1 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
3 * bb-reorder.c (max_entry_frequency): Remove.
4 (find_traces, rotate_loop, mark_bb_visited, connect_better_edge_p,
5 connect_traces, push_to_next_round_p): Remove prototypes.
6 (find_traces_1_round): Use counts only.
7 (push_to_next_round_p): Likewise.
8 (find_traces): Likewise.
9 (rotate_loop): Likewise.
10 (find_traces_1_round): Likewise.
11 (connect_traces): Likewise.
12 (edge_order): Likewise.
14 2017-11-09 Thomas Preud'homme <thomas.preudhomme@arm.com>
16 * config/arm/arm.c (output_return_instruction): Add comments to
17 indicate requirement for cmse_nonsecure_entry return to account
18 for the size of clearing instruction output here.
19 (thumb_exit): Likewise.
20 * config/arm/thumb2.md (thumb2_cmse_entry_return): Fix length for
21 return in hardfloat mode.
23 2017-11-09 Segher Boessenkool <segher@kernel.crashing.org>
25 * config/rs6000/rs6000.c (machine_function): Add a bool,
26 "toc_is_wrapped_separately".
27 (rs6000_option_override_internal): Enable OPTION_MASK_SAVE_TOC_INDIRECT
28 if it wasn't explicitly set or unset, we are optimizing for speed, and
29 doing separate shrink-wrapping.
30 (rs6000_get_separate_components): Enable the TOC component if
31 saving the TOC register in the prologue.
32 (rs6000_components_for_bb): Handle the TOC component.
33 (rs6000_emit_prologue_components): Store the TOC register where needed.
34 (rs6000_set_handled_components): Mark TOC as handled, if handled.
35 (rs6000_emit_prologue): Don't save the TOC if that is already done.
37 2017-11-09 Martin Jambor <mjambor@suse.cz>
39 * ipa-param-manipulation.c: New file.
40 * ipa-param-manipulation.h: Likewise.
41 * Makefile.in (OBJS): Add ipa-param-manipulation.o.
42 (PLUGIN_HEADERS): Addded ipa-param-manipulation.h
43 * ipa-param.h (ipa_parm_op): Moved to ipa-param-manipulation.h.
44 (ipa_parm_adjustment): Likewise.
45 (ipa_parm_adjustment_vec): Likewise.
46 (ipa_get_vector_of_formal_parms): Moved declaration to
47 ipa-param-manipulation.h.
48 (ipa_get_vector_of_formal_parm_types): Likewise.
49 (ipa_modify_formal_parameters): Likewise.
50 (ipa_modify_call_arguments): Likewise.
51 (ipa_combine_adjustments): Likewise.
52 (ipa_dump_param_adjustments): Likewise.
53 (ipa_modify_expr): Likewise.
54 (ipa_get_adjustment_candidate): Likewise.
55 * ipa-prop.c (ipa_get_vector_of_formal_parms): Moved to
56 ipa-param-manipulation.c.
57 (ipa_get_vector_of_formal_parm_types): Likewise.
58 (ipa_modify_formal_parameters): Likewise.
59 (ipa_modify_call_arguments): Likewise.
60 (ipa_modify_expr): Likewise.
61 (get_ssa_base_param): Likewise.
62 (ipa_get_adjustment_candidate): Likewise.
63 (index_in_adjustments_multiple_times_p): Likewise.
64 (ipa_combine_adjustments): Likewise.
65 (ipa_dump_param_adjustments): Likewise.
66 * tree-sra.c: Also include ipa-param-manipulation.h
67 * omp-simd-clone.c: Include ipa-param-manipulation.h instead of
70 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
71 Alan Hayward <alan.hayward@arm.com>
72 David Sherwood <david.sherwood@arm.com>
74 * doc/sourcebuild.texi (vect_masked_store): Document.
76 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
77 Alan Hayward <alan.hayward@arm.com>
78 David Sherwood <david.sherwood@arm.com>
80 * doc/sourcebuild.texi (vect_align_stack_vars): Document.
82 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
83 Alan Hayward <alan.hayward@arm.com>
84 David Sherwood <david.sherwood@arm.com>
86 * doc/sourcebuild.texi (vect_variable_length): Document.
88 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
89 Alan Hayward <alan.hayward@arm.com>
90 David Sherwood <david.sherwood@arm.com>
92 * doc/sourcebuild.texi (vect_unaligned_possible): Document.
94 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
95 Alan Hayward <alan.hayward@arm.com>
96 David Sherwood <david.sherwood@arm.com>
98 * doc/sourcebuild.texi (vect_element_align_preferred): Document.
100 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
101 Alan Hayward <alan.hayward@arm.com>
102 David Sherwood <david.sherwood@arm.com>
104 * doc/sourcebuild.texi (vect_perm_short, vect_perm_byte): Document
105 previously undocumented selectors.
106 (vect_perm3_byte, vect_perm3_short, vect_perm3_int): Document.
108 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
110 * doc/rtl.texi (const_vector): Say that elements can be
112 * emit-rtl.h (valid_for_const_vec_duplicate_p): Declare.
113 * emit-rtl.c (valid_for_const_vec_duplicate_p): New function.
114 (gen_vec_duplicate): Use it instead of CONSTANT_P.
115 * optabs.c (expand_vector_broadcast): Likewise.
117 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
118 Alan Hayward <alan.hayward@arm.com>
119 David Sherwood <david.sherwood@arm.com>
121 * tree-ssa-loop-ivopts.c (get_address_cost): Try using a
122 scaled index even if the unscaled address was invalid.
123 Don't increase the complexity of using a scale in that case.
125 2017-11-09 Richard Sandiford <richard.sandiford@linaro.org>
126 Alan Hayward <alan.hayward@arm.com>
127 David Sherwood <david.sherwood@arm.com>
129 * doc/rtl.texi: Rewrite the subreg rules so that they partition
130 the inner register into REGMODE_NATURAL_SIZE bytes rather than
131 UNITS_PER_WORD bytes.
132 * emit-rtl.c (validate_subreg): Divide subregs into blocks
133 based on REGMODE_NATURAL_SIZE of the inner mode.
134 (gen_lowpart_common): Split the SCALAR_FLOAT_MODE_P and
135 !SCALAR_FLOAT_MODE_P cases. Use REGMODE_NATURAL_SIZE for the latter.
136 * expmed.c (lowpart_bit_field_p): Divide the value up into
137 chunks of REGMODE_NATURAL_SIZE rather than UNITS_PER_WORD.
138 * expr.c (store_constructor): Use REGMODE_NATURAL_SIZE to test
139 whether something is likely to occupy more than one register.
141 2017-11-09 Jan Hubicka <hubicka@ucw.cz>
144 * ipa-inline-transform.c (update_noncloned_frequencies): Use
145 profile_count::adjust_for_ipa_scaling.
146 * tree-inline.c (copy_bb, copy_cfg_body): Likewise.
147 * profile-count.c (profile_count::adjust_for_ipa_scaling): New member
149 * profile-count.h (profile_count::adjust_for_ipa_scaling): Declare.
151 2017-11-09 Jakub Jelinek <jakub@redhat.com>
153 * gimple-ssa-store-merging.c (count_multiple_uses): New function.
154 (split_group): Add total_orig and total_new arguments, estimate the
155 number of statements related to the store group without store merging
156 and with store merging.
157 (imm_store_chain_info::output_merged_store): Adjust split_group
158 callers, punt if estimated number of statements with store merging
159 is not smaller than estimated number of statements without it.
161 (handled_load): Remove has_single_use checks.
162 (pass_store_merging::process_store): Likewise.
164 2017-11-09 Richard Biener <rguenther@suse.de>
166 PR tree-optimization/82902
167 * tree-ssa-phiprop.c (propagate_with_phi): Test proper type.
169 2017-11-09 Martin Liska <mliska@suse.cz>
172 * emit-rtl.c (init_emit_regs): Initialize split_branch_probability to
175 2017-11-09 Martin Liska <mliska@suse.cz>
177 PR tree-optimization/82669
178 * sbitmap.h (bmp_iter_set_init): Remove non needed check.
180 2017-11-09 Martin Liska <mliska@suse.cz>
182 PR gcov-profile/48463
183 * coverage.c (coverage_begin_function): Output also end locus
184 of a function and information whether the function is
186 * gcov-dump.c (tag_function): Parse and print the information.
187 * gcov.c (INCLUDE_MAP): Add include.
188 (INCLUDE_SET): Likewise.
189 (struct line_info): Move earlier in the source file because
190 of vector<line_info> in function_info structure.
191 (line_info::line_info): Likewise.
192 (line_info::has_block): Likewise.
193 (struct source_info): Add new member index.
194 (source_info::get_functions_at_location): New function.
195 (function_info::group_line_p): New function.
196 (output_intermediate_line): New function.
197 (output_intermediate_file): Use the mentioned function.
198 (struct function_start): New.
199 (struct function_start_pair_hash): Likewise.
200 (process_file): Add code that identifies group functions.
201 Assign lines either to global or function scope.
202 (generate_results): Skip artificial functions.
203 (find_source): Assign index for each source file.
204 (read_graph_file): Read new flag artificial and end_line.
205 (add_line_counts): Assign it either to global of function scope.
206 (accumulate_line_counts): Isolate core of the function to
207 accumulate_line_info and call it for both function and global
209 (accumulate_line_info): New function.
210 (output_line_beginning): Fix GNU coding style.
211 (print_source_line): New function.
212 (output_line_details): Likewise.
213 (output_function_details): Likewise.
214 (output_lines): Iterate both source (global) scope and function
216 (struct function_line_start_cmp): New class.
217 * doc/gcov.texi: Reflect changes in documentation.
219 2017-11-09 Jakub Jelinek <jakub@redhat.com>
222 * dwarf2out.c (const_ok_for_output_1): Reject NEG in addition to NOT.
223 (mem_loc_descriptor): Handle (const (neg (...))) as (neg (const (...)))
224 and similarly for not instead of neg.
226 2017-11-08 Andi Kleen <ak@linux.intel.com>
228 * config/i386/i386.opt: Add -mforce-indirect-call.
229 * config/i386/predicates.md: Check for flag_force_indirect_call.
230 * doc/invoke.texi: Document -mforce-indirect-call
232 2017-11-08 Kito Cheng <kito.cheng@gmail.com>
234 * config/riscv/riscv-protos.h (riscv_slow_unaligned_access_p):
236 (MOVE_RATIO): Use riscv_slow_unaligned_access_p.
237 config/riscv/riscv.c (predict.h): New include.
238 (riscv_slow_unaligned_access_p): No longer static.
239 (riscv_block_move_straight): Add require.
240 config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
242 2017-11-08 Jakub Jelinek <jakub@redhat.com>
245 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>,
246 <avx512>_eq<mode>3<mask_scalar_merge_name>_1): Use
247 nonimmediate_operand predicate for operand 1 instead of
250 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
252 * config/aarch64/aarch64-simd.md (store_pair_lanes<mode>):
254 * config/aarch64/constraints.md (Uml): New constraint.
255 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): New
258 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
260 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
261 of two vec_duplicates into a vec_concat.
263 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
265 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
266 Simplify vec_merge of vec_duplicate and vec_concat.
267 * config/aarch64/constraints.md (Utq): New constraint.
268 * config/aarch64/aarch64-simd.md (load_pair_lanes<mode>): New
271 2017-11-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
273 * simplify-rtx.c (simplify_ternary_operation, VEC_MERGE):
274 Simplify vec_merge of vec_duplicate and const_vector.
275 * config/aarch64/predicates.md (aarch64_simd_or_scalar_imm_zero):
277 * config/aarch64/aarch64-simd.md (*aarch64_combinez<mode>): Use VDC
278 mode iterator. Update predicate on operand 1 to
279 handle non-const_vec constants. Delete constraints.
280 (*aarch64_combinez_be<mode>): Likewise for operand 2.
282 2017-11-08 Jakub Jelinek <jakub@redhat.com>
284 PR tree-optimization/78821
285 * gimple-ssa-store-merging.c (struct store_operand_info): Add bit_not_p
287 (store_operand_info::store_operand_info): Initialize it to false.
288 (pass_store_merging::terminate_all_aliasing_chains): Rewritten to use
289 ref_maybe_used_by_stmt_p and stmt_may_clobber_ref_p on lhs of each
290 store in the group, and if chain_info is non-NULL, to ignore altogether
292 (compatible_load_p): Fail if bit_not_p does not match.
293 (imm_store_chain_info::output_merged_store): Handle bit_not_p loads.
294 (handled_load): Fill in bit_not_p. Handle BIT_NOT_EXPR.
295 (pass_store_merging::process_store): Adjust
296 terminate_all_aliasing_chains calls to pass NULL in all current spots,
297 call terminate_all_aliasing_chains newly when adding a store into
298 a chain with non-NULL chain_info.
300 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
302 * config/aarch64/aarch64.c (aarch64_can_eliminate): Simplify logic.
304 2017-11-08 Wilco Dijkstra <wdijkstr@arm.com>
306 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
308 (aarch64_layout_frame): Initialise emit_frame_chain.
309 (aarch64_can_eliminate): Remove omit leaf frame pointer code.
310 (TARGET_FRAME_POINTER_REQUIRED): Remove define.
312 2017-11-08 Martin Liska <mliska@suse.cz>
314 * gimplify.c (expand_FALLTHROUGH_r): Simplify usage
315 of gimple_call_internal_p.
317 2017-11-07 Tom de Vries <tom@codesourcery.com>
319 * config/mips/mips.h (ASM_OUTPUT_LABELREF): Wrap in "do {} while (0)".
321 2017-11-07 Tom de Vries <tom@codesourcery.com>
323 * config/mips/mips.h (ASM_OUTPUT_CASE_END): Remove semicolon after
326 2017-11-08 Martin Liska <mliska@suse.cz>
329 * gimplify.c (expand_FALLTHROUGH_r): Skip IFN_ASAN_MARK.
331 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
333 * gimple-pretty-print.c (dump_profile): Return "" instead of NULL.
335 2017-11-07 Jakub Jelinek <jakub@redhat.com>
338 * config/i386/i386.md (SWI1248_AVX512BWDQ2_64): New mode iterator.
339 (*cmp<mode>_ccz_1): New insn with $k alternative.
342 * config/i386/i386.c (ix86_swap_binary_operands_p): Treat
343 RTX_COMM_COMPARE as commutative as well.
344 (ix86_binary_operator_ok): Formatting fix.
345 * config/i386/sse.md (*mul<mode>3<mask_name><round_name>,
346 *<code><mode>3<mask_name><round_saeonly_name>,
347 *<code><mode>3<mask_name>, *<code>tf3, *mul<mode>3<mask_name>,
348 *<s>mul<mode>3_highpart<mask_name>,
349 *vec_widen_umult_even_v16si<mask_name>,
350 *vec_widen_umult_even_v8si<mask_name>,
351 *vec_widen_umult_even_v4si<mask_name>,
352 *vec_widen_smult_even_v16si<mask_name>,
353 *vec_widen_smult_even_v8si<mask_name>, *sse4_1_mulv2siv2di3<mask_name>,
354 *avx2_pmaddwd, *sse2_pmaddwd, *<sse4_1_avx2>_mul<mode>3<mask_name>,
355 *avx2_<code><mode>3, *avx512f_<code><mode>3<mask_name>,
356 *sse4_1_<code><mode>3<mask_name>, *<code>v8hi3,
357 *sse4_1_<code><mode>3<mask_name>, *<code>v16qi3, *avx2_eq<mode>3,
358 <avx512>_eq<mode>3<mask_scalar_merge_name>_1, *sse4_1_eqv2di3,
359 *sse2_eq<mode>3, <mask_codefor><code><mode>3<mask_name>,
360 *<code><mode>3, *<sse2_avx2>_uavg<mode>3<mask_name>,
361 *<ssse3_avx2>_pmulhrsw<mode>3<mask_name>, *ssse3_pmulhrswv4hi3): Use
362 !(MEM_P (operands[1]) && MEM_P (operands[2])) condition instead of
363 ix86_binary_operator_ok. Formatting fixes.
