve alignment of accesses.
26732 2017-04-06 Richard Biener <rguenther@suse.de>
26734 PR tree-optimization/80262
26735 * tree-sra.c (build_ref_for_offset): Preserve address-space
26737 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26738 Drop useless address-space information on MEM_REF offsets.
26740 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26742 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26744 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26746 PR rtl-optimization/70703
26747 * ira-color.c (update_conflict_hard_regno_costs): Use
26748 HOST_WIDE_INT instead of long.
26750 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26753 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26754 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26756 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26757 for x86_64 target. Handle -m3dnowa option.
26759 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26761 PR rtl-optimization/70703
26762 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26763 (update_conflict_hard_regno_costs): Use long instead of unsigned
26764 arithmetic for cost calculation.
26766 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26767 Bernd Edlinger <bernd.edlinger@hotmail.de>
26770 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26773 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26776 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26777 ptr_mode with Pmode throughout.
26778 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26779 into probe_stack_range and use DImode.
26781 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26784 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26785 call_eh_return is true.
26787 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26789 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26790 Initialize last_match_fntype_index.
26792 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26795 * tree-nvr.c: Include internal-fn.h.
26796 (pass_return_slot::execute): Ignore internal calls without
26799 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26800 Richard Biener <rguenther@suse.de>
26803 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26804 captures used multiple times, except for the last use.
26805 * generic-match-head.c: Include gimplify.h.
26807 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26809 PR tree-optimization/79390
26810 * target.h (struct noce_if_info): Declare.
26811 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26812 * target.def (noce_conversion_profitable_p): New target hook.
26813 * ifcvt.h (struct noce_if_info): New type, moved from ...
26814 * ifcvt.c (struct noce_if_info): ... here.
26815 (noce_conversion_profitable_p): Renamed to ...
26816 (default_noce_conversion_profitable_p): ... this. No longer
26818 (noce_try_store_flag_constants, noce_try_addcc,
26819 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26820 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26821 instead of noce_conversion_profitable_p.
26822 * config/i386/i386.c: Include ifcvt.h.
26823 (ix86_option_override_internal): Don't override
26824 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26825 (ix86_noce_conversion_profitable_p): New function.
26826 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26827 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26828 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26829 * doc/tm.texi: Regenerated.
26831 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26833 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26836 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26839 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26840 instructions for small multiply cores.
26842 2017-04-04 Jeff Law <law@redhat.com>
26844 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26846 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26847 that are not set by the loop over the elements.
26849 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26852 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26853 int mode, convert_modes it to mode as unsigned, otherwise use
26854 lowpart_subreg to mode rather than SImode.
26855 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26856 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26857 Use DImode instead of SImode for the shift count operand.
26858 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26861 2017-04-04 Richard Biener <rguenther@suse.de>
26863 PR middle-end/80281
26864 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26865 arithmetic done for the negate or the plus. Simplify.
26866 (A - (-B) -> A + B): Likewise.
26867 * fold-const.c (split_tree): Make sure to not negate pointers.
26869 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26871 PR rtl-optimization/60818
26872 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26873 a compare of comparisons with the thing compared if this results
26874 in a different machine mode.
26876 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26878 * alias.c (base_alias_check): Fix typo in comment.
26879 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26880 * cgraphunit.c (symbol_table::compile): Likewise.
26881 * collect2.c (maybe_run_lto_and_relink): Likewise.
26882 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26883 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26884 * config/avr/avr.c (avr_map_op_t): Likewise.
26885 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26886 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26887 * config/epiphany/epiphany.md (movcc): Likewise.
26888 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26889 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26891 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26892 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26893 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26894 * config/sh/sh.c (sh_rtx_costs): Likewise.
26895 * fold-const.c (fold_truth_andor): Likewise.
26896 * genautomata.c (collapse_flag): Likewise.
26897 * gengtype.h (struct type::u::s): Likewise.
26898 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26899 * input.c (FORMAT_AMOUNT): Likewise.
26900 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26901 (known_aggs_to_agg_replacement_list): Likewise.
26902 * ipa-inline-analysis.c: Likewise.
26903 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26904 * ipa-polymorphic-call.c
26905 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26906 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26907 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26909 * modulo-sched.c (apply_reg_moves): Likewise.
26910 * omp-expand.c (build_omp_regions_1): Likewise.
26911 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26912 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26913 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26914 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26915 * value-prof.c: Likewise.
26916 * var-tracking.c (val_reset): Likewise.
26918 2017-04-03 Richard Biener <rguenther@suse.de>
26920 PR tree-optimization/80275
26921 * fold-const.c (split_address_to_core_and_offset): Handle
26924 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26926 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26927 descriptors is at least equal to that of functions.
26929 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26931 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26933 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26936 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26937 (mov<IMOD4:mode>): New expander.
26938 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26940 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26942 PR rtl-optimization/79405
26943 * fwprop.c (propagations_left): New variable.
26944 (forward_propagate_into): Decrement it.
26945 (fwprop_init): Initialize it.
26946 (fw_prop): If the variable has reached zero, stop propagating.
26947 (fwprop_addr): Ditto.
26949 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26952 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26953 a FUNCTION_DECL, pass it as decl instead of origin to
26956 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26958 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26961 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26964 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26965 TARGET_VSX_SMALL_INTEGER.
26967 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26969 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26970 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26972 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26974 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26975 extraction from odd-numbered MSA register.
26977 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26979 PR middle-end/80173
26980 * expmed.c (store_bit_field_1): Don't attempt to create
26981 a word subreg out of hard registers wider than word if they
26982 have HARD_REGNO_NREGS of 1 for their mode.
26984 PR middle-end/80163
26985 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26986 conversions to integer types wider than word and pointer.
26989 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26990 (rtx_equal_for_cselib_p): Pass 0 to it.
26991 * cselib.c (cselib_hasher::equal): Likewise.
26992 (rtx_equal_for_cselib_1): Add depth argument. If depth
26993 is 128, don't look up VALUE locs and punt. Increment
26994 depth in recursive calls when walking VALUE locs.
26996 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26998 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26999 (make_gcov_file_name): Use the canonical path name for generating
27001 (read_line): Fix handling of files with ascii null bytes.
27003 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
27005 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
27006 to initialise a vector register instead
27007 of using a const_int.
27009 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27011 PR translation/80189
27012 * gimplify.c (omp_default_clause): Use %qs instead of %s in
27013 diagnostic messages.
27015 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
27018 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
27019 (dfp_diex_<mode>): Update mode of operand 1.
27020 * doc/extend.texi (dxex, dxexq): Document change to return type.
27021 (diex, diexq): Document change to argument type.
27023 2017-03-30 Martin Jambor <mjambor@suse.cz>
27026 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27027 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27028 it reflects the signature changes performed at the callee side.
27029 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27030 to cgraph_build_function_type_skip_args.
27031 (build_function_decl_skip_args): Adjust call to the above function.
27033 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27036 * config/i386/sse.md
27037 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27038 register as dest whenever it is a MEM not rtx_equal_p to the
27039 corresponding dup operand, and when forcing into reg move the
27040 reg into the memory afterwards.
27041 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27042 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27043 for the force_reg mode.
27044 (avx512vl_vextractf128<mode>): Use register as dest either
27045 always when a MEM, or when it is a MEM not rtx_equal_p to the
27046 corresponding dup operand, or even not when it is a CONST_VECTOR
27047 depending on the mode and lo vs. hi.
27048 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27050 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27051 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27052 Likewise. Require that operands[2] is even.
27053 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27054 Remove extraneous parens. Require that operands[2] is a multiple
27056 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27057 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27058 disallow memory then.
27060 2017-03-30 Richard Biener <rguenther@suse.de>
27062 PR tree-optimization/77498
27063 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27064 to non-constants over backedges.
27066 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27068 PR rtl-optimization/80233
27069 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27070 as last_combined_insn. Do not test for BARRIER_P separately.
27072 2017-03-29 Andreas Schwab <schwab@suse.de>
27075 * calls.c (prepare_call_address): Convert funexp to Pmode before
27076 copying to temp reg.
27078 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27080 PR tree-optimization/80158
27081 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27082 Handle possible future case of more than one alternate
27084 (replace_rhs_if_not_dup): Likewise.
27085 (replace_one_candidate): Likewise.
27087 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27089 PR rtl-optimization/80193
27090 * ira.c (ira): Do not check allocation for LRA.
27092 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27094 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27095 (nvptx_output_simt_exit): Declare.
27096 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27097 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27098 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27099 (nvptx_declare_function_name): ...here. Emit declaration of local
27100 memory space buffer for omp_simt_enter insn.
27101 (nvptx_output_unisimt_switch): New.
27102 (nvptx_output_softstack_switch): New.
27103 (nvptx_output_simt_enter): New.
27104 (nvptx_output_simt_exit): New.
27105 * config/nvptx/nvptx.h (struct machine_function): New fields
27106 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27107 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27108 (UNSPECV_SIMT_EXIT): Ditto.
27109 (omp_simt_enter_insn): New insn.
27110 (omp_simt_enter): New expansion.
27111 (omp_simt_exit): New insn.
27112 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27114 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27115 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27116 (expand_GOMP_SIMT_EXIT): New.
27117 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27118 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27119 (GOMP_SIMT_EXIT): Ditto.
27120 * target-insns.def (omp_simt_enter): New insn.
27121 (omp_simt_exit): Ditto.
27122 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27124 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27125 (lower_rec_input_clauses): Likewise.
27126 (lower_lastprivate_clauses): Handle SIMT privatization.
27128 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27129 (ompdevlow_adjust_simt_enter): New.
27130 (find_simtpriv_var_op): New.
27131 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27132 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27134 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27135 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27136 (copy_decl_for_dup_finish): Ditto.
27138 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27140 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27143 * config/i386/i386.c (ix86_option_override_internal): Always
27144 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27146 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27148 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27150 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27152 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27153 mark new edge's irreducible flag accordign to it.
27154 (vect_do_peeling): Check loop preheader edge's irreducible flag
27155 and pass it to function slpeel_add_loop_guard.
27157 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27159 PR tree-optimization/80218
27160 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27161 Update block frequencies and counts.
27163 2017-03-28 Richard Biener <rguenther@suse.de>
27165 PR tree-optimization/78644
27166 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27167 of a simplification result we may not use it at all.
27169 2017-03-28 Richard Biener <rguenther@suse.de>
27172 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27173 without arguments, generate default definition of a SSA name.
27175 2017-03-28 Richard Biener <rguenther@suse.de>
27177 PR middle-end/80222
27178 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27179 TYPE_REF_CAN_ALIAS_ALL references.
27180 * fold-const.c (fold_indirect_ref_1): Likewise.
27182 2017-03-28 Martin Liska <mliska@suse.cz>
27185 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27186 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27188 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27189 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27191 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27192 (EXTRA_SPECS): Define.
27193 (SUBTARGET_EXTRA_SPECS): Likewise.
27194 (SUBTARGET_CPP_SPEC): Likewise.
27195 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27196 SUBTARGET_EXTRA_SPECS.
27197 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27199 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27201 * config/arc/simdext.md (vst64_insn): Update pattern.
27202 (vld32wh_insn): Likewise.
27203 (vld32wl_insn): Likewise.
27204 (vld64_insn): Likewise.
27205 (vld32_insn): Likewise.
27207 2017-03-28 Marek Polacek <polacek@redhat.com>
27210 * fold-const.c (fold_comparison): Use protected_set_expr_location
27211 instead of SET_EXPR_LOCATION.
27213 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27215 * tree.c (add_expr): Avoid name lookup warning.
27217 2017-03-27 Jeff Law <law@redhat.com>
27219 PR tree-optimization/80216
27220 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27221 function name. Limit recursion depth.
27222 (record_temporary_equivalences): Corresponding changes.
27224 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27226 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27229 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27232 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27234 * cfgcleanup.c (reg_note_cfa_p): New array.
27235 (insns_have_identical_cfa_notes): New function.
27236 (old_insns_match_p): Don't cross-jump in between /f
27237 and non-/f instructions. If both i1 and i2 are frame related,
27238 verify all CFA notes, their order and content.
27240 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27243 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27244 HImode and SImode with zero extend to DImode to one insn.
27245 (bswap<mode>2_extenddi): Likewise.
27246 (bswapsi2_extenddi): Likewise.
27247 (bswaphi2_extendsi): Likewise.
27248 (bswaphi2): Combine bswap HImode and SImode into one insn.
27249 Separate memory insns from swapping register.
27250 (bswapsi2): Likewise.
27251 (bswap<mode>2): Likewise.
27252 (bswaphi2_internal): Delete, no longer used.
27253 (bswapsi2_internal): Likewise.
27254 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27255 store, and gpr<-gpr swap insns.
27256 (bswap<mode>2_store): Likewise.
27257 (bswaphi2_reg): Register only splitter, combine with the splitter.
27258 (bswaphi2 splitter): Likewise.
27259 (bswapsi2_reg): Likewise.
27260 (bswapsi2 splitter): Likewise.
27261 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27262 the insns into load, store, and register/register insns.
27263 (bswapdi2_ldbrx): Likewise.
27264 (bswapdi2_load): Likewise.
27265 (bswapdi2_store): Likewise.
27266 (bswapdi2_reg): Likewise.
27268 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27270 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27271 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27273 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27276 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27278 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27279 special handling for target option conflicts between dform
27280 options (-mpower9-dform, -mpower9-dform-vector,
27281 -mpower9-dform-scalar) and -mno-direct-move.
27283 2017-03-27 Richard Biener <rguenther@suse.de>
27285 PR tree-optimization/80181
27286 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27288 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27290 * config/arc/predicates.md (move_double_src_operand): Replace the
27291 call to move_double_src_operand with a call to address_operand.
27293 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27295 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27296 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27297 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27299 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27301 * config/arc/predicates.md (long_immediate_loadstore_operand):
27302 Consider scaled addresses cases.
27304 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27306 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27307 restored when in interrupt.
27308 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27309 doesn't have delay slot.
27311 2017-03-27 Richard Biener <rguenther@suse.de>
27314 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27315 inlined thunk clones.
27317 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27320 * asan.c (instrument_derefs): Copy over last operand from
27321 original COMPONENT_REF to the new COMPONENT_REF with
27322 DECL_BIT_FIELD_REPRESENTATIVE.
27323 * ubsan.c (instrument_object_size): Likewise.
27325 2017-03-27 Richard Biener <rguenther@suse.de>
27327 PR tree-optimization/80170
27328 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27329 sure DR/SCEV didnt fold in constants we do not see when looking
27330 at the reference base alignment.
27332 2017-03-27 Richard Biener <rguenther@suse.de>
27334 PR middle-end/80171
27335 * gimple-fold.c (fold_ctor_reference): Properly guard against
27336 NULL return value from canonicalize_constructor_val.
27338 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27341 * config/i386/i386.c (ix86_expand_builtin)
27342 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27343 flags reg setting and flags reg using instructions.
27344 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27345 clobbering instructions to zero extend op2.
27347 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27349 * doc/install.texi (Configuration) <--with-aix-soname>:
27350 Update link to AIX ld.
27352 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27354 PR rtl-optimization/80160
27355 PR rtl-optimization/80159
27356 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27357 reg_alternate_class into account.
27359 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27362 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27363 to consider in curr_insn_transform.
27365 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27367 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27368 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27369 and emit_mode_inner.
27371 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27373 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27374 argument to the overloaded builtin variants. Use the new flag to
27375 deprecate certain builtin variants.
27376 * config/s390/s390-builtin-types.def: Add new builtin types.
27377 * config/s390/s390-builtins.h: Support new flags field for
27378 overloaded builtins.
27379 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27380 (s390_macro_to_expand): Enable vector float data type.
27381 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27382 builtins by incrementing the __VEC__ version number.
27383 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27385 (s390_resolve_overloaded_builtin): Emit error messages depending
27386 on the builtin flags.
27387 * config/s390/s390.c (s390_expand_builtin): Support additional
27388 flags argument. Change error message to match the messages
27389 emitted in s390-c.c.
27390 * config/s390/s390.md: New UNSPEC_* constants.
27391 (op_type): Add new instruction types.
27392 * config/s390/vecintrin.h: Add new builtins and test data class
27394 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27395 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27396 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27397 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27398 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27399 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27401 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27402 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27403 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27404 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27406 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27407 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27408 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27409 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27411 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27412 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27413 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27414 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27415 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27416 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27417 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27419 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27420 ("vec_scatter_element<V_HW_4:mode>_DI")
27421 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27422 ("vec_fpint<mode>", "vflls")
27423 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27424 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27425 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27426 ("*vec_cmphe<mode>_cc"): ... these.
27428 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27429 mode constant instead of magic value.
27431 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27433 * config/s390/s390.c (s390_expand_vec_compare): Support other
27434 vector floating point modes than just V2DF.
27435 (s390_expand_vcond): Likewise.
27436 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27437 (s390_cannot_change_mode_class): Prevent mode changes between TF
27438 and V1TF in vector registers.
27439 * config/s390/s390.md (DF, SF): New mode attributes.
27440 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27441 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27442 SFmode support for VRs.
27443 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27445 (VFT, VF_HW): New mode iterators.
27446 (vw, sdx): New mode attributes.
27447 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27448 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27449 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27450 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27451 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27452 also the new vector floating point modes. Renaming to ...
27454 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27455 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27456 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27457 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27458 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27459 ("vec_unordered<mode>"): ... these.
27461 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27462 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27463 ("*vec_extendv2df"): New insn definitions.
27465 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27467 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27468 ("mulditi3_2", "*muldi3_sign"): New patterns.
27469 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27470 rename the pattern definition.
27472 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27474 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27476 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27478 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27480 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27481 instruction if possible.
27482 * config/s390/vector.md (vec_halfnumelts): New mode
27484 ("*vec_vllezlf<mode>"): New pattern.
27486 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27488 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27489 ("popcountv4si2", "popcountv2di2"): Rename to ...
27490 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27491 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27493 ("popcount<mode>2_vxe"): New pattern.
27495 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27497 * common/config/s390/s390-common.c (processor_flags_table): Add
27499 * config.gcc: Add arch12.
27500 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27501 Default to arch12 for unknown CPU model numbers.
27502 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27503 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27504 PROCESSOR_max sanity check.
27505 * config/s390/s390-opts.h (enum processor_type): Add
27507 * config/s390/s390.c (processor_table): Add arch12.
27508 (s390_expand_builtin): Add check for B_VXE flag.
27509 (s390_issue_rate): Add PROCESSOR_ARCH12.
27510 (s390_get_sched_attrmask): Likewise.
27511 (s390_get_unit_mask): Likewise.
27512 (s390_sched_score): Enable z13 scheduling for arch12.
27513 (s390_sched_reorder): Likewise.
27514 (s390_sched_variable_issue): Likewise.
27515 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27517 (s390_tune_attr): Use z13 scheduling also for arch12.
27518 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27519 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27520 (TARGET_VXE_P): New macros.
27521 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27522 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27523 * config/s390/s390.opt: Add arch12 as processor_type.
27525 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27527 * config/s390/s390.md
27528 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27529 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27530 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27532 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27533 Rename expanders to ...
27535 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27536 ("fixuns_truncdddi2_emu"): ... these.
27538 ("fixuns_trunc<mode>si2_emu"): New expander.
27540 ("*fixuns_truncdfdi2_z13"): Rename to ...
27541 ("*fixuns_truncdfdi2_vx"): ... this.
27543 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27545 * config/s390/2964.md: Remove the single element vector compare
27546 instructions which are no longer used.
27547 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27549 (s390_canonicalize_comparison): Remove handling of DFmode
27551 (s390_expand_vec_compare_scalar): Remove function.
27552 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27553 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27555 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27557 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27559 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27561 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27562 will anyway by matched by mov<mode>_64dfp.
27564 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27566 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27567 vlef/vstef. Add missing operand to vleif.
27569 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27571 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27572 pair for all vector types with 64 bit elements.
27573 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27574 * config/s390/vector.md (V_HW_64): ... here.
27575 (V_128_NOSINGLE): New mode iterator.
27576 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27577 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27578 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27579 ("*vec_load_pairv2di"): Change to ...
27580 ("*vec_load_pair<mode>"): ... this one.
27582 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27584 * config/s390/constraints.md: Add comments.
27585 (jKK): Reject element sizes > 8 bytes.
27586 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27588 * config/s390/s390.md: Add the s_operand checks formerly in
27589 s390_split_ok_p to various splitters where they are still
27591 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27592 for 128 bit vectors. Plus two splitters.
27594 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27596 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27599 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27602 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27603 error if the boundary argument is not constant.
27605 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27607 PR rtl-optimization/80112
27608 * loop-doloop.c (doloop_condition_get): Don't check condition
27609 if cmp isn't SET with IF_THEN_ELSE src.
27611 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27613 PR tree-optimization/80158
27614 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27615 replacing a candidate statement, also replace it for the
27616 candidate's alternate interpretation.
27617 (replace_rhs_if_not_dup): Likewise.
27618 (replace_one_candidate): Likewise.
27620 2017-03-24 Richard Biener <rguenther@suse.de>
27622 PR tree-optimization/80167
27623 * graphite-isl-ast-to-gimple.c
27624 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27626 (translate_isl_ast_to_gimple::get_rename): Likewise.
27628 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27630 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27631 handling of certain combinations of target options, including the
27632 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27633 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27635 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27638 * config/arm/arm.md (*load_multiple): Add reload_completed to
27639 matching condition.
27641 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27642 Richard Biener <rguenth@suse.de>
27644 PR tree-optimization/79908
27645 PR tree-optimization/80136
27646 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27647 been cast away, gimplify_and_add suffices.
27649 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27651 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27653 2017-03-23 Richard Biener <rguenther@suse.de>
27655 PR tree-optimization/80032
27656 * gimplify.c (gimple_push_cleanup): Forced unconditional
27657 cleanups still have to go to the conditional_cleanups
27660 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27662 PR tree-optimization/80072
27663 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27665 (next_operand_entry_id): Change type to unsigned int.
27666 (sort_by_operand_rank): Make sure to return the right return value
27667 even if unsigned fields are bigger than INT_MAX.
27668 (struct oecount): Change cnt and id type to unsigned int.
27669 (oecount_hasher::equal): Formatting fix.
27670 (oecount_cmp): Make sure to return the right return value
27671 even if unsigned fields are bigger than INT_MAX.
27672 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27675 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27676 TREE_READONLY on result if writing it more than once.
27679 * doc/invoke.texi (-fsanitize=thread): Document that with
27680 -fnon-call-exceptions atomics are not able to throw
27684 * tsan.c: Include tree-eh.h.
27685 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27686 maybe_clean_or_replace_eh_stmt where needed.
27687 (instrument_memory_accesses): Add cfg_changed argument.
27688 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27689 if it returned true.
27690 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27692 PR rtl-optimization/63191
27693 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27694 wrapper function, moved the whole old content into ...
27695 (ix86_delegitimize_address_1): ... this. New inline function.
27696 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27697 true as last argument instead of ix86_delegitimize_address.
27699 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27701 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27702 cortexa57_branch_cost.
27704 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27706 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27708 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27711 * doc/md.texi (Constraints): Document wA constraint.
27712 * config/rs6000/constraints.md (wA): New.
27713 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27714 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27715 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27716 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27718 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27721 * gimplify.c (is_oacc_declared): New function.
27722 (oacc_default_clause): Use it to set default flags for acc declared
27723 variables inside parallel regions.
27724 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27725 declared variables.
27726 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27727 declare attribute to any decl as necessary.
27729 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27732 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27733 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27734 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27735 (arm_arch_lpae): This.
27736 * config/arm/arm.c (arm_arch7ve): Rename into ...
27737 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27738 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27741 2017-03-22 Martin Liska <mliska@suse.cz>
27744 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27745 error message instead of an ICE.
27747 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27749 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27751 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27753 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27755 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27758 2017-03-21 Martin Sebor <msebor@redhat.com>
27760 * doc/extend.texi: Use "cannot" instead of "can't."