364 (*<plusminus_insn><mode>3<mask_name><round_name>,
365 *<plusminus_insn><mode>3, *<plusminus_insn><mode>3_m): Formatting
368 2017-11-07 Segher Boessenkool <segher@kernel.crashing.org>
370 * config/rs6000/rs6000.md (GPR2): New mode_iterator.
371 ("cstore<mode>4"): Don't always expand with rs6000_emit_int_cmove for
372 eq and ne if TARGET_ISEL.
373 (cmp): New code_iterator.
374 (UNS, UNSU_, UNSIK): New code_attrs.
375 (<code><GPR:mode><GPR2:mode>2_isel): New define_insn_and_split.
376 ("eq<mode>3"): New define_expand, rename the define_insn_and_split
378 ("eq<mode>3"): ... this.
379 ("ne<mode>3"): New define_expand, rename the define_insn_and_split
381 ("ne<mode>3"): ... this.
383 2017-11-07 Julia Koval <julia.koval@intel.com>
386 * common/config/i386/i386-common.c
387 (OPTION_MASK_ISA_GENERAL_REGS_ONLY_UNSET): Remove MPX from flag.
388 (ix86_handle_option): Move MPX to isa_flags2 and GFNI to isa_flags.
389 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto.
390 * config/i386/i386.opt: Ditto.
391 * config/i386/i386.c (ix86_target_string): Ditto.
392 (ix86_option_override_internal): Ditto.
393 (ix86_init_mpx_builtins): Move MPX to args2.
394 (ix86_expand_builtin): Special handling for OPTION_MASK_ISA_GFNI.
395 * config/i386/i386-builtin.def (__builtin_ia32_vgf2p8affineinvqb_v64qi,
396 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
397 __builtin_ia32_vgf2p8affineinvqb_v32qi,
398 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
399 __builtin_ia32_vgf2p8affineinvqb_v16qi,
400 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): Move to ARGS array.
402 2017-11-07 Uros Bizjak <ubizjak@gmail.com>
405 * config/i386.i386.md (*zero_extendsidi2): Change (?r,*Yj), (?*Yi,r)
406 and (*x,m) to ($r,Yj), ($Yi,r) and ($x,m).
407 (zero-extendsidi peephole2): Remove peephole.
409 2017-11-07 Eric Botcazou <ebotcazou@adacore.com>
412 * stor-layout.c: Include attribs.h.
413 (handle_warn_if_not_align): Replace test on TYPE_USER_ALIGN with
414 explicit lookup of "aligned" attribute.
416 2017-11-07 Andrew Waterman <andrew@sifive.com>
418 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): New prototype.
419 (riscv_expand_block_move): Likewise.
420 * config/riscv/riscv.h (MOVE_RATIO): Tune cost to movmemsi
422 (RISCV_MAX_MOVE_BYTES_PER_LOOP_ITER): New define.
423 (RISCV_MAX_MOVE_BYTES_STRAIGHT): New define.
424 * config/riscv/riscv.c (riscv_block_move_straight): New function.
425 (riscv_adjust_block_mem): Likewise.
426 (riscv_block_move_loop): Likewise.
427 (riscv_expand_block_move): Likewise.
428 gcc/config/riscv/riscv.md (movmemsi): New pattern.
430 2017-11-07 Michael Clark <michaeljclark@mac.com>
432 * config/riscv/linux.h (MUSL_ABI_SUFFIX): New define.
433 (MUSL_DYNAMIC_LINKER): Likewise.
435 2017-11-07 Richard Sandiford <richard.sandiford@linaro.org>
437 * config/aarch64/aarch64.md (ashl<mode>3, ashr<mode>3, lshr<mode>3)
438 (rotr<mode>3, rotl<mode>3): Use aarch64_reg_or_imm instead of
441 2017-11-07 Richard Biener <rguenther@suse.de>
443 * match.pd: Fix build.
445 2017-11-07 Wilco Dijkstra <wdijkstr@arm.com>
446 Jackson Woodruff <jackson.woodruff@arm.com>
448 PR tree-optimization/71026
449 * match.pd: Canonicalize negate in division.
451 2017-11-07 Sudakshina Das <sudi.das@arm.com>
454 * match.pd: Simplify 1 << (C - x) where C = precision (x) - 1.
456 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
458 * match.pd ((a&~b)|(a^b),(a&~b)^~a,(a|b)&~(a^b),a|~(a^b),
459 (a|b)|(a&^b),(a&b)|~(a^b),~(~a&b),~X^Y): New transformations.
461 2017-11-07 Marc Glisse <marc.glisse@inria.fr>
463 * fold-const.c (negate_expr_p) [PLUS_EXPR, MINUS_EXPR]: Handle
464 non-scalar integral types.
465 * match.pd (negate_expr_p): Handle MINUS_EXPR.
466 (-(A-B), -(~A)): New transformations.
468 2017-11-07 Tom de Vries <tom@codesourcery.com>
470 * config/powerpcspe/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove
471 semicolon after "do {} while (0)".
472 * config/powerpcspe/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
473 * config/powerpcspe/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
474 * config/powerpcspe/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
475 * config/powerpcspe/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
476 * config/powerpcspe/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
478 2017-11-07 Tom de Vries <tom@codesourcery.com>
480 * config/rs6000/aix43.h (SUBTARGET_OVERRIDE_OPTIONS): Remove semicolon
481 after "do {} while (0)".
482 * config/rs6000/aix51.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
483 * config/rs6000/aix52.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
484 * config/rs6000/aix53.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
485 * config/rs6000/aix61.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
486 * config/rs6000/aix71.h (SUBTARGET_OVERRIDE_OPTIONS): Same.
488 2017-11-07 Tom de Vries <tom@codesourcery.com>
491 * config/arm/arm.c (HANDLE_NARROW_SHIFT_ARITH): Remove semicolon after
493 (arm_rtx_costs_internal): Add missing semicolon after
494 HANDLE_NARROW_SHIFT_ARITH call.
496 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
498 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
499 disable isel if it was not set explicitly.
501 2017-11-06 James Bowman <james.bowman@ftdichip.com>
503 * gcc/dwarf2out.c (modified_type_die): Retain ADDR_SPACE
505 (add_type_attribute) likewise.
507 2017-11-06 H.J. Lu <hongjiu.lu@intel.com>
509 * config/i386/i386.c (ix86_can_use_return_insn_p): Use reference
510 of struct ix86_frame.
511 (ix86_initial_elimination_offset): Likewise.
512 (ix86_expand_split_stack_prologue): Likewise.
514 2017-11-06 Marc Glisse <marc.glisse@inria.fr>
516 * tree-vrp.h (enum value_range_type): Update stale comment.
518 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
519 Alan Hayward <alan.hayward@arm.com>
520 David Sherwood <david.sherwood@arm.com>
522 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm)
523 (aarch64_expand_vec_perm_const): Take the number of units too.
524 * config/aarch64/aarch64.c (aarch64_expand_vec_perm)
525 (aarch64_expand_vec_perm_const): Likewise.
526 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>)
527 (vec_perm<mode>): Update accordingly.
529 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
530 Alan Hayward <alan.hayward@arm.com>
531 David Sherwood <david.sherwood@arm.com>
533 * config/aarch64/aarch64-protos.h (aarch64_simd_vect_par_cnst_half):
534 Take the number of units too.
535 * config/aarch64/aarch64.c (aarch64_simd_vect_par_cnst_half): Likewise.
536 (aarch64_simd_check_vect_par_cnst_half): Update call accordingly,
537 but check for a vector mode before rather than after the call.
538 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
539 (move_hi_quad_<mode>, vec_unpack<su>_hi_<mode>)
540 (vec_unpack<su>_lo_<mode, vec_widen_<su>mult_lo_<mode>)
541 (vec_widen_<su>mult_hi_<mode>, vec_unpacks_lo_<mode>)
542 (vec_unpacks_hi_<mode>, aarch64_saddl2<mode>, aarch64_uaddl2<mode>)
543 (aarch64_ssubl2<mode>, aarch64_usubl2<mode>, widen_ssum<mode>3)
544 (widen_usum<mode>3, aarch64_saddw2<mode>, aarch64_uaddw2<mode>)
545 (aarch64_ssubw2<mode>, aarch64_usubw2<mode>, aarch64_sqdmlal2<mode>)
546 (aarch64_sqdmlsl2<mode>, aarch64_sqdmlal2_lane<mode>)
547 (aarch64_sqdmlal2_laneq<mode>, aarch64_sqdmlsl2_lane<mode>)
548 (aarch64_sqdmlsl2_laneq<mode>, aarch64_sqdmlal2_n<mode>)
549 (aarch64_sqdmlsl2_n<mode>, aarch64_sqdmull2<mode>)
550 (aarch64_sqdmull2_lane<mode>, aarch64_sqdmull2_laneq<mode>)
551 (aarch64_sqdmull2_n<mode>): Update accordingly.
553 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
554 Alan Hayward <alan.hayward@arm.com>
555 David Sherwood <david.sherwood@arm.com>
557 * config/aarch64/aarch64-protos.h (aarch64_reverse_mask): Take
558 the number of units too.
559 * config/aarch64/aarch64.c (aarch64_reverse_mask): Likewise.
560 * config/aarch64/aarch64-simd.md (vec_load_lanesoi<mode>)
561 (vec_store_lanesoi<mode>, vec_load_lanesci<mode>)
562 (vec_store_lanesci<mode>, vec_load_lanesxi<mode>)
563 (vec_store_lanesxi<mode>): Update accordingly.
565 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
566 Alan Hayward <alan.hayward@arm.com>
567 David Sherwood <david.sherwood@arm.com>
569 * config/aarch64/aarch64-protos.h (aarch64_endian_lane_rtx): Declare.
570 * config/aarch64/aarch64.c (aarch64_endian_lane_rtx): New function.
571 * config/aarch64/aarch64.h (ENDIAN_LANE_N): Take the number
572 of units rather than the mode.
573 * config/aarch64/iterators.md (nunits): New mode attribute.
574 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args):
575 Use aarch64_endian_lane_rtx instead of GEN_INT (ENDIAN_LANE_N ...).
576 * config/aarch64/aarch64-simd.md (aarch64_dup_lane<mode>)
577 (aarch64_dup_lane_<vswap_width_name><mode>, *aarch64_mul3_elt<mode>)
578 (*aarch64_mul3_elt_<vswap_width_name><mode>): Likewise.
579 (*aarch64_mul3_elt_to_64v2df, *aarch64_mla_elt<mode>): Likewise.
580 (*aarch64_mla_elt_<vswap_width_name><mode>, *aarch64_mls_elt<mode>)
581 (*aarch64_mls_elt_<vswap_width_name><mode>, *aarch64_fma4_elt<mode>)
582 (*aarch64_fma4_elt_<vswap_width_name><mode>):: Likewise.
583 (*aarch64_fma4_elt_to_64v2df, *aarch64_fnma4_elt<mode>): Likewise.
584 (*aarch64_fnma4_elt_<vswap_width_name><mode>): Likewise.
585 (*aarch64_fnma4_elt_to_64v2df, reduc_plus_scal_<mode>): Likewise.
586 (reduc_plus_scal_v4sf, reduc_<maxmin_uns>_scal_<mode>): Likewise.
587 (reduc_<maxmin_uns>_scal_<mode>): Likewise.
588 (*aarch64_get_lane_extend<GPI:mode><VDQQH:mode>): Likewise.
589 (*aarch64_get_lane_zero_extendsi<mode>): Likewise.
590 (aarch64_get_lane<mode>, *aarch64_mulx_elt_<vswap_width_name><mode>)
591 (*aarch64_mulx_elt<mode>, *aarch64_vgetfmulx<mode>): Likewise.
592 (aarch64_sq<r>dmulh_lane<mode>, aarch64_sq<r>dmulh_laneq<mode>)
593 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_lane<mode>): Likewise.
594 (aarch64_sqrdml<SQRDMLH_AS:rdma_as>h_laneq<mode>): Likewise.
595 (aarch64_sqdml<SBINQOPS:as>l_lane<mode>): Likewise.
596 (aarch64_sqdml<SBINQOPS:as>l_laneq<mode>): Likewise.
597 (aarch64_sqdml<SBINQOPS:as>l2_lane<mode>_internal): Likewise.
598 (aarch64_sqdml<SBINQOPS:as>l2_laneq<mode>_internal): Likewise.
599 (aarch64_sqdmull_lane<mode>, aarch64_sqdmull_laneq<mode>): Likewise.
600 (aarch64_sqdmull2_lane<mode>_internal): Likewise.
601 (aarch64_sqdmull2_laneq<mode>_internal): Likewise.
602 (aarch64_vec_load_lanesoi_lane<mode>): Likewise.
603 (aarch64_vec_store_lanesoi_lane<mode>): Likewise.
604 (aarch64_vec_load_lanesci_lane<mode>): Likewise.
605 (aarch64_vec_store_lanesci_lane<mode>): Likewise.
606 (aarch64_vec_load_lanesxi_lane<mode>): Likewise.
607 (aarch64_vec_store_lanesxi_lane<mode>): Likewise.
608 (aarch64_simd_vec_set<mode>): Update use of ENDIAN_LANE_N.
609 (aarch64_simd_vec_setv2di): Likewise.
611 2017-11-06 Carl Love <cel@us.ibm.com>
613 * config/rs6000/rs6000-c.c (P8V_BUILTIN_VEC_REVB): Add power 8
615 (P9V_BUILTIN_VEC_REVB): Remove the power 9 instance definitions.
616 * config/rs6000/altivec.h (vec_revb): Change the #define from power 9
618 * config/rs6000/r6000-protos.h (swap_endian_selector_for_mode): Add new
620 * config/rs6000/rs6000.c (swap_endian_selector_for_mode): Add function.
621 * config/rs6000/rs6000-builtin.def (BU_P8V_VSX_1, BU_P8V_OVERLOAD_1):
622 Add power 8 macro expansions.
623 (BU_P9V_OVERLOAD_1): Remove power 9 overload expansion.
624 * config/rs6000/vsx.md (revb_<mode>): Add define_expand to generate
625 power 8 instructions. (VSX_XXBR): Add iterator.
627 2017-11-06 Wilco Dijkstra <wdijkstr@arm.com>
629 * config/arm/arm.md (predicable_short_it): Change default to "no",
630 improve documentation, remove uses that are identical to the default.
631 (enabled_for_depr_it): Rename to enabled_for_short_it.
632 * gcc/config/arm/arm-fixed.md (predicable_short_it): Remove default uses.
633 * gcc/config/arm/ldmstm.md (predicable_short_it): Likewise.
634 * gcc/config/arm/sync.md (predicable_short_it): Likewise.
635 * gcc/config/arm/thumb2.md (predicable_short_it): Likewise.
636 * gcc/config/arm/vfp.md (predicable_short_it): Likewise.
638 2017-11-06 Michael Meissner <meissner@linux.vnet.ibm.com>
641 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1): Delete
642 float128 helper macros, which are no longer used after deleting
643 the old 'q' built-in functions, and moving the round to odd
644 built-in functions to being special built-in functions.
645 (BU_FLOAT128_2): Likewise.
646 (BU_FLOAT128_1_HW): Likewise.
647 (BU_FLOAT128_2_HW): Likewise.
648 (BU_FLOAT128_3_HW): Likewise.
649 (FABSQ): Delete old 'q' built-in functions.
650 (COPYSIGNQ): Likewise.
651 (SQRTF128_ODD): Move round to odd built-in functions to be
652 special built-in functions, so that we can handle
653 -mabi=ieeelongdouble.
654 (TRUNCF128_ODD): Likewise.
655 (ADDF128_ODD): Likewise.
656 (SUBF128_ODD): Likewise.
657 (MULF128_ODD): Likewise.
658 (DIVF128_ODD): Likewise.
659 (FMAF128_ODD): Likewise.
660 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Map old 'q'
661 built-in names to 'f128'.
662 * config/rs6000/rs6000.c (rs6000_fold_builtin): Remove folding the
663 old 'q' built-in functions, as the machine independent code for
664 'f128' built-in functions handles this.