27761 * doc/hostconfig.texi: Same.
27762 * doc/install.texi: Same.
27763 * doc/invoke.texi: Same.
27764 * doc/loop.texi: Same.
27765 * doc/md.texi: Same.
27766 * doc/objc.texi: Same.
27767 * doc/rtl.texi: Same.
27768 * doc/tm.texi: Same.
27769 * doc/tm.texi.in: Same.
27770 * doc/trouble.texi: Same.
27772 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27775 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27776 (collect_checksum_attributes): Set it.
27777 (die_checksum_ordered): Use it.
27779 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27781 PR tree-optimization/79908
27782 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27783 change: For a VA_ARG whose LHS has been cast away, use
27784 force_gimple_operand to construct the side effects.
27786 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27788 PR translation/80001
27789 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27790 more amenable to translation.
27791 (oacc_loop_auto_partitions): Likewise.
27793 2017-03-21 Marek Polacek <polacek@redhat.com>
27794 Martin Sebor <msebor@redhat.com>
27796 PR tree-optimization/80109
27797 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27798 on INTEGRAL_TYPE_P.
27800 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27801 Segher Boessenkool <segher@kernel.crashing.org>
27804 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27805 check reg_used_between_p between insn and one of succ or succ2
27806 depending on if succ is artificial insn not inserted into insn
27809 2017-03-21 Martin Liska <mliska@suse.cz>
27811 PR gcov-profile/80081
27812 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27813 * doc/gcc.texi: Include gcov-dump stuff.
27814 * doc/gcov-dump.texi: New file.
27816 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27818 PR rtl-optimization/79150
27819 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27820 conditional jump, if the jump is the last insn of the loop.
27822 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27823 Richard Biener <rguenth@suse.de>
27825 PR tree-optimization/79908
27826 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27827 been cast away, use force_gimple_operand to construct the side
27830 2017-03-21 Martin Liska <mliska@suse.cz>
27832 PR libfortran/79956
27833 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27836 2017-03-21 Brad Spengler <spender@grsecurity.net>
27839 * plugin.c (htab_hash_plugin): New function.
27840 (add_new_plugin): Use it and adjust.
27841 (parse_plugin_arg_opt): Adjust.
27842 (init_one_plugin): Likewise.
27844 2017-03-21 Richard Biener <rguenther@suse.de>
27846 PR tree-optimization/80032
27847 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27848 if set force the cleanup to happen unconditionally.
27849 (gimplify_target_expr): Push inserted clobbers with force_uncond
27850 to avoid them being removed by control-dependent DCE.
27852 2017-03-21 Richard Biener <rguenther@suse.de>
27854 PR tree-optimization/80122
27855 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27856 va_arg_pack_len when the inlined call stmt requires pack
27858 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27860 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27863 * tsan.c (instrument_builtin_call): If the memory model argument
27864 is not a constant, assume it is valid.
27867 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27870 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27872 PR rtl-optimization/79910
27873 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27874 if its dest is used by an insn before I2 (other than the combined
27875 insns themselves, which are properly handled already).
27877 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27880 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27882 * combine.c (record_used_regs): New static function.
27883 (try_combine): Handle situations where there is an additional
27884 instruction between I2 and I3 which needs to have a LOG_LINK
27888 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27890 * combine.c (try_combine): Delete redundant i1 test. Call
27891 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27893 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27896 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27897 alternatives 13/14.
27899 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27901 PR tree-optimization/80054
27902 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27903 the optimization if a PHI or any of its arguments is not dominated
27904 by the candidate's basis. Use gphi* rather than gimple* as
27906 (replace_profitable_candidates): Clean up a gimple* variable that
27907 should be a gphi* variable.
27909 2017-03-20 Martin Sebor <msebor@redhat.com>
27912 * doc/extend.texi (attribute constructor): Document present limitation.
27914 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27917 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27918 __POWER9_VECTOR__ #ifdef control, change template definition to
27919 use Power9-specific built-in function.
27920 (vec_any_eq): Likewise.
27921 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27922 to control outcomes from this test.
27923 (vector_ae_<mode>p): For VEC_F modes, likewise.
27925 2017-03-20 Ian Lance Taylor <iant@google.com>
27927 * config/i386/i386.c (ix86_function_regparm): Save an extra
27928 register for -fsplit-stack with DECL_STATIC_CHAIN.
27930 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27933 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27934 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27936 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27938 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27940 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27943 2017-03-20 Marek Polacek <polacek@redhat.com>
27946 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27948 2017-03-20 Richard Biener <rguenther@suse.de>
27950 PR tree-optimization/80113
27951 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27952 allocate extra SSA name for PHI def.
27953 (add_close_phis_to_outer_loops): Likewise.
27954 (add_close_phis_to_merge_points): Likewise.
27955 (copy_loop_close_phi_args): Likewise.
27956 (copy_cond_phi_nodes): Likewise.
27958 2017-03-20 Martin Liska <mliska@suse.cz>
27960 PR middle-end/79753
27961 * tree-chkp.c (chkp_build_returned_bound): Do not build
27962 returned bounds for a LHS that's not a BOUNDED_P type.
27964 2017-03-20 Martin Liska <mliska@suse.cz>
27968 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27969 COMPLEX_CST and VECTOR_CST.
27971 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27974 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27975 target operand. A new splitter adds the clobber statement in case
27976 the target operand is dead anyway.
27978 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27980 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27981 to age-old versions of binutils and glibc.
27983 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27985 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27987 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27989 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27991 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27993 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27994 requirement for binutils 2.13.
27996 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27998 * combine.c (try_combine): Delete redundant i1 test. Call
27999 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
28001 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
28003 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
28004 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
28006 <riscv64-*-elf>: Re-arrange section
28007 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
28008 <riscv32-*-linux>: Likewise.
28009 <riscv64-*-elf>: Likewise
28010 <riscv64-*-linux>: Likewise.
28012 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
28015 * aarch64.opt(verbose-cost-dump): Fix typo.
28017 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
28020 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
28021 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
28023 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
28025 * reload.c (find_reloads): When reloading a nonoffsettable address,
28026 use RELOAD_OTHER for it and its address reloads.
28028 PR rtl-optimization/79910
28029 * combine.c (record_used_regs): New static function.
28030 (try_combine): Handle situations where there is an additional
28031 instruction between I2 and I3 which needs to have a LOG_LINK
28034 2017-03-17 Jeff Law <law@redhat.com>
28036 PR tree-optimization/71437
28037 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28038 conditional in the hash table first.
28039 (vrp_dom_walker::before_dom_children): Extract condition from
28040 ASSERT_EXPR. Record condition, its inverion and any implied
28041 conditions as well.
28043 2017-03-17 Marek Polacek <polacek@redhat.com>
28044 Markus Trippelsdorf <markus@trippelsdorf.de>
28046 PR tree-optimization/80079
28047 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28050 2017-03-17 Richard Biener <rguenther@suse.de>
28052 PR middle-end/80075
28053 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28054 Properly verify the LHS before the RHS possibly claims to be
28056 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28059 2017-03-17 Martin Jambor <mjambor@suse.cz>
28061 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28062 (List of -O2 options): Likewise.
28063 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28066 2017-03-17 Tom de Vries <tom@codesourcery.com>
28068 * gcov-dump.c (print_usage): Print bug_report_url.
28070 2017-03-17 Richard Biener <rguenther@suse.de>
28072 PR middle-end/80050
28073 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28074 (parser::peek): Likewise.
28076 2017-03-17 Richard Biener <rguenther@suse.de>
28078 PR tree-optimization/80048
28079 * sese.c (free_sese_info): Properly release rename_map and
28080 copied_bb_map elements.
28082 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28084 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28085 Add linked-list forward and backlinks. Insert on
28086 construction, remove on destruction.
28087 (class pass_store_merging): Add m_stores_head field.
28088 (pass_store_merging::terminate_and_process_all_chains):
28089 Iterate over m_stores_head list.
28090 (pass_store_merging::terminate_all_aliasing_chains):
28092 (pass_store_merging::execute): Check for debug stmts first.
28093 Push new chains onto the m_stores_head stack.
28095 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28098 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28099 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28100 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28102 2017-03-16 Jeff Law <law@redhat.com>
28104 PR tree-optimization/71437
28105 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28106 member function. Implementation moved into after_dom_children
28107 member function and into the threader's thread_outgoing_edges
28109 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28110 some code into new thread_outgoing_edges.
28111 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28112 definition. Simplify marker handling (do it here). Assume we always
28113 have the available expression and the const/copies tables.
28114 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28116 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28117 * tree-vrp.c (equiv_stack): No longer file scoped.
28118 (vrp_dom_walker): New class.
28119 (vrp_dom_walker::before_dom_children): New member function.
28120 (vrp_dom_walker::after_dom_children): Likewise.
28121 (identify_jump_threads): Setup domwalker. Use it rather than
28122 walking edges in a random order by hand. Simplify setup/finalization.
28123 (finalize_jump_threads): Remove.
28124 (vrp_finalize): Do not call identify_jump_threads here.
28125 (execute_vrp): Do it here instead and call thread_through_all_blocks
28128 PR tree-optimization/71437
28129 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28131 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28133 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28134 (dom_opt_dom_walker::thread_across_edge): Remove
28135 handle_dominating_asserts argument. All callers changed.
28136 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28137 changes. Remove calls to lhs_of_dominating_assert. Other
28138 uses of handle_dominating_asserts turn into unconditional code
28139 (simplify_control_stmt_condition_1): Likewise.
28140 (simplify_control_stmt_condition): Likewise.
28141 (thread_through_normal_block, thread_across_edge): Likewise.
28142 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28143 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28144 object if it is not an SSA_NAME.
28145 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28146 before calling into the VRP specific simplifiers.
28147 (identify_jump_threads): Remove handle_dominating_asserts
28150 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28153 * tree-diagnostic.c (default_tree_printer): No longer static.
28154 * tree-diagnostic.h (default_tree_printer): New prototype.
28156 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28158 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28159 Change ins into fmov.
28161 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28163 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28164 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28165 Use h_con constraint for operand 1.
28166 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28167 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28169 2017-03-15 Jeff Law <law@redhat.com>
28171 PR tree-optimization/71437
28172 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28173 (record_temporary_equivalences): Use it.
28175 PR tree-optimization/71437
28176 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28177 tree-ssa-scopedtables.
28178 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28179 (record_conditions, record_cond, vuse_eq): Likewise.
28180 (record_edge_info): Adjust to API tweak of record_conditions.
28181 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28182 (record_temporary_equivalences, optimize_stmt): Likewise.
28183 (eliminate_redundant_computations): Likewise.
28184 (record_equivalences_from_stmt): Likewise.
28185 * tree-ssa-scopedtables.c: Include options.h and params.h.
28186 (vuse_eq): New function, moved from tree-ssa-dom.c
28187 (build_and_record_new_cond): Likewise.
28188 (record_conditions): Likewise. Accept vector of conditions rather
28189 than edge_equivalence structure for first argument.
28190 for the first argument.
28191 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28192 from tree-ssa-dom.c.
28193 (avail_exprs_stack::record_cond): Likewise.
28194 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28195 from tree-ssa-dom.c.
28196 (avail_exprs_stack): Add new member functions lookup_avail_expr
28198 (record_conditions): Declare.
28200 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28203 * lra-constraints.c (process_alt_operands): Increase reject for
28204 reloading an input/output operand.
28206 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28209 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28210 insns to convert from signed/unsigned char/short to IEEE 128-bit
28212 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28214 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28217 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28218 subreg of inner mode for values already in registers.
28220 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28222 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28223 iteration reg is used after the loop.
28225 2017-03-14 Martin Sebor <msebor@redhat.com>
28227 PR tree-optimization/79800
28228 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28229 precision in negative-positive range.
28230 (format_floating): Call non-const overload with adjusted precision.
28232 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28235 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28238 2017-03-14 Martin Sebor <msebor@redhat.com>
28240 PR middle-end/80020
28241 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28242 * builtins.def (aligned_alloc): Use it.
28245 * Makefile.in (GTFILES): Add calls.c.
28246 * calls.c: Include "gt-calls.h".
28248 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28250 PR rtl-optimization/79728
28251 * regs.h (struct target_regs): New field
28252 x_contains_allocatable_regs_of_mode.
28253 (contains_allocatable_regs_of_mode): New macro.
28254 * reginfo.c (init_reg_sets_1): Initialize it, and change
28255 contains_reg_of_mode so it includes global regs as well.
28256 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28257 rather than contains_regs_of_mode.
28259 2017-03-14 Martin Liska <mliska@suse.cz>
28261 * doc/invoke.texi: Document options that can't be combined with
28262 -fcheck-pointer-bounds.
28264 2017-03-14 Martin Liska <mliska@suse.cz>
28266 PR middle-end/79831
28267 * doc/invoke.texi (-Wchkp): Document the option.
28269 2017-03-14 Martin Liska <mliska@suse.cz>
28271 * Makefile.in: Install gcov-dump.
28273 2017-03-14 Martin Liska <mliska@suse.cz>
28275 * multiple_target.c (expand_target_clones): Bail out for
28276 an invalid attribute.
28278 2017-03-14 Richard Biener <rguenther@suse.de>
28280 * alias.c (struct alias_set_entry): Pack properly.
28281 * cfgloop.h (struct loop): Likewise.
28282 * cse.c (struct set): Likewise.
28283 * ipa-utils.c (struct searchc_env): Likewise.
28284 * loop-invariant.c (struct invariant): Likewise.
28285 * lra-remat.c (struct cand): Likewise.
28286 * recog.c (struct change_t): Likewise.
28287 * rtl.h (struct address_info): Likewise.
28288 * symbol-summary.h (function_summary): Likewise.
28289 * tree-loop-distribution.c (struct partition): Likewise.
28290 * tree-object-size.c (struct object_size_info): Likewise.
28291 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28292 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28293 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28294 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28295 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28296 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28297 (struct _stmt_vec_info): Likewise.
28299 2017-03-14 Martin Liska <mliska@suse.cz>
28302 * multiple_target.c (create_dispatcher_calls): Check that
28303 a target can create a function dispatcher.
28305 2017-03-14 Martin Liska <mliska@suse.cz>
28308 * multiple_target.c (expand_target_clones): Drop local.local
28309 flag for default implementation.
28311 2017-03-14 Richard Biener <rguenther@suse.de>
28313 PR tree-optimization/80030
28314 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28316 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28318 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28319 gcc_fallthrough() instead of __attribute__((fallthrough));
28321 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28323 * doc/gcc.texi: Remove "up" link to (DIR).
28324 * doc/gccint.texi: Ditto.
28326 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28328 * doc/install.texi (Specific) <avr>: Remove reference to
28331 2017-03-13 Jeff Law <law@redhat.com>
28333 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28334 attribute rather than comments.
28336 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28337 match_scratch operand is highest.
28339 2017-03-13 Martin Liska <mliska@suse.cz>
28341 PR middle-end/78339
28342 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28343 is a CHKP clone, use original declaration.
28345 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28347 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28348 (arc_conditional_register_usage): Use a different allocation order
28349 when optimizing for size.
28350 * common/config/arc/arc-common.c (arc_option_optimization_table):
28351 Section anchors default on when optimizing for size.
28353 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28355 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28357 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28359 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28360 * config/arc/arc.md (cpu_facility): Add cd variant.
28361 (*movqi_insn): Add code density variant.
28362 (*movhi_insn): Likewise.
28363 (*movqi_insn): Likewise.
28364 (*addsi3_mixed): Likewise.
28365 (subsi3_insn): Likewise.
28367 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28369 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28371 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28373 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28374 expressions with MINUS and UNARY ops.
28376 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28379 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28381 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28382 between vec_select and vector argument.
28383 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28384 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28385 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28386 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28387 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28388 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28390 2017-03-13 Richard Biener <rguenther@suse.de>
28393 * params.def (vect-max-peeling-for-alignment): Fix typo.
28395 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28397 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28398 issue that only occurred with binutils below 2.18.
28400 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28402 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28403 refer to binutils 2.11/2.12 minimum.
28405 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28407 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28408 ftp.kernel.org and simplify binutils requirement.
28410 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28412 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28414 (Optimize Options): Ditto. Also remove redundancy.
28416 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28418 PR translation/79848
28419 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28421 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28422 to G_ to avoid double translation.
28424 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28426 PR translation/79923
28427 * auto-profile.c (get_combined_location): Convert leading
28428 character of diagnostics to lower case and remove trailing period.
28429 (read_profile): Likewise for various diagnostics.
28430 * config/arm/arm.c (arm_option_override): Remove trailing period
28431 from various diagnostics.
28432 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28433 (msp430_expand_delay_cycles): Likewise.
28435 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28438 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28439 full command-line argument, rather than just "str".
28440 (aarch64_validate_march): Likewise.
28441 (aarch64_validate_mtune): Likewise.
28443 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28445 PR rtl-optimization/78911
28446 * lra-assigns.c (must_not_spill_p): New function.
28447 (spill_for): Use it.
28449 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28451 PR tree-optimization/79981
28452 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28453 ATOMIC_COMPARE_EXCHANGE ifn result.
28454 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28455 IFN_ATOMIC_COMPARE_EXCHANGE.
28457 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28460 * opts.c (parse_sanitizer_options): Add missing question mark to
28461 "did you mean" message.
28463 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28465 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28467 (VMULEUH_UNS): Likewise.
28468 (VMULOUB_UNS): Likewise.
28469 (VMULOUH_UNS): Likewise.
28470 * config/rs6000/rs6000.c (builtin_function_type): Remove
28471 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28473 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28476 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28477 x with result of extra_parsing_for_operand_code_0.
28478 (function_reader::extra_parsing_for_operand_code_0): Convert
28479 return type from void to rtx, returning x. When reading
28480 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28481 larger size containing struct block_symbol.
28483 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28485 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28486 -mfloat128-hardware without -m64.
28488 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28491 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28492 entries to the case statement that marks unsigned arguments to
28493 overloaded functions.
28495 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28497 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28498 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28500 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28503 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28504 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28506 2017-03-10 Martin Liska <mliska@suse.cz>
28510 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28511 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28514 2017-03-10 Olivier Hainque <hainque@adacore.com>
28516 * tree-switch-conversion (array_value_type): Start by resetting
28517 candidate type to it's main variant.
28519 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28521 PR rtl-optimization/79909
28522 * combine.c (try_combine): Use simplify_replace_rtx on individual
28523 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28524 of the whole CALL_INSN_FUNCTION_USAGE.
28526 PR tree-optimization/79972
28527 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28528 get_range_info on SSA_NAMEs. Formatting fixes.
28530 2017-03-10 Richard Biener <rguenther@suse.de>
28531 Jakub Jelinek <jakub@redhat.com>
28533 PR tree-optimization/77975
28534 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28535 edge to be constant.
28536 (get_val_for): For constant x return it. Formatting fix.
28537 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28538 would use the same bases as the current one.
28540 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28542 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28543 instead of vec_select for V1TImode.
28544 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28546 (VSX_LE_128): Add V1TI to this mode iterator.
28547 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28548 (*vsx_le_perm_store_<mode>): Likewise.
28549 (pre-reload splitter for VSX stores): Likewise.
28550 (post-reload splitter for VSX stores): Likewise.
28551 (*vsx_xxpermdi2_le_<mode>): Likewise.
28552 (*vsx_lxvd2x2_le_<mode>): Likewise.
28553 (*vsx_stxvd2x2_le_<mode>): Likewise.
28555 2017-03-09 Michael Eager <eager@eagercon.com>
28557 Correct failures with --enable-checking=yes,rtl.
28559 * config/microblaze/microblaze.c (microblaze_expand_shift):
28560 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28561 test for const0_rtx.
28562 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28563 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28565 2017-03-09 Richard Biener <rguenther@suse.de>
28567 PR tree-optimization/79977
28568 * graphite-scop-detection.c (scop_detection::merge_sese):
28569 Handle the case of extra exits to blocks dominating the entry.
28571 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28573 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28576 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28578 PR rtl-optimization/79949
28579 * lra-constraints.c (process_alt_operands): Check memory when
28580 trying to predict a cycle. Print about the overall increase.
28582 2017-03-09 Richard Biener <rguenther@suse.de>
28584 PR middle-end/79971
28585 * gimple-expr.c (useless_type_conversion_p): Preserve
28586 TYPE_SATURATING for fixed-point types.
28588 2017-03-09 Richard Biener <rguenther@suse.de>
28591 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28592 alignment of BLKmode params.
28594 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28597 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28598 (VALL_NO_V2Q): Likewise.
28600 * config/aarch64/aarch64-simd.md
28601 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28603 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28604 VALL_NO_V2Q mode iterator.
28605 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28607 2017-03-09 Martin Liska <mliska@suse.cz>
28609 PR tree-optimization/79631
28610 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28611 tree_int_cst_sign_bit just for INTEGER constants.
28613 2017-03-09 Martin Liska <mliska@suse.cz>
28617 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28620 2017-03-09 Marek Polacek <polacek@redhat.com>
28623 * tree.c (inchash::add_expr): Handle TREE_VEC.
28625 2017-03-09 Martin Liska <mliska@suse.cz>
28628 (chkp_narrow_size_and_offset): New function.
28629 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28630 (void chkp_parse_bit_field_ref): New function.
28631 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28632 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28634 2017-03-09 Martin Liska <mliska@suse.cz>
28637 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28638 (chkp_find_bounds_1): Remove gcc_unreachable.
28640 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28643 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28644 BUILT_IN_SYNC*, determine the access type from the size suffix and
28645 always build a MEM_REF with that type. Handle forgotten
28646 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28649 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28650 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28651 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28652 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28653 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28654 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28655 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28656 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28657 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28658 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28659 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28660 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28661 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28662 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28663 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28664 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28665 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28666 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28667 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28668 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28669 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28670 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28671 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28672 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28673 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28674 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28675 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28676 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28677 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28678 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28679 definitions outside of __OPTIMIZE__ guarded section.
28682 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28683 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28684 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28685 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28688 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28690 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28691 ("vfenez<mode>"): Add missing constraints.
28693 2017-03-08 Martin Sebor <msebor@redhat.com>
28696 * config/nds32/nds32.c (nds32_option_override):
28697 Fix misspelled diagnostic.
28699 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28702 * gimplify.c (gimplify_omp_for): Replace index var in outer
28703 taskloop statement with an artificial variable and add
28704 OMP_CLAUSE_PRIVATE clause for it.
28706 2017-03-08 Richard Biener <rguenther@suse.de>
28708 PR tree-optimization/79955
28709 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28710 for accesses that are completely outside of the variable.
28712 2017-03-08 Andrew Haley <aph@redhat.com>
28714 PR tree-optimization/79943
28715 * tree-ssa-loop-split.c (compute_new_first_bound): When
28716 calculating the new upper bound, (END-BEG) should be added, not
28719 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28721 * config/avr/avr.md (setmemhi): Make sure match_dup
28722 operand number comes before match_scratch.
28724 2017-03-08 Richard Biener <rguenther@suse.de>
28726 PR tree-optimization/79920
28727 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28728 with ncopies == 1 to ...
28729 (vect_transform_slp_perm_load): ... here. Properly compute
28730 all element loads by iterating VF times over the group. Do
28731 not handle ncopies (computed in a broken way) in
28732 vect_create_mask_and_perm.
28734 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28737 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28738 is a uniform vector, use uniform_vector_p return value instead of
28739 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28741 2017-03-07 Marek Polacek <polacek@redhat.com>
28743 PR middle-end/79809
28744 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28745 (alloca_call_type): Likewise.
28747 2017-03-07 Martin Liska <mliska@suse.cz>
28749 * gcov.c (process_args): Put comment to correct location.
28751 2017-03-07 Martin Liska <mliska@suse.cz>
28753 PR middle-end/68270
28754 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28755 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28756 (chkp_narrow_bounds_for_field): Likewise.
28757 (chkp_parse_array_and_component_ref): Pass one more argument to
28760 2017-03-07 Richard Biener <rguenther@suse.de>
28762 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28765 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28767 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28768 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28770 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28773 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28774 to end of description.
28775 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28777 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28779 PR rtl-optimization/79901
28780 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28782 (*avx512f_<code><mode>3<mask_name>): ... this.
28783 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28784 iterator instead of VI8_AVX2_AVX512BW.
28786 PR rtl-optimization/79901
28787 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28788 min/max expander, expand it using expand_vec_cond_expr.
28791 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28794 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28797 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28798 to void * for PCH reasons.
28799 * dwarf2out.c (output_loc_operands, output_die): Cast
28800 v.val_vec.array to unsigned char *.
28802 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28805 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28808 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28810 PR rtl-optimization/79571
28811 * lra-constraints.c (process_alt_operands): Calculate static
28812 reject and subtract it from overall when only addresses will be
28815 2017-03-06 Julia Koval <julia.koval@intel.com>
28818 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28819 incoming stack boundary to 128 for 64-bit targets.
28821 2017-03-06 Richard Biener <rguenther@suse.de>
28823 PR tree-optimization/79894
28824 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28825 to NULL after folding it.
28827 2017-03-06 Richard Biener <rguenther@suse.de>
28829 PR tree-optimization/79824
28830 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28831 check disabling peeling for gaps.
28833 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28835 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28836 attributes): Document gettimeofday.
28838 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28840 * config/s390/s390.c (s390_option_override_internal): Set
28841 PARAM_MIN_VECT_LOOP_BOUND
28843 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28845 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28846 * config/s390/s390.md: Likewise.
28848 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28851 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28852 (<avx2_avx512>_perm<mode>): Rename to ...
28853 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28855 (<avx512>_perm<mode>_mask): Rename to ...
28856 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28858 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28859 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28860 instead of VI8F_256_512.
28861 (avx512f_perm<mode>): New define_expand.
28862 (avx512f_perm<mode>_mask): Likewise.
28863 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28864 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28866 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28868 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28869 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28871 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28873 2017-03-06 Martin Liska <mliska@suse.cz>
28876 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28877 when having a SSA NAME w/o VAR_DECL assigned to it.
28879 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28881 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28882 msa_dpsub_<su>_d): Fix MODE for vec_select.
28884 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28886 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28888 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28890 2017-03-06 Richard Biener <rguenther@suse.de>
28892 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28893 * plugin.c (register_plugin_info): Likewise.
28894 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28896 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28898 * config/i386/sse.md (sse_storehps, sse_storelps,
28899 avx_<castmode><avxsizesuffix>_<castmode>,
28900 avx512f_<castmode><avxsizesuffix>_<castmode>,
28901 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28902 in condition that at least one operand is not a MEM.
28904 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28906 PR middle-end/79805
28907 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28908 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28910 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28911 gimple_call_nothrow_p flag based on whether original builtin can throw.
28912 If it can, emit following stmts on the fallthrough edge.
28913 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28914 don't create new bb if inserting just debug stmts on the edge, try to
28915 insert them on the fallthru bb or just reset debug stmts.
28917 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28920 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28921 restore recog_data (including the operand rtxes inside it) around
28922 the call to get_insn_template.
28924 2017-03-03 Martin Sebor <msebor@redhat.com>
28926 PR tree-optimization/79699
28927 * context.c (context::~context): Free MPFR caches to avoid
28928 a memory leak on program exit.
28930 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28932 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28933 Use wide_int::ulow () instead of .elt (0).
28935 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28937 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28938 (*pushxf): Limit oF constraint to 32bit targets and add oC
28939 constraint for 64bit targets.
28940 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28941 (*pushdf): Change rmF constraint to rmC.
28943 2017-03-03 Martin Liska <mliska@suse.cz>
28945 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28946 Remove unused variable.
28948 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28951 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28952 is a memory operand, increase num_memory.
28953 (ix86_expand_args_builtin): Likewise.
28955 2017-03-03 Jan Hubicka <jh@suse.cz>
28958 * ipa-devirt.c (maybe_record_node): Properly handle
28959 __cxa_pure_virtual visibility.
28961 2017-03-03 Martin Liska <mliska@suse.cz>
28963 PR tree-optimization/79803
28964 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28966 (pass_loop_prefetch::execute): Disabled optimization if an
28967 assumption about L1 cache size is not met.
28969 2017-03-03 Martin Liska <mliska@suse.cz>
28971 PR rtl-optimization/79574
28972 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28973 (hash_scan_set): Likewise.
28974 (dump_hash_table): Likewise.
28975 (hoist_code): Likewise.
28977 2017-03-03 Richard Biener <rguenther@suse.de>
28979 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28980 in place of uhigh ().
28981 (fixed_convert_from_real): Likewise.
28983 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28986 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28988 2017-03-03 Richard Biener <rguenther@suse.de>
28990 PR middle-end/79818
28991 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28992 TYPE_OVERFLOW_UNDEFINED check.
28994 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28996 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28998 (vector_ae_<mode>_p): Likewise.
28999 (vector_nez_<mode>_p): Likewise.
29000 (vector_ne_v2di_p): Likewise.
29001 (vector_ae_v2di_p): Likewise.
29002 (vector_ne_<mode>_p): Likewise.
29003 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
29005 (vsx_tsqrt<mode>2_fe): Likewise.
29007 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
29010 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
29012 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29014 PR rtl-optimization/79780
29015 * cprop.c (one_cprop_pass): When second and further conditional trap
29016 in a single basic block is turned into an unconditional trap, turn it
29017 into a deleted note to avoid RTL verification failures.
29019 2017-03-02 Richard Biener <rguenther@suse.de>
29021 * fold-const.c (const_binop): Use ulow () instead of elt (0).
29023 2017-03-02 Richard Biener <rguenther@suse.de>
29025 PR tree-optimization/79345
29027 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29028 param and abort the walk, returning -1 if it is hit.
29029 (walk_aliased_vdefs): Take a limit param and pass it on.
29030 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29031 defaulting to 0 and return a signed int.
29032 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29033 (check_defs): New helper.
29034 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29035 about uninitialized memory.
29036 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29037 bogus uninitialized warning.
29038 (fixed_convert_from_real): Likewise.
29040 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29042 PR tree-optimization/66768
29043 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29044 iv_use if base object can't be determined.
29046 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29048 PR tree-optimization/79345
29049 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29050 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29051 (get_pattern_stats): Initialize it.
29052 * genemit.c (gen_expand): Verify match_scratch numbers come after
29053 match_operand/match_dup numbers.
29054 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29055 match_scratch numbers.
29056 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29058 * config/s390/s390.md (trunctdsd2): Likewise.
29060 2017-03-02 Richard Biener <rguenther@suse.de>
29062 * wide-int.h (wide_int_storage::operator=): Implement in terms
29065 2017-03-02 Richard Biener <rguenther@suse.de>
29067 PR tree-optimization/79777
29068 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29069 the to insert expression to sth existing.
29071 2017-03-01 Martin Sebor <msebor@redhat.com>
29073 PR middle-end/79692
29074 * gimple-ssa-sprintf.c
29075 (directive::known_width_and_precision): New function.
29076 (format_integer): Use it.
29077 (get_mpfr_format_length): Consider the full range of precision
29078 when computing %g output with the # flag. Set the likely byte
29079 count to 3 rather than 1 when precision is indeterminate.
29080 (format_floating): Correct the lower bound of precision.
29082 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29084 * doc/invoke.texi: Document default code model for 64-bit Linux.
29086 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29089 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29090 udiv rather than div since input pattern is unsigned.
29092 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29094 * config/i386/i386.c (print_reg): Warn for values of
29095 unsupported size in integer register.
29097 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29100 * config/rs6000/predicates.md (current_file_function_operand): Do
29101 not allow self calls to be local if the function is replaceable.
29103 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29106 * config/rs6000/altivec.h (vec_ctz and others): Change the
29107 preprocessor macro that controls conditional compilation from
29108 _ARCH_PWR9 to __POWER9_VECTOR__.
29109 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29110 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29111 control (instead of _ARCH_PWR9 control) so that template
29112 definition uses power9-specific function.
29113 (vec_any_eq): Likewise.
29114 (vec_all_ne): Change macro definition to use a power9-specific
29115 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29116 _ARCH_PWR9 control).
29117 (vec_any_eq) Likewise.
29118 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29119 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29120 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29121 support for xvcmpnedp instruction.
29122 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29123 macro expansion so that Power9 implementation of vec_all_ne does
29124 not use the AltiVec predicate framework.
29125 (VCMPNEH_P): Likewise.
29126 (VCMPNEW_P): Likewise.
29127 (VCMPNED_P): Likewise.
29128 (VCMPNEFP_P): Likewise.
29129 (VCMPNEDP_P): Likewise.
29130 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29131 implementation of vec_any_eq to not use AltiVec predicate
29133 (VCMPAEH_P): Likewise.
29134 (VCMPAEW_P): Likewise.
29135 (VCMPAED_P): Likewise.
29136 (VCMPAEFP_P): Likewise.
29137 (VCMPAEDP_P): Likewise.
29138 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29139 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29140 not use the AltiVec predicate framework.
29141 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29142 of vec_any_eq to not use AltiVec predicate framework.
29143 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29144 support for predefined __POWER9_VECTOR__ macro to indicate that
29145 Power9 instruction selection is enabled.
29146 (altivec_overloaded_builtins): Remove extraneous
29147 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29148 function argument types RS6000_BTI_bool_V16QI and
29149 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29150 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29151 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29152 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29153 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29154 Power9 for implementations of vec_cmpne. Change the signature for
29155 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29156 (representing vec_all_ne) to remove the previously described first
29157 argument of type RS6000_BTI_INTSI, as this was an artifact of
29158 reliance on the AltiVec predicate framework, which is no longer
29159 used in the implementation of these functions. Add
29160 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29161 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29162 since, unlike the AltiVec predicate framework implementation, we
29163 do not share function descriptors between vec_alle and vec_anyeq.
29164 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29165 set of modes that receive special treatment even when
29166 TARGET_P9_VECTOR is true. The special treatment emits code that
29167 does not depend on Power9 instructions.
29168 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29169 define_expand to not rely on AltiVec predicate framework.
29170 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29172 (vector_ne_v2di_p): Change this define_expand to not rely on
29173 AltiVec predicate framework.
29174 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29176 (vector_ne_<mode>_p): Change this define_expand to not rely on
29177 AltiVec predicate framework.
29178 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29180 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29181 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29182 define_insn pattern.
29183 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29184 define_insn pattern because the xvcmpne<VSs>. instruction is not
29186 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29187 instruction is not supported.
29189 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29191 * config/nvptx/nvptx.c: Include intl.h.
29193 2017-03-01 Martin Jambor <mjambor@suse.cz>
29196 * ipa-prop.h (ipa_bits): Removed field known.
29197 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29198 to pointers. Adjusted their comments to warn about their sharing.
29199 (ipcp_transformation_summary): Change bits to a vector of pointers.
29200 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29201 (ipa_get_ipa_bits_for_value): Declare.
29202 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29203 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29204 (ipa_bits_hash_table): Likewise.
29205 (ipa_vr_ggc_hash_traits): Likewise.
29206 (ipa_vr_hash_table): Likewise.
29207 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29208 being pointers and vr_known being removed.
29209 (ipa_set_jf_unknown): Likewise.
29210 (ipa_get_ipa_bits_for_value): New function.
29211 (ipa_set_jfunc_bits): Likewise.
29212 (ipa_get_value_range): New overloaded functions.
29213 (ipa_set_jfunc_vr): Likewise.
29214 (ipa_compute_jump_functions_for_edge): Use the above functions to
29215 construct bits and vr parts of jump functions.
29216 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29217 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29219 (ipcp_grow_transformations_if_necessary): Also allocate
29220 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29222 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29223 them. Fix too long lines.
29224 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29225 vr_known being removed.
29226 (ipa_read_jump_function): Use new setter functions to construct bits
29227 and vr parts of jump functions or set them to NULL.
29228 (write_ipcp_transformation_info): Adjust for bits being pointers.
29229 (read_ipcp_transformation_info): Likewise.
29230 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29232 Include gt-ipa-prop.h.
29233 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29235 (ipcp_store_bits_results): Likewise.
29236 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29237 Do not write to existing jump functions but use a temporary instead.
29239 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29242 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29243 attempt to use its first operand as BIT_FIELD_REF base.
29245 2017-03-01 Richard Biener <rguenther@suse.de>
29247 PR middle-end/79721
29248 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29249 interpolating formula in wrapping arithmetic.
29250 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29252 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29254 PR tree-optimization/79734
29255 * tree-vect-generic.c (expand_vector_condition): Optimize
29256 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29257 Handle VEC_COND_EXPR where comparison has different inner width from
29258 type's inner width.
29260 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29262 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29263 markup, and similar issues. Remove @opindex entries for things
29264 that aren't options. Add missing -mmpy-option entries.
29266 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29268 PR tree-optimization/79737
29269 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29270 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29271 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29272 instead of byte_size. Formatting fix.
29273 (shift_bytes_in_array_right): Formatting fix.
29275 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29278 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29279 condition on optimize for the leaf function test.
29281 2017-02-28 Martin Liska <mliska@suse.cz>
29284 * read-rtl-function.c (function_reader::handle_unknown_directive):
29285 Bail out when one uses -flto.
29287 2017-02-28 Martin Liska <mliska@suse.cz>
29289 * common.opt: Replace space with tabular for options of <number>
29291 * config/i386/i386.opt: Show <number> value for
29292 -mlarge-data-threshold.
29293 * opts.c (print_filtered_help): Do not display number in hexadecimal
29296 2017-02-28 Martin Liska <mliska@suse.cz>
29298 * common.opt: Fix --help=option -Q for options which are of
29301 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29303 * config/i386/i386.c (print_reg): Error out for values
29304 of 8-bit size in invalid integer register.
29306 2017-02-28 Martin Sebor <msebor@redhat.com>
29308 PR tree-optimization/79691
29309 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29311 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29314 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29315 gcc_unreachable with output_operand_lossage.
29317 2017-02-28 Richard Biener <rguenther@suse.de>
29319 PR tree-optimization/79740
29320 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29322 (visit_nary_op): Insert the nary into the hashtable if we
29323 pattern-matched sth.
29324 * tree-ssa-pre.c (eliminate_insert): Robustify.
29326 2017-02-28 Richard Biener <rguenther@suse.de>
29328 PR middle-end/79731
29329 * fold-const.c (decode_field_reference): Reject out-of-bound
29332 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29334 * config/i386/i386.c: Include intl.h.
29335 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29336 instead of just cond ? "..." : "...".
29337 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29338 * coverage.c (read_counts_file): Likewise.
29339 * omp-offload.c: Include intl.h.
29340 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29341 of just cond ? "..." : "...".
29342 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29343 of just cond ? "..." : "...".
29345 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29348 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29350 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29351 'tune for' CPU name.
29352 * config/arm/arm-cpu-data.h: Regenerated.
29354 2017-02-28 Richard Biener <rguenther@suse.de>
29356 PR tree-optimization/79732
29357 * tree-inline.c (expand_call_inline): Do not shadow var.
29359 2017-02-28 Richard Biener <rguenther@suse.de>
29361 PR tree-optimization/79723
29362 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29363 address-space properly.
29365 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29367 * doc/optinfo.texi (Optimization groups): Fix option used for
29369 * doc/invoke.texi (-fopt-info): Document "omp".
29370 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29371 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29372 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29373 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29374 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29376 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29378 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29379 OPTGROUP_OMP with "omp".
29381 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29384 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29385 for arithmetic shift of unsigned V2DI.
29387 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29389 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29390 arc/linux.h headers.
29391 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29392 (LINK_SPEC): Likewise.
29393 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29394 (EXTRA_SPECS): Likewise.
29395 (STARTFILE_SPEC): Likewise.
29396 (ENDFILE_SPEC): Likewise.
29397 (LIB_SPEC): Likewise.
29398 (TARGET_SDATA_DEFAULT): Likewise.
29399 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29400 (MULTILIB_DEFAULTS): Likewise.
29401 (DWARF2_UNWIND_INFO): Likewise.
29402 * config/arc/big.h: New file.
29403 * config/arc/elf.h: Likewise.
29404 * config/arc/linux.h: Likewise.
29405 * config/arc/t-uClibc: Remove.
29407 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29409 PR tree-optimization/77536
29410 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29411 (tree_transform_and_unroll_loop): Use above function to compute the
29412 estimated niter of unrolled loop and use it when scaling profile.
29413 Also use count info rather than frequency if it's non-zero.
29414 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29415 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29416 (vect_transform_loop): Call above function.
29418 2017-02-27 Richard Biener <rguenther@suse.de>
29420 PR tree-optimization/45397
29421 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29422 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29423 (visit_nary_op): Add pattern matching for CSEing sign-changed
29424 or truncated operations with wider ones.
29426 2017-02-27 Richard Biener <rguenther@suse.de>
29428 PR tree-optimization/79690
29429 * tree-vect-stmts.c (vectorizable_store): Use vector type
29430 built from the DR with address-space.
29432 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29434 * doc/invoke.texi (Optimize Options): Refine the description
29435 of asan-use-after-return.
29437 2017-02-25 Alan Modra <amodra@gmail.com>
29439 PR rtl-optimization/79584
29440 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29441 base, not ad->base_term, the reg within base. Remove assertion
29442 that ad->base == ad->base_term. Replace gen_int_mode using
29443 bogus mode with const0_rtx.
29445 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29447 PR middle-end/79396
29448 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29449 FMA_EXPR like tcc_binary or tcc_unary.
29451 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29454 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29456 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29457 (output_loc_operands): Handle DW_OP_call_ref and
29458 DW_OP_GNU_variable_value.
29459 (struct variable_value_struct): New type.
29460 (struct variable_value_hasher): Likewise.
29461 (variable_value_hash): New variable.
29462 (string_types): Remove.
29463 (copy_loc_descr): New function.
29464 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29465 (prepend_loc_descr_to_each): New function.
29466 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29467 instead of add_loc_descr_to_each if the first argument is single
29468 location list and the second has multiple.
29469 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29470 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29471 when looking for variable value which doesn't have other location info.
29472 (loc_list_from_tree): Formatting fix.
29473 (gen_array_type_die): Simplify DW_AT_string_length handling.
29474 (adjust_string_types): Remove.
29475 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29476 string_types. Call resolve_variable_values.
29477 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29478 (resolve_addr_in_expr): Likewise. Add A argument.
29479 (copy_deref_exprloc): Remove deref argument. Adjust for the
29480 original expression being DW_OP_GNU_variable_value with optionally
29481 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29482 optionally after it.
29483 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29484 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29485 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29486 (variable_value_hasher::hash, variable_value_hasher::equal): New
29488 (resolve_variable_value_in_expr, resolve_variable_value,
29489 resolve_variable_values, note_variable_value_in_expr,
29490 note_variable_value): New functions.
29491 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29494 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29497 * opts.h (handle_generated_option): Add GENERATED_P argument.
29498 * opts-common.c (handle_option): Adjust function comment.
29499 (handle_generated_option): Add GENERATED_P argument, pass it to
29501 (control_warning_option): Pass false to handle_generated_option
29503 * opts.c (maybe_default_option): Pass true to handle_generated_option
29505 * optc-gen.awk: Likewise.
29507 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29509 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29510 a REG, look at the REG it is a SUBREG of.
29511 (splitter for cmpeqsi_t): Ditto.
29513 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29515 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29516 the special USEs with the pattern of the insn, not the insn itself.
29518 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29521 * doc/invoke.texi: Document -mload-store-pairs.
29523 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29524 Sandra Loosemore <sandra@codesourcery.com>
29526 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29527 argument isn't a CONST_INT.
29528 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29529 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29530 (nios2_validate_compare): Bypass alternate compare logic if *op2
29531 is not a CONST_INT.
29532 (ldstwm_operation_p): Return false if first_base is not a REG or
29533 if first_offset is not a CONST_INT.
29535 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29537 * config/cris/cris.md: Use correct operand in a define_peephole2.
29539 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29541 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29543 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29545 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29546 this_insn if it is an INSN or JUMP_INSN.
29547 (force_offsettable): Look at base, not at addr.
29548 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29549 on things that aren't necessarily CONST_INTs.
29551 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29553 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29554 -mfpmath=sse is the default also for x86-32 targets with SSE2
29555 instruction set when @option{-ffast-math} is enabled
29557 2017-02-24 Jeff Law <law@redhat.com>
29559 PR rtl-optimizatoin/79286
29560 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29563 2017-02-24 Richard Biener <rguenther@suse.de>
29565 PR tree-optimization/79389
29566 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29569 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29571 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29572 function comment to reflect reality.
29573 (loop_exits_before_overflow): Fix typo in function description.
29575 2017-02-24 Richard Biener <rguenther@suse.de>
29577 PR tree-optimization/79389
29578 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29579 properly that a threading opportunity exists. Detect conditional
29580 copy/constant propagation opportunities.
29582 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29584 * config/visium/visium.md (type): Add trap.
29585 (b): New mode attribute.
29586 (*btst): Rename into...
29587 (*btst<mode>): ...this and adjust.
29588 (*cbranchsi4_btst_insn): Rename into...
29589 (*cbranch<mode>4_btst_insn): ...this and adjust.
29590 (trap): New define_insn.
29592 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29594 PR tree-optimization/79389
29595 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29596 (noce_reversed_cond_code): New function.
29597 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29598 reversed_comparison_code. Formatting fix.
29599 (noce_try_store_flag): Test rev_cond != NULL in addition to
29600 reversed_comparison_code.
29601 (noce_try_store_flag_constants): Likewise.
29602 (noce_try_store_flag_mask): Likewise.
29603 (noce_try_addcc): Use rev_cond if non-NULL instead of
29604 reversed_comparison_code.
29605 (noce_try_cmove_arith): Likewise. Formatting fixes.
29606 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29607 (noce_find_if_block): Initialize rev_cond.
29608 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29609 instead of false as last argument never attempt to reverse it
29612 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29614 PR tree-optimization/79663
29615 * tree-predcom.c (combine_chains): Process refs in reverse order
29616 only for ZERO length chains, and add explaining comment.
29618 2017-02-23 Jeff Law <law@redhat.com>
29620 PR tree-optimization/79578
29621 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29622 in call to operand_equal_p.
29624 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29627 * config/i386/cpuid.h: Fix another undefined behavior.
29629 2017-02-23 Richard Biener <rguenther@suse.de>
29631 PR tree-optimization/79683
29632 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29633 vector types for data-refs.
29635 2017-02-23 Martin Liska <mliska@suse.cz>
29637 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29639 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29641 PR middle-end/79665
29642 * internal-fn.c (get_range_pos_neg): Moved to ...
29643 * tree.c (get_range_pos_neg): ... here. No longer static.
29644 * tree.h (get_range_pos_neg): New prototype.
29645 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29646 are known to be in between 0 and signed maximum inclusive, try to
29647 expand both unsigned and signed divmod and use the cheaper one from
29650 2017-02-22 Jeff Law <law@redhat.com>
29652 PR tree-optimization/79578
29653 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29654 to compare base operands.
29656 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29659 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29660 gpc_reg_operand instead of fpr_reg_operand.
29662 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29664 * config/mips/mips.c (mips_return_in_memory): Force FP
29665 vector types to be returned in memory for o32 ABI.
29667 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29669 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29670 instead of DW_TAG_member for static data member declarations and don't
29671 set no_linkage_name for static inline data members.
29672 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29675 2017-02-22 Martin Liska <mliska@suse.cz>
29677 * doc/invoke.texi: Replace inequality signs with square brackets
29680 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29683 * lra-constraints.c (simplify_operand_subreg): Handle
29684 WORD_REGISTER_OPERATIONS targets.
29686 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29689 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29690 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29691 elimination by swapping fld*.