665 (rs6000_expand_builtin): Add expansion for float128 round to odd
666 functions, keying off on -mabi=ieeelongdouble of whether to use
667 the KFmode or TFmode variant.
668 (rs6000_init_builtins): Initialize the _Float128 round to odd
670 * doc/extend.texi (PowerPC Built-in Functions): Document the old
671 _Float128 'q' built-in functions are now mapped into the new
672 'f128' built-in functions.
674 2017-11-06 David Edelsohn <dje.gcc@gmail.com>
676 * collect2.c (add_lto_object): Compile for OBJECT_COFF.
677 (scan_prog_file): Don't skip PASS_LTOINFO. Scan for LTO objects.
679 2017-11-06 David Malcolm <dmalcolm@redhat.com>
682 * ipa-fnsummary.c (ipa_fnsummary_c_finalize): New function.
683 * ipa-fnsummary.h (ipa_fnsummary_c_finalize): New decl.
684 * toplev.c: Include "ipa-fnsummary.h".
685 (toplev::finalize): Call ipa_fnsummary_c_finalize.
687 2017-11-06 Jakub Jelinek <jakub@redhat.com>
689 PR tree-optimization/82838
690 * gimple-ssa-store-merging.c
691 (imm_store_chain_info::output_merged_store): Call force_gimple_operand_1
692 on a separate gimple_seq which is then appended to seq.
694 2017-11-06 Jeff Law <law@redhat.com>
697 * config/i386/i386.c (PROBE_INTERVAL): Remove.
698 (get_probe_interval): New functions.
699 (ix86_adjust_stack_and_probe_stack_clash): Use get_probe_interval.
700 (ix86_adjust_stack_and_probe): Likewise.
701 (output_adjust_stack_and_probe): Likewise.
702 (ix86_emit_probe_stack_range): Likewise.
703 (ix86_expand_prologue): Likewise.
705 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
707 PR tree-optimization/82816
708 * tree-ssa-math-opts.c (convert_mult_to_widen): Return false
709 if the modes of the two types are the same.
710 (convert_plusminus_to_widen): Likewise.
712 2017-11-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
714 * config/rs6000/altivec.md (*p9_vadu<mode>3) Rename to
716 (usadv16qi): New define_expand.
717 (usadv8hi): New define_expand.
719 2017-11-06 Jan Hubicka <hubicka@ucw.cz>
722 * ipa-inline-transform.c (update_noncloned_frequencies): Always
724 (inline_transform): Likewise.
725 * predict.c (counts_to_freqs): Remove useless conditional.
726 * profile-count.h (profile_count::apply_scale): Move sanity check.
727 * tree-inline.c (copy_bb): Always scale.
728 (copy_cfg_body): Likewise.
730 2017-11-06 Christophe Lyon <christophe.lyon@linaro.org>
733 * config/arm/arm.md (*sub_shiftsi): Add predicable_short_it
735 (*cmp_ite0): Add enabled_for_depr_it attribute.
736 (*cmp_ite1): Likewise.
738 2017-11-06 Segher Boessenkool <segher@kernel.crashing.org>
740 * config/rs6000/rs6000.c (rs6000_insn_cost): Handle TYPE_MFCR and
743 2017-11-06 Richard Sandiford <richard.sandiford@linaro.org>
745 * tree-vrp.c (vrp_int_const_binop): Return true on success and
746 return the value by pointer.
747 (extract_range_from_multiplicative_op_1): Update accordingly.
748 Return as soon as an operation fails.
750 2017-11-05 Tom de Vries <tom@codesourcery.com>
753 * asan.c (DEF_SANITIZER_BUILTIN_1): Factor out of ...
754 (DEF_SANITIZER_BUILTIN): ... here.
755 (initialize_sanitizer_builtins): Use DEF_SANITIZER_BUILTIN_1 instead of
756 DEF_SANITIZER_BUILTIN in if stmt. Add missing semicolon.
758 2017-11-05 Tom de Vries <tom@codesourcery.com>
761 * config/elfos.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon after
763 (ASM_OUTPUT_CASE_LABEL): Add semicolon after
764 ASM_OUTPUT_BEFORE_CASE_LABEL call.
765 * config/arc/arc.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove semicolon
767 * config/m68k/m68kelf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
768 * config/mips/mips.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
769 * config/v850/v850.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Same.
771 2017-11-05 Tom de Vries <tom@codesourcery.com>
774 * graphite-scop-detection.c (DEBUG_PRINT): Remove semicolon after
777 2017-11-04 Michael Clark <michaeljclark@mac.com>
779 * config/riscv/riscv.c (riscv_print_operand): Add a 'i' format.
780 config/riscv/riscv.md (addsi3): Use 'i' for immediates.
782 (*addsi3_extended): Likewise.
783 (*addsi3_extended2): Likewise.
784 (<optab>si3): Likewise.
785 (<optab>di3): Likewise.
786 (<optab><mode>3): Likewise.
787 (<*optabe>si3_internal): Likewise.
788 (zero_extendqi<SUPERQI:mode>2): Likewise.
789 (*add<mode>hi3): Likewise.
790 (*xor<mode>hi3): Likewise.
791 (<optab>di3): Likewise.
792 (*<optab>si3_extend): Likewise.
793 (*sge<u>_<X:mode><GPR:mode>): Likewise.
794 (*slt<u>_<X:mode><GPR:mode>): Likewise.
795 (*sle<u>_<X:mode><GPR:mode>): Likewise.
797 2017-11-04 Andrew Waterman <andrew@sifive.com>
799 * config/riscv/riscv.c (riscv_option_override): Conditionally set
800 TARGET_STRICT_ALIGN based upon -mtune argument.
802 2017-11-04 Andrew Waterman <andrew@sifive.com>
804 * config/riscv/riscv.h (SLOW_BYTE_ACCESS): Change to 1.
806 2017-11-04 Daniel Santos <daniel.santos@pobox.com>
808 config/i386/i386.c (choose_basereg): Use optional scratch
809 register and add assertion.
810 (x86_emit_outlined_ms2sysv_save): Use scratch register when
811 needed, and don't allocate stack.
812 (ix86_expand_prologue): Rearrange where SSE saves/stub call is
813 emitted, correct wrong allocation with -mcall-ms2sysv-xlogues.
814 (ix86_emit_outlined_ms2sysv_restore): Fix non-immediate offsets.
816 2017-11-03 Jeff Law <law@redhat.com>
818 * config/i386/i386.c (ix86_emit_restore_reg_using_pop): Prototype.
819 (ix86_adjust_stack_and_probe_stack_clash): Use a push/pop sequence
820 to probe at the start of a noreturn function.
822 2017-11-03 Jakub Jelinek <jakub@redhat.com>
824 PR tree-optimization/78821
825 * gimple-ssa-store-merging.c: Update the file comment.
826 (MAX_STORE_ALIAS_CHECKS): Define.
827 (struct store_operand_info): New type.
828 (store_operand_info::store_operand_info): New constructor.
829 (struct store_immediate_info): Add rhs_code and ops data members.
830 (store_immediate_info::store_immediate_info): Add rhscode, op0r
831 and op1r arguments to the ctor, initialize corresponding data members.
832 (struct merged_store_group): Add load_align_base and load_align
834 (merged_store_group::merged_store_group): Initialize them.
835 (merged_store_group::do_merge): Update them.
836 (merged_store_group::apply_stores): Pick the constant for
837 encode_tree_to_bitpos from one of the two operands, or skip
838 encode_tree_to_bitpos if neither operand is a constant.
839 (class pass_store_merging): Add process_store method decl. Remove
840 bool argument from terminate_all_aliasing_chains method decl.
841 (pass_store_merging::terminate_all_aliasing_chains): Remove
842 var_offset_p argument and corresponding handling.
843 (stmts_may_clobber_ref_p): New function.
844 (compatible_load_p): New function.
845 (imm_store_chain_info::coalesce_immediate_stores): Terminate group
846 if there is overlap and rhs_code is not INTEGER_CST. For
847 non-overlapping stores terminate group if rhs is not mergeable.
848 (get_alias_type_for_stmts): Change first argument from
849 auto_vec<gimple *> & to vec<gimple *> &. Add IS_LOAD, CLIQUEP and
850 BASEP arguments. If IS_LOAD is true, look at rhs1 of the stmts
851 instead of lhs. Compute *CLIQUEP and *BASEP in addition to the
853 (get_location_for_stmts): Change first argument from
854 auto_vec<gimple *> & to vec<gimple *> &.
855 (struct split_store): Remove orig_stmts data member, add orig_stores.
856 (split_store::split_store): Create orig_stores rather than orig_stmts.
857 (find_constituent_stmts): Renamed to ...
858 (find_constituent_stores): ... this. Change second argument from
859 vec<gimple *> * to vec<store_immediate_info *> *, push pointers
860 to info structures rather than the statements.
861 (split_group): Rename ALLOW_UNALIGNED argument to
862 ALLOW_UNALIGNED_STORE, add ALLOW_UNALIGNED_LOAD argument and handle
863 it. Adjust find_constituent_stores caller.
864 (imm_store_chain_info::output_merged_store): Handle rhs_code other
865 than INTEGER_CST, adjust split_group, get_alias_type_for_stmts and
866 get_location_for_stmts callers. Set MR_DEPENDENCE_CLIQUE and
867 MR_DEPENDENCE_BASE on the MEM_REFs if they are the same in all stores.
868 (mem_valid_for_store_merging): New function.
869 (handled_load): New function.
870 (pass_store_merging::process_store): New method.
871 (pass_store_merging::execute): Use process_store method. Adjust
872 terminate_all_aliasing_chains caller.
874 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
876 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
877 Return true for more constants, symbols and label references.
878 (aarch64_valid_floating_const): Remove unused function.
880 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
883 * config/aarch64/aarch64.c (aarch64_layout_frame):
884 Undo forcing of LR at bottom of frame.
886 2017-11-03 Jeff Law <law@redhat.com>
889 * config/i386/i386.c (ix86_expand_prologue): Tighten assert
890 for int_registers_saved.
892 * cfganal.c (single_pred_edge_ignoring_loop_edges): New function
893 extracted from tree-ssa-dom.c.
894 * cfganal.h (single_pred_edge_ignoring_loop_edges): Prototype.
895 * tree-ssa-dom.c (single_incoming_edge_ignoring_loop_edges): Remove.
896 (record_equivalences_from_incoming_edge): Add additional argument
897 to single_pred_edge_ignoring_loop_edges call.
898 * tree-ssa-uncprop.c (single_incoming_edge_ignoring_loop_edges): Remove.
899 (uncprop_dom_walker::before_dom_children): Add additional argument
900 to single_pred_edge_ignoring_loop_edges call.
901 * tree-ssa-sccvn.c (sccvn_dom_walker::before_dom_children): Use
902 single_pred_edge_ignoring_loop_edges rather than open coding.
903 * tree-vrp.c (evrp_dom_walker::before_dom_children): Similarly.
905 2017-11-03 Marc Glisse <marc.glisse@inria.fr>
907 * match.pd (-(-A)): Rewrite.
909 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
911 * config/rs6000/rs60000-protos.h (rs6000_emit_sISEL): Delete.
912 (rs6000_emit_int_cmove): New declaration.
913 * config/rs6000/rs6000.c (rs6000_emit_int_cmove): Delete declaration.
914 (rs6000_emit_sISEL): Delete.
915 (rs6000_emit_int_cmove): Make non-static.
916 * config/rs6000/rs6000.md (cstore<mode>4): Use rs6000_emit_int_cmove
917 instead of rs6000_emit_sISEL.
919 2017-11-03 Jan Hubicka <hubicka@ucw.cz>
921 * asan.c (create_cond_insert_point): Maintain profile.
922 * ipa-utils.c (ipa_merge_profiles): Be sure only IPA profiles are
924 * basic-block.h (struct basic_block_def): Remove frequency.
925 (EDGE_FREQUENCY): Use to_frequency
926 * bb-reorder.c (push_to_next_round_p): Use only IPA counts for global
928 (find_traces): Update to use to_frequency.
929 (find_traces_1_round): Likewise; use only IPA counts.
930 (bb_to_key): Likewise.
931 (connect_traces): Use IPA counts only.
932 (copy_bb_p): Update to use to_frequency.
933 (fix_up_crossing_landing_pad): Likewise.
934 (sanitize_hot_paths): Likewise.
935 * bt-load.c (basic_block_freq): Likewise.
936 * cfg.c (init_flow): Set count_max to uninitialized.
937 (check_bb_profile): Remove frequencies; check counts.
938 (dump_bb_info): Do not dump frequencies.
939 (update_bb_profile_for_threading): Update counts only.
940 (scale_bbs_frequencies_int): Likewise.
941 (MAX_SAFE_MULTIPLIER): Remove.
942 (scale_bbs_frequencies_gcov_type): Update counts only.
943 (scale_bbs_frequencies_profile_count): Update counts only.
944 (scale_bbs_frequencies): Update counts only.
945 * cfg.h (struct control_flow_graph): Add count-max.
946 (update_bb_profile_for_threading): Update prototype.
947 * cfgbuild.c (find_bb_boundaries): Do not update frequencies.
948 (find_many_sub_basic_blocks): Likewise.
949 * cfgcleanup.c (try_forward_edges): Likewise.
950 (try_crossjump_to_edge): Likewise.
951 * cfgexpand.c (expand_gimple_cond): Likewise.
952 (expand_gimple_tailcall): Likewise.
953 (construct_init_block): Likewise.
954 (construct_exit_block): Likewise.
955 * cfghooks.c (verify_flow_info): Check consistency of counts.
956 (dump_bb_for_graph): Do not dump frequencies.
957 (split_block_1): Do not update frequencies.
958 (split_edge): Do not update frequencies.
959 (make_forwarder_block): Do not update frequencies.
960 (duplicate_block): Do not update frequencies.
961 (account_profile_record): Do not update frequencies.
962 * cfgloop.c (find_subloop_latch_edge_by_profile): Use IPA counts
963 for global heuristics.
964 * cfgloopanal.c (average_num_loop_insns): Update to use to_frequency.
965 (expected_loop_iterations_unbounded): Use counts only.
966 * cfgloopmanip.c (scale_loop_profile): Simplify.
967 (create_empty_loop_on_edge): Simplify
969 (duplicate_loop_to_header_edge): Simplify
970 * cfgrtl.c (force_nonfallthru_and_redirect): Update profile.
971 (update_br_prob_note): Take care of removing note when profile
973 (relink_block_chain): Do not dump frequency.
974 (rtl_account_profile_record): Use to_frequency.
975 * cgraph.c (symbol_table::create_edge): Convert count to ipa count.
976 (cgraph_edge::redirect_call_stmt_to_calle): Conver tcount to ipa count.
977 (cgraph_update_edges_for_call_stmt_node): Likewise.
978 (cgraph_edge::verify_count_and_frequency): Update.
979 (cgraph_node::verify_node): Temporarily disable frequency verification.
980 * cgraphbuild.c (compute_call_stmt_bb_frequency): Use
982 (cgraph_edge::rebuild_edges): Convert to ipa counts.
983 * cgraphunit.c (init_lowered_empty_function): Do not initialize
985 (cgraph_node::expand_thunk): Update profile.
986 * except.c (dw2_build_landing_pads): Do not update frequency.
987 * final.c (compute_alignments): Use to_frequency.
988 (dump_basic_block_info): Do not dump frequency.
989 * gimple-pretty-print.c (dump_profile): Do not dump frequency.
990 (dump_gimple_bb_header): Do not dump frequency.
991 * gimple-ssa-isolate-paths.c (isolate_path): Do not update frequency;
993 * gimple-streamer-in.c (input_bb): Do not stream frequency.
994 * gimple-streamer-out.c (output_bb): Do not stream frequency.
995 * haifa-sched.c (sched_pressure_start_bb): Use to_freuqency.
996 (init_before_recovery): Do not update frequency.
997 (sched_create_recovery_edges): Do not update frequency.
998 * hsa-gen.c (convert_switch_statements): Do not update frequency.
999 * ipa-cp.c (ipcp_propagate_stage): Update search for max_count.
1000 (ipa_cp_c_finalize): Set max_count to uninitialized.