29693 2017-02-22 Richard Biener <rguenther@suse.de>
29695 PR tree-optimization/79673
29696 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29697 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29698 irrelevant address-space qualifiers and avoiding a
29699 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29701 2017-02-22 Richard Biener <rguenther@suse.de>
29703 PR tree-optimization/79666
29704 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29705 to not symbolically negate if that may introduce undefined
29708 2017-02-22 Martin Liska <mliska@suse.cz>
29711 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29712 * data-streamer-out.c (streamer_write_gcov_count_stream):
29714 * value-prof.c (stream_out_histogram_value): Make assert more
29715 precise based on type of counter.
29717 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29720 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29721 Use nonimmediate_operand instead of memory_operand for operand 1.
29722 (float-extend standard_x87sse_constant_load splitter): Ditto.
29724 2017-02-21 Jeff Law <law@redhat.com>
29726 PR tree-optimization/79621
29727 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29728 blocks with edges to themselves.
29730 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29733 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29734 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29735 Use gimple_call_builtin_p.
29738 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29739 on temporarily removed DEBUG_INSNs.
29741 PR tree-optimization/79649
29742 * tree-loop-distribution.c (classify_partition): Give up on
29743 non-generic address space loads/stores.
29745 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29747 * doc/loop.texi (Loop manipulation): Remove nonexistent
29748 tree_ssa_loop_version from the documentation.
29749 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29751 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29754 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29755 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29756 * config/rs6000/rs6000.c: Include except.h.
29757 (rs6000_expand_split_stack_prologue): Call
29758 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29760 2017-02-21 Martin Jambor <mjambor@suse.cz>
29763 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29764 have been analyzed.
29766 2017-02-21 Martin Jambor <mjambor@suse.cz>
29768 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29769 for backward compatibility only.
29770 * doc/invoke.texi (Option Summary): Remove all references to
29771 -fipa-cp-alignment.
29773 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29777 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29779 * lra-constraints.c (curr_insn_transform): Handle
29780 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29782 2017-02-21 Martin Liska <mliska@suse.cz>
29784 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29786 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29788 PR translation/79638
29789 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29791 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29794 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29795 (arm_function_ok_for_sibcall): Return false for an indirect call by
29796 descriptor if all the argument registers are used.
29797 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29798 alignment of the function.
29800 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29802 PR tree-optimization/61441
29803 * simplify-rtx.c (simplify_const_unary_operation): For
29804 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29805 the sNaN unmodified.
29807 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29809 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29810 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29811 instead of SYSTEM_HEADER_DIR.
29813 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29814 Martin Liška <mliska@suse.cz>
29816 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29817 Fix typos and grammar, use active voice, and clarify.
29819 2017-02-20 Marek Polacek <polacek@redhat.com>
29821 PR middle-end/79537
29822 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29825 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29827 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29830 * config/i386/i386.c (ix86_expand_builtin): Handle
29831 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29832 ix86_builtins_isa[fcode].isa as a requirement of those
29833 flags and any other flag in the bitmask.
29834 (ix86_init_mmx_sse_builtins): Use 0 instead of
29835 ~OPTION_MASK_ISA_64BIT as mask.
29836 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29837 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29838 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29839 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29841 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29844 * lra-constraints.c (split_reg): Check requested split mode
29845 is supported by the register.
29847 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29849 * lra-constraints.c (simplify_operand_subreg): Remove early
29852 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29855 * lra-constraints.c (curr_insn_transform): Tighten condition
29856 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29858 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29861 * lra-constraints.c (curr_insn_transform): Handle
29862 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29864 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29867 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29869 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29871 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29874 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29877 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29879 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29882 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29884 * final.c (last_columnnum, override_columnnum): New variables.
29885 (final_start_function): Set last_columnnum, pass it to begin_prologue
29886 hook and pass 0 to dwarf2out_begin_prologue.
29887 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29888 to source_line debug hook.
29889 (notice_source_line): Compute last_columnnum and for debug_column_info
29890 return true on column changes.
29891 * debug.h (struct gcc_debug_hooks): Add column argument to
29892 source_line and begin_prologue hooks.
29893 (debug_nothing_int_charstar_int_bool): Remove prototype.
29894 (debug_nothing_int_int_charstar,
29895 debug_nothing_int_int_charstar_int_bool): New prototypes.
29896 (dwarf2out_begin_prologue): Add column argument.
29897 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29898 begin_prologue hooks.
29899 (debug_nothing_int_charstar_int_bool): Remove.
29900 (debug_nothing_int_int_charstar,
29901 debug_nothing_int_int_charstar_int_bool): New functions.
29902 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29903 through to dwarf2out_source_line.
29904 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29905 (dwarf2out_source_line): Add column argument, emit it if requested.
29906 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29908 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29909 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29910 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29911 through to dwarf2out_begin_prologue.
29912 (vmsdbgout_source_line): Add column argument, pass it through to
29913 dwarf2out_source_line.
29914 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29915 dbxout_source_line caller.
29916 (dbxout_source_line): Add column argument.
29918 * common.opt (gno-column-info, gcolumn-info): New options.
29919 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29920 (check_die): Also test for multiple DW_AT_decl_column attributes.
29921 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29922 DW_AT_decl_column if requested.
29923 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29925 (gen_variable_die): Likewise.
29926 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29927 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29930 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29931 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29932 (ix86_handle_option): Handle OPT_m3dnowa.
29933 * doc/invoke.texi (-m3dnowa): Document.
29934 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29935 -m3dnowa instead of -m3dnow -march=athlon.
29938 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29939 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29941 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29944 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29945 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29946 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29947 generator for vsx_xxpermdi_<mode>_be.
29948 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29949 force big-endian semantics.
29950 (vsx_xxpermdi_<mode>_be): New define_expand with same
29951 implementation as previous version of vsx_xxpermdi_<mode>.
29953 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29955 PR tree-optimization/79327
29956 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29957 variable, its initialization and use.
29959 2017-02-17 Julia Koval <julia.koval@intel.com>
29961 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29962 (OPTION_MASK_ISA_PKU_UNSET): New.
29963 (ix86_handle_option): Handle -mrdpid.
29964 * config/i386/cpuid.h (bit_RDPID): New.
29965 * config/i386/driver-i386.c (host_detect_local_cpu):
29966 Detect RDPID feature.
29967 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29968 * config/i386/i386-c.c (ix86_target_macros_internal):
29970 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29971 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29972 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29973 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29974 * config/i386/i386.md (define_insn "rdpid"): New.
29975 * config/i386/i386.opt Add -mrdpid.
29976 * config/i386/immintrin.h (_rdpid_u32): New.
29978 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29980 PR rtl-optimization/79541
29981 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29982 instead of transforming it into USE.
29984 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29986 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29987 If HONOR_SNANS (SFmode) force the input to a register.
29988 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29989 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29990 an frsp or similar insn.
29992 2017-02-17 Martin Liska <mliska@suse.cz>
29994 PR rtl-optimization/79577
29995 * params.def (selsched-max-sched-times): Increase minimum to 1.
29997 2017-02-17 Martin Liska <mliska@suse.cz>
29999 PR rtl-optimization/79574
30000 * gcse.c (want_to_gcse_p): Prevent integer overflow.
30002 2017-02-17 Martin Liska <mliska@suse.cz>
30004 PR tree-optimization/79529
30005 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
30006 ssa_defined_default_def_p to handle cases which are implicitly
30008 * tree-ssa.c (ssa_defined_default_def_p): New function.
30009 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
30010 which are implicitly defined.
30011 * tree-ssa.h (ssa_defined_default_def_p): Declare.
30013 2017-02-17 Richard Biener <rguenther@suse.de>
30015 PR middle-end/79576
30016 * params.def (max-ssa-name-query-depth): Limit to 10.
30018 2017-02-17 Richard Biener <rguenther@suse.de>
30020 PR tree-optimization/79552
30021 * tree-ssa-structalias.c (visit_loadstore): Properly verify
30024 2017-02-17 Richard Biener <rguenther@suse.de>
30027 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30029 2017-02-17 Marek Polacek <polacek@redhat.com>
30031 PR middle-end/79536
30032 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30033 (fold_negate_expr): New wrapper.
30035 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30037 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30038 Correct terminology and de-emphasize pre-standard behavior.
30040 2017-02-16 Alan Modra <amodra@gmail.com>
30042 PR rtl-optimization/79286
30043 * ira.c (def_dominates_uses): New function.
30044 (update_equiv_regs): Don't create an equivalence for insns that
30045 may trap where the register def does not dominate the use.
30047 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30049 PR rtl-optimization/78127
30050 * lra.c (lra): Call lra_eliminate before finish the loop after
30053 2017-02-16 Richard Biener <rguenther@suse.de>
30055 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30057 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30058 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30059 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30060 (isl_val_int_from_wi): New function.
30061 (extract_affine_gmp): Rename to ...
30062 (extract_affine_wi): ... this, take a widest_int.
30063 (extract_affine_int): Just wrap extract_affine_wi.
30064 (add_param_constraints): Use isl_val_int_from_wi.
30065 (add_loop_constraints): Likewise, and extract_affine_wi.
30067 2017-02-15 Jeff Law <law@redhat.com>
30069 PR middle-end/79521
30070 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30071 ira_init_register_move_cost_if_necessary.
30073 2017-02-15 Martin Sebor <msebor@redhat.com>
30075 PR middle-end/32003
30076 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30077 removed in a prior commit.
30079 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30081 PR tree-optimization/79347
30082 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30083 counters during peeling.
30085 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30087 * Makefile.in (site.exp): Remove "set ISLVER".
30089 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30092 * real.c (real_from_integer): Call real_convert even for decimal.
30094 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30097 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30099 2017-02-14 Andrew Pinski <apinski@cavium.com>
30101 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30102 cores and change the partno/implementer to be correct.
30103 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30104 the 'B" as the implementer.
30105 * config/aarch64/aarch64-tune.md: Regenerate.
30107 2017-02-14 Carl Love <cel@us.ibm.com>
30109 * config/rs6000/rs6000.c: Add case statement entry to make the
30110 xvcvuxdsp built-in argument unsigned.
30111 * config/rs6000/vsx.md: Fix the source and return operand types so they
30112 match the instruction definitions from the ISA document. Fix typo
30113 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30116 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30119 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30120 member early_clobber_alts.
30121 * lra-lives.c (reg_early_clobber_p): New.
30122 (process_bb_lives): Use it.
30123 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30124 (debug_operand_data): Initialize early_clobber_alts.
30125 (setup_operand_alternative): Set up early_clobber_alts.
30126 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30127 alternatives to new_insn_reg.
30128 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30130 (lra_update_insn_regno_info): Pass the new arg.
30132 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30134 PR middle-end/79505
30135 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30136 (new_oacc_loop_raw): Don't clear already cleared fields.
30139 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30140 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30141 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30143 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30146 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30148 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30151 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30152 the extra instruction to the right place to store 128-bit constant
30155 2017-02-14 Martin Sebor <msebor@redhat.com>
30157 PR middle-end/79448
30158 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30159 warning for strings of unknown length.
30161 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30163 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30165 2017-02-14 Jeff Law <law@redhat.com>
30168 * ira-costs.c (scan_one_insn): Initialize register move costs
30169 for pseudos seen in USE/CLOBBER insns.
30171 PR tree-optimization/79095
30172 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30173 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30174 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30175 if the operands are known to be not equal, then the resulting range
30177 (intersect_ranges): If the new range is ~[0,0] and the old range is
30178 wide, then prefer ~[0,0].
30179 * tree-vrp.c (overflow_comparison_p_1): New function.
30180 (overflow_comparison_p): New function.
30181 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30182 if NAME is used in an overflow test.
30183 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30184 overflow check that can be expressed as an equality test, then adjust
30185 ops to be that equality test.
30187 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30189 * config/s390/s390-builtin-types.def: Remove flags argument.
30190 * config/s390/s390.c (s390_init_builtins): Likewise.
30192 2017-02-14 Martin Liska <mliska@suse.cz>
30194 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30195 vector. Fix trailing white spaces.
30197 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30199 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30202 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30204 PR rtl-optimization/68664
30205 * config/arm/arm.c (arm_sched_can_speculate_insn):
30206 New function. Declare prototype.
30207 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30209 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30211 PR rtl-optimization/68664
30212 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30214 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30216 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30218 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30219 max skip bytes for function, loop and jump.
30221 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30223 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30224 ABS_EXPR for gimple dump.
30226 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30229 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30231 PR tree-optimization/79408
30232 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30233 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30234 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30235 also if rhs1 is INTEGER_CST.
30237 2017-02-14 Richard Biener <rguenther@suse.de>
30239 PR middle-end/79432
30240 * tree-into-ssa.c (insert_phi_nodes): When the function can
30241 have abnormal edges rewrite SSA names with broken use-def
30242 dominance out of SSA and register them for PHI insertion.
30244 2017-02-13 Martin Sebor <msebor@redhat.com>
30246 PR middle-end/79496
30247 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30248 clearing info.nowrite flag when snprintf size argument is a range.
30250 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30252 * cprop.c (cprop_jump): Add missing space in string literal.
30253 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30254 (get_constraint_for_component_ref): Likewise.
30255 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30256 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30257 * lra-constraints.c (process_alt_operands): Likewise.
30258 * ipa-inline.c (inline_small_functions): Likewise.
30259 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30260 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30261 * trans-mem.c (diagnose_tm_1_op): Likewise.
30262 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30263 (grid_parallel_clauses_gridifiable): Likewise.
30265 * config/nvptx/mkoffload.c (process): Add space in between
30268 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30269 "MOD4_SSE_REGS" and "ALL_REGS".
30271 * spellcheck.c (test_data): Add , in between "foo" and "food".
30273 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30276 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30277 boundary crossing check and subsequent code generation agree.
30279 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30281 * config/aarch64/aarch64.c (has_memory_op): Delete.
30282 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30285 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30287 PR rtl-optimization/79388
30288 PR rtl-optimization/79450
30289 * combine.c (distribute_notes): When removing TEM_INSN for which
30290 corresponding dest has last value recorded, invalidate that last
30293 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30295 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30296 of explicit '@'. Add missing assembly comment marker on branch costs
30299 2017-02-13 Nathan Sidwell <nathan@acm.org>
30301 * gengtype-lex.l (<in_struct>): Add '/'.
30303 2017-02-13 Martin Liska <mliska@suse.cz>
30306 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30308 2017-02-13 Richard Biener <rguenther@suse.de>
30310 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30312 * configure: Re-generate.
30313 * config.in: Likewise.
30314 * graphite-dependences.c: Simplify as if
30315 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30316 * graphite-isl-ast-to-gimple.c: Likewise.
30317 * graphite-optimize-isl.c: Likewise.
30318 * graphite-poly.c: Likewise.
30319 * graphite-sese-to-poly.c: Likewise.
30320 * graphite.h: Likewise.
30321 * toplev.c: Include isl/version.h and use isl_version () for
30322 printing the ISL version.
30323 * doc/install.texi: Update ISL requirement.
30325 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30327 * doc/standards.texi (Standards): Update reference to
30330 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30332 * doc/extend.texi (Named Address Spaces): sourceware.org now
30334 * doc/install.texi (Binaries): Ditto.
30337 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30339 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30340 standard terminology "stringize"/"stringizing" throughout.
30341 * doc/cppinternals.texi: Likewise.
30343 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30345 * doc/extend.texi: Fix some spelling mistakes and typos.
30346 * doc/invoke.texi: Likewise.
30348 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30351 * params.def (inline-min-speedup) Change from 10 to 8.
30353 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30355 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30358 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30361 * ipa-inline-analysis.c (get_minimal_bb): New function.
30362 (record_modified): Use it.
30363 (remap_edge_change_prob): Handle also ancestor functions.
30365 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30367 * doc/contrib.texi (Contributors): Remove broken link into
30368 the Mauve CVS repository.
30370 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30372 PR middle-end/79454
30373 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30374 result computation whenever lhs doesn't have vector mode, not
30375 just when it has BLKmode.
30377 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30379 * doc/makefile.texi (profiledbootstrap): Refer to the
30380 installation instructions only in textual form.
30382 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30385 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30387 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30389 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30390 (Specific): Update mingw-w64 reference.
30392 (Specific): Remove broken link to Renesas RX processor.
30394 2017-02-10 Richard Biener <rguenther@suse.de>
30396 * toplev.c (process_options): Do not mention obsolete graphite
30397 options when printing sorry message about missing graphite support.
30398 Mention -floop-nest-optimize.
30400 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30402 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30403 (vtst_p16): Likewise.
30404 (vtstq_p8): Likewise.
30405 (vtstq_p16): Likewise.
30407 (vtstq_p64): Likewise.
30408 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30409 (vset_lane_p64): New.
30410 (vsetq_lane_p64): New.
30412 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30414 PR tree-optimization/79411
30415 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30416 stmt operands are SSA_NAMEs used in abnormal phis.
30417 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30420 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30423 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30426 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30428 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30430 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30432 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30436 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30437 not arbitrary TREE_CONSTANT.
30440 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30441 "omp declare target link" attribute unless is_global_var.
30442 * omp-offload.c (find_link_var_op): Likewise.
30444 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30445 Chung-Lin Tang <cltang@codesourcery.com>
30447 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30449 (gimplify_adjust_omp_clauses): Don't delete TILE.
30450 (gimplify_omp_for): Deal with TILE.
30451 * internal-fn.c (expand_GOACC_TILE): New function.
30452 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30454 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30455 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30457 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30458 avoid DIV for outermost collapse var.
30459 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30460 Remove out of date comments, fix whitespace.
30461 * omp-general.c (omp_extract_for_data): Deal with tiling.
30462 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30463 adjust OLF_DIM_BASE value.
30464 (struct omp_for_data): Add tiling field.
30465 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30466 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30467 for auto loops. Remove default auto determining, moved to
30468 oacc_loop_fixed_partitions.
30469 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30470 stmts, add e_mask field.
30471 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30472 (oacc_thread_numbers): Use oacc_dim_call.
30473 (oacc_xform_tile): New.
30474 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30475 (finish_oacc_loop): Adjust for ifns vector.
30476 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30477 add case for GOACC_TILE fns.
30478 (oacc_loop_xform_loop): Delete.
30479 (oacc_loop_process): Iterate over call list directly, and add
30480 handling for GOACC_TILE fns.
30481 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30483 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30484 vector partitioning to outer loops. Assign 2 partitions to loops
30485 when available. Add TILE handling.
30486 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30487 (execite_oacc_device_lower): Process GOACC_TILE fns,
30488 ignore unknown specs.
30489 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30490 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30491 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30493 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30495 * configure.ac (ACX_BUGURL): Update.
30496 * configure: Regenerate.
30498 2017-02-09 Richard Biener <rguenther@suse.de>
30500 PR tree-optimization/69823
30501 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30502 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30504 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30506 * config/arc/arc-c.def: Add __NPS400__ definition.
30507 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30508 (TARGET_NPS400): Define.
30510 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30512 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30514 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30515 pointer, arch_info.
30516 (arc_cpu_types): Fill the arch_info field with a pointer into the
30517 arc_arch_types table.
30518 (arc_selected_cpu): Declare.
30519 * config/arc/arc.c (arc_selected_cpu): Make global.
30520 (arc_selected_arch): Delete.
30521 (arc_base_cpu): Delete.
30522 (arc_override_options): Remove references to deleted variables,
30523 update access to arch information.
30524 (ARC_OPT): Update access to arch information.
30525 (ARC_OPTX): Likewise.
30526 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30527 (TARGET_ARC600): Update access to arch information.
30528 (TARGET_ARC601): Likewise.
30529 (TARGET_ARC700): Likewise.
30530 (TARGET_EM): Likewise.
30531 (TARGET_HS): Likewise.
30532 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30535 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30538 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30539 condition/operands for integer GE/LE/GEU/LEU operations.
30541 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30543 PR translation/79397
30544 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30547 2017-02-08 Martin Jambor <mjambor@suse.cz>
30550 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30551 whether allocation happened.
30552 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30553 nothing was allocated.
30555 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30557 PR tree-optimization/79408
30558 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30559 constant, but SSA_NAME with a known integer range, use the minimum
30560 of that range instead of op1 to determine if modulo can be replaced
30561 with its first operand.
30563 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30565 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30567 2017-02-08 Richard Biener <rguenther@suse.de>
30569 PR tree-optimization/71824
30570 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30571 Check all loops contained in the merged region.
30573 2017-02-07 Andrew Pinski <apinski@cavium.com>
30575 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30577 2017-02-07 Andrew Pinski <apinski@cavium.com>
30579 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30580 (thunderxt88): Likewise.
30581 (thunderxt81): Disable LSE and change v8.1 to v8.
30582 (thunderxt83): Likewise.
30584 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30585 Richard Biener <rguenther@suse.de>
30587 PR middle-end/79399
30588 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30589 type from int to size_t.
30590 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30592 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30594 PR rtl-optimization/79386
30595 * cprop.c (bypass_conditional_jumps): Initialize
30596 bypass_last_basic_block already before splitting bbs after
30597 unconditional traps...
30598 (bypass_conditional_jumps): ... rather than here.
30601 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30602 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30603 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30604 fix -masm=intel patterns.
30606 2017-02-07 Richard Biener <rguenther@suse.de>
30608 PR tree-optimization/79256
30609 PR middle-end/79278
30610 * builtins.c (get_object_alignment_2): Use min_align_of_type
30611 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30612 and ADJUST_FIELD_ALIGN.
30614 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30616 * doc/tm.texi: Regenerate.
30617 * stor-layout.c (layout_decl): Adjust.
30618 (update_alignment_for_field): Likewise.
30619 (place_field): Likewise.
30620 (min_align_of_type): Likewise.
30621 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30622 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30623 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30624 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30625 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30626 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30627 * config/i386/i386.c (x86_field_alignment): Likewise.
30628 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30629 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30630 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30631 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30632 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30633 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30637 2017-01-30 Richard Biener <rguenther@suse.de>
30639 PR tree-optimization/79256
30640 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30641 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30644 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30646 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30647 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30648 builtins to SImode and emit a zero-extend, if necessary.
30650 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30652 * docs/invoke.texi (RISC-V Options): Alphabetize.
30654 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30656 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30659 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30661 * config/riscv/riscv.c: New file.
30662 * common/config/riscv/riscv-common.c: Likewise.
30663 * config.gcc: Likewise.
30664 * config/riscv/constraints.md: Likewise.
30665 * config/riscv/elf.h: Likewise.
30666 * config/riscv/generic.md: Likewise.
30667 * config/riscv/linux.h: Likewise.
30668 * config/riscv/multilib-generator: Likewise.
30669 * config/riscv/peephole.md: Likewise.
30670 * config/riscv/pic.md: Likewise.
30671 * config/riscv/predicates.md: Likewise.
30672 * config/riscv/riscv-builtins.c: Likewise.
30673 * config/riscv/riscv-c.c: Likewise.
30674 * config/riscv/riscv-ftypes.def: Likewise.
30675 * config/riscv/riscv-modes.def: Likewise.
30676 * config/riscv/riscv-opts.h: Likewise.
30677 * config/riscv/riscv-protos.h: Likewise.
30678 * config/riscv/riscv.h: Likewise.
30679 * config/riscv/riscv.md: Likewise.
30680 * config/riscv/riscv.opt: Likewise.
30681 * config/riscv/sync.md: Likewise.
30682 * config/riscv/t-elf-multilib: Likewise.
30683 * config/riscv/t-linux: Likewise.
30684 * config/riscv/t-linux-multilib: Likewise.
30685 * config/riscv/t-riscv: Likewise.
30686 * configure.ac: Likewise.
30687 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30688 Waterman as RISC-V maintainers.
30689 * doc/install.texi: Add RISC-V entries.
30690 * doc/invoke.texi: Add RISC-V options section.
30691 * doc/md.texi: Add RISC-V constraints section.
30692 * configure: Regenerated.
30694 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30697 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30698 false values to be constant vectors with all 0 or all 1 bits set.