1001 * ipa-fnsummary.c (get_minimal_bb): Use counts.
1002 (param_change_prob): Use counts.
1003 * ipa-profile.c (ipa_profile_generate_summary): Do not summarize
1005 * ipa-split.c (consider_split): Use to_frequency.
1006 (split_function): Use to_frequency.
1007 * ira-build.c (loop_compare_func): Likewise.
1008 (mark_loops_for_removal): Likewise.
1009 (mark_all_loops_for_removal): Likewise.
1010 * loop-doloop.c (doloop_modify): Do not update frequency.
1011 * loop-unroll.c (unroll_loop_runtime_iterations): Do not update
1013 * lto-streamer-in.c (input_function): Update count_max.
1014 * omp-expand.c (expand_omp_taskreg): Update count_max.
1015 * omp-simd-clone.c (simd_clone_adjust): Update profile.
1016 * predict.c (maybe_hot_frequency_p): Use to_frequency.
1017 (maybe_hot_count_p): Use ipa counts only.
1018 (maybe_hot_bb_p): Simplify.
1019 (maybe_hot_edge_p): Simplify.
1020 (probably_never_executed): Do not take frequency argument.
1021 (probably_never_executed_bb_p): Do not pass frequency.
1022 (probably_never_executed_edge_p): Likewise.
1023 (combine_predictions_for_bb): Check that profile is nonzero.
1024 (propagate_freq): Do not set frequency.
1025 (drop_profile): Simplify.
1026 (counts_to_freqs): Simplify.
1027 (expensive_function_p): Use to_frequency.
1028 (propagate_unlikely_bbs_forward): Simplify.
1029 (determine_unlikely_bbs): Simplify.
1030 (estimate_bb_frequencies): Add hack to silence graphite issues.
1031 (compute_function_frequency): Use ipa counts.
1032 (pass_profile::execute): Update.
1033 (rebuild_frequencies): Use counts only.
1034 (force_edge_cold): Use counts only.
1035 * profile-count.c (profile_count::dump): Dump new count types.
1036 (profile_count::differs_from_p): Check compatiblity.
1037 (profile_count::to_frequency): New function.
1038 (profile_count::to_cgraph_frequency): New function.
1039 * profile-count.h (struct function): Declare.
1040 (enum profile_quality): Add profile_guessed_local and
1041 profile_guessed_global0.
1042 (class profile_proability): Decrease number of bits to 29;
1043 update from_reg_br_prob_note and to_reg_br_prob_note.
1044 (class profile_count: Update comment; decrease number of bits
1045 to 61. Check compatibility.
1046 (profile_count::compatible_p): New private member function.
1047 (profile_count::ipa_p): New member function.
1048 (profile_count::operator<): Handle global zero correctly.
1049 (profile_count::operator>): Handle global zero correctly.
1050 (profile_count::operator<=): Handle global zero correctly.
1051 (profile_count::operator>=): Handle global zero correctly.
1052 (profile_count::nonzero_p): New member function.
1053 (profile_count::force_nonzero): New member function.
1054 (profile_count::max): New member function.
1055 (profile_count::apply_scale): Handle IPA scalling.
1056 (profile_count::guessed_local): New member function.
1057 (profile_count::global0): New member function.
1058 (profile_count::ipa): New member function.
1059 (profile_count::to_frequency): Declare.
1060 (profile_count::to_cgraph_frequency): Declare.
1061 * profile.c (OVERLAP_BASE): Delete.
1062 (compute_frequency_overlap): Delete.
1063 (compute_branch_probabilities): Do not use compute_frequency_overlap.
1064 * regs.h (REG_FREQ_FROM_BB): Use to_frequency.
1065 * sched-ebb.c (rank): Use counts only.
1066 * shrink-wrap.c (handle_simple_exit): Use counts only.
1067 (try_shrink_wrapping): Use counts only.
1068 (place_prologue_for_one_component): Use counts only.
1069 * tracer.c (find_best_predecessor): Use to_frequency.
1070 (find_trace): Use to_frequency.
1071 (tail_duplicate): Use to_frequency.
1072 * trans-mem.c (expand_transaction): Do not update frequency.
1073 * tree-call-cdce.c: Do not update frequency.
1074 * tree-cfg.c (gimple_find_sub_bbs): Likewise.
1075 (gimple_merge_blocks): Likewise.
1076 (gimple_split_edge): Likewise.
1077 (gimple_duplicate_sese_region): Likewise.
1078 (gimple_duplicate_sese_tail): Likewise.
1079 (move_sese_region_to_fn): Likewise.
1080 (gimple_account_profile_record): Likewise.
1081 (insert_cond_bb): Likewise.
1082 * tree-complex.c (expand_complex_div_wide): Likewise.
1083 * tree-eh.c (lower_resx): Update profile.
1084 * tree-inline.c (copy_bb): Simplify count scaling; do not scale
1086 (initialize_cfun): Do not initialize frequencies
1087 (freqs_to_counts): Delete.
1088 (copy_cfg_body): Ignore count parameter.
1089 (copy_body): Update.
1090 (expand_call_inline): Update count_max.
1091 (optimize_inline_calls): Update count_max.
1092 (tree_function_versioning): Update count_max.
1093 * tree-ssa-coalesce.c (coalesce_cost_bb): Use to_frequency.
1094 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Do not update
1096 * tree-ssa-loop-im.c (execute_sm_if_changed): Use counts only.
1097 * tree-ssa-loop-ivcanon.c (unloop_loops): Do not update freuqency.
1098 (try_peel_loop): Likewise.
1099 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Use
1101 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Pass -1.
1102 (tree_transform_and_unroll_loop): Do not use frequencies
1103 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations):
1104 Use reliable prediction only.
1105 * tree-ssa-loop-unswitch.c (hoist_guard): Do not use frequencies.
1106 * tree-ssa-sink.c (select_best_block): Use to_frequency.
1107 * tree-ssa-tail-merge.c (replace_block_by): Temporarily disable
1108 probability scaling.
1109 * tree-ssa-threadupdate.c (create_block_for_threading): Do
1110 not update frequency
1111 (any_remaining_duplicated_blocks): Likewise.
1112 (update_profile): Likewise.
1113 (estimated_freqs_path): Delete.
1114 (freqs_to_counts_path): Delete.
1115 (clear_counts_path): Delete.
1116 (ssa_fix_duplicate_block_edges): Likewise.
1117 (duplicate_thread_path): Likewise.
1118 * tree-switch-conversion.c (gen_inbound_check): Use counts.
1119 * tree-tailcall.c (decrease_profile): Do not update frequency.
1120 (eliminate_tail_call): Likewise.
1121 * tree-vect-loop-manip.c (vect_do_peeling): Likewise.
1122 * tree-vect-loop.c (scale_profile_for_vect_loop): Likewise.
1123 (optimize_mask_stores): Likewise.
1124 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
1125 * ubsan.c (ubsan_expand_null_ifn): Update profile.
1126 (ubsan_expand_ptr_ifn): Update profile.
1127 * value-prof.c (gimple_ic): Simplify.
1128 * value-prof.h (gimple_ic): Update prototype.
1129 * ipa-inline-transform.c (inline_transform): Fix scaling conditoins.
1130 * ipa-inline.c (compute_uninlined_call_time): Be sure that
1132 (want_inline_self_recursive_call_p): Likewise.
1133 (resolve_noninline_speculation): Only cummulate defined counts.
1134 (inline_small_functions): Use nonzero_p.
1135 (ipa_inline): Do not access freed node.
1137 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
1139 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1140 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
1142 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
1144 * config/riscv/riscv.c (riscv_legitimize_move): Handle
1145 non-legitimate address.
1147 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1149 * config/rs6000/rs6000.md (*lt0_disi): Delete.
1150 (*lt0_<mode>di, *lt0_<mode>si): New.
1152 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
1154 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
1155 TARGET_PAIRED_FLOAT.
1157 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
1158 Jim Wilson <jim.wilson@linaro.org>
1160 * config/aarch64/aarch64-cores.def (saphira): New CPU.
1161 * config/aarch64/aarch64-tune.md: Regenerated.
1162 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
1163 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
1165 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
1167 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
1168 (arc_expand_prologue): Restore blink for millicode.
1169 * config/arc/linux.h (LINK_EH_SPEC): Defined.
1171 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
1174 * config/i386/i386.c (ix86_vector_duplicate_value): Use
1175 gen_vec_duplicate after forcing the scalar into a register.
1177 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
1179 * combine (try_combine): Print the insns input to try_combine to the
1182 2017-11-02 Steve Ellcey <sellcey@cavium.com>
1185 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
1186 Remove second argument from aarch64_process_target_attr call.
1187 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
1189 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
1191 (aarch64_handle_attr_arch): Remove second argument.
1192 (aarch64_handle_attr_cpu): Ditto.
1193 (aarch64_handle_attr_tune): Ditto.
1194 (aarch64_handle_attr_isa_flags): Ditto.
1195 (aarch64_process_one_target_attr): Ditto.
1196 (aarch64_process_target_attr): Ditto.
1197 (aarch64_option_valid_attribute_p): Remove second argument.
1198 on aarch64_process_target_attr call.
1200 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1202 * diagnostic.c: Include "selftest-diagnostic.h".
1203 (selftest::assert_location_text): New function.
1204 (selftest::test_diagnostic_get_location_text): New function.
1205 (selftest::diagnostic_c_tests): Call it.
1207 2017-11-02 David Malcolm <dmalcolm@redhat.com>
1209 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
1210 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
1211 (class selftest::test_diagnostic_context): Move to...
1212 * selftest-diagnostic.c: New file.
1213 * selftest-diagnostic.h: New file.
1215 2017-11-02 James Bowman <james.bowman@ftdichip.com>
1217 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
1218 offset range for FT32B.
1219 * config/ft32/ft32.h: option "mcompress" enables relaxation.
1220 * config/ft32/ft32.md: Add TARGET_NOPM.
1221 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
1222 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
1224 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
1226 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
1228 2017-11-02 Jeff Law <law@redhat.com>
1230 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
1231 virtual keyword on FINAL OVERRIDE members.
1233 * tree-ssa-propagate.h (ssa_propagation_engine): Group
1234 virtuals together. Add virtual destructor.
1235 (substitute_and_fold_engine): Similarly.
1237 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
1239 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
1241 2017-11-02 Richard Biener <rguenther@suse.de>
1243 PR tree-optimization/82795
1244 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
1246 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1248 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
1250 (gcc_GAS_CHECK_FEATURE): Remove.
1251 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
1252 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
1253 * configure: Regenerate.
1255 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
1257 * config/arc/arc.c (hwloop_optimize): Account for empty
1260 2017-11-02 Richard Biener <rguenther@suse.de>
1263 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
1264 Truncate ARRAY_REF index and element size.
1266 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1268 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
1270 2017-11-01 Jeff Law <law@redhat.com>
1272 * tree-ssa-ccp.c (ccp_folder): New class derived from
1273 substitute_and_fold_engine.
1274 (ccp_folder::get_value): New member function.
1275 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
1276 (ccp_fold_stmt): Remove prototype.
1277 (ccp_finalize): Call substitute_and_fold from the ccp_class.
1278 * tree-ssa-copy.c (copy_folder): New class derived from
1279 substitute_and_fold_engine.
1280 (copy_folder::get_value): Renamed from get_value.
1281 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
1282 * tree-vrp.c (vrp_folder): New class derived from
1283 substitute_and_fold_engine.
1284 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
1285 (vrp_folder::get_value): New member function.
1286 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
1287 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
1288 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
1289 provide a class interface to folder/substitute routines.
1290 (ssa_prop_fold_stmt_fn): Remove typedef.
1291 (ssa_prop_get_value_fn): Likewise.
1292 (subsitute_and_fold): Remove prototype.
1293 (replace_uses_in): Likewise.
1294 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
1295 Renamed from replace_uses_in. Call the virtual member function
1296 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
1297 (substitute_and_fold_dom_walker): Remove initialization of
1298 data member entries for calbacks. Add substitute_and_fold_engine
1299 member and initialize it.
1300 (substitute_and_fold_dom_walker::before_dom_children0: Use the
1301 member functions for get_value, replace_phi_args_in c
1302 replace_uses_in, and fold_stmt calls.
1303 (substitute_and_fold_engine::substitute_and_fold): Renamed from
1304 substitute_and_fold. Remove assert. Update ctor call.
1306 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
1307 (ssa_prop_visit_phi_fn): Likewise.
1308 (class ssa_propagation_engine): New class to provide an interface
1310 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
1312 (ssa_prop_visit_phi): Likewise.
1313 (ssa_propagation_engine::simulate_stmt): Moved into class.
1314 Call visit_phi/visit_stmt from the class rather than via
1315 file scoped static variables.
1316 (ssa_propagation_engine::simulate_block): Moved into class.
1317 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
1318 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
1319 set file scoped statics for the visit_stmt/visit_phi callbacks.
1320 * tree-complex.c (complex_propagate): New class derived from
1321 ssa_propagation_engine.
1322 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
1323 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
1324 (tree_lower_complex): Call ssa_propagate via the complex_propagate
1326 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
1327 ssa_propagation_engine.
1328 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
1329 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
1330 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
1331 * tree-ssa-copy.c (copy_prop): New class derived from
1332 ssa_propagation_engine.
1333 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
1334 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
1335 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
1336 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
1337 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
1338 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
1339 (execute_vrp): Call ssa_propagate from the vrp_prop class.
1341 2017-11-01 Jakub Jelinek <jakub@redhat.com>
1343 PR rtl-optimization/82778
1344 PR rtl-optimization/82597
1345 * compare-elim.c (struct comparison): Add in_a_setter field.
1346 (find_comparison_dom_walker::before_dom_children): Remove killed
1347 bitmap and df_simulate_find_defs call, instead walk the defs.
1348 Compute last_setter and initialize in_a_setter. Merge definitions
1349 with first initialization for a few variables.
1350 (try_validate_parallel): Use insn_invalid_p instead of
1351 recog_memoized. Return insn rather than just the pattern.
1352 (try_merge_compare): Fix up comment. Don't uselessly test if
1353 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
1355 (execute_compare_elim_after_reload): Remove df_chain_add_problem
1358 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1359 Alan Hayward <alan.hayward@arm.com>
1360 David Sherwood <david.sherwood@arm.com>
1362 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
1363 aarch64_hard_regno_nregs to get the number of registers
1366 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1367 Alan Hayward <alan.hayward@arm.com>
1368 David Sherwood <david.sherwood@arm.com>
1370 * config/aarch64/constraints.md (Upl): Rename to...
1372 * config/aarch64/aarch64.md
1373 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
1376 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1377 Alan Hayward <alan.hayward@arm.com>
1378 David Sherwood <david.sherwood@arm.com>
1380 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
1381 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
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_evpc_trn, aarch64_evpc_uzp)
1389 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
1390 (aarch64_evpc_dup): Generate rtl direcly, rather than using
1392 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
1393 of a single element.
1394 * config/aarch64/iterators.md: Add a comment above the permute
1395 unspecs to say that they are generated directly by
1396 aarch64_expand_vec_perm_const.
1397 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
1399 2017-11-01 Nathan Sidwell <nathan@acm.org>
1401 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
1403 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
1405 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
1406 and medany code models, and describe what they do.
1408 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1410 Revert accidental duplicate:
1412 * combine.c (can_change_dest_mode): Reject changes in
1413 REGMODE_NATURAL_SIZE.
1415 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
1417 PR rtl-optimization/64682
1418 PR rtl-optimization/69567
1419 PR rtl-optimization/69737
1420 PR rtl-optimization/82683
1421 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
1422 register mentioned in the note, drop the note, unless it came from I3,
1423 in which case it should go to I3 again.
1425 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1427 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
1428 and return false if not.
1429 (clear_bytes_written_by, live_bytes_read): Update accordingly.
1431 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1433 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
1434 range is known to be empty.
1436 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1437 Alan Hayward <alan.hayward@arm.com>
1438 David Sherwood <david.sherwood@arm.com>
1440 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
1441 and CONST_VECTOR_NUNITS instead of computing the number of units from
1442 the byte sizes of the vector and element.
1443 (simplify_binary_operation_1): Likewise.
1444 (simplify_const_binary_operation): Likewise.
1445 (simplify_ternary_operation): Likewise.
1447 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1448 Alan Hayward <alan.hayward@arm.com>
1449 David Sherwood <david.sherwood@arm.com>
1451 * var-tracking.c (INT_MEM_OFFSET): Replace with...
1452 (int_mem_offset): ...this new function.
1453 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
1454 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
1455 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
1458 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1459 Alan Hayward <alan.hayward@arm.com>
1460 David Sherwood <david.sherwood@arm.com>
1462 * lower-subreg.c (interesting_mode_p): New function.
1463 (compute_costs, find_decomposable_subregs, decompose_register)
1464 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
1465 (resolve_clobber, dump_choices): Use it.
1467 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1468 Alan Hayward <alan.hayward@arm.com>
1469 David Sherwood <david.sherwood@arm.com>
1471 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
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 * alias.c (find_base_value, find_base_term): Only process integer
1478 truncations. Check the precision rather than the size.
1480 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1481 Alan Hayward <alan.hayward@arm.com>
1482 David Sherwood <david.sherwood@arm.com>
1484 * machmode.h (is_narrower_int_mode): New function
1485 * optabs.c (expand_float, expand_fix): Use it.
1486 * dwarf2out.c (rotate_loc_descriptor): Likewise.
1488 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1489 Alan Hayward <alan.hayward@arm.com>
1490 David Sherwood <david.sherwood@arm.com>
1492 * rtl.h (narrower_subreg_mode): New function.
1493 * ira-color.c (update_costs_from_allocno): Use it.
1495 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1496 Alan Hayward <alan.hayward@arm.com>
1497 David Sherwood <david.sherwood@arm.com>
1499 * optabs-query.h (convert_optab_p): New function, split out from...
1500 (convert_optab_handler): ...here.
1501 (widening_optab_handler): Delete.
1502 (find_widening_optab_handler): Remove permit_non_widening parameter.
1503 (find_widening_optab_handler_and_mode): Likewise. Provide an
1504 override that operates on mode class wrappers.
1505 * optabs-query.c (widening_optab_handler): Delete.
1506 (find_widening_optab_handler_and_mode): Remove permit_non_widening
1507 parameter. Assert that the two modes are the same class and that
1508 the "from" mode is narrower than the "to" mode. Use
1509 convert_optab_handler instead of widening_optab_handler.
1510 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
1511 instead of widening_optab_handler.
1512 * expr.c (expand_expr_real_2): Update calls to
1513 find_widening_optab_handler.
1514 * optabs.c (expand_widen_pattern_expr): Likewise.
1515 (expand_binop_directly): Take the insn_code as a parameter.
1516 (expand_binop): Only call find_widening_optab_handler for
1517 conversion optabs; use optab_handler otherwise. Update calls
1518 to find_widening_optab_handler and expand_binop_directly.
1519 Use convert_optab_handler instead of widening_optab_handler.
1520 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
1521 find_widening_optab_handler and use scalar_mode rather than
1523 (convert_plusminus_to_widen): Likewise.
1525 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1526 Alan Hayward <alan.hayward@arm.com>
1527 David Sherwood <david.sherwood@arm.com>
1529 * machmode.h (fixed_size_mode): New class.
1530 * rtl.h (get_pool_mode): Return fixed_size_mode.
1531 * gengtype.c (main): Add fixed_size_mode.
1532 * target.def (get_raw_result_mode): Return a fixed_size_mode.
1533 (get_raw_arg_mode): Likewise.
1534 * doc/tm.texi: Regenerate.
1535 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
1536 * targhooks.c (default_get_reg_raw_mode): Likewise.
1537 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
1538 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
1539 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
1540 (msp430_get_raw_result_mode): Likewise.
1541 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
1542 * dbxout.c (dbxout_parms): Require fixed-size modes.
1543 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
1544 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
1545 * omp-low.c (lower_oacc_reductions): Likewise.
1546 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
1547 (simplify_subreg): Update accordingly.
1548 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
1549 (force_const_mem): Update accordingly. Return NULL_RTX for modes
1550 that aren't fixed-size.
1551 (get_pool_mode): Return a fixed_size_mode.
1552 (output_constant_pool_2): Take a fixed_size_mode.
1554 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1555 Alan Hayward <alan.hayward@arm.com>
1556 David Sherwood <david.sherwood@arm.com>
1558 * doc/rtl.texi (vec_series): Document.
1559 (const): Say that the operand can be a vec_series.
1560 * rtl.def (VEC_SERIES): New rtx code.
1561 * rtl.h (const_vec_series_p_1): Declare.
1562 (const_vec_series_p): New function.
1563 * emit-rtl.h (gen_const_vec_series): Declare.
1564 (gen_vec_series): Likewise.
1565 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
1566 (gen_vec_series): Likewise.
1567 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
1568 * simplify-rtx.c (simplify_unary_operation): Handle negations
1570 (simplify_binary_operation_series): New function.
1571 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
1572 (test_vector_ops_series): New function.
1573 (test_vector_ops): Call it.
1574 * config/powerpcspe/altivec.md (altivec_lvsl): Use
1575 gen_const_vec_series.
1576 (altivec_lvsr): Likewise.
1577 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
1579 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1580 Alan Hayward <alan.hayward@arm.com>
1581 David Sherwood <david.sherwood@arm.com>
1583 * doc/rtl.texi (const): Update description of address constants.
1584 Say that vector constants are allowed too.
1585 * common.md (E, F): Use CONSTANT_P instead of checking for
1587 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
1588 checking for CONST_VECTOR.
1589 * expmed.c (make_tree): Use build_vector_from_val for a CONST
1591 * expr.c (expand_expr_real_2): Check for vector modes instead
1592 of checking for CONST_VECTOR.
1593 * rtl.h (const_vec_p): New function.
1594 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
1595 (unwrap_const_vec_duplicate): Handle them here too.
1597 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1598 David Malcolm <dmalcolm@redhat.com>
1599 Alan Hayward <alan.hayward@arm.com>
1600 David Sherwood <david.sherwood@arm.com>
1602 * rtl.h (vec_duplicate_p): New function.
1603 * selftest-rtl.c (assert_rtx_eq_at): New function.
1604 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
1605 (assert_rtx_eq_at): Declare.
1606 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
1607 * selftest-run-tests.c (selftest::run_tests): Call it.
1608 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
1609 (simplify_unary_operation_1): Recursively handle vector duplicates.
1610 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
1612 (simplify_subreg): Handle subregs of vector duplicates.
1613 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
1614 (selftest::simplify_rtx_c_tests): New functions.
1616 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1617 Alan Hayward <alan.hayward@arm.com>
1618 David Sherwood <david.sherwood@arm.com>
1620 * emit-rtl.h (gen_const_vec_duplicate): Declare.
1621 (gen_vec_duplicate): Likewise.
1622 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
1624 (gen_const_vector): ...here.
1625 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
1626 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
1627 whose elements are all equal.
1628 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
1629 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
1630 (simplify_relational_operation): Likewise.
1631 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
1633 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
1634 (aarch64_expand_vector_init): Likewise.
1635 * config/arm/arm.c (neon_vdup_constant): Likewise.
1636 (neon_expand_vector_init): Likewise.
1637 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
1638 (arm_block_set_unaligned_vect): Likewise.
1639 (arm_block_set_aligned_vect): Likewise.
1640 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
1641 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
1642 (expand_vec_perm_even_odd_pack): Likewise.
1643 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
1644 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
1645 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
1646 gen_const_vec_duplicate.
1647 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
1648 * config/mips/mips.c (mips_gen_const_int_vector): Use
1649 gen_const_vec_duplicate.
1650 (mips_expand_vector_init): Use CONST0_RTX.
1651 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
1652 (define_split): Use gen_const_vec_duplicate.
1653 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
1654 (define_split): Use gen_const_vec_duplicate.
1655 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
1656 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
1657 * config/spu/spu.c (spu_const): Likewise.
1659 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
1660 Alan Hayward <alan.hayward@arm.com>
1661 David Sherwood <david.sherwood@arm.com>
1663 * combine.c (can_change_dest_mode): Reject changes in
1664 REGMODE_NATURAL_SIZE.
1666 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
1668 * configure.ac (--enable-libssp): New.
1669 (gcc_cv_libc_provides_ssp): Check for explicit setting before
1670 trying to determine target-specific default. Adjust indentation.
1671 * configure: Regenerated.
1672 * doc/install.texi (Configuration): Expand --disable-libssp
1675 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
1677 config/i386/i386.c (ix86_expand_epilogue): Correct stack
1680 2017-10-31 Martin Jambor <mjambor@suse.cz>
1683 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
1685 2017-10-31 David Malcolm <dmalcolm@redhat.com>
1687 * auto-profile.c (autofdo_source_profile::read): Use
1688 UNKNOWN_LOCATION rather than 0.
1689 * diagnostic-core.h (warning_at_rich_loc): Rename to...
1690 (warning_at): ...this overload.
1691 (warning_at_rich_loc_n): Rename to...
1692 (warning_n): ...this overload.
1693 (error_at_rich_loc): Rename to...
1694 (error_at): ...this overload.
1695 (pedwarn_at_rich_loc): Rename to...
1696 (pedwarn): ...this overload.
1697 (permerror_at_rich_loc): Rename to...
1698 (permerror): ...this overload.
1699 (inform_at_rich_loc): Rename to...
1700 (inform): ...this overload.
1701 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
1702 (diagnostic_n_impl_richloc): Rename to...
1703 (diagnostic_n_impl): ...this rich_location *-based decl.
1704 (inform_at_rich_loc): Rename to...
1705 (inform): ...this, and add an assertion.
1706 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
1707 (warning_at_rich_loc): Rename to...
1708 (warning_at): ...this, and add an assertion.
1709 (warning_at_rich_loc_n): Rename to...
1710 (warning_n): ...this, and add an assertion.
1711 (warning_n): Update location_t-based implementation for removal of
1712 location_t-based diagnostic_n_impl.
1713 (pedwarn_at_rich_loc): Rename to...
1714 (pedwarn): ...this, and add an assertion.
1715 (permerror_at_rich_loc): Rename to...
1716 (permerror): ...this, and add an assertion.
1717 (error_n): Update for removal of location_t-based diagnostic_n_impl.
1718 (error_at_rich_loc): Rename to...
1719 (error_at): ...this, and add an assertion.
1720 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
1721 (driver::do_spec_on_infiles): Likewise.
1722 * substring-locations.c (format_warning_va): Update for renaming
1723 of inform_at_rich_loc.
1725 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
1727 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
1728 _Float<N>X built-in functions so that the variant without the
1729 "__builtin_" prefix is only enabled for the GNU C and Objective C
1730 languages when they are in non-strict ANSI/ISO mode.
1731 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
1732 * target.def (floatn_builtin_p): Add a target hook to control
1733 whether _Float<N> and _Float<N>X built-in functions without the
1734 "__builtin_" prefix are enabled, and return true for C and
1735 Objective C in the default hook. Include langhooks.h in
1737 * targhooks.h (default_floatn_builtin_p): Likewise.
1738 * targhooks.c (default_floatn_builtin_p): Likewise.
1739 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
1740 floatn_builtin_p target hook.
1741 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
1743 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
1744 Eric Botcazou <ebotcazou@adacore.com>
1746 PR rtl-optimization/81803
1747 * lra-constraints.c (curr_insn_transform): Also reload the whole
1748 register for a strict subreg no wider than a word if this is for
1749 a WORD_REGISTER_OPERATIONS target.
1751 2017-10-31 Jason Merrill <jason@redhat.com>
1753 * gdbinit.in: Skip over inlines from timevar.h.
1755 2017-10-31 Martin Liska <mliska@suse.cz>
1757 * doc/gcov.texi: Document new option.
1758 * gcov.c (print_usage): Likewise print it.
1759 (process_args): Support the argument.
1760 (format_count): New function.
1761 (format_gcov): Use the function.
1763 2017-10-31 Martin Liska <mliska@suse.cz>
1765 * gcov.c (struct name_map): do not use typedef.
1766 Define operator== and operator<.
1767 (name_search): Remove.
1768 (name_sort): Remove.
1769 (main): Do not allocate names.
1770 (process_file): Add vertical space.
1771 (generate_results): Use std::find.
1772 (release_structures): Do not release memory.
1773 (find_source): Use std::find.
1775 2017-10-31 Martin Liska <mliska@suse.cz>
1777 * gcov.c (struct line_info): Remove it's typedef.
1778 (line_info::line_info): Add proper ctor.
1779 (line_info::has_block): Do not use a typedef.
1780 (struct source_info): Do not use typedef.
1781 (circuit): Likewise.
1782 (get_cycles_count): Likewise.
1783 (output_intermediate_file): Iterate via vector iterator.
1784 (add_line_counts): Use std::vector methods.
1785 (accumulate_line_counts): Likewise.
1786 (output_lines): Likewise.
1788 2017-10-31 Martin Liska <mliska@suse.cz>
1790 * gcov.c (struct source_info): Remove typedef.
1791 (source_info::source_info): Add proper ctor.
1792 (accumulate_line_counts): Use struct, not it's typedef.
1793 (output_gcov_file): Likewise.
1794 (output_lines): Likewise.
1795 (main): Do not allocate an array.
1796 (output_intermediate_file): Use size of vector container.
1797 (process_file): Resize the vector.
1798 (generate_results): Do not preallocate, use newly added vector
1800 (release_structures): Do not release sources.
1801 (find_source): Use vector methods.
1802 (add_line_counts): Do not use typedef.
1804 2017-10-31 Martin Liska <mliska@suse.cz>
1806 * doc/gcov.texi: Document that.
1807 * gcov.c (add_line_counts): Mark lines with a non-executed
1809 (output_line_beginning): Handle such lines.
1810 (output_lines): Pass new argument.
1811 (output_intermediate_file): Print it in intermediate format.
1813 2017-10-31 Martin Liska <mliska@suse.cz>
1815 * color-macros.h: New file.
1816 * diagnostic-color.c: Factor out color related to macros to
1818 * doc/gcov.texi: Document -k option.
1819 * gcov.c (INCLUDE_STRING): Include string.h.
1820 (print_usage): Add -k option.
1821 (process_args): Parse it.
1822 (pad_count_string): New function.
1823 (output_line_beginning): Likewise.
1824 (DEFAULT_LINE_START): New macro.
1825 (output_lines): Support color output.
1827 2017-10-31 Martin Liska <mliska@suse.cz>
1829 PR gcov-profile/82633
1830 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
1831 their interaction with GCOV infrastructure.
1832 * configure.ac: Add -fkeep-{inline,static}-functions to
1834 * configure: Regenerate.
1836 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
1839 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
1841 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
1844 * config/rs6000/rs6000.md (allocate_stack): Force update interval
1845 into a register if it does not fit into an immediate offset field.
1847 2017-10-31 Olivier Hainque <hainque@adacore.com>
1849 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
1851 2017-10-31 Julia Koval <julia.koval@intel.com>
1853 * config.gcc: Add gfniintrin.h.
1854 * config/i386/gfniintrin.h: New.
1855 * config/i386/i386-builtin-types.def
1856 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
1857 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
1858 __builtin_ia32_vgf2p8affineinvqb_v32qi,
1859 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
1860 __builtin_ia32_vgf2p8affineinvqb_v16qi,
1861 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
1862 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
1863 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
1864 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
1865 V64QI_FTYPE_V64QI_V64QI_INT): New types.
1866 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
1867 * config/i386/immintrin.h: Include gfniintrin.h.
1868 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
1870 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
1872 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
1874 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1876 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
1877 (arm_ashldi3_1bit): Remove pattern.
1878 (ashrdi3): Remove shift by 1 expansion.
1879 (arm_ashrdi3_1bit): Remove pattern.
1880 (lshrdi3): Remove shift by 1 expansion.
1881 (arm_lshrdi3_1bit): Remove pattern.
1882 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
1883 cost of ashldi3 by 1.
1884 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
1885 (<shift>di3_neon): Likewise.
1887 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
1889 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
1890 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
1891 and (*aarch64_simd_mov<VQ:mode>).