30699 (vcondu<mode><mode>): Likewise.
30700 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30702 (fpmask_comparison_operator): Update comment.
30703 (vecint_comparison_operator): New predicate.
30704 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30705 vector conditionals when the true and false values are constant
30706 vectors with all 0 bits or all 1 bits set.
30708 2017-02-06 Martin Sebor <msebor@redhat.com>
30710 PR tree-optimization/79376
30711 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30713 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30715 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30716 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30717 to simplify split condition.
30719 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30721 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30722 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30725 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30727 PR rtl-optimization/68664
30728 * target.def (can_speculate_insn): New hook.
30729 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30730 * doc/tm.texi: Regenerate.
30731 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30732 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30733 (rs6000_sched_can_speculate_insn): New function.
30735 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30737 PR tree-optimization/79284
30738 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30739 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30740 vectorizable_mask_load_store, vectorizable_operation,
30741 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30742 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30743 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30744 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30745 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30746 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30747 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30748 is_gimple_assign (stmt). Replace another such test with
30749 is_gimple_assign (stmt).
30751 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30754 * config/avr/avr.c (rtl-iter.h): Include it.
30755 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30756 (avr_legitimate_combined_insn): ...and implementation.
30758 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30760 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30761 * config/s390/s390.c (s390_const_operand_ok)
30762 (s390_canonicalize_comparison, s390_extract_part)
30763 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30764 (s390_contiguous_bitmask_p, s390_rtx_costs)
30765 (legitimize_pic_address): Likewise.
30766 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30767 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30768 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30769 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30770 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30772 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30774 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30775 REGNO($0) == REGNO($1).
30777 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30779 * config/s390/linux.h(SIZE_TYPE): Add comment.
30781 2017-02-06 Julian Brown <julian@codesourcery.com>
30782 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30783 Virendra Pathak <virendra.pathak@broadcom.com>
30785 * config/aarch64/aarch64-cores.def: Change the scheduler
30787 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30788 * config/aarch64/thunderx2t99.md: New file.
30790 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30792 * doc/standards.texi (Go Language): Update link to language
30795 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30797 * tree-eh.c (lower_resx): Sanitize profile.
30798 (cleanup_empty_eh_move_lp): Likewise.
30800 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30803 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30805 * cfgloopmanip.h (loop_version): Update prototype.
30806 * modulo-sched.c (sms_schedule): Update call of loop_version.
30807 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30808 * tree-parloops.c (gen_parallel_loop): Likewise.
30809 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30810 * tree-ssa-loop-split.c (split_loop): Likewise.
30811 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30812 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30814 2017-02-05 Martin Liska <mliska@suse.cz>
30817 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30819 (print_operand_address): Initialize a struct to zero.
30821 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30823 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30824 garbage collector only in textual form.
30826 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30828 * doc/extend.texi (x86 specific memory model extensions for
30829 transactional memory): Simplify a phrase.
30831 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30834 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30835 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30836 (atomic_storedi_1): Likewise.
30838 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30840 PR tree-optimization/79338
30841 * tree-parloops.c (gather_scalar_reductions): Don't call
30842 vect_analyze_loop_form for loop->inner before destroying loop's
30845 2017-02-03 Martin Sebor <msebor@redhat.com>
30847 PR tree-optimization/79327
30848 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30849 when precision has resulted in leading zeros.
30850 (format_integer): Adjust the likely counter to assume an unknown
30851 argument that may be zero is non-zero.
30853 2017-02-03 Jason Merrill <jason@redhat.com>
30856 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30857 avoid copying non-taken branch.
30859 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30861 PR tree-optimization/79340
30862 * tree-vect-loop.c (vectorizable_reduction): Release
30863 vec_defs elements after safe_splicing them into other vectors.
30866 PR tree-optimization/79327
30867 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30868 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30870 (format_integer): Use wide_int_to_tree instead of build_int_cst
30871 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30872 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30873 of shortest and longest sequence.
30875 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30877 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30878 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30880 2017-02-03 Walter Lee <walt@tilera.com>
30883 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30884 after initial stackframe link reg save.
30885 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30887 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30890 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30891 wu for stxssp alternative.
30893 2017-02-03 Martin Sebor <msebor@redhat.com>
30895 PR tree-optimization/79352
30896 * gimple-fold.c (get_range_strlen): Add argument.
30897 (get_range_strlen): Change return type to bool.
30898 (get_maxval_strlen): Pass in a dummy argument.
30899 * gimple-fold.h (get_range_strlen): Change return type to bool.
30900 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30901 * tree.h (array_at_struct_end_p): Add argument.
30902 * tree.c (array_at_struct_end_p): Handle it.
30904 2017-02-03 Martin Liska <mliska@suse.cz>
30907 * multiple_target.c (create_dispatcher_calls): Redirect edge
30908 from a caller of a dispatcher.
30909 (expand_target_clones): Make the clones local.
30910 (ipa_target_clone): Do both target clones and resolvers.
30911 (ipa_dispatcher_calls): Remove the pass.
30912 (pass_dispatcher_calls::gate): Likewise.
30913 (make_pass_dispatcher_calls): Likewise.
30914 * passes.def (pass_target_clone): Put as very first IPA early
30917 2017-02-03 Martin Liska <mliska@suse.cz>
30919 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30920 in case of a function with ifunc attribute.
30922 2017-02-03 Martin Liska <mliska@suse.cz>
30924 * cgraph.c (cgraph_node::dump): Dump function version info.
30925 * symtab.c (symtab_node::dump_base): Add missing new line.
30927 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30929 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30930 (ifcombine_ifandif): Use it.
30932 2017-02-03 Martin Liska <mliska@suse.cz>
30934 * doc/invoke.texi: Document default value for
30935 use-after-scope-direct-emission-threshold.
30937 2017-02-03 Martin Liska <mliska@suse.cz>
30939 PR tree-optimization/79339
30940 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30941 (format_floating): Likewise.
30943 2017-02-03 Martin Liska <mliska@suse.cz>
30946 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30948 (ipa_node_params_t::remove): Likewise.
30949 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30950 initialization from removed ipa_node_params_t::insert.
30951 (ipa_node_params::~ipa_node_params): Move from removed
30952 ipa_node_params_t::release.
30953 * symbol-summary.h (symbol_summary::m_released): New member.
30954 Do not release a summary twice. Do not allow to call finalizer
30955 for types of a summary that live in GGC memory.
30957 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30959 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30962 2017-02-02 Martin Sebor <msebor@redhat.com>
30964 PR middle-end/79275
30965 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30966 (format_string): Tighten up the range of output for non-constant
30967 strings and correct the expected range for wide non-constant strings.
30969 2017-02-02 Martin Sebor <msebor@redhat.com>
30971 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30973 PR middle-end/32003
30974 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30976 (-fdump-tree-@var): Add to index and document how to come up
30977 with pass-specific option and dump file names.
30978 (-fdump-passes): Clarify where to look for output.
30980 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30982 PR middle-end/77445
30983 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30984 statistics of the analyzed path; allow threading for speed when
30985 any of BBs along the path are optimized for speed.
30987 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30989 PR middle-end/78468
30990 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30991 settings of the virtual registers.
30994 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30996 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30997 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
31000 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31002 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
31003 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
31005 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31007 * config/s390/s390.md: Add missing comments with the expanded
31009 * config/s390/vector.md: Likewise.
31010 * config/s390/vx-builtins.md: Likewise.
31012 2017-02-02 Jakub Jelinek <jakub@redhat.com>
31015 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
31016 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
31017 conditions on a single line.
31019 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31021 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31022 __S390_VX__ to __VX__.
31024 2017-02-01 Andrew Pinski <apinski@cavium.com>
31026 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31027 stmt_info to record_stmt_cost.
31028 (vect_get_known_peeling_cost): Pass stmt_info if known to
31030 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31031 cpu_vector_cost field into
31032 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31033 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31034 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31035 splitting of scalar_stmt_cost and vec_stmt_cost.
31036 (thunderx_vector_cost): Likewise.
31037 (cortexa57_vector_cost): LIkewise.
31038 (exynosm1_vector_cost): Likewise.
31039 (xgene1_vector_cost): Likewise.
31040 (thunderx2t99_vector_cost): Improve after the splitting of the two
31042 (aarch64_builtin_vectorization_cost): Update for the splitting of
31043 scalar_stmt_cost and vec_stmt_cost.
31045 2017-02-01 Torvald Riegel <triegel@redhat.com>
31046 Richard Henderson <rth@redhat.com>
31048 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31049 conditional on existance of a fast atomic load.
31050 * optabs-query.c (can_atomic_load_p): New function.
31051 * optabs-query.h (can_atomic_load_p): Declare it.
31052 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31053 no fast atomic load is available for the particular size of access.
31054 (expand_atomic_compare_and_swap): Likewise.
31055 (expand_atomic_load): Likewise.
31056 (expand_atomic_store): Likewise.
31057 (expand_atomic_fetch_op): Likewise.
31058 * testsuite/lib/target-supports.exp
31059 (check_effective_target_sync_int_128): Remove x86 because it provides
31060 no fast atomic load.
31061 (check_effective_target_sync_int_128_runtime): Likewise.
31063 2017-02-01 Richard Biener <rguenther@suse.de>
31065 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31066 (graphite_transform_loops): Provide opt-info for optimized nests.
31067 * tree-parloop.c (parallelize_loops): Provide opt-info for
31068 parallelized loops.
31070 2017-02-01 Richard Biener <rguenther@suse.de>
31072 PR middle-end/79315
31073 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31074 was not set before.
31076 2017-02-01 Richard Biener <rguenther@suse.de>
31078 PR tree-optimization/71824
31079 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31080 Verify the loops are valid in the merged SESE region.
31081 (scop_detection::can_represent_loop_1): Check analyzing the
31082 evolution of the number of iterations in the region succeeds.
31084 2017-01-31 Ian Lance Taylor <iant@golang.org>
31086 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31087 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31089 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31091 PR preprocessor/79210
31092 * input.c (get_substring_ranges_for_loc): Replace line_width
31093 assertion with error-handling.
31095 2017-01-31 Richard Biener <rguenther@suse.de>
31097 PR tree-optimization/77318
31098 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31099 (create_pw_aff_from_tree): Take loop parameter.
31100 (add_condition_to_pbb): Pass loop of the condition to
31101 create_pw_aff_from_tree.
31103 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31105 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31106 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31108 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31112 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31114 (convert_int_to_float128): Likewise.
31115 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31116 (convert_int_to_float128): Likewise.
31117 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31118 (UNSPEC_IEEE128_CONVERT): Likewise.
31119 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31120 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31121 Use local variables for IBM extended format.
31122 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31123 (fix_trunc<mode>si2_fprs): Likewise.
31124 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31125 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31126 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31127 to know that we can now have integers of all sizes in vector
31129 (fix<uns>_<mode>di2_hw): Likewise.
31130 (float<uns>_<mode>si2_hw): Likewise.
31131 (fix_<mode>si2_hw): Likewise.
31132 (fixuns_<mode>si2_hw): Likewise.
31133 (float<uns>_<mode>di2_hw): Likewise.
31134 (float_<mode>di2_hw): Likewise.
31135 (float_<mode>si2_hw): Likewise.
31136 (floatuns_<mode>di2_hw): Likewise.
31137 (floatuns_<mode>si2_hw): Likewise.
31138 (xscvqp<su>wz_<mode>): Delete, no longer used.
31139 (xscvqp<su>dz_<mode>): Likewise.
31140 (xscv<su>dqp_<mode>): Likewise.
31141 (ieee128_mfvsrd_64bit): Likewise.
31142 (ieee128_mfvsrd_32bit): Likewise.
31143 (ieee128_mfvsrwz): Likewise.
31144 (ieee128_mtvsrw): Likewise.
31145 (ieee128_mtvsrd_64bit): Likewise.
31146 (ieee128_mtvsrd_32bit): Likewise.
31148 2017-01-31 Martin Liska <mliska@suse.cz>
31151 * ipa-prop.c (ipa_free_all_node_params): Call release method
31152 instead of ~sumbol_summary to not to trigger double times
31155 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31157 PR tree-optimization/71691
31158 * bitmap.h (class auto_bitmap): New.
31159 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31160 is_maybe_undefined instead of ssa_undefined_value_p.
31162 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31164 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31165 __S390_ARCH_LEVEL__ to __ARCH__.
31167 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31169 PR tree-optimization/79267
31170 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31171 if should_remove_lhs_p is true.
31173 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31176 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31177 (add_alignment_attribute): New.
31178 (base_type_die): Add alignment attribute.
31179 (subrange_type_die): Likewise.
31180 (modified_type_die): Likewise.
31181 (gen_array_type_die): Likewise.
31182 (gen_descr_array_type_die: Likewise.
31183 (gen_enumeration_type_die): Likewise.
31184 (gen_subprogram_die): Likewise.
31185 (gen_variable_die): Likewise.
31186 (gen_field_die): Likewise.
31187 (gen_ptr_to_mbr_type_die): Likewise.
31188 (gen_struct_or_union_type_die): Likewise.
31189 (gen_subroutine_type_die): Likewise.
31190 (gen_typedef_die): Likewise.
31191 (base_type_cmp): Compare alignment attribute.
31193 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31196 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31197 (setb_unsigned) New pattern for setb with CCUNS.
31198 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31199 subfc./subfe sequence to avoid overflow problems. Generate a
31200 shorter sequence with cmpld/setb for power9.
31201 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31202 for generating subfc. instruction.
31203 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31204 now uses this instruction.
31206 2017-01-30 Ian Lance Taylor <iant@google.com>
31209 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31210 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31212 2017-01-30 Martin Sebor <msebor@redhat.com>
31214 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31215 Move constant to the right of a relational operator.
31216 (get_mpfr_format_length, format_character, format_string): Ditto.
31217 (should_warn_p, maybe_warn): Same.
31219 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31221 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31224 * asan.c (get_translation_unit_decl): Remove function.
31225 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31227 2017-01-30 Martin Liska <mliska@suse.cz>
31229 PR gcov-profile/79259
31230 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31231 -fprofile-generate.
31233 2017-01-30 Martin Liska <mliska@suse.cz>
31236 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31237 Initialize variables with NULL value.
31239 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31242 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31244 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31246 2017-01-30 Richard Biener <rguenther@suse.de>
31248 PR tree-optimization/79276
31249 * tree-vrp.c (process_assert_insertions): Properly adjust common
31250 when removing a duplicate.
31252 2017-01-30 Richard Biener <rguenther@suse.de>
31254 PR tree-optimization/79256
31255 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31256 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31258 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31260 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31263 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31264 ("*r<noxa>sbg_<mode>_sll_bitmask")
31265 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31266 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31267 Use contiguous_bitmask_nowrap_operand.
31269 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31272 * config/rs6000/altivec.h (vec_xl): Revise #define.
31273 (vec_xst): Likewise.
31275 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31277 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31279 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31281 PR rtl-optimization/79194
31282 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31283 traps before call to bypass_conditional_jumps.
31285 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31287 PR tree-optimization/71374
31288 * lra-constraints.c (check_conflict_input_operands): New.
31289 (match_reload): Use it.
31291 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31294 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31295 account to calculate conflict_set.
31297 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31299 PR rtl-optimization/78559
31300 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31301 other_insn in combine.
31303 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31305 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31306 uint16_type_node for BT_UINT16.
31308 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31310 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31311 "RTL Tests" to menu.
31312 (GIMPLE Tests): New node.
31313 (RTL Tests): New node.
31315 2017-01-27 Richard Biener <rguenther@suse.de>
31317 PR tree-optimization/79245
31318 * tree-loop-distribution.c (distribute_loop): Apply cost
31319 modeling also to detected patterns.
31321 2017-01-27 Richard Biener <rguenther@suse.de>
31323 PR tree-optimization/71433
31324 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31325 (compare_assert_loc): New function.
31326 (process_assert_insertions): Sort and optimize assert locations
31327 to remove duplicates and push down identical assertions on
31328 edges to their destination block.
31330 2017-01-27 Richard Biener <rguenther@suse.de>
31332 PR tree-optimization/79244
31333 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31334 out SSA names even if abnormal.
31336 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31338 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31339 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31340 instead of MPFR_RNDN.
31342 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31345 * arm.c (arm_option_override): Don't call build_target_option_node
31346 until after doing all option overrides.
31347 (arm_valid_target_attribute_tree): Likewise.
31349 2017-01-27 Martin Liska <mliska@suse.cz>
31351 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31352 for {cd}tors and C++ {cd}tors.
31354 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31356 * config/s390/s390.md ("*setmem_long_and")
31357 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31359 2017-01-26 Martin Sebor <msebor@redhat.com>
31361 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31364 2017-01-26 Martin Sebor <msebor@redhat.com>
31366 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31367 HAVE_DFmode before using XFmode or DFmode.
31368 (parse_directive): Avoid using the z length modifier to avoid
31369 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31371 PR middle-end/78703
31372 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31373 to accept adjustment as an array.
31374 (get_int_range): New function.
31375 (struct directive): Make width and prec arrays.
31376 (directive::set_width, directive::set_precision): Call get_int_range.
31377 (format_integer, format_floating): Handle width and precision ranges.
31378 (format_string, parse_directive): Same.
31380 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31383 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31384 comdat_type_p set, just clone them, but keep the children in the
31388 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31389 which have direct callers with -fvar-tracking-assignments enabled
31391 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31392 inside of type units.
31394 2017-01-26 Martin Sebor <msebor@redhat.com>
31396 PR middle-end/78703
31397 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31399 (struct format_result): Replace number_chars, number_chars_min,
31400 and number_chars_max with a single member of struct result_range.
31402 (format_result::operator+=): Adjust.
31403 (struct fmtresult): Remove bounded. Handle likely and unlikely
31405 (fmtresult::adjust_for_width_or_precision): New function.
31406 (fmtresult:type_max_digits): New function.
31407 (bytes_remaining): Handle likely and unlikely counters.
31408 (min_bytes_remaining): Remove.
31409 (format_percent): Simplify.
31410 (format_integer, format_floating): Set likely and unlikely counters.
31411 (get_string_length, format_character, format_string): Same.
31412 (format_plain, should_warn_p): New function.
31413 (maybe_warn): Call should_warn_p. Update diagnostic messages
31414 and handle those for all directives, including plain strings.
31415 (format_directive): Handle likely and unlikely counters.
31416 Remove unnecessary quoting from diagnostics. Add an informational
31418 (add_bytes): Remove.
31419 (pass_sprintf_length::compute_format_length): Simplify.
31420 (try_substitute_return_value): Handle likely and unlikely counters.
31422 2017-01-26 Carl Love <cel@us.ibm.com>
31424 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31425 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31427 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31430 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31431 endianess for subregs into account.
31432 * lra-constraints.c (lra_constraints): Do risky transformations
31433 always on the first iteration.
31434 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31435 last_call_used_reg_set.
31436 (process_bb_lives): Define and use last_call_used_reg_set.
31437 * lra.c (lra): Always continue after lra_constraints on the first
31440 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31442 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31444 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31446 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31448 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31449 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31450 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31451 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31452 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31453 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31454 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31455 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31456 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31458 2017-01-26 Marek Polacek <polacek@redhat.com>
31461 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31462 for the third operand.
31464 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31466 PR middle-end/79236
31467 * omp-low.c (struct omp_context): Add simt_stmt field.
31468 (scan_omp_for): Return omp_context *.
31469 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31470 context to the _simt_ SIMD stmt.
31471 (lower_omp_for): For combined SIMD with sibling _simt_
31472 SIMD, make sure to use the same decls in _looptemp_
31473 clauses as in the sibling.
31475 2017-01-26 David Sherwood <david.sherwood@arm.com>
31477 PR middle-end/79212
31478 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31481 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31484 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31485 emit fld b; fld a; if possible.
31487 * brig-builtins.def: Update copyright years.
31488 * config/arm/arm_acle_builtins.def: Update copyright years.
31490 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31493 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31494 constraint instead of o for the stxsd instruction.
31496 2017-01-25 Carl Love <cel@us.ibm.com>
31498 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31499 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31501 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31503 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31505 2017-01-25 Richard Biener <rguenther@suse.de>
31507 PR tree-optimization/69264
31508 * target.def (vector_alignment_reachable): Improve documentation.
31509 * doc/tm.texi: Regenerate.
31510 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31512 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31513 earlier changes with respect to TYPE_USER_ALIGN.
31514 (vector_alignment_reachable_p): Likewise. Improve dumping.
31516 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31519 * config/arm/arm.md (xordi3): Force constant operand into a register
31522 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31524 * doc/invoke.texi (-fstore-merging): Correct default optimization
31525 levels at which it is enabled.
31526 (-O): Move -fstore-merging from list to...
31529 2017-01-25 Richard Biener <rguenther@suse.de>
31532 * omp-expand.c: Include debug.h.
31533 (expand_omp_taskreg): Make sure to generate early debug before
31534 outlining anything from a function.
31535 (expand_omp_target): Likewise.
31536 (grid_expand_target_grid_body): Likewise.
31538 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31541 * asan.c (get_translation_unit_decl): New function.
31542 (asan_add_global): Extract modules file name from globals
31543 TRANSLATION_UNIT_DECL name.
31545 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31548 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31549 for long calls with APCS frame and VFP.
31551 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31553 * cfg.c (original_copy_tables_initialized_p): New function.
31554 * cfg.h (original_copy_tables_initialized_p): New decl.
31555 * cfgrtl.c (relink_block_chain): Guard the call to
31556 free_original_copy_tables with a call to
31557 original_copy_tables_initialized_p.
31558 * cgraph.h (symtab_node::native_rtl_p): New decl.
31559 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31560 (symtab_node::needed_p): Don't assert for early assembly output
31561 for __RTL functions.
31562 (cgraph_node::finalize_function): Set "force_output" for __RTL
31564 (cgraph_node::analyze): Bail out early for __RTL functions.
31565 (analyze_functions): Update assertion to support __RTL functions.
31566 (cgraph_node::expand): Bail out early for __RTL functions.
31567 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31569 * function.h (struct function): Update comment for field
31571 * gimple-expr.c: Include "tree-pass.h".
31572 (gimple_has_body_p): Return false for __RTL functions.
31573 * Makefile.in (OBJS): Add run-rtl-passes.o.
31574 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31576 (gcc::pass_manager::get_clean_slate): New accessor.
31577 * passes.c: Include "insn-addr.h".
31578 (should_skip_pass_p): Add logging. Update logic for running
31579 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31580 property-provider override so it is only done for gimple passes.
31582 (skip_pass): New function.
31583 (execute_one_pass): Call skip_pass when skipping passes.
31584 * read-md.c (md_reader::read_char): Support filtering
31585 the input to a subset of line numbers.
31586 (md_reader::md_reader): Initialize fields
31587 m_first_line and m_last_line.
31588 (md_reader::read_file_fragment): New function.
31589 * read-md.h (md_reader::read_file_fragment): New decl.
31590 (md_reader::m_first_line): New field.
31591 (md_reader::m_last_line): New field.
31592 * read-rtl-function.c (function_reader::create_function): Only
31593 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31594 curr_properties. Set DECL_INITIAL to a dummy block.
31595 (read_rtl_function_body_from_file_range): New function.
31596 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31598 * run-rtl-passes.c: New file.
31599 * run-rtl-passes.h: New file.
31601 2017-01-24 Jeff Law <law@redhat.com>
31603 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31606 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31608 PR tree-optimization/79159
31609 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31610 (record_nonwrapping_iv): Improve boundary using above function if no
31611 value range information.
31613 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31614 Martin Jambor <mjambor@suse.cz>
31616 * brig-builtins.def: New file.
31617 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31618 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31619 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31620 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31621 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31622 * builtin-types.def (BT_INT8): New.
31623 (BT_INT16): Likewise.
31624 (BT_UINT8): Likewise.
31625 (BT_UINT16): Likewise.
31626 (BT_FN_ULONG): Likewise.
31627 (BT_FN_UINT_INT): Likewise.