1892 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
1893 pattern alternative.
1894 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
1895 attributes to match pattern alternative.
1897 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
1899 * config.gcc (powerpc*-*-*): Add emmintrin.h.
1900 * config/rs6000/emmintrin.h: New file.
1901 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
1903 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
1905 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
1906 * (movdi_vfp_cortexa8): Remove pattern.
1908 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1910 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
1912 (Specific, alpha*-dec-osf5.1): Remove.
1913 (Specific, mips-sgi-irix5): Remove.
1914 (Specific, mips-sgi-irix6): Remove.
1916 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1919 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
1920 arguments to clear_bit_region_be.
1922 2017-10-30 Jim Wilson <wilson@tuliptree.org>
1924 * gimplify.c: Include memmodel.h.
1926 2017-10-30 Martin Jambor <mjambor@suse.cz>
1928 * omp-grid.c (grid_attempt_target_gridification): Also insert a
1929 condition whether loop should be executed at all.
1931 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
1933 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
1934 gimple folding of vec_madd() intrinsics.
1935 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
1936 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
1937 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
1939 2017-10-30 Richard Biener <rguenther@suse.de>
1941 PR tree-optimization/82762
1943 2017-10-23 Richard Biener <rguenther@suse.de>
1945 PR tree-optimization/82129
1947 2017-08-01 Richard Biener <rguenther@suse.de>
1949 PR tree-optimization/81181
1950 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1951 (compute_antic): ... end of iteration here.
1953 2017-10-30 Joseph Myers <joseph@codesourcery.com>
1955 * doc/invoke.texi (C Dialect Options): Document -std=c17,
1956 -std=iso9899:2017 and -std=gnu17.
1957 * doc/standards.texi (C Language): Document C17 support.
1958 * doc/cpp.texi (Overview): Mention -std=c17.
1959 (Standard Predefined Macros): Document C11 and C17 values of
1960 __STDC_VERSION__. Do not refer to C99 support as incomplete.
1961 * doc/extend.texi (Inline): Do not list individual options for
1962 standards newer than C99.
1963 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
1965 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
1968 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
1970 * asan.c (asan_finish_file): Align asan globals array by shadow
1973 2017-10-30 Jakub Jelinek <jakub@redhat.com>
1976 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
1977 (struct store_immediate_info): Add bitregion_start and bitregion_end
1979 (store_immediate_info::store_immediate_info): Add brs and bre
1980 arguments and initialize bitregion_{start,end} from those.
1981 (struct merged_store_group): Add bitregion_start, bitregion_end,
1982 align_base and mask fields. Drop unnecessary struct keyword from
1983 struct store_immediate_info. Add do_merge method.
1984 (clear_bit_region_be): Use memset instead of loop storing zeros.
1985 (merged_store_group::do_merge): New method.
1986 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
1987 stores as long as the surrounding bitregions have no gaps.
1988 (merged_store_group::merge_overlapping): Use do_merge.
1989 (merged_store_group::apply_stores): Test that bitregion_{start,end}
1990 is byte aligned, rather than requiring that start and width are
1991 byte aligned. Drop unnecessary struct keyword from
1992 struct store_immediate_info. Allocate and populate also mask array.
1993 Make start of the arrays relative to bitregion_start rather than
1994 start and size them according to bitregion_{end,start} difference.
1995 (struct imm_store_chain_info): Drop unnecessary struct keyword from
1996 struct store_immediate_info.
1997 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
1998 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
1999 struct keyword from struct store_immediate_info.
2000 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
2001 between stores as long as the surrounding bitregions have no gaps.
2003 (struct split_store): Add orig non-static data member.
2004 (split_store::split_store): Initialize orig to false.
2005 (find_constituent_stmts): Return store_immediate_info *, non-NULL
2006 if there is exactly a single original stmt. Change stmts argument
2007 to pointer from reference, if NULL, don't push anything to it. Add
2008 first argument, use it to optimize skipping over orig stmts that
2009 are known to be before bitpos already. Simplify.
2010 (split_group): Return unsigned int count how many stores are or
2011 would be needed rather than a bool. Add allow_unaligned argument.
2012 Change split_stores argument from reference to pointer, if NULL,
2013 only do a dry run computing how many stores would be produced.
2014 Rewritten algorithm to use both alignment and misalign if
2015 !allow_unaligned and handle bitfield stores with gaps.
2016 (imm_store_chain_info::output_merged_store): Set start_byte_pos
2017 from bitregion_start instead of start. Compute allow_unaligned
2018 here, if true, do 2 split_group dry runs to compute which one
2019 produces fewer stores and prefer aligned if equal. Punt if
2020 new count is bigger or equal than original before emitting any
2021 statements, rather than during that. Remove no longer needed
2022 new_ssa_names tracking. Replace num_stmts with
2023 split_stores.length (). Use 32-bit stack allocated entries
2024 in split_stores auto_vec. Try to reuse original store lhs/rhs1
2025 if possible. Handle bitfields with gaps.
2026 (pass_store_merging::execute): Ignore bitsize == 0 stores.
2027 Compute bitregion_{start,end} for the stores and construct
2028 store_immediate_info with that. Formatting fixes.
2030 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
2033 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
2034 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
2036 2017-10-29 Jim Wilson <wilson@tuliptree.org>
2038 * gimplify.c: Include tm_p.h.
2040 * common.opt (gcoff): Re-add as ignored option.
2041 (gcoff1, gcoff2, gcoff3): Likewise.
2043 * Makefile.in (OBJS): Delete sdbout.o.
2044 (GTFILES): Delete $(srcdir)/sdbout.c.
2045 * debug.h: Delete sdb_debug_hooks.
2046 * final.c: Delete sdbout.h include.
2047 (final_scan_insn): Delete SDB_DEBUG check.
2048 (rest_of_clean_state): Likewise.
2049 * output.h: Delete sdb_begin_function_line.
2052 * toplev.c: Delete sdbout.h include.
2053 (process_options): Delete SDB_DEBUG check.
2054 * tree-core.h (tree_type_common): Delete pointer field of
2056 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
2057 TYPE_SYMTAB_POINTER.
2058 * tree.h (TYPE_SYMTAB_POINTER): Delete.
2059 (TYPE_SYMTAB_IS_POINTER): Delete.
2060 (TYPE_SYMTAB_IS_DIE): Renumber.
2061 * xcoffout.c: Refer to former sdbout.c file.
2062 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
2064 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
2065 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
2066 (Debugging Options): Delete -gcoff.
2067 (-gstabs): Delete SDB reference.
2069 (-gcoff@var{level}): Delete.
2070 * doc/passes.texi (Debugging information output): Delete SDB and
2071 sdbout.c references.
2072 * doc/tm.texi: Regenerate.
2073 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
2074 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
2076 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
2077 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
2078 (SDB_DEBUGGING_INFO): Delete.
2079 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
2080 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
2081 * target.def (output_source_filename): Delete COFF reference.
2083 * common.opt (gcoff): Delete.
2084 (gxcoff+): Update Negative chain.
2085 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
2087 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
2088 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
2089 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
2091 (expand_function_start): Change sdb reference to past tense.
2092 (expand_function_end): Change sdb reference to past tense.
2093 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
2094 * opts.c (debug_type_names): Delete coff entry.
2095 (common_handle_option): Delete OPT_gcoff case.
2096 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
2098 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
2099 * config/cris/cris.h: Delete SDB reference in comment.
2100 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
2101 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
2102 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
2103 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
2105 (ix86_expand_prologue): Likewise.
2106 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
2107 * config/ia64/ia64.h: Likewise.
2108 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
2109 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
2111 * config/mmix/mmix.h: Likewise.
2112 * config/nds32/nds32.c: Likewise.
2113 * config/stormy/storym16.h: Likewise.
2114 * config/visium/visium.h: Likewise.
2115 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
2117 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
2119 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
2120 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
2121 FRAME_POINTER_REGNUM point at high end of local var area.
2123 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
2125 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
2126 Move comment around. Do not reset best_edge for a copiable
2127 destination if the copy would cause a partition change.
2128 (better_edge_p): Remove redundant check.
2130 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2132 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
2134 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
2136 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
2137 for math functions that have _Float<N> and _Float<N>X variants.
2138 (mathfn_built_in_2): Add support for math functions that have
2139 _Float<N> and _Float<N>X variants.
2140 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
2141 (expand_builtin_mathfn_ternary): Add support for fma with
2142 _Float<N> and _Float<N>X variants.
2143 (expand_builtin): Likewise.
2144 (fold_builtin_3): Likewise.
2145 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
2146 create math function _Float<N> and _Float<N>X variants as external
2148 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
2149 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
2150 the __builtin_ prefix and if not strict ansi, without the prefix.
2151 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
2152 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
2153 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
2154 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
2155 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
2156 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
2157 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
2158 function signatures for fma _Float<N> and _Float<N>X variants.
2159 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
2160 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
2161 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
2162 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
2163 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
2164 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
2165 * gencfn-macros.c (print_case_cfn): Add support for math functions
2166 that have _Float<N> and _Float<N>X variants.
2167 (print_define_operator_list): Likewise.
2168 (fltfn_suffixes): Likewise.
2170 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
2171 for math functions that have _Float<N> and _Float<N>X variants.
2172 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
2173 and _Float<N>X variants.
2174 (COPYSIGN): Likewise.
2177 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
2178 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
2180 (integer_valued_read_call_p): Likewise.
2181 * fold-const-call.c (fold_const_call_ss): Likewise.
2182 (fold_const_call_sss): Add support for copysign, fmin, and fmax
2183 _Float<N> and _Float<N>X variants.
2184 (fold_const_call_ssss): Add support for fma _Float<N> and
2185 _Float<N>X variants.
2186 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
2187 support for copysign and fma _Float<N> and _Float<N>X variants.
2188 (backprop::process_builtin_call_use): Likewise.
2189 * tree-call-cdce.c (can_test_argument_range); Add support for
2190 sqrt _Float<N> and _Float<N>X variants.
2191 (edom_only_function): Likewise.
2192 (get_no_error_domain): Likewise.
2193 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
2194 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
2195 copysign _Float<N> and _Float<N>X variants.
2196 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
2197 handled by machine independent code.
2198 (FMAF128): Likewise.
2199 * doc/cpp.texi (Common Predefined Macros): Document defining
2200 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
2201 fma _Float<N> and _Float<N>X variants.
2203 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
2206 * config/i386/i386-modes.def (CCFPU): Remove definition.
2207 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
2208 (ix86_cc_modes_compatible): Ditto.
2209 (ix86_expand_carry_flag_compare): Ditto.
2210 (ix86_expand_int_movcc): Ditto.
2211 (ix86_expand_int_addcc): Ditto.
2212 (ix86_reverse_condition): Ditto.
2213 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
2214 Return true/false for unordered/ordered fp comparisons.
2215 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
2216 (ix86_prepare_fp_compare_args): Update for rename.
2217 (ix86_expand_fp_compare): Update for rename. Generate unordered
2218 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
2219 (ix86_expand_sse_compare_and_jump): Ditto.
2220 * config/i386/predicates.md (fcmov_comparison_operator):
2221 Remove CCFPU mode handling.
2222 (ix86_comparison_operator): Ditto.
2223 (ix86_carry_flag_operator): Ditto.
2224 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
2225 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
2226 (*cmpu<mode>_cc_i387): Ditto.
2227 (FPCMP): Remove mode iterator.
2228 (unord): Remove mode attribute.
2229 (unord_subst): New define_subst transformation
2230 (unord): New define_subst attribute.
2232 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
2233 (*cmpi<unord>xf_i387): Ditto.
2234 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
2235 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
2236 using unord_subst transformation.
2237 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
2238 (round_saeonly): Also handle CCFP mode.
2239 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
2240 Remove UNSPEC_SAHF unspec handling.
2242 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2244 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
2246 2017-10-27 Jeff Law <law@redhat.com>
2248 * gimple-ssa-sprintf.c: Include domwalk.h.
2249 (class sprintf_dom_walker): New class, derived from dom_walker.
2250 (sprintf_dom_walker::before_dom_children): New function.
2251 (struct call_info): Moved into sprintf_dom_walker class
2252 (compute_formath_length, handle_gimple_call): Likewise.
2253 (sprintf_length::execute): Call the dominator walker rather
2254 than walking the statements.
2256 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
2257 gimple statement locations.
2258 (check_array_bounds): Corresponding changes. Get the statement's
2259 location directly from wi->stmt.
2261 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
2264 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
2266 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
2268 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
2269 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
2271 2017-10-27 Jakub Jelinek <jakub@redhat.com>
2274 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
2275 * config/i386/i386.c (maybe_get_pool_constant): Removed.
2276 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
2277 maybe_get_pool_constant.
2278 * config/i386/predicates.md (zero_extended_scalar_load_operand):
2281 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2283 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
2284 2.26 caveat. Update gas and gld versions.
2285 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
2288 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2290 * cgraph.h (set_malloc_flag): Declare.
2291 * cgraph.c (set_malloc_flag_1): New function.
2292 (set_malloc_flag): Likewise.
2293 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
2294 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
2296 (read_ipa_call_summary): Add support for reading is_return_callee.
2297 (write_ipa_call_summary): Stream is_return_callee.
2298 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
2299 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
2300 ipa-prop.h, ipa-fnsummary.h.
2301 (pure_const_names): Change to static.
2302 (malloc_state_e): Define.
2303 (malloc_state_names): Define.
2304 (funct_state_d): Add field malloc_state.
2305 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
2306 (check_retval_uses): New function.
2307 (malloc_candidate_p): Likewise.
2308 (analyze_function): Add support for malloc attribute.
2309 (pure_const_write_summary): Stream malloc_state.
2310 (pure_const_read_summary): Add support for reading malloc_state.
2311 (dump_malloc_lattice): New function.
2312 (propagate_malloc): New function.
2313 (warn_function_malloc): New function.
2314 (ipa_pure_const::execute): Call propagate_malloc and
2315 ipa_free_fn_summary.
2316 (pass_local_pure_const::execute): Add support for malloc attribute.
2317 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
2318 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
2320 2017-10-27 Martin Liska <mliska@suse.cz>
2322 PR gcov-profile/82457
2323 * doc/invoke.texi: Document that one needs a non-strict ISO mode
2324 for fork-like functions to be properly instrumented.
2326 2017-10-27 Richard Biener <rguenther@suse.de>
2329 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
2330 info when we redirected EH.
2332 2017-10-26 Michael Collison <michael.collison@arm.com>
2334 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
2336 (<optab>_trunchf<GPI:mode>2: New pattern.
2337 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
2338 * config/aarch64/iterators.md (wv): New mode attribute.
2339 (vf, VF): New mode attributes.
2340 (vgp, VGP): New mode attributes.
2341 (s): Update attribute with SImode and DImode prefixes.
2343 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2345 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
2346 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
2347 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
2348 (nios2_option_overide): Initialize it. Don't allow R0-relative
2349 addressing with PIC.
2350 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
2351 (nios2_symbolic_constant_p): Likewise.
2352 (nios2_legitimate_address_p): Likewise.
2353 (nios2_r0rel_section_name_p): New.
2354 (nios2_symbol_ref_in_r0rel_data_p): New.
2355 (nios2_emit_move_sequence): Handle r0rel_constant_p.
2356 (r0rel_constant_p): New.
2357 (nios2_print_operand_address): Handle r0rel_constant_p.
2358 (nios2_cdx_narrow_form_p): Likewise.
2359 * config/nios2/nios2.opt (mr0rel-sec=): New option.
2360 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
2361 (Nios II Options): Document -mr0rel-sec.
2363 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
2365 * config/nios2/nios2.c: Include xregex.h.
2366 (nios2_gprel_sec_regex): New.
2367 (nios2_option_overide): Initialize it. Don't allow GP-relative
2368 addressing with PIC.
2369 (nios2_small_section_name_p): Check for regex match.
2370 * config/nios2/nios2.opt (mgprel-sec=): New option.
2371 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
2372 (Nios II Options): Document -mgprel-sec.