31628 (BT_FN_UINT_ULONG): Likewise.
31629 (BT_FN_UINT_LONG): Likewise.
31630 (BT_FN_UINT_PTR): Likewise.
31631 (BT_FN_ULONG_PTR): Likewise.
31632 (BT_FN_INT8_FLOAT): Likewise.
31633 (BT_FN_INT16_FLOAT): Likewise.
31634 (BT_FN_UINT32_FLOAT): Likewise.
31635 (BT_FN_UINT16_FLOAT): Likewise.
31636 (BT_FN_UINT8_FLOAT): Likewise.
31637 (BT_FN_UINT64_FLOAT): Likewise.
31638 (BT_FN_UINT16_UINT32): Likewise.
31639 (BT_FN_UINT32_UINT16): Likewise.
31640 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31641 (BT_FN_INT_PTR_INT): Likewise.
31642 (BT_FN_UINT_PTR_UINT): Likewise.
31643 (BT_FN_LONG_PTR_LONG): Likewise.
31644 (BT_FN_ULONG_PTR_ULONG): Likewise.
31645 (BT_FN_VOID_UINT64_UINT64): Likewise.
31646 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31647 (BT_FN_INT8_INT8_INT8): Likewise.
31648 (BT_FN_INT16_INT16_INT16): Likewise.
31649 (BT_FN_INT_INT_INT): Likewise.
31650 (BT_FN_UINT_FLOAT_UINT): Likewise.
31651 (BT_FN_FLOAT_UINT_UINT): Likewise.
31652 (BT_FN_ULONG_UINT_UINT): Likewise.
31653 (BT_FN_ULONG_UINT_PTR): Likewise.
31654 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31655 (BT_FN_UINT_UINT_UINT): Likewise.
31656 (BT_FN_VOID_UINT_PTR): Likewise.
31657 (BT_FN_UINT_UINT_PTR: Likewise.
31658 (BT_FN_UINT32_UINT64_PTR): Likewise.
31659 (BT_FN_INT_INT_UINT_UINT): Likewise.
31660 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31661 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31662 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31663 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31664 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31665 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31666 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31667 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31668 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31669 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31670 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31671 * doc/frontends.texi: List BRIG FE.
31672 * doc/install.texi (Testing): Add BRIG tesring requirements.
31673 * doc/invoke.texi (Overall Options): Mention BRIG.
31674 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31676 2017-01-24 Richard Biener <rguenther@suse.de>
31678 PR translation/79208
31679 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31681 2017-01-24 Martin Jambor <mjambor@suse.cz>
31684 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31685 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31686 and known_contexts.
31688 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31690 PR middle-end/79123
31691 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31692 casts from signed to unsigned really don't have a range.
31694 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31696 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31697 GMP_RNDx for compatiblity.
31699 2017-01-24 Martin Liska <mliska@suse.cz>
31702 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31703 that would prevent us to call alloca with -1 as argument.
31705 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31707 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31708 Avoid -Wformat-security warning.
31710 2017-01-23 Andrew Pinski <apinski@cavium.com>
31712 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31715 2017-01-23 Martin Sebor <msebor@redhat.com>
31717 PR middle-end/78703
31718 * gimple-ssa-sprintf.c (warn_level): New global.
31719 (format_integer): Use it here and throughout the rest of the file.
31720 Use the same switch to compute sign as base.
31721 (maybe_warn): New function.
31722 (format_directive): Factor out warnings into maybe_warn.
31723 Add debugging output. Use warn_level.
31724 (add_bytes): Use warn_level.
31725 (pass_sprintf_length::compute_format_length): Add debugging output.
31726 (try_substitute_return_value): Same.
31727 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31729 PR middle-end/78703
31730 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31731 (struct fmtresult, format_integer, format_floating): Adjust.
31732 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31733 (get_string_length, format_string,format_directive): Same.
31734 (pass_sprintf_length::compute_format_length): Same.
31735 (try_substitute_return_value): Simplify slightly.
31737 PR middle-end/78703
31738 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31739 (fmtresult::operator+=): Outlined.
31740 (struct fmtresult): Add ctors.
31741 (struct conversion_spec): Rename...
31742 (struct directive): ...to this. Add and remove data members.
31743 (directive::set_width, directive::set_precision): New functions.
31744 (format_percent): Use fmtresult ctor.
31745 (get_width_and_precision): Remove.
31746 (format_integer): Make naming changes. Avoid computing width and
31748 (format_floating): Same. Adjust indentation.
31749 (format_character, format_none): New functions.
31750 (format_string): Moved character handling to format_character.
31751 (format_directive): Remove arguments, change return type.
31752 (parse_directive): New function.
31753 (pass_sprintf_length::compute_format_length): Move directive
31754 parsing to parse_directive.
31756 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31758 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31759 (assign_assembler_name_if_needed): ... this.
31760 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31761 (assign_assembler_name_if_needed): ... this.
31762 (free_lang_data_in_cgraph): Adjust callers.
31763 * cgraphunit.c (cgraph_node::analyze): Likewise.
31764 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31767 2017-01-23 Richard Biener <rguenther@suse.de>
31769 PR tree-optimization/79088
31770 PR tree-optimization/79188
31771 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31772 resetting loop bounds after last path deletion. Reset loop
31773 bounds of the target loop, make code match the comments.
31774 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31775 Make sure loops need no fixups.
31777 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31779 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31780 exponent support with double type for first argument.
31781 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31782 type returned by __builtin_vec_extract_sig,
31783 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31784 functions from "vector int" to "vector unsigned int" or from
31785 "vector long long int" to "vector unsigned long long int".
31786 Changed type returned by __builtin_vec_extract_exp,
31787 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31788 functions from "vector int" to "vector unsigned int" or from
31789 "vector long long int" to "vector unsigned long long int".
31790 Changed return type of __builtin_vec_test_data_class,
31791 __builtin_vec_test_data_class_sp, and
31792 __builtin_vec_test_data_class_dp from "vector int" to
31793 "vector bool int" or from "vector long long int" to "vector bool
31794 long long int" and changed second argument type from "unsigned
31795 int" to "int". Added new overloaded function forms "vector float
31796 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31797 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31798 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31799 double, vector unsigned long long int)" and "vector double
31800 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31801 long int)". Changed return type of
31802 __builtin_scalar_test_data_class and
31803 __builtin_scalar_test_data_class_sp and
31804 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31805 int" and changed second argument from "unsigned int" to "int".
31806 Changed type returned by __builtin_scalar_test_neg,
31807 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31808 from "int" to "bool int". Added new overloaded function form
31809 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31810 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31811 exponent double-precision with floating point first argument.
31812 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31813 documentation of scalar_test_data_class, scalar_test_neg,
31814 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31815 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31816 vec_test_data_class built-in functions to reflect refinements in
31817 their type signatures.
31819 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31821 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31823 (aarch64_elf_asm_destructor): Likewise.
31825 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31827 PR rtl-optimization/78634
31828 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31829 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31830 * ifcvt.c (noce_try_cmove): Add missing cost check.
31832 PR rtl-optimization/71724
31833 * combine.c (if_then_else_cond): Look for situations where it is
31834 beneficial to undo the work of one of the recursive calls.
31836 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31838 PR tree-optimization/70754
31839 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31840 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31841 combined stmt before it if not NULL.
31842 (combine_chains): Process refs reversely and compute dominance point
31845 2017-01-23 Martin Liska <mliska@suse.cz>
31847 PR tree-optimization/79196
31848 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31849 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31851 (strlen_optimize_stmt): Call the renamed function.
31853 2017-01-23 Michael Matz <matz@suse.de>
31855 PR tree-optimization/78384
31856 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31858 2017-01-23 Richard Biener <rguenther@suse.de>
31860 PR tree-optimization/79186
31861 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31862 both incoming edges before moving an assert.
31864 2017-01-23 Martin Jambor <mjambor@suse.cz>
31866 * ipa-prop.c (load_from_param_1): Removed.
31867 (load_from_unmodified_param): Bits from load_from_param_1 put back
31869 (load_from_param): Removed.
31870 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31871 with stmt. Reverted back to use of load_from_unmodified_param.
31873 2017-01-23 Martin Jambor <mjambor@suse.cz>
31876 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31877 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31878 field a pointer to garbage collected vector, mark lattices and
31879 ipcp_orig_node with GTY((skip)).
31880 (ipa_get_param_count): Adjust to descriptors being a pointer.
31881 (ipa_get_param): Likewise.
31882 (ipa_get_type): Likewise.
31883 (ipa_get_param_move_cost): Likewise.
31884 (ipa_set_param_used): Likewise.
31885 (ipa_get_controlled_uses): Likewise.
31886 (ipa_set_controlled_uses): Likewise.
31887 (ipa_is_param_used): Likewise.
31888 (ipa_node_params_t): Move into garbage collector. New methods insert
31890 (ipa_node_params_sum): Annotate wth GTY(()).
31891 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31893 (ipa_load_from_parm_agg): Adjust declaration.
31894 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31895 * ipa-profile.c (ipa_profile): Likewise.
31896 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31897 (ipa_populate_param_decls): Make descriptors parameter garbage
31899 (ipa_dump_param): Adjust to descriptors being a pointer.
31900 (ipa_alloc_node_params): Likewise.
31901 (ipa_initialize_node_params): Likewise.
31902 (load_from_param_1): Make descriptors parameter garbage collected.
31903 (load_from_unmodified_param): Likewise.
31904 (load_from_param): Likewise.
31905 (ipa_load_from_parm_agg): Likewise.
31906 (ipa_node_params::~ipa_node_params): Removed.
31907 (ipa_free_all_node_params): Remove call to delete operator.
31908 (ipa_node_params_t::insert): New.
31909 (ipa_node_params_t::remove): Likewise.
31910 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31911 copy known_csts and known_contexts vectors.
31912 (ipa_read_node_info): Adjust to descriptors being a pointer.
31913 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31915 (ipcp_transform_function): Make descriptors variable garbage
31918 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31920 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31921 * config/i386/avx512dqintrin.h: Ditto.
31922 * config/i386/avx512fintrin.h: Ditto.
31923 * config/i386/i386.c: Handle new builtins.
31924 * config/i386/i386-builtin.def: Add new builtins.
31925 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31926 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31928 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31929 Martin Liska <mliska@suse.cz>
31931 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31932 * asan.c (asan_expand_poison_ifn): Support stores and use
31933 appropriate ASAN report function.
31934 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31935 * internal-fn.def (ASAN_POISON_USE): Declare.
31936 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31937 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31938 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31939 ASAN_POISON calls w/o LHS.
31940 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31941 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31942 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31943 * gimplify.c (asan_poison_variables): Add attribute
31944 use_after_scope_memory to variables that really needs to live
31946 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31947 having the attribute.
31949 2017-01-23 Martin Liska <mliska@suse.cz>
31951 * asan.c (create_asan_shadow_var): New function.
31952 (asan_expand_poison_ifn): Likewise.
31953 * asan.h (asan_expand_poison_ifn): New declaration.
31954 * internal-fn.c (expand_ASAN_POISON): Likewise.
31955 * internal-fn.def (ASAN_POISON): New builtin.
31956 * sanopt.c (pass_sanopt::execute): Expand
31957 asan_expand_poison_ifn.
31958 * tree-inline.c (copy_decl_for_dup_finish): Make function
31960 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31961 * tree-ssa.c (is_asan_mark_p): New function.
31962 (execute_update_addresses_taken): Rewrite local variables
31963 (identified just by use-after-scope as addressable) into SSA.
31965 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31967 * doc/install.texi (Specific): opensource.apple.com uses https
31968 now. Remove trailing slash.
31970 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31972 * README.Portability: Remove note on an Irix compatibility issue.
31974 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31976 * gcov.c (INCLUDE_ALGORITHM): Define.
31977 (INCLUDE_VECTOR): Define.
31978 No longer include <vector> and <algorithm> directly.
31980 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31982 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31984 * doc/invoke.texi (Code Gen Options): Ditto.
31986 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31989 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31991 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31993 rtl-optimization/79125
31994 * cprop.c (local_cprop_pass): Handle cases where we make an
31995 unconditional trap.
31997 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
32001 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
32002 read from, for big endian.
32004 2017-01-20 Jiong Wang <jiong.wang@arm.com>
32006 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
32007 register pauth builtins for LP64 only.
32009 2017-01-20 Marek Polacek <polacek@redhat.com>
32012 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
32015 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32017 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
32019 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
32020 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
32021 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
32023 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32026 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32027 in big-endian mode when they are not a single duplicated value.
32029 2017-01-20 Richard Biener <rguenther@suse.de>
32031 * BASE-VER: Bump to 7.0.1.
32033 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32035 * omp-low.c (omplow_simd_context): New struct. Use it...
32036 (lower_rec_simd_input_clauses): ...here and...
32037 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32038 references to idx, lane, max_vf, is_simt.
32040 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32042 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32045 2017-01-20 Martin Jambor <mjambor@suse.cz>
32047 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32048 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32050 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32051 (GTFILES): Rename hsa.c to hsa-common.c.
32052 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32053 * hsa-dump.c: Likewise.
32054 * hsa-gen.c: Likewise.
32055 * hsa-regalloc.c: Likewise.
32056 * ipa-hsa.c: Likewise.
32057 * omp-expand.c: Likewise.
32058 * omp-low.c: Likewise.
32059 * toplev.c: Likewise.
32061 2017-01-20 Marek Polacek <polacek@redhat.com>
32064 * doc/invoke.texi: Document -Wduplicated-branches.
32065 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32066 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32067 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32068 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32069 return 0 only when not OEP_LEXICOGRAPHIC.
32070 (fold_build_cleanup_point_expr): Use the expression
32071 location when building CLEANUP_POINT_EXPR.
32072 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32073 * tree.c (add_expr): Handle error_mark_node.
32075 2017-01-20 Martin Liska <mliska@suse.cz>
32078 * tree-profile.c (init_ic_make_global_vars): Do not call
32080 (gimple_init_gcov_profiler): Likewise.
32082 2017-01-20 Martin Liska <mliska@suse.cz>
32085 * cgraph.h (maybe_create_reference): Remove argument and
32087 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32089 * ipa-cp.c (create_specialized_node): Likewise.
32090 * symtab.c (symtab_node::maybe_create_reference): Handle
32091 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32093 2017-01-20 Martin Liska <mliska@suse.cz>
32095 * read-rtl-function.c (function_reader::create_function): Use
32096 build_decl instread of build_decl_stat.
32098 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32100 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32101 * config/i386/avx512dqintrin.h: Ditto.
32102 * config/i386/avx512fintrin.h: Ditto.
32103 * config/i386/i386-builtin-types.def: Add new types.
32104 * config/i386/i386.c: Handle new types.
32105 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32106 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32107 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32108 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32109 (__builtin_ia32_kshiftridi): New.
32110 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32112 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32116 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32117 define to rs6000_init_stack_protect_guard.
32118 (rs6000_init_stack_protect_guard): New function.
32120 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32121 Yunqiang Su <yunqiang.su@imgtec.com>
32123 * config.gcc (supported_defaults): Add madd4.
32124 (with_madd4): Add validation.
32125 (all_defaults): Add madd4.
32126 * config/mips/mips.opt (mmadd4): New option.
32127 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32129 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32131 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32132 (ISA_HAS_FUSED_MADD4): Likewise.
32133 * doc/invoke.texi (-mmadd4): Document the new option.
32134 * doc/install.texi (--with-madd4): Document the new option.
32136 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32138 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32139 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32140 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32141 (aarch64_init_pauth_hint_builtins): New.
32142 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32143 (aarch64_expand_builtin): Expand new builtins.
32145 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32147 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32148 * combine-stack-adj.c (no_unhandled_cfa): Handle
32149 REG_CFA_TOGGLE_RA_MANGLE.
32150 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32151 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32152 info for return address signing.
32153 (aarch64_expand_epilogue): Likewise.
32155 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32157 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32158 * config/aarch64/aarch64-protos.h
32159 (aarch64_return_address_signing_enabled): New declaration.
32160 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32162 (aarch64_expand_prologue): Sign return address before it's pushed onto
32164 (aarch64_expand_epilogue): Authenticate return address fetched from
32166 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32167 (aarch64_attributes): New function attributes for "sign-return-address".
32168 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32169 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32170 ("*do_return"): Generate combined instructions according to key index.
32171 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32172 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32174 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32175 * config/aarch64/aarch64.opt (msign-return-address=): New.
32176 * doc/extend.texi (AArch64 Function Attributes): Documents
32177 "sign-return-address=".
32178 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32180 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32182 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32183 overall option summary.
32185 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32187 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32188 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32189 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32190 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32192 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32194 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32195 -mpower9-minmax by default for -mcpu=power9.
32196 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32197 128-bit floating point.
32199 2017-01-20 Alan Modra <amodra@gmail.com>
32201 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32202 optimizing for size.
32204 2017-01-20 Alan Modra <amodra@gmail.com>
32207 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32208 for strcmp and strncmp from corresponding builtin decl.
32210 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32212 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32213 instead of i386/rtems-64.h.
32214 * config/i386/rtems-64.h: Remove.
32216 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32220 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32222 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32224 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32226 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32227 Change int to HOST_WIDE_INT.
32228 * config/aarch64/aarch64-protos.h
32229 (aarch64_simd_gen_const_vector_dup): Likewise.
32230 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32232 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32234 * langhooks-def.h (lhd_type_for_size): New decl.
32235 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32236 * langhooks.c (lhd_type_for_size): New function, taken from
32239 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32241 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32242 define_bypass for CR latency.
32243 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32244 (power9-alu2): Add define_bypass for CR latency.
32246 (power9-mul): Update insn latency.
32247 (power9-mul-compare): Update insn latency, bypass latency and remove
32250 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32252 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32254 * config/aarch64/aarch64.md
32255 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32256 aarch64_nopcrelative_literal_loads.
32257 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32259 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32261 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32262 TARGET_LOONGSON_3A.
32263 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32265 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32268 * config.gcc (supported_defaults): Add lxc1-sxc1.
32269 (with_lxc1_sxc1): Add validation.
32270 (all_defaults): Add lxc1-sxc1.
32271 * config/mips/mips.opt (mlxc1-sxc1): New option.
32272 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32274 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32275 __mips_no_lxc1_sxc1.
32276 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32277 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32278 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32280 2017-01-19 Richard Biener <rguenther@suse.de>
32282 PR tree-optimization/72488
32283 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32284 sure to restore SSA info.
32285 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32287 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32289 PR rtl-optimization/79121
32290 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32291 of the inner type when shifting an extended value.
32293 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32296 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32297 interposable aliases.
32299 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32302 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32303 Use the evmergelohi instruction.
32304 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32305 (mov_sitf_e500_subreg8_2_be): Likewise.
32306 (mov_sitf_e500_subreg12_2_le): Likewise.
32307 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32308 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32309 (mov_sitf_e500_subreg8_2_le): Likewise.
32310 (mov_sitf_e500_subreg12_2_be): Likewise.
32312 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32314 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32315 attribute from vecsimple to vecperm.
32316 (altivec_vbpermq2): Likewise.
32318 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32321 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32323 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32324 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32325 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32326 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32327 case where N arg is SIZE_MAX.
32328 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32329 (cmpstrsi): Add pattern.
32331 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32333 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32334 __builtin_vec_revb builtins.
32335 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32336 built-in functions to support generation of the ISA 3.0 XXBR<x>
32337 vector byte reverse instructions.
32338 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32339 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32340 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32341 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32342 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32343 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32344 (P9V_BUILTIN_VEC_REVB): Likewise.
32345 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32346 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32347 (p9_xxbrq_v16qi): Likewise.
32348 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32349 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32350 (p9_xxbrh_v8hi): Likewise.
32351 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32352 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32353 vec_revb built-in functions.
32355 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32357 PR rtl-optimization/78952
32358 * config/i386/i386.md (any_extract): New code iterator.
32359 (*insvqi_2): Use any_extract for source operand.
32360 (*insvqi_3): Use any_shiftrt for source operand.
32362 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32364 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32366 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32368 2017-01-18 Matthias Klose <doko@ubuntu.com>
32370 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32372 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32374 * config/rs6000/altivec.h (vec_bperm): Change #define.
32375 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32376 (altivec_vbpermq2): New define_insn.
32377 (altivec_vbpermd): Likewise.
32378 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32379 function interface.
32380 (VBPERMD): Likewise.
32381 (VBPERM): New polymorphic function interface.
32382 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32383 Add entries for P9V_BUILTIN_VEC_VBPERM.
32384 * doc/extend.texi: Add interfaces for vec_bperm.
32386 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32388 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32389 first letter of error messages.
32390 (s390_resolve_overloaded_builtin): Likewise.
32391 * config/s390/s390.c (s390_expand_builtin): Likewise.
32392 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32393 (s390_valid_target_attribute_inner_p): Likewise.
32394 * config/s390/s390.md ("tabort"): Likewise.
32396 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32398 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32399 (ISA_AVOID_DIV_HILO): New macro.
32400 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32401 (ISA_HAS_DDIV): Likewise.
32403 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32405 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32407 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32409 * doc/invoke.texi (fabi-version): Spelling fix.
32411 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32414 * doc/invoke.texi (fabi-version): Mention mangling fix for
32417 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32420 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32422 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32425 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32426 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32428 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32430 (stack_protect_test): Ditto.
32431 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32432 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32434 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32435 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32436 -mstack-protector-guard-offset=.
32437 (RS/6000 and PowerPC Options): Ditto.
32439 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32441 * config/i386/i386.h (MASK_CLASS_P): New define.
32442 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32443 there are no registers from different register sets also when
32444 mask registers are used. Update function comment.
32445 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32446 to (*k/*r) and (*k/*km) alternatives.
32448 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32450 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32451 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32452 (EH_RETURN_HANDLER_RTX): New define.
32453 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32454 Force frame pointer in EH return functions.
32455 (aarch64_expand_epilogue): Add barrier for eh_return.
32456 (aarch64_final_eh_return_addr): Remove.
32457 (aarch64_eh_return_handler_rtx): New function.
32458 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32460 (aarch64_eh_return_handler_rtx): New prototype.
32462 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32464 * config/rs6000/altivec.h (vec_rlmi): New #define.
32465 (vec_vrlnm): Likewise.
32466 (vec_rlnm): Likewise.
32467 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32468 (UNSPEC_VRLNM): Likewise.
32469 (VIlong): New mode iterator.
32470 (altivec_vrl<VI_char>mi): New define_insn.
32471 (altivec_vrl<VI_char>nm): Likewise.
32472 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32474 (VRLDNM): Likewise.
32475 (RLNM): New polymorphic function entry.
32476 (VRLWMI): New monomorphic function entry.
32477 (VRLDMI): Likewise.
32478 (RLMI): New polymorphic function entry.
32479 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32480 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32481 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32484 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32487 * dwarf2out.c (field_byte_offset): Restore the
32488 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32489 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32490 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32493 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32496 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32498 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32501 * dwarf2out.c (add_data_member_location_attribute): For constant
32502 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32503 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32504 DW_AT_byte_size attributes.
32506 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32508 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32509 after forcing to constant memory when the code model is medium.
32511 2017-01-17 Julia Koval <julia.koval@intel.com>
32514 * config/i386/avx512fintrin.h
32515 (_mm512_i32gather_ps): Change __addr type to void const*.
32516 (_mm512_mask_i32gather_ps): Ditto.
32517 (_mm512_i32gather_pd): Ditto.
32518 (_mm512_mask_i32gather_pd): Ditto.
32519 (_mm512_i64gather_ps): Ditto.
32520 (_mm512_mask_i64gather_ps): Ditto.
32521 (_mm512_i64gather_pd): Ditto.
32522 (_mm512_mask_i64gather_pd): Ditto.
32523 (_mm512_i32gather_epi32): Ditto.
32524 (_mm512_mask_i32gather_epi32): Ditto.
32525 (_mm512_i32gather_epi64): Ditto.
32526 (_mm512_mask_i32gather_epi64): Ditto.
32527 (_mm512_i64gather_epi32): Ditto.
32528 (_mm512_mask_i64gather_epi32): Ditto.
32529 (_mm512_i64gather_epi64): Ditto.
32530 (_mm512_mask_i64gather_epi64): Ditto.
32531 (_mm512_i32scatter_ps): Change __addr type to void*.
32532 (_mm512_mask_i32scatter_ps): Ditto.
32533 (_mm512_i32scatter_pd): Ditto.
32534 (_mm512_mask_i32scatter_pd): Ditto.
32535 (_mm512_i64scatter_ps): Ditto.
32536 (_mm512_mask_i64scatter_ps): Ditto.
32537 (_mm512_i64scatter_pd): Ditto.
32538 (_mm512_mask_i64scatter_pd): Ditto.
32539 (_mm512_i32scatter_epi32): Ditto.
32540 (_mm512_mask_i32scatter_epi32): Ditto.
32541 (_mm512_i32scatter_epi64): Ditto.
32542 (_mm512_mask_i32scatter_epi64): Ditto.
32543 (_mm512_i64scatter_epi32): Ditto.
32544 (_mm512_mask_i64scatter_epi32): Ditto.
32545 (_mm512_i64scatter_epi64): Ditto.
32546 (_mm512_mask_i64scatter_epi64): Ditto.
32547 * config/i386/avx512pfintrin.h
32548 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32549 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32550 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32551 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32552 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32553 (_mm512_prefetch_i32scatter_ps): Ditto.
32554 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32555 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32556 (_mm512_prefetch_i64scatter_pd): Ditto.
32557 (_mm512_prefetch_i64scatter_ps): Ditto.
32558 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32559 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32560 * config/i386/avx512vlintrin.h
32561 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32562 (_mm_mmask_i32gather_ps): Ditto.
32563 (_mm256_mmask_i32gather_pd): Ditto.
32564 (_mm_mmask_i32gather_pd): Ditto.
32565 (_mm256_mmask_i64gather_ps): Ditto.
32566 (_mm_mmask_i64gather_ps): Ditto.
32567 (_mm256_mmask_i64gather_pd): Ditto.
32568 (_mm_mmask_i64gather_pd): Ditto.
32569 (_mm256_mmask_i32gather_epi32): Ditto.
32570 (_mm_mmask_i32gather_epi32): Ditto.
32571 (_mm256_mmask_i32gather_epi64): Ditto.
32572 (_mm_mmask_i32gather_epi64): Ditto.
32573 (_mm256_mmask_i64gather_epi32): Ditto.
32574 (_mm_mmask_i64gather_epi32): Ditto.
32575 (_mm256_mmask_i64gather_epi64): Ditto.
32576 (_mm_mmask_i64gather_epi64): Ditto.
32577 (_mm256_i32scatter_ps): Change __addr type to void*.
32578 (_mm256_mask_i32scatter_ps): Ditto.
32579 (_mm_i32scatter_ps): Ditto.
32580 (_mm_mask_i32scatter_ps): Ditto.
32581 (_mm256_i32scatter_pd): Ditto.
32582 (_mm256_mask_i32scatter_pd): Ditto.
32583 (_mm_i32scatter_pd): Ditto.
32584 (_mm_mask_i32scatter_pd): Ditto.
32585 (_mm256_i64scatter_ps): Ditto.
32586 (_mm256_mask_i64scatter_ps): Ditto.
32587 (_mm_i64scatter_ps): Ditto.
32588 (_mm_mask_i64scatter_ps): Ditto.
32589 (_mm256_i64scatter_pd): Ditto.
32590 (_mm256_mask_i64scatter_pd): Ditto.
32591 (_mm_i64scatter_pd): Ditto.
32592 (_mm_mask_i64scatter_pd): Ditto.
32593 (_mm256_i32scatter_epi32): Ditto.
32594 (_mm256_mask_i32scatter_epi32): Ditto.
32595 (_mm_i32scatter_epi32): Ditto.
32596 (_mm_mask_i32scatter_epi32): Ditto.
32597 (_mm256_i32scatter_epi64): Ditto.
32598 (_mm256_mask_i32scatter_epi64): Ditto.
32599 (_mm_i32scatter_epi64): Ditto.
32600 (_mm_mask_i32scatter_epi64): Ditto.
32601 (_mm256_i64scatter_epi32): Ditto.
32602 (_mm256_mask_i64scatter_epi32): Ditto.
32603 (_mm_i64scatter_epi32): Ditto.
32604 (_mm_mask_i64scatter_epi32): Ditto.
32605 (_mm256_i64scatter_epi64): Ditto.
32606 (_mm256_mask_i64scatter_epi64): Ditto.
32607 (_mm_i64scatter_epi64): Ditto.
32608 (_mm_mask_i64scatter_epi64): Ditto.
32609 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32610 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32611 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32612 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32613 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32614 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32615 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32616 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32617 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32618 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32619 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32620 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32621 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32622 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32623 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32624 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32625 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32626 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32627 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32628 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32629 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32630 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32631 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32632 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32633 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32634 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32635 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32636 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32637 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32638 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32639 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32640 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32641 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32642 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32643 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32644 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32645 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32646 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32647 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32648 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32649 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32650 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32651 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32652 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32653 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32654 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32655 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32656 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32657 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32658 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32659 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32660 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32661 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32662 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32663 definitions accordingly.
32665 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32666 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32669 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32672 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32675 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32676 hard regs for inner regno.
32678 2017-01-17 Martin Liska <mliska@suse.cz>
32681 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32682 assumption and add comment.
32684 2017-01-17 Nathan Sidwell <nathan@acm.org>
32686 * ipa-visibility.c (localize_node): New function, broken out of ...
32687 (function_and_variable_visibility): ... here. Call it.
32689 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32691 PR middle-end/77445
32692 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32693 correctly set frequency of oudgoing edge.
32694 (duplicate_thread_path): Fix profile updating.
32696 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32699 * configure.ac: Add GCC_BASE_VER.
32700 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32701 version from BASE-VER file.
32702 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32703 (gcc.o): Depend on $(BASEVER).
32704 * common.opt (dumpfullversion): New option.
32705 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32706 * doc/invoke.texi: Document -dumpfullversion.
32707 * doc/install.texi: Document --with-gcc-major-version-only.
32708 * configure: Regenerated.
32710 2017-01-17 Richard Biener <rguenther@suse.de>
32712 PR tree-optimization/71433
32713 * tree-vrp.c (register_new_assert_for): Merge same asserts
32714 on all incoming edges.
32715 (process_assert_insertions_for): Handle insertions at the
32718 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32720 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32721 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32723 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32726 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32729 2017-01-17 Alan Modra <amodra@gmail.com>
32732 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32733 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32734 symbolic stack limit when pic.
32736 2017-01-16 Martin Sebor <msebor@redhat.com>
32738 PR tree-optimization/78608
32739 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32741 2017-01-16 Jeff Law <law@redhat.com>
32744 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32745 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32746 for several include directories that may be relative to sysroot.
32747 * config/i386/x-mingw32 (gplus_includedir): Define.
32748 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32749 (native_system_includedir): Likewise.
32750 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32751 override if TARGET_SYSTEM_ROOT is defined.
32752 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32754 PR tree-optimization/79090
32755 PR tree-optimization/33562
32756 PR tree-optimization/61912
32757 PR tree-optimization/77485
32758 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32759 and computed trims into the dump file.
32761 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32763 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32765 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32768 * gimplify.c (gimplify_init_constructor): If want_value and
32769 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32773 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32774 sequence. Formatting fixes.
32775 (doloop_optimize): Formatting fixes.
32778 * gcc.c (debug_level_greater_than_spec_func): New function.
32779 (static_spec_functions): Add debug-level-gt spec function.
32780 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32782 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32783 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32784 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32785 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32786 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32787 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32789 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32791 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32792 QImode fixups to general and mask registers only.
32794 2017-01-16 Carl Love <cel@us.ibm.com>
32796 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32797 for built-in functions
32798 vector signed char vec_nabs (vector signed char)
32799 vector signed short vec_nabs (vector signed short)
32800 vector signed int vec_nabs (vector signed int)
32801 vector signed long long vec_nabs (vector signed long long)
32802 vector float vec_nabs (vector float)
32803 vector double vec_nabs (vector double)
32804 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32806 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32807 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32808 * doc/extend.texi: Update the documentation file for the new built-in
32811 2017-01-16 Martin Sebor <msebor@redhat.com>
32813 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32816 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32818 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32819 UNSPEC_VSX__XXSPLTD to require special splat handling.
32821 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32824 * system.h: Poison strndup.
32826 2017-01-16 Alan Modra <amodra@gmail.com>
32829 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32832 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32834 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32836 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32838 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32839 call recog here. Assert that INSN_CODE (insn) is non-negative.
32841 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32844 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32846 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32847 in the currently scheduled RTL fragment.
32849 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32851 PR rtl-optimization/78751
32852 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32855 2017-01-14 Jeff Law <law@redhat.com>
32857 PR tree-optimization/79090
32858 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32859 variable length stores.
32860 (compute_trims): Delete dead assignment to *trim_tail.
32861 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32864 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32866 PR rtl-optimization/78626
32867 PR rtl-optimization/78727
32868 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32869 of a block, and split such blocks after everything else is finished.
32871 2017-01-14 Alan Modra <amodra@gmail.com>
32874 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32875 target legitimate_combined_insn.
32876 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32877 (rs6000_legitimate_combined_insn): New function.
32878 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32880 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32881 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32882 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32884 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32886 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32888 2017-01-13 Jeff Law <law@redhat.com>
32890 PR tree-optimization/33562
32891 PR tree-optimization/61912
32892 PR tree-optimization/77485
32893 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32895 (delete_dead_assignment): Likewise.
32896 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32897 statement to delete_dead_call and delete_dead_assignment.
32899 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32902 * substring-locations.c (format_warning_va): Strengthen case 1 so
32903 that both endpoints of the substring must be within the format
32904 range for just the substring to be printed.
32906 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32908 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32909 * config/i386/i386.c (ix86_target_string): Add missing options
32910 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32911 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32912 flags_other and ix86_target_other to flags2_other. Display unknown
32914 (ix86_valid_target_attribute_inner_p): Add missing options and
32915 reorder options by implied ISAs, as in ix86_target_string.
32917 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32919 * hash-table.h (hash_table::too_empty_p): New function.
32920 (hash_table::expand): Use it.
32921 (hash_table::traverse): Likewise.
32922 (hash_table::empty_slot): Use sizeof (value_type) instead of
32923 sizeof (PTR) to convert bytes to elements. Shrink the table
32924 if the current size is excessive for the current number of
32927 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32929 * ira-costs.c (record_reg_classes): Break from the inner loop
32930 early once alt_fail is known to be true. Update outer loop
32931 handling accordingly.
32933 2017-01-13 Jeff Law <law@redhat.com>
32935 * tree-ssa-dse.c (decrement_count): New function.
32936 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32937 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32938 when we know the partially dead statement is a mem* function.
32940 PR tree-optimization/61912
32941 PR tree-optimization/77485
32942 * tree-ssa-dse.c: Include expr.h.
32943 (maybe_trim_constructor_store): New function.
32944 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32946 PR tree-optimization/33562
32947 PR tree-optimization/61912
32948 PR tree-optimization/77485
32949 * doc/invoke.texi: Document new dse-max-object-size param.
32950 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32951 * tree-ssa-dse.c: Include params.h.
32952 (dse_store_status): New enum.
32953 (initialize_ao_ref_for_dse): New, partially extracted from
32955 (valid_ao_ref_for_dse, normalize_ref): New.
32956 (setup_live_bytes_from_ref, compute_trims): Likewise.
32957 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32958 (maybe_trim_partially_dead_store): Likewise.
32959 (maybe_trim_complex_store): Likewise.
32960 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32961 Track what bytes live from the original store. Return tri-state
32962 for dead, partially dead or live.
32963 (dse_dom_walker): Add constructor, destructor and new private members.
32964 (delete_dead_call, delete_dead_assignment): New extracted from
32966 (dse_optimize_stmt): Make a member of dse_dom_walker.
32967 Use initialize_ao_ref_for_dse.
32969 PR tree-optimization/33562
32970 PR tree-optimization/61912
32971 PR tree-optimization/77485
32972 * sbitmap.h (bitmap_count_bits): Prototype.
32973 (bitmap_clear_range, bitmap_set_range): Likewise.
32974 * sbitmap.c (bitmap_clear_range): New function.
32975 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32977 2017-01-13 Martin Liska <mliska@suse.cz>
32980 * function.c (set_cfun): Add new argument force.
32981 * function.h (set_cfun): Likewise.
32982 * ipa-inline-transform.c (inline_call): Use the function when
32983 strict alising from is dropped for function we inline to.
32985 2017-01-13 Richard Biener <rguenther@suse.de>
32987 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32988 for dumping GIMPLE INTEGER_CSTs.
32990 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32992 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32993 to 201112L since C++17.
32995 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32998 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32999 if -fsanitize=kernel-address is present.
33001 2017-01-13 Richard Biener <rguenther@suse.de>
33003 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
33004 as _Literal ( type ) number in case usual suffixes do not
33005 preserve all information.
33007 2017-01-13 Richard Biener <rguenther@suse.de>
33009 PR tree-optimization/77283
33010 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
33011 and ssa-iterators.h.
33012 (is_feasible_trace): Implement a cost model based on joiner
33015 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
33018 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
33019 char or short to __float128/_Float128 directly.
33021 2017-01-12 Martin Sebor <msebor@redhat.com>
33023 to -Wformat-overflow.
33024 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
33025 (min_bytes_remaining): Same.
33026 (get_string_length): Same.
33027 (format_string): Same.
33028 (format_directive): Same.
33030 (pass_sprintf_length::handle_gimple_call): Same.
33032 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33034 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33035 info.nowrite calls with no lhs that can't throw. Return bool
33036 whether gsi_remove has been called or not.
33037 (pass_sprintf_length::handle_gimple_call): Return bool whether
33038 try_substitute_return_value called gsi_remove. Formatting fix.
33039 (pass_sprintf_length::execute): Don't use gsi_remove if
33040 handle_gimple_call returned true.
33043 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33044 be removed due to side-effects, don't remove following barrier nor
33045 turn the successor edge into fallthru edge.
33047 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33050 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33051 element-reversing loads and stores as not swappable.
33053 2017-01-12 Nathan Sidwell <nathan@acm.org>
33054 Nicolai Stange <nicstange@gmail.com>
33056 * combine.c (try_combine): Don't ignore result of overlap checking
33057 loop. Combine overlap & asm check into single loop.
33059 2017-01-12 Richard Biener <rguenther@suse.de>
33061 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33062 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33064 2017-01-12 Richard Biener <rguenther@suse.de>
33066 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33067 and TS_TARGET_OPTION directly derive from TS_BASE.
33068 * tree-core.h (tree_optimization_option): Derive from tree_base.
33069 (tree_target_option): Likewise.
33071 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33073 * config/i386/i386.c (memory_address_length): Increase len
33074 only when rip_relative_addr_p returns false.
33076 2017-01-11 Julia Koval <julia.koval@intel.com>
33078 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33079 (OPTION_MASK_ISA_SGX_SET): New.
33080 (ix86_handle_option): Handle OPT_msgx.
33081 * config.gcc: Added sgxintrin.h.
33082 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33083 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33084 * config/i386/i386.c (ix86_target_string): Add -msgx.
33086 (ix86_option_override_internal): Handle new options.
33087 (ix86_valid_target_attribute_inner_p): Add sgx.
33088 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33089 * config/i386/i386.opt: Add msgx.
33090 * config/i386/sgxintrin.h: New file.
33091 * config/i386/x86intrin.h: Add sgxintrin.h.
33093 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33096 * fold-const.c (maybe_nonzero_address): Return 1 for function
33098 (tree_single_nonzero_warnv_p): Don't handle function local objects
33102 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33105 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33108 * opts.c: Include "spellcheck.h"
33109 (struct string_fragment): New struct.
33110 (struct edit_distance_traits<const string_fragment &>): New
33112 (get_closest_sanitizer_option): New function.
33113 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33115 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33117 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33119 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33120 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33121 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33122 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33123 for initial die_offset if dwarf_split_debug_info.
33124 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33125 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33127 (output_skeleton_debug_sections): Formatting fix. Use
33128 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33129 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33131 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33133 * config/arm/cortex-a53.md: Add bypasses for
33134 cortex_a53_r2f_cvt.
33135 (cortex_a53_r2f): Only use for transfers.
33136 (cortex_a53_f2r): Likewise.
33137 (cortex_a53_r2f_cvt): Add reservation for conversions.
33138 (cortex_a53_f2r_cvt): Likewise.
33140 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33142 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33143 to all inlined functions, change static to extern.
33145 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33148 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33150 (arm_assemble_integer): Likewise.
33152 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33154 * config.gcc: Use new awk script to check CPU, FPU and architecture
33155 parameters for --with-... options.
33156 * config/arm/parsecpu.awk: New file
33157 * config/arm/arm-cpus.in: New file.
33158 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33160 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33162 * config/arm/t-arm: Update dependency rules.
33163 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33164 of processing .def files.
33165 * config/arm/genopt.sh: Deleted.
33166 * config/arm/gentune.sh: Deleted.
33167 * config/arm/arm-cores.def: Deleted.
33168 * config/arm/arm-arches.def: Deleted.
33169 * config/arm/arm-fpus.def: Deleted.
33170 * config/arm/arm-tune.md: Regenerated.
33171 * config/arm/arm-tables.opt: Regenerated.
33172 * config/arm/arm-cpu.h: New generated file.
33173 * config/arm/arm-cpu-data.h: New generated file.
33174 * config/arm/arm-cpu-cdata.h: New generated file.
33176 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33179 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33181 (input_varpool_node): Unpack dynamically_initialized bit.
33183 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33185 PR rtl-optimization/79032
33186 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33187 the alignment of the adjusted memory reference against that of MODE,
33188 instead of the alignment of the original memory reference.
33190 2017-01-11 Martin Jambor <mjambor@suse.cz>
33192 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33194 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33195 decorated functions.
33197 2017-01-11 Richard Biener <rguenther@suse.de>
33199 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33200 set range/nonnull info for PHI results. Do not set it on
33201 stmts marked for removal.
33203 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33205 * expr.c (store_field): In the bitfield case, fetch the return value
33206 from the registers before applying a single big-endian adjustment.
33207 Always do a final load for a BLKmode value not larger than a word.
33209 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33212 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33213 that we correctly handle column numbers greater than
33214 LINE_MAP_MAX_COLUMN_NUMBER.
33216 2017-01-10 Martin Sebor <msebor@redhat.com>
33218 PR middle-end/78245
33219 * gimple-ssa-sprintf.c (get_destination_size): Call
33220 {init,fini}object_sizes.
33221 * tree-object-size.c (addr_object_size): Adjust.
33222 (pass_through_call): Adjust.
33223 (pass_object_sizes::execute): Adjust.
33224 * tree-object-size.h (fini_object_sizes): Declare.
33226 2017-01-10 Martin Sebor <msebor@redhat.com>
33228 PR tree-optimization/78775
33229 * builtins.c (get_size_range): Move...
33230 * calls.c: ...to here.
33231 (alloc_max_size): Accept zero argument.
33232 (operand_signed_p): Remove.
33233 (maybe_warn_alloc_args_overflow): Call get_size_range.
33234 * calls.h (get_size_range): Declare.
33236 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33238 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33239 from TI's devices.csv file as of September 2016.
33240 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33242 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33244 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33245 * doc/invoke.texi: Likewise.
33246 * doc/md.texi: Likewise.
33247 * doc/objc.texi: Likewise.
33249 2017-01-10 Joshua Conner <joshconner@google.com>
33251 * config/arm/fuchsia-elf.h: New file.
33252 * config/fuchsia.h: New file.
33253 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33254 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33256 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33258 2016-01-10 Richard Biener <rguenther@suse.de>
33260 PR tree-optimization/79034
33261 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33262 Propagate out degenerate PHIs in the joiner.
33264 2017-01-10 Martin Liska <mliska@suse.cz>
33266 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33267 (sort_congruence_classes_by_decl_uid): Likewise.
33268 (sort_congruence_class_groups_by_decl_uid): Likewise.
33269 (sem_item_optimizer::merge_classes): Sort class, groups in these
33270 classes and members in the groups by DECL_UID of declarations.
33271 This would make merge operations stable.
33273 2017-01-10 Martin Liska <mliska@suse.cz>
33275 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33276 usage of m_classes_vec.
33277 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33278 (sem_item_optimizer::get_group_by_hash): Likewise.
33279 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33280 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33281 (sem_item_optimizer::verify_classes): Likewise.
33282 (sem_item_optimizer::process_cong_reduction): Likewise.
33283 (sem_item_optimizer::dump_cong_classes): Likewise.
33284 (sem_item_optimizer::merge_classes): Likewise.
33285 * ipa-icf.h (congruence_class_hash): Rename from
33286 congruence_class_group_hash. Remove declaration of m_classes_vec.
33288 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33290 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33291 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33292 * config.gcc: Add avx512vpopcntdqintrin.h.
33293 * config/i386/avx512vpopcntdqintrin.h: New.
33294 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33295 * config/i386/i386-builtin-types.def: Add new types.
33296 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33297 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33298 __builtin_ia32_vpopcountq_v8di_mask): New.
33299 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33300 __AVX512VPOPCNTDQ__.
33301 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33302 (PTA_AVX512VPOPCNTDQ): Define.
33303 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33304 TARGET_AVX512VPOPCNTDQ_P): Define.
33305 * config/i386/i386.opt: Add mavx512vpopcntdq.
33306 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33307 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33309 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33311 PR middle-end/77484
33312 * predict.def (PRED_CALL): Set to 67.
33314 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33316 * expr.c (store_field): In the bitfield case, if the value comes from
33317 a function call and is of an aggregate type returned in registers, do
33318 not modify the field mode; extract the value in all cases if the mode
33319 is BLKmode and the size is not larger than a word.
33321 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33324 * config/i386/cpuid.h: Fix undefined behavior.
33326 2017-01-04 Jeff Law <law@redhat.com>
33328 PR tree-optimization/79007
33329 PR tree-optimization/67955
33330 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33331 conservative for pt.null when flag_non_call_exceptions is on.
33333 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33335 PR translation/79019
33336 PR translation/79020
33337 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33338 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33339 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33341 * config/avr/avr.opt (maccumulate-args): Likewise.
33342 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33343 * common.opt (freport-bug): Likewise.
33344 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33345 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33346 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33347 translatable string.
33348 * config/i386/i386.c (function_value_32): Likewise.
33349 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33350 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33352 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33353 * common/config/msp430/msp430-common.c (msp430_handle_option):
33355 * symtab.c (symtab_node::verify_base): Likewise.
33356 * opts.c (set_debug_level): Likewise.
33357 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33358 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33359 missing whitespace to translatable strings.
33360 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33361 * config/sh/superh.h: Likewise.
33362 * config/i386/xopintrin.h: Likewise.
33363 * config/i386/znver1.md: Likewise.
33364 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33365 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33366 * double-int.h (struct double_int): Likewise.
33367 * double-int.c (div_and_round_double): Likewise.
33368 * wide-int.cc: Likewise.
33369 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33370 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33371 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33372 (crossjumps_occurred): ... this.
33373 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33376 PR tree-optimization/78899
33377 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33378 returning bool return struct loop *, NULL for failure and the new
33380 (versionable_outer_loop_p): Don't version outer loop if it has
33381 dont_vectorized bit set.
33382 (tree_if_conversion): When versioning outer loop, ensure
33383 tree_if_conversion is performed also on the inner loop of the
33384 non-vectorizable outer loop copy.
33385 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33386 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33387 prevent vectorization of it.
33388 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33389 the outer loop vectorization of the non-scalar version is attempted
33390 before vectorization of the inner loop in scalar version. If
33391 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33392 vectorization of its inner loop.