2374 2017-10-26 Jim Wilson <wilson@tuliptree.org>
2376 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
2378 2017-10-26 Tom de Vries <tom@codesourcery.com>
2380 PR tree-optimization/82707
2381 * gimple.c (gimple_copy): Fix unsharing of
2382 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
2384 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
2386 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
2387 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
2388 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
2389 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
2390 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
2391 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
2392 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
2393 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
2394 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
2395 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
2396 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
2397 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
2398 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
2399 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
2400 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
2401 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
2402 _mm512_mask_cmpunord_ps_mask): New intrinsics.
2404 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
2406 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
2408 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
2409 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
2410 warning to rs6000.c. Remove the Undocumented flag, since it has
2412 (-mabi=ibmlongdouble): Likewise.
2413 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
2414 already set, set the default format for long double.
2415 (rs6000_debug_reg_global): Print whether long double is IBM or
2417 (rs6000_option_override_internal): Rework setting long double
2418 format. Only warn if the user is changing the long double default
2419 and they did not use -Wno-psabi.
2420 * doc/invoke.texi (PowerPC options): Update the documentation for
2421 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
2423 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2424 Alan Hayward <alan.hayward@arm.com>
2425 David Sherwood <david.sherwood@arm.com>
2427 * rtl.h (wider_subreg_mode): New function.
2428 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
2429 rather than an unsigned int *.
2430 * ira-color.c (regno_max_ref_width): Replace with...
2431 (regno_max_ref_mode): ...this new variable.
2432 (coalesced_pseudo_reg_slot_compare): Update accordingly.
2433 Use wider_subreg_mode.
2434 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
2435 rather than an unsigned int *.
2436 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
2437 (process_alt_operands): Likewise.
2438 (invariant_p): Likewise.
2439 * lra-spills.c (assign_mem_slot): Likewise.
2440 (add_pseudo_to_slot): Likewise.
2441 * lra.c (collect_non_operand_hard_regs): Likewise.
2442 (add_regs_to_insn_regno_info): Likewise.
2443 * reload1.c (regno_max_ref_width): Replace with...
2444 (regno_max_ref_mode): ...this new variable.
2445 (reload): Update accordingly. Update call to
2446 ira_sort_regnos_for_alter_reg.
2447 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
2448 (init_eliminable_invariants): Update to use regno_max_ref_mode.
2449 (scan_paradoxical_subregs): Likewise.
2451 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2453 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
2454 (aarch64_frame): Add emit_frame_chain boolean.
2455 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2456 Move eh_return case to aarch64_layout_frame.
2457 (aarch64_layout_frame): Initialize emit_frame_chain.
2458 (aarch64_expand_prologue): Use emit_frame_chain.
2460 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2462 * config/aarch64/aarch64.c (aarch64_layout_frame):
2463 Ensure LR is always stored at the bottom of the callee-saves.
2464 Remove rarely used frame layout which saves callee-saves at top of
2465 frame, so the store of LR can be used as a valid probe in all cases.
2467 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
2469 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
2470 Improve unaligned TImode/TFmode base/offset split.
2472 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2473 Alan Hayward <alan.hayward@arm.com>
2474 David Sherwood <david.sherwood@arm.com>
2476 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
2477 * combine.c (find_single_use_1): Likewise.
2478 (expand_field_assignment): Likewise.
2479 (move_deaths): Likewise.
2480 * lra-constraints.c (simplify_operand_subreg): Likewise.
2481 (curr_insn_transform): Likewise.
2482 * lra.c (collect_non_operand_hard_regs): Likewise.
2483 (add_regs_to_insn_regno_info): Likewise.
2484 * rtlanal.c (reg_referenced_p): Likewise.
2485 (covers_regno_no_parallel_p): Likewise.
2487 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2489 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
2490 Don't print any bits outside the precision of the value.
2491 * wide-int.cc (test_printing): Add some new tests.
2493 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2495 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
2496 supports -xbrace_comment option.
2497 * configure: Regenerate.
2498 * config.in: Regenerate.
2499 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
2500 (ASM_CPU_SPEC): Use it.
2502 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
2504 * target.def (static_rtx_alignment): New hook.
2505 * targhooks.h (default_static_rtx_alignment): Declare.
2506 * targhooks.c (default_static_rtx_alignment): New function.
2507 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
2508 * doc/tm.texi: Regenerate.
2509 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
2510 instead of targetm.constant_alignment. Remove call to
2512 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2513 (cris_preferred_mininum_alignment): New function, split out from...
2514 (cris_constant_alignment): ...here.
2515 (cris_static_rtx_alignment): New function.
2516 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
2518 (ix86_constant_alignment): ...here.
2519 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2520 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2521 (mmix_static_rtx_alignment): New function.
2522 * config/spu/spu.c (spu_static_rtx_alignment): New function.
2523 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
2525 2017-10-26 Tamar Christina <tamar.christina@arm.com>
2528 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
2529 Add flag_trapping_math and flag_fp_int_builtin_inexact.
2531 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
2533 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
2534 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
2535 mark as a sign-extending load.
2536 (local_pic_load_u): Define.
2538 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
2541 * fold-const.c (operand_equal_for_comparison_p): Also return true
2542 if ARG0 is a simple variant of ARG1 with narrower precision.
2543 (fold_ternary_loc): Always pass unstripped operands to the predicate.
2545 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
2547 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
2549 * i386.h (processor_costs): Add gather_static, gather_per_elt,
2550 scatter_static, scatter_per_elt.
2551 * x86-tune-costs.h: Add new cost entries.
2553 2017-10-25 Richard Biener <rguenther@suse.de>
2555 * tree-ssa-sccvn.h (vn_eliminate): Declare.
2556 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
2557 class pass_fre): Move to ...
2558 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
2559 class pass_fre): ... here and adjust for statistics.
2561 2017-10-25 Jakub Jelinek <jakub@redhat.com>
2564 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
2565 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
2566 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
2567 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
2570 2017-10-25 Richard Biener <rguenther@suse.de>
2572 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
2573 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
2574 eliminate_push_avail, eliminate_insert): Move inside...
2575 (class eliminate_dom_walker): ... this class in preparation
2577 (fini_eliminate): Remove by merging with ...
2578 (eliminate): ... this function. Adjust for class changes.
2579 (pass_pre::execute): Remove fini_eliminate call.
2580 (pass_fre::execute): Likewise.
2582 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2585 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
2586 (VPERMI2, VPERMI2I): New mode iterators.
2587 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
2588 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
2590 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
2591 mode iterator. Remove 3 old define_insn patterns.
2592 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
2593 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
2594 VPERMI2 mode iterator, remove the other two expanders.
2595 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
2596 to use VPERMI2 mode iterator, add another alternative for vpermi2*
2597 instructions, remove the other two patterns.
2598 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
2599 mode iterator, remove the other two patterns.
2600 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
2601 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
2602 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
2603 and adjust argument order accordingly.
2604 (ix86_expand_vec_perm): Adjust caller.
2605 (expand_vec_perm_1): Likewise.
2606 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
2607 (expand_vec_perm_vpermt2_vpshub2): ... this.
2608 (ix86_expand_vec_perm_const_1): Adjust caller.
2609 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
2612 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
2613 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
2614 (vec_shl_<mode>): Remove unused expander.
2615 (avx512bw_<shift_insn><mode>3): New define_insn.
2616 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
2617 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
2619 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
2622 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
2625 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2627 PR rtl-optimization/82396
2628 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
2629 (autopref_multipass_init): Simplify initialization.
2630 (autopref_rank_data): Simplify sort order.
2631 * gcc/sched-int.h (autopref_multipass_data_): Remove
2632 multi_mem_insn_p, min_offset and max_offset.
2634 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
2637 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
2638 Check special value of flag_omit_frame_pointer.
2639 (aarch64_can_eliminate): Likewise.
2640 (aarch64_override_options_after_change_1): Simplify handling of
2641 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
2643 2017-10-24 Richard Biener <rguenther@suse.de>
2645 PR tree-optimization/82697
2646 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
2647 zero for conditional load and unconditional store.
2649 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2651 * doc/install.texi: Document bootstrap-cet.
2653 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
2656 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
2657 ENDBR instruction at function entrance if function is only
2660 2017-10-24 Jakub Jelinek <jakub@redhat.com>
2663 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
2664 patterns to better describe from which operation the CF is computed.
2665 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
2666 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
2667 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
2668 is 0, use _0 suffixed expanders instead of emitting a comparison
2671 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
2673 * config/i386/i386.md(*movsf_internal, *movdf_internal):
2674 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
2676 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
2679 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
2680 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
2681 * loop-iv.c (iv_get_reaching_def): Likewise.
2682 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
2683 variable is promoted and the partition contains undefined values.
2685 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2687 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
2689 (nios2_address_cost): Define.
2690 (nios2_legitimize_address): Recognize (exp + constant) directly.
2691 (TARGET_ADDRESS_COST): Define.
2693 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2695 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
2696 (nios2_symbolic_memory_operand_p): Declare.
2697 (nios2_split_large_constant): Declare.
2698 (nios2_split_symbolic_memory_operand): Declare.
2699 * config/nios2/nios2.c: Adjust includes.
2700 (nios2_symbolic_constant_allowed): New.
2701 (nios2_symbolic_constant_p): New.
2702 (nios2_plus_symbolic_constant_p): New.
2703 (nios2_valid_addr_expr_p): Recognize addresses involving
2705 (nios2_legitimate_address_p): Likewise, also LO_SUM.
2706 (nios2_symbolic_memory_operand_p): New.
2707 (nios2_large_constant_p): New.
2708 (nios2_split_large_constant): New.
2709 (nios2_split_plus_large_constant): New.
2710 (nios2_split_symbolic_memory_operand): New.
2711 (nios2_legitimize_address): Code refactoring. Handle addresses
2712 involving symbolic constants.
2713 (nios2_emit_move_sequence): Likewise.
2714 (nios2_print_operand): Improve error output.
2715 (nios2_print_operand_address): Handle LO_SUM.
2716 (nios2_cdx_narrow_form_p): Likewise.
2717 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
2718 operands involving symbolic constants.
2719 (movhi_internal, movsi_internal): Likewise.
2720 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
2721 (extendhisi2, extendqi<mode>2): Likewise.
2723 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2725 * tree-pass.h (PROP_rtl_split_insns): Define.
2726 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
2728 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
2730 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
2732 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2735 * target.def (const_not_ok_for_debug_p): Default to
2736 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
2737 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
2738 * targhooks.c (default_const_not_ok_for_debug_p): New function.
2739 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
2740 which targetm.const_not_ok_for_debug_p returned true.
2741 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
2743 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
2745 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
2746 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
2747 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
2749 (ix86_const_not_ok_for_debug_p): New function.
2750 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
2751 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
2753 2017-10-23 David Malcolm <dmalcolm@redhat.com>
2756 * system.h: Conditionally include "unique-ptr.h" if
2757 INCLUDE_UNIQUE_PTR is defined.
2758 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
2759 of defining INCLUDE_UNIQUE_PTR before including "system.h".
2761 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
2763 * config/rl78/rl78.md: New define_expand "subdi3".
2765 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
2768 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
2769 DF_REF_INSN if DF_REF_INSN_INFO is false.
2771 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
2773 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
2774 xmm_move instead of sse_move.
2775 (sse_store_index): New function.
2776 (ix86_register_move_cost): Be more sensible about mismatch stall;
2777 model AVX moves correctly; make difference between sse->integer and
2779 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
2780 moves; make difference between SSE and AVX.
2781 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
2782 and zmm_move. Increase size of sse load and store tables;
2783 add unaligned load and store tables; add ssemmx_to_integer.
2784 * x86-tune-costs.h: Update all entries according to real
2785 move latencies from Agner Fog's manual and chip documentation.
2787 2017-10-23 Jakub Jelinek <jakub@redhat.com>
2790 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
2791 * config/i386/constraints.md (Wf): New constraint.
2792 * config/i386/i386.md (UNSPEC_SBB): New unspec.
2793 (cmp<dwi>_doubleword): Removed.
2794 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
2795 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
2796 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
2797 expand with cmp<dwi>_doubleword. For LTU and GEU use
2798 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
2800 * common.opt (gcolumn-info): Enable by default.
2801 * doc/invoke.texi (gcolumn-info): Document new default.
2803 2017-10-23 Richard Biener <rguenther@suse.de>
2805 PR tree-optimization/82672
2806 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
2807 Fold the stmt if we propagated into it.
2809 2017-10-23 Richard Biener <rguenther@suse.de>
2811 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
2812 (bitmap_remove_expr_from_set): ... this. All callers call this
2813 for non-constant values.
2814 (bitmap_set_subtract): Rename to...
2815 (bitmap_set_subtract_expressions): ... this. Adjust and
2817 (bitmap_set_contains_value): Remove superfluous check.
2818 (bitmap_set_replace_value): Inline into single caller ...
2819 (bitmap_value_replace_in_set): ... here and simplify.
2820 (dependent_clean): Merge into ...
2821 (clean): ... this using an overload. Adjust.
2822 (prune_clobbered_mems): Adjust.
2823 (compute_antic_aux): Likewise.
2824 (compute_partial_antic_aux): Likewise.
2826 2017-10-23 Richard Biener <rguenther@suse.de>
2828 PR tree-optimization/82129
2830 2017-08-01 Richard Biener <rguenther@suse.de>
2832 PR tree-optimization/81181
2833 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
2834 (compute_antic): ... end of iteration here.
2836 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2838 * target.def (starting_frame_offset): New hook.
2839 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
2840 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
2841 * doc/tm.texi.in: Regenerate.
2842 * hooks.h (hook_hwi_void_0): Declare.
2843 * hooks.c (hook_hwi_void_0): New function.
2844 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
2845 STARTING_FRAME_OFFSET.
2846 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
2847 * reload1.c (reload): Likewise.
2848 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
2849 instead of STARTING_FRAME_OFFSET.
2850 * function.c (try_fit_stack_local): Likewise.
2851 (assign_stack_local_1): Likewise
2852 (instantiate_virtual_regs): Likewise.
2853 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
2854 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
2855 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
2856 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
2857 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
2858 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
2859 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
2860 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
2861 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
2862 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
2863 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
2864 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
2865 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
2866 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
2867 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
2868 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
2869 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
2870 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
2871 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
2872 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
2873 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
2874 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
2875 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
2876 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
2877 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
2878 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
2879 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
2880 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
2881 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
2882 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
2883 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
2884 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
2885 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
2886 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
2887 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
2888 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
2889 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
2890 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
2891 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
2892 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
2893 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
2894 * config/avr/avr.c (avr_starting_frame_offset): Make static and
2895 return a HOST_WIDE_INT.
2896 (avr_builtin_setjmp_frame_value): Use it instead of
2897 STARTING_FRAME_OFFSET.
2898 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2899 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
2900 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
2902 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2903 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
2904 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
2905 (TARGET_CONSTANT_ALIGNMENT): Redefine.
2906 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
2907 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
2908 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2909 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
2910 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
2911 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2912 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
2913 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
2915 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2916 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
2917 * config/mips/mips.c (mips_compute_frame_info): Refer to
2918 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2919 (mips_starting_frame_offset): New function.
2920 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2921 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
2922 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
2923 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
2924 and return a HOST_WIDE_INT.
2925 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2926 (mmix_initial_elimination_offset): Refer to
2927 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
2928 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
2929 * config/pa/pa.c (pa_starting_frame_offset): New function.
2930 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
2931 (pa_expand_prologue): Likewise.
2932 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2933 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
2934 !FRAME_GROWS_DOWNWARD handling to...
2935 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2936 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
2937 !FRAME_GROWS_DOWNWARD handling to...
2938 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2939 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
2940 !FRAME_GROWS_DOWNWARD handling to...
2941 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2942 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
2944 (rs6000_starting_frame_offset): New function.
2945 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
2946 !FRAME_GROWS_DOWNWARD handling to...
2947 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2948 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
2949 !FRAME_GROWS_DOWNWARD handling to...
2950 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2951 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
2952 !FRAME_GROWS_DOWNWARD handling to...
2953 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
2954 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
2955 (rs6000_starting_frame_offset): New function.
2956 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
2957 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
2958 * config/vax/vax.c (vax_starting_frame_offset): New function.