33393 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33394 has 2 inner loops, rename also on edges from bb whose single pred
33395 is outer_loop->header. Fix typo in function comment.
33397 2017-01-09 Martin Sebor <msebor@redhat.com>
33400 * asan.c (asan_emit_stack_protection): Increase local buffer size
33401 to avoid snprintf truncation warning.
33403 2017-01-09 Andrew Pinski <apinski@cavium.com>
33405 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33406 to reference thunderx2t99 for the tuning structure
33407 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33409 (thunderx2t99_extra_costs): This.
33410 * config/aarch64/aarch64-tune.md: Regenerate.
33411 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33412 (vulcan_addrcost_table): This.
33413 (vulcan_regmove_cost): Rename to ...
33414 (thunderx2t99_regmove_cost): This.
33415 (vulcan_vector_cost): Rename to ...
33416 (thunderx2t99_vector_cost): this.
33417 (vulcan_branch_cost): Rename to ...
33418 (thunderx2t99_branch_cost): This.
33419 (vulcan_tunings): Rename to ...
33420 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33421 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33423 2017-01-09 Martin Jambor <mjambor@suse.cz>
33427 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33428 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33429 (propagate_vr_accross_jump_function): Use the above function for all
33430 value range computations for pass-through jump functions and type
33431 converasion from explicit value range values.
33432 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33433 parameters from TYPE_ARG_TYPES.
33434 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33435 (ipa_write_node_info): Stream type of the actual argument.
33436 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33438 2017-01-09 Martin Liska <mliska@suse.cz>
33441 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33442 (lookup_compiler): Do not show error message with have_E.
33444 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33446 PR tree-optimization/78938
33447 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33448 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33449 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33450 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33453 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33455 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33458 2017-01-09 Richard Biener <rguenther@suse.de>
33460 PR tree-optimization/78997
33461 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33462 name condition properly.
33464 2017-01-09 Richard Biener <rguenther@suse.de>
33467 * dwarf2out.c (is_cxx): New overload with context.
33468 (is_naming_typedef_decl): Use it.
33470 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33472 * invoke.texi (Option Summary): Correct spacing in option lists
33473 and add line breaks to fix over-long lines.
33475 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33477 PR middle-end/17660
33479 * extend.texi (Common Variable Attributes): Add xref to GCC
33480 Internals manual to explain mode attribute keywords.
33482 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33485 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33486 and Preprocessor Options.
33487 (Options for Linking): Document -pthread here....
33488 (RS/6000 and PowerPC Options): ...not here.
33489 (Solaris 2 Options): ...or here.
33490 * doc/cppopts.texi: Document -pthread.
33492 2017-01-08 Martin Sebor <msebor@redhat.com>
33494 PR middle-end/77708
33495 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33496 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33497 New member functions.
33498 (format_directive): Used them.
33500 (pass_sprintf_length::handle_gimple_call): Same.
33501 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33502 to avoid truncation for any argument.
33503 (extract_affine_mul): Same.
33504 * tree.c (get_file_function_name): Same.
33506 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33508 PR middle-end/77484
33509 * predict.def (PRED_INDIR_CALL): Set to 86.
33511 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33513 PR preprocessor/54124
33514 * doc/cppopts.texi: Reformat -d subtable to list the full name
33515 of the options. Add cross-reference to the docs for the general
33516 compiler -d options.
33517 * doc/invoke.texi (Developer Options): Add cross-reference to the
33518 preprocessor-specific -d option documentation.
33520 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33522 PR preprocessor/13498
33523 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33524 redudant material, and reflect new command-line options.
33525 (System Headers): Likewise.
33527 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33529 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33530 -isystem, and -idirafter. Copy-edit.
33531 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33532 default for -ftrack-macro-expansion. Delete obsolete and
33533 badly-formatted implementation details about -fdebug-cpp output.
33534 * doc/cppwarnopts.texi: Copy-edit.
33536 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33539 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33540 that the transition from a max line width >= 1<<10 to narrower
33541 lines works correctly.
33543 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33545 * doc/options.texi (PerFunction): New.
33546 * opt-functions.awk (switch_flags): Map both Optimization and
33547 PerFunction to CL_OPTIMIZATION.
33548 * opth-gen.awk: Test for PerFunction flag along with
33550 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33551 it only when the latter is present. Skip those that don't in
33552 the hash function generator.
33553 * common.opt (fvar-tracking): Mark as PerFunction instead of
33555 (fvar-tracking-assignments): Likewise.
33556 (fvar-tracking-assignments-toggle): Likewise.
33557 (fvar-tracking-uninit): Likewise.
33559 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33561 PR translation/79018
33562 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33565 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33568 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33569 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33570 TARGET_LONG_JUMP_TABLE_OFFSETS.
33571 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33572 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33573 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33574 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33575 * config/m68k/m68k.md (tablejump expander): Likewise.
33576 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33577 TARGET_LONG_JUMP_TABLE_OFFSETS.
33578 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33579 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33581 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33582 David Holsgrove <david.holsgrove@xilinx.com>
33584 * common/config/microblaze/microblaze-common.c
33585 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33586 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33588 * config/microblaze/microblaze.c (microblaze_must_save_register)
33589 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33591 (microblaze_eh_return): New function.
33592 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33593 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33594 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33595 * config/microblaze/microblaze.md (eh_return): New pattern.
33597 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33599 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33600 GCC_DIAGNOSTIC_STRINGIFY): Define.
33602 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33604 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33606 * config/arm/arm.md (<mcrr>): New.
33608 * config/arm/arm.c (arm_arch5te): New.
33609 (arm_option_override): Set arm_arch5te.
33610 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33612 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33613 (arm_mcrr_qualifiers): ... this. New.
33614 (MRRC_QUALIFIERS): Define to...
33615 (arm_mrrc_qualifiers): ... this. New.
33616 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33618 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33619 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33620 (MRRCI, mrrc, MRRC): New.
33621 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33622 VUNSPEC_MRRC2): New.
33624 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33626 * config/arm/arm.md (<mcr>): New.
33628 * config/arm/arm.c (arm_coproc_builtin_available): Add
33629 support for mcr, mrc, mcr2 and mrc2.
33630 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33631 (arm_mcr_qualifiers): ... this. New.
33632 (MRC_QUALIFIERS): Define to ...
33633 (arm_mrc_qualifiers): ... this. New.
33634 (MCR_QUALIFIERS): Define to ...
33635 (arm_mcr_qualifiers): ... this. New.
33636 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33638 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33639 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33640 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33641 VUNSPEC_MRC2): New.
33643 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33645 * config/arm/arm.md (*ldc): New.
33649 * config/arm/arm.c (arm_coproc_builtin_available): Add
33650 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33651 (arm_coproc_ldc_stc_legitimate_address): New.
33652 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33653 'qualifier_const_pointer'.
33654 (LDC_QUALIFIERS): Define to...
33655 (arm_ldc_qualifiers): ... this. New.
33656 (STC_QUALIFIERS): Define to...
33657 (arm_stc_qualifiers): ... this. New.
33658 * config/arm/arm-protos.h
33659 (arm_coproc_ldc_stc_legitimate_address): New.
33660 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33661 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33662 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33663 stc2, stcl, stc2l): New.
33664 * config/arm/constraints.md (Uz): New.
33665 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33666 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33667 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33668 VUNSPEC_STC2L): New.
33670 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33672 * config/arm/arm.md (<cdp>): New.
33673 * config/arm/arm.c (neon_const_bounds): Rename this ...
33674 (arm_const_bounds): ... this.
33675 (arm_coproc_builtin_available): New.
33676 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33677 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33678 (CDP_QUALIFIERS): Define to...
33679 (arm_cdp_qualifiers): ... this. New.
33681 (arm_expand_builtin_args): Add case for 6 arguments.
33682 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33683 (arm_const_bounds): ... this.
33684 (arm_coproc_builtin_available): New.
33685 * config/arm/arm_acle.h (__arm_cdp): New.
33687 * config/arm/arm_acle_builtins.def (cdp): New.
33689 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33690 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33691 'arm_const_bounds'.
33692 * config/arm/types.md (coproc): New.
33693 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33694 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33695 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33696 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33698 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33700 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33701 (UBINOP_QUALIFIERS): New.
33703 (acle_builtin_data): New. Change comment.
33704 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33705 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33706 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33707 arm_acle_builtins.def.
33708 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33709 (arm_init_acle_builtins): New.
33710 (CRC32_BUILTIN): Remove.
33711 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33712 crc32cb, crc32ch and crc32cw.
33713 (arm_init_crc32_builtins): Remove.
33714 (arm_init_builtins): Use arm_init_acle_builtins rather
33715 than arm_init_crc32_builtins.
33716 (arm_expand_acle_builtin): New.
33717 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33718 * config/arm/arm_acle_builtins.def: New.
33720 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33722 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33723 (arm_builtin_datum): ... this.
33724 (arm_init_neon_builtin): Rename to ...
33725 (arm_init_builtin): ... this. Add a new parameters PREFIX
33726 and USE_SIG_IN_NAME.
33727 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33728 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33729 'arm_builtin_datum'.
33730 (arm_init_vfp_builtins): Likewise.
33731 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33732 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33733 (arm_expand_neon_args): Rename to ...
33734 (arm_expand_builtin_args): ... this. Rename builtin_arg
33735 enum values and differentiate between ARG_BUILTIN_MEMORY
33736 and ARG_BUILTIN_NEON_MEMORY.
33737 (arm_expand_neon_builtin_1): Rename to ...
33738 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33739 values, arm_expand_builtin_args and add bool parameter NEON.
33740 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33741 (arm_expand_vfp_builtin): Likewise.
33742 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33744 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33746 PR middle-end/77484
33747 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33748 * predict.c (tree_estimate_probability_bb): Reverse direction of
33749 polymorphic call predictor.
33751 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33753 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33754 (determine_pass_name_match): ...this new function and...
33755 (should_skip_pass_p): ...this new function.
33757 2017-01-06 Nathan Sidwell <nathan@acm.org>
33759 * ipa-visibility.c (function_and_variable_visibility): Reformat
33760 comments and long lines. Remove extrneous if.
33761 * symtab.c (symtab_node::make_decl_local): Fix code format.
33762 (symtab_node::set_section_for_node): Fix comment typo.
33764 2017-01-06 Martin Liska <mliska@suse.cz>
33767 * lra-constraints.c: Rename invariant to lra_invariant.
33768 * predict.c (set_even_probabilities): Initialize e to NULL.
33770 2017-01-05 Martin Sebor <msebor@redhat.com>
33772 PR tree-optimization/78910
33773 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33774 (format_integer): Correct off-by-one error in the handling
33775 of precision with negative numbers in signed conversions..
33777 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33779 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33781 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33783 PR tree-optimization/71016
33784 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33785 factor_out_conditional_conversion. Formatting fix.
33786 (factor_out_conditional_conversion): Add cond_stmt argument.
33787 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33788 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33791 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33793 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33794 read-rtl-function.o, and selftest-rtl.o.
33795 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33796 (selftest::aarch64_test_loading_full_dump): New function.
33797 (selftest::aarch64_run_selftests): New function.
33798 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33799 selftest::aarch64_run_selftests.
33800 * config/i386/i386.c
33801 (selftest::ix86_test_loading_dump_fragment_1): New function.
33802 (selftest::ix86_test_loading_call_insn): New function.
33803 (selftest::ix86_test_loading_full_dump): New function.
33804 (selftest::ix86_test_loading_unspec): New function.
33805 (selftest::ix86_run_selftests): Call the new functions.
33806 * emit-rtl.c (maybe_set_max_label_num): New function.
33807 * emit-rtl.h (maybe_set_max_label_num): New decl.
33808 * function.c (instantiate_decls): Guard call to
33809 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33810 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33812 * gensupport.c (gen_reader::gen_reader): Pass "false"
33813 for new "compact" param of rtx_reader.
33814 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33815 rather than an empty string for NULL strings.
33816 * read-md.c: Potentially include config.h rather than bconfig.h.
33817 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33818 (have_error): New global, copied from errors.c.
33819 (md_reader::read_name): Rename to...
33820 (md_reader::read_name_1): ...this, adding "out_loc" param,
33821 and converting "missing name or number" to returning false, rather
33823 (md_reader::read_name): Reimplement in terms of read_name_1.
33824 (md_reader::read_name_or_nil): New function.
33825 (md_reader::read_string): Handle "(nil)" by returning NULL.
33826 (md_reader::md_reader): Add new param "compact".
33827 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33828 (md_reader::read_file): New method.
33829 * read-md.h (md_reader::md_reader): Add new param "compact".
33830 (md_reader::read_file): New method.
33831 (md_reader::is_compact): New accessor.
33832 (md_reader::read_name): Convert return type from void to file_location.
33833 (md_reader::read_name_or_nil): New decl.
33834 (md_reader::read_name_1): New decl.
33835 (md_reader::m_compact): New field.
33836 (noop_reader::noop_reader): Pass "false" for new "compact" param
33838 (rtx_reader::rtx_reader): Add new "compact" param.
33839 (rtx_reader::read_rtx_operand): Make virtual and convert return
33840 type from void to rtx.
33841 (rtx_reader::read_until): New decl.
33842 (rtx_reader::handle_any_trailing_information): New virtual function.
33843 (rtx_reader::postprocess): New virtual function.
33844 (rtx_reader::finalize_string): New virtual function.
33845 (rtx_reader::m_in_call_function_usage): New field.
33846 (rtx_reader::m_reuse_rtx_by_id): New field.
33847 * read-rtl-function.c: New file.
33848 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33849 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33850 (selftest::verify_three_block_rtl_cfg): New decl.
33851 * read-rtl-function.h: New file.
33852 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33853 For host, include function.h, memmodel.h, and emit-rtl.h.
33854 (one_time_initialization): New function.
33855 (struct compact_insn_name): New struct.
33856 (compact_insn_names): New array.
33857 (find_code): Handle insn codes in compact dumps.
33858 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33859 (bind_subst_iter_and_attr): Likewise.
33860 (add_condition_to_string): Likewise.
33861 (add_condition_to_rtx): Likewise.
33862 (apply_attribute_uses): Likewise.
33863 (add_current_iterators): Likewise.
33864 (apply_iterators): Likewise.
33865 (initialize_iterators): Guard usage of apply_subst_iterator with
33866 #ifdef GENERATOR_FILE.
33867 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33868 (md_reader::read_mapping): Likewise.
33869 (add_define_attr_for_define_subst): Likewise.
33870 (add_define_subst_attr): Likewise.
33871 (read_subst_mapping): Likewise.
33872 (check_code_iterator): Likewise.
33873 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33875 (one_time_initialization): New function.
33876 (rtx_reader::read_until): New method.
33877 (read_flags): New function.
33878 (parse_reg_note_name): New function.
33879 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33880 Handle reuse_rtx ids.
33881 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33882 Add parsing support for RTL dumps, mirroring the special-cases in
33883 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33884 values, and calling handle_any_trailing_information.
33885 (rtx_reader::read_rtx_operand): Convert return type from void
33886 to rtx, returning return_rtx. Handle case 'e'. Call
33887 finalize_string on XSTR and XTMPL fields.
33888 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33889 "(nil)" values were omitted. Call the postprocess vfunc on the
33891 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33892 class ctor. Initialize m_in_call_function_usage. Call
33893 one_time_initialization.
33894 * rtl-tests.c (selftest::test_uncond_jump): Call
33895 set_new_first_and_last_insn.
33896 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33897 * selftest-rtl.c: New file.
33898 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33899 (selftest::get_insn_by_uid): New decl.
33900 * selftest-run-tests.c (selftest::run_tests): Call
33901 read_rtl_function_c_tests.
33902 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33903 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33906 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33908 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33909 operands in a special way. Assert that pos+len <= mode precision.
33911 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33913 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33914 3 argument Alias with unlimited for the negative form.
33915 (fno-vect-cost-model): Removed.
33917 2017-01-05 Martin Liska <mliska@suse.cz>
33919 * hsa-gen.c (gen_hsa_divmod): New function.
33920 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33922 2017-01-05 Martin Liska <mliska@suse.cz>
33925 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33928 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33930 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33931 small constant length operands.
33933 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33935 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33936 between loop iterations.
33938 2017-01-05 Martin Liska <mliska@suse.cz>
33941 * gimplify.c (gimplify_decl_expr): Compare to
33942 asan_poisoned_variables instread of checking flags.
33943 (gimplify_target_expr): Likewise.
33944 (gimplify_expr): Likewise.
33945 (gimplify_function_tree): Conditionally initialize
33946 asan_poisoned_variables.
33948 2017-01-04 Jeff Law <law@redhat.com>
33950 PR tree-optimizatin/78812
33951 * rtl.h (contains_mem_rtx_p): Prototype.
33952 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33953 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33954 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33955 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33956 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33958 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33960 * input.c (assert_char_at_range): Default-initialize actual_range.
33962 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33964 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33965 to match the caller.
33967 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33969 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33970 insns after final jump in test to emit dummy move.
33972 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33974 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33975 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33977 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33979 * multiple_target.c (create_dispatcher_calls): Init e_next.
33980 * tree-ssa-loop-split.c (split_loop): Init border.
33981 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33984 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33989 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33990 (altivec_register_operand): Do not return true if the operand
33991 contains a SUBREG mixing SImode and SFmode.
33992 (vsx_register_operand): Likewise.
33993 (vsx_reg_sfsubreg_ok): New predicate.
33994 (vfloat_operand): Do not return true if the operand contains a
33995 SUBREG mixing SImode and SFmode.
33996 (vint_operand): Likewise.
33997 (vlogical_operand): Likewise.
33998 (gpc_reg_operand): Likewise.
33999 (int_reg_operand): Likewise.
34000 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
34001 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
34002 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
34004 (rs6000_emit_move_si_sf_subreg): New helper function.
34005 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
34006 fixup SUBREGs involving SImode and SFmode.
34007 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
34008 numbers for the new peephole2 optimization.
34009 (peephole2 for SFmode unions): New peephole2 to optimize cases in
34010 the GLIBC math library that do AND/IOR/XOR operations on single
34011 precision floating point.
34012 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
34013 target macros to say whether we need to avoid SUBREGs mixing
34015 (TARGET_ALLOW_SF_SUBREG): Likewise.
34016 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
34017 (UNSPEC_SI_FROM_SF): Likewise.
34018 (iorxor): Change spacing.
34019 (and_ior_xor): New iterator for AND, IOR, and XOR.
34020 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
34021 (movdi_from_sf_zero_ext): Likewise.
34022 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
34023 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
34024 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
34025 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34026 (fms<mode>4): Likewise.
34027 (fnma<mode>4): Likewise.
34028 (fnms<mode>4): Likewise.
34029 (nfma<mode>4): Likewise.
34030 (nfms<mode>4): Likewise.
34032 2017-01-04 Marek Polacek <polacek@redhat.com>
34035 * doc/invoke.texi: Document -Wpointer-compare.
34037 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34039 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34042 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34043 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34046 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34049 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34050 documentation of the powerpc_popcntb_ok attribute.
34051 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34052 code to issue warning messages if a requested CPU configuration is
34053 not supported by the binary (assembler and loader) toolchain.
34054 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34055 made to define a built-in function that has been disabled.
34056 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34057 made to define a built-in function that has been disabled.
34058 (altivec_init_builtins): Add comment explaining why definition
34059 of the DST built-in functions is not preceded by an assertion
34060 check. Add assertions to prevent ICE if attempts are made to
34061 define an altivec predicate or an abs* built-in function that has
34063 (htm_init_builtins): Add comment explaining why definition of the
34064 htm built-in functions is not preceded by an assertion check.
34066 2017-01-04 Jeff Law <law@redhat.com>
34068 PR tree-optimizatin/67955
34069 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34070 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34071 the points-to solution does not include pt_null. Use DECL_PT_UID
34074 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34076 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34077 Use gen_int_mode instead of gen_lopwart for const_int operands.
34079 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34081 PR tree-optimization/71563
34082 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34083 out of range value - has low bits known to be zero.
34085 2017-01-04 Alan Modra <amodra@gmail.com>
34087 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34088 * configure: Regenerate.
34089 * config.in: Regenerate.
34091 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34094 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34095 a substring of the message, but strcmp with the whole message. Ifdef
34096 ENABLE_NLS, translate the message first using dgettext.
34098 2017-01-03 Jeff Law <law@redhat.com>
34100 PR tree-optimizatin/78856
34101 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34102 (mark_threaded_blocks): Remove code to truncate thread paths that
34103 cross multiple loop headers. Instead invalidate the cached loop
34104 iteration information and handle case of a thread path walking
34105 into an irreducible region.
34107 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34110 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34111 assertions. Add support for doing the signbit if the IEEE 128-bit
34112 floating point value is in a GPR.
34113 * config/rs6000/rs6000.md (Fsignbit): Delete.
34114 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34115 Update the length attribute if the value is in a GPR.
34116 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34117 the sign or zero extension instruction, since the value is always 0/1.
34118 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34121 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34122 extracting SImode to a GPR register so that we can generate a
34123 store, limit the vector to be in a traditional Altivec register
34124 for the vextuwrx instruction.
34126 2017-01-03 Ian Lance Taylor <iant@google.com>
34128 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34130 2017-01-03 Martin Sebor <msebor@redhat.com>
34132 PR tree-optimization/78696
34133 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34134 precision. Use MPFR for %f for greater fidelity. Correct handling
34136 (pass_sprintf_length::compute_format_length): Set width and precision
34137 specified by asrerisk to void_node for vararg functions.
34138 (try_substitute_return_value): Adjust dump output.
34140 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34142 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34144 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34146 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34147 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34148 -mlra/-mno-lra was passed to the compiler.
34150 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34152 PR rtl-optimization/65618
34153 * emit-rtl.c (try_split): Move initialization of "before" and
34154 "after" to just before the call to emit_insn_after_setloc.
34156 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34158 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34160 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34162 * dwarf2out.c (gen_enumeration_type_die): When
34163 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34165 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34167 PR tree-optimization/78965
34168 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34169 Change first argument from const call_info & to call_info &. For %n
34170 set info.nowrite to false.
34172 PR middle-end/78901
34173 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34174 possibly throwing calls.
34176 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34177 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34178 and fns handling, rather than in a separate case SSA_NAME.
34180 2017-01-02 Jeff Law <law@redhat.com>
34182 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34183 fixes for first_period and second_period variables.
34185 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34188 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34189 (*insvqi_1): New insn pattern.
34190 (*insvqi_1_mem_rex64): Ditto.
34191 (*insvqi_2): Ditto.
34192 (*insvqi_3): Rename from *insvqi.
34194 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34196 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34198 * doc/cfg.texi (Edges): Remove reference to Java.
34199 (Maintaining the CFG): Ditto.
34201 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34203 PR middle-end/77674
34204 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34205 transparent aliases.
34207 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34209 PR middle-end/77484
34210 * predict.def (PRED_CALL): Update hitrate.
34211 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34212 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34213 into direct/indirect/polymorphic variants.
34215 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34217 Update copyright years.
34219 * gcc.c (process_command): Update copyright notice dates.
34220 * gcov-dump.c (print_version): Ditto.
34221 * gcov.c (print_version): Ditto.
34222 * gcov-tool.c (print_version): Ditto.
34223 * gengtype.c (create_file): Ditto.
34224 * doc/cpp.texi: Bump @copying's copyright year.
34225 * doc/cppinternals.texi: Ditto.
34226 * doc/gcc.texi: Ditto.
34227 * doc/gccint.texi: Ditto.
34228 * doc/gcov.texi: Ditto.
34229 * doc/install.texi: Ditto.
34230 * doc/invoke.texi: Ditto.
34232 Copyright (C) 2017 Free Software Foundation, Inc.
34234 Copying and distribution of this file, with or without modification,
34235 are permitted in any medium without royalty provided the copyright
34236 notice and this notice are preserved.