2959 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
2960 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2961 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
2962 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
2963 (TARGET_STARTING_FRAME_OFFSET): Redefine.
2964 * system.h (STARTING_FRAME_OFFSET): Poison.
2966 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2968 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2969 SCALAR_TYPE_MODE instead of TYPE_MODE.
2971 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2972 Alan Hayward <alan.hayward@arm.com>
2973 David Sherwood <david.sherwood@arm.com>
2975 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
2977 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2978 Alan Hayward <alan.hayward@arm.com>
2979 David Sherwood <david.sherwood@arm.com>
2981 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
2983 2017-10-23 Richard Biener <rguenther@suse.de>
2985 PR tree-optimization/82129
2986 * tree-ssa-pre.c (bitmap_set_and): Remove.
2987 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
2988 canonicalizing expressions in the set to those with lowest
2989 ID rather than taking that from the first edge.
2991 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
2993 * combine.c (rtx_equal_for_field_assignment_p): Use
2994 byte_lowpart_offset.
2996 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
2997 Alan Hayward <alan.hayward@arm.com>
2998 David Sherwood <david.sherwood@arm.com>
3000 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
3001 to a SUBREG_PROMOTED_VAR.
3003 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3004 Alan Hayward <alan.hayward@arm.com>
3005 David Sherwood <david.sherwood@arm.com>
3007 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
3008 (expand_debug_source_expr): Likewise.
3009 * combine.c (combine_simplify_rtx): Likewise.
3010 * cse.c (fold_rtx): Likewise.
3011 * optabs.c (expand_float): Likewise.
3012 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3013 (simplify_binary_operation_1): Likewise.
3015 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3016 Alan Hayward <alan.hayward@arm.com>
3017 David Sherwood <david.sherwood@arm.com>
3019 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
3020 (record_promoted_value): Likewise.
3021 * expr.c (expand_expr_real_2): Likewise.
3022 * ree.c (update_reg_equal_equiv_notes): Likewise.
3023 (combine_set_extension): Likewise.
3024 * rtlanal.c (low_bitmask_len): Likewise.
3025 * simplify-rtx.c (neg_const_int): Likewise.
3026 (simplify_binary_operation_1): Likewise.
3028 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3029 Alan Hayward <alan.hayward@arm.com>
3030 David Sherwood <david.sherwood@arm.com>
3032 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
3033 * regcprop.c (maybe_mode_change): Likewise.
3034 * reload1.c (alter_reg): Likewise.
3036 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3038 * inchash.h (inchash::hash::add_wide_int): New function.
3039 * lto-streamer-out.c (hash_tree): Use it.
3041 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
3043 * inchash.h (inchash::hash::add_wide_int): Rename to...
3044 (inchash::hash::add_hwi): ...this.
3045 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
3046 (polymorphic_call_target_hasher::hash): Likewise.
3047 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
3048 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
3049 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
3050 * lto-streamer-out.c (hash_tree): Likewise.
3051 * optc-save-gen.awk: Likewise.
3052 * tree.c (add_expr): Likewise.
3054 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3057 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
3058 for ordered inequality comparisons even with TARGET_IEEE_FP.
3060 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
3063 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
3064 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3065 Expand with cmp<dwi>_doubleword.
3067 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3069 * extend.texi: Add x86 specific to 'nocf_check' attribute.
3070 List CET intrinsics.
3071 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
3072 specific to -fcf-protection option.
3074 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3076 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
3077 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
3078 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
3079 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
3080 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
3081 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
3082 (extra_objs): Add cet.o for Linux/x86 targets.
3083 (tmake_file): Add i386/t-cet for Linux/x86 targets.
3084 * config/i386/cet.c: New file.
3085 * config/i386/cetintrin.h: Likewise.
3086 * config/i386/t-cet: Likewise.
3087 * config/i386/cpuid.h (bit_SHSTK): New.
3088 (bit_IBT): Likewise.
3089 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
3090 pass IBT and SHSTK bits.
3091 * config/i386/i386-builtin-types.def
3092 (VOID_FTYPE_UNSIGNED_PVOID): New.
3093 (VOID_FTYPE_UINT64_PVOID): Likewise.
3094 * config/i386/i386-builtin.def: Add CET intrinsics.
3095 * config/i386/i386-c.c (ix86_target_macros_internal): Add
3096 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
3097 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
3098 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
3100 * config/i386/i386.c (rest_of_insert_endbranch): New.
3101 (pass_data_insert_endbranch): Likewise.
3102 (pass_insert_endbranch): Likewise.
3103 (make_pass_insert_endbranch): Likewise.
3104 (ix86_notrack_prefixed_insn_p): Likewise.
3105 (ix86_target_string): Add -mibt, -mshstk flags.
3106 (ix86_option_override_internal): Add flag_cf_protection
3108 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
3109 (ix86_print_operand): Add 'notrack' prefix output.
3110 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
3111 (ix86_expand_builtin): Expand CET intrinsics.
3112 (x86_output_mi_thunk): Add 'endbranch' instruction.
3113 * config/i386/i386.h (TARGET_IBT): New.
3114 (TARGET_IBT_P): Likewise.
3115 (TARGET_SHSTK): Likewise.
3116 (TARGET_SHSTK_P): Likewise.
3117 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
3118 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
3119 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
3120 (builtin_setjmp_setup): New pattern.
3121 (builtin_longjmp): Likewise.
3122 (rdssp<mode>): Likewise.
3123 (incssp<mode>): Likewise.
3124 (saveprevssp): Likewise.
3125 (rstorssp): Likewise.
3126 (wrss<mode>): Likewise.
3127 (wruss<mode>): Likewise.
3128 (setssbsy): Likewise.
3129 (clrssbsy): Likewise.
3130 (nop_endbr): Likewise.
3131 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
3133 * config/i386/immintrin.h: Include <cetintrin.h>.
3134 * config/i386/linux-common.h
3135 (file_end_indicate_exec_stack_and_cet): New prototype.
3136 (TARGET_ASM_FILE_END): New.
3138 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3140 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
3141 latencies instead of having separate table; make difference between
3142 integer and float costs.
3143 * i386.h (processor_costs): Remove scalar_stmt_cost,
3144 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
3145 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
3147 * x86-tune-costs.h: Remove entries which has been removed in
3148 procesor_costs from all tables; make cond_taken_branch_cost
3149 and cond_not_taken_branch_cost COST_N_INSNS based.
3151 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
3153 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
3155 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3157 * config/i386/i386.md (isa): Remove fma_avx512f.
3158 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
3159 <avx512>_fmadd_<mode>_mask3<round_name>,
3160 <avx512>_fmsub_<mode>_mask<round_name>,
3161 <avx512>_fmsub_<mode>_mask3<round_name>,
3162 <avx512>_fnmadd_<mode>_mask<round_name>,
3163 <avx512>_fnmadd_<mode>_mask3<round_name>,
3164 <avx512>_fnmsub_<mode>_mask<round_name>,
3165 <avx512>_fnmsub_<mode>_mask3<round_name>,
3166 <avx512>_fmaddsub_<mode>_mask<round_name>,
3167 <avx512>_fmaddsub_<mode>_mask3<round_name>,
3168 <avx512>_fmsubadd_<mode>_mask<round_name>,
3169 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
3170 (*vec_widen_umult_even_v16si<mask_name>,
3171 *vec_widen_smult_even_v16si<mask_name>): Likewise.
3172 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
3174 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3176 * extend.texi: Add 'nocf_check' documentation.
3177 * gimple.texi: Add second parameter to
3178 gimple_build_call_from_tree.
3179 * invoke.texi: Add -fcf-protection documentation.
3180 * rtl.texi: Add REG_CALL_NOTRACK documenation.
3182 2017-10-20 Richard Biener <rguenther@suse.de>
3184 PR tree-optimization/82473
3185 * tree-vect-loop.c (vectorizable_reduction): Properly get at
3186 the largest input type.
3188 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
3190 * c-attribs.c (handle_nocf_check_attribute): New function.
3191 (c_common_attribute_table): Add 'nocf_check' handling.
3192 * gimple-parser.c: Add second argument NULL to
3193 gimple_build_call_from_tree.
3194 * attrib.c (comp_type_attributes): Check nocf_check attribute.
3195 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
3197 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
3198 * common.opt: Add fcf-protection flag.
3199 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
3200 * flag-types.h: Add enum cf_protection_level.
3201 * gimple.c (gimple_build_call_from_tree): Add second parameter.
3202 Add 'nocf_check' attribute propagation to gimple call.
3203 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
3204 (gimple_build_call_from_tree): Update prototype.
3205 (gimple_call_nocf_check_p): New function.
3206 (gimple_call_set_nocf_check): Likewise.
3207 * gimplify.c: Add second argument to gimple_build_call_from_tree.
3208 * ipa-icf.c: Add nocf_check attribute in statement hash.
3209 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
3210 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
3211 * toplev.c (process_options): Add flag_cf_protection handling.
3213 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3215 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
3217 2017-10-20 Richard Biener <rguenther@suse.de>
3219 PR tree-optimization/82603
3220 * tree-if-conv.c (predicate_mem_writes): Make sure to only
3221 remove false predicated stores.
3223 2017-10-20 Richard Biener <rguenther@suse.de>
3225 * graphite-isl-ast-to-gimple.c
3226 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
3227 Remove return value and simplify, dump copied stmt after lhs
3229 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
3230 Reduce dump verbosity.
3231 (gsi_insert_earliest): Likewise.
3232 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
3233 * graphite.c (print_global_statistics): Adjust dumping.
3234 (print_graphite_scop_statistics): Likewise.
3235 (print_graphite_statistics): Do not dump loops here.
3236 (graphite_transform_loops): But here.
3238 2017-10-20 Nicolas Roche <roche@adacore.com>
3240 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
3241 * configure: Regenerate.
3243 2017-10-20 Jakub Jelinek <jakub@redhat.com>
3246 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
3247 functions when optimizing replace GIMPLE_RETURN stmts with
3248 calls to __builtin_unreachable ().
3251 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
3252 for -fsanitize=thread link of executables.
3253 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
3254 link of executables.
3257 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
3259 (<shift_insn><mode>3<mask_name>): Change the last of the 3
3260 define_insns for logical vector shifts to use VI248_AVX512BW
3261 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
3262 condition, useless isa and prefix attributes. Change the first
3263 2 of these define_insns to ...
3264 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
3265 define_insn for avx512vl.
3266 (<shift_insn><mode>3): ... and this, new define_insn without
3267 masking for non-avx512vl.
3270 * config/i386/sse.md (*andnot<mode>3,
3271 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
3272 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
3273 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
3274 not applied use empty suffix even for TARGET_AVX512VL.
3275 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
3276 is applied, supply evex,evex or evex,evex,evex instead of just
3279 2017-10-20 Julia Koval <julia.koval@intel.com>
3281 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
3282 (OPTION_MASK_ISA_GFNI_UNSET): New.
3283 (ix86_handle_option): Handle OPT_mgfni.
3284 * config/i386/cpuid.h (bit_GFNI): New.
3285 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
3286 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
3287 * config/i386/i386.c (ix86_target_string): Add -mgfni.
3288 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
3289 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
3290 * config/i386/i386.opt: Add mgfni.
3292 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
3294 * config/msp430/msp430.c (msp430_option_override): Disable
3295 -fdelete-null-pointer-checks.
3296 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
3298 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3300 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
3301 of x87 and SSE instructions.
3303 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
3305 * asan.c (create_cond_insert_point): Do not update edge count.
3306 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
3307 (afdo_propagate_circuit): Likewise.
3308 (afdo_calculate_branch_prob): Likewise.
3309 (afdo_annotate_cfg): Likewise.
3310 * basic-block.h (struct edge_def): Remove count.
3311 (edge_def::count): New accessor.
3312 * bb-reorder.c (rotate_loop): Update.
3313 (find_traces_1_round): Update.
3314 (connect_traces): Update.
3315 (sanitize_hot_paths): Update.
3316 * cfg.c (unchecked_make_edge): Update.
3317 (make_single_succ_edge): Update.
3318 (check_bb_profile): Update.
3319 (dump_edge_info): Update.
3320 (update_bb_profile_for_threading): Update.
3321 (scale_bbs_frequencies_int): Update.
3322 (scale_bbs_frequencies_gcov_type): Update.
3323 (scale_bbs_frequencies_profile_count): Update.
3324 (scale_bbs_frequencies): Update.
3325 * cfganal.c (connect_infinite_loops_to_exit): Update.
3326 * cfgbuild.c (compute_outgoing_frequencies): Update.
3327 (find_many_sub_basic_blocks): Update.
3328 * cfgcleanup.c (try_forward_edges): Update.
3329 (try_crossjump_to_edge): Update
3330 * cfgexpand.c (expand_gimple_cond): Update
3331 (expand_gimple_tailcall): Update
3332 (construct_exit_block): Update
3333 * cfghooks.c (verify_flow_info): Update
3334 (redirect_edge_succ_nodup): Update
3335 (split_edge): Update
3336 (make_forwarder_block): Update
3337 (duplicate_block): Update
3338 (account_profile_record): Update
3339 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
3340 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
3341 * cfgloopmanip.c (scale_loop_profile): Update.
3343 (lv_adjust_loop_entry_edge): Update.
3344 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
3345 (force_nonfallthru_and_redirect): Update.
3346 (purge_dead_edges): Update.
3347 (rtl_flow_call_edges_add): Update.
3348 * cgraphunit.c (init_lowered_empty_function): Update.
3349 (cgraph_node::expand_thunk): Update.
3350 * gimple-pretty-print.c (dump_probability): Update.
3351 (dump_edge_probability): Update.
3352 * gimple-ssa-isolate-paths.c (isolate_path): Update.
3353 * haifa-sched.c (sched_create_recovery_edges): Update.
3354 * hsa-gen.c (convert_switch_statements): Update.
3355 * ifcvt.c (dead_or_predicable): Update.
3356 * ipa-inline-transform.c (inline_transform): Update.
3357 * ipa-split.c (split_function): Update.
3358 * ipa-utils.c (ipa_merge_profiles): Update.
3359 * loop-doloop.c (add_test): Update.
3360 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
3361 * lto-streamer-in.c (input_cfg): Update.
3362 (input_function): Update.
3363 * lto-streamer-out.c (output_cfg): Update.
3364 * modulo-sched.c (sms_schedule): Update.
3365 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
3366 * predict.c (maybe_hot_edge_p): Update.
3367 (unlikely_executed_edge_p): Update.
3368 (probably_never_executed_edge_p): Update.
3369 (dump_prediction): Update.
3370 (drop_profile): Update.
3371 (propagate_unlikely_bbs_forward): Update.
3372 (determine_unlikely_bbs): Update.
3373 (force_edge_cold): Update.
3374 * profile.c (compute_branch_probabilities): Update.
3375 * reg-stack.c (better_edge): Update.
3376 * shrink-wrap.c (handle_simple_exit): Update.
3377 * tracer.c (better_p): Update.
3378 * trans-mem.c (expand_transaction): Update.
3379 (split_bb_make_tm_edge): Update.
3380 * tree-call-cdce.c: Update.
3381 * tree-cfg.c (gimple_find_sub_bbs): Update.
3382 (gimple_split_edge): Update.
3383 (gimple_duplicate_sese_region): Update.
3384 (gimple_duplicate_sese_tail): Update.
3385 (gimple_flow_call_edges_add): Update.
3386 (insert_cond_bb): Update.
3387 (execute_fixup_cfg): Update.
3388 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
3389 * tree-complex.c (expand_complex_div_wide): Update.
3390 * tree-eh.c (lower_resx): Update.
3391 (unsplit_eh): Update.
3392 (cleanup_empty_eh_move_lp): Update.
3393 * tree-inline.c (copy_edges_for_bb): Update.
3394 (freqs_to_counts): Update.
3395 (copy_cfg_body): Update.
3396 * tree-ssa-dce.c (remove_dead_stmt): Update.
3397 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
3398 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
3399 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
3400 (unloop_loops): Update.
3401 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
3402 * tree-ssa-loop-split.c (connect_loops): Update.
3403 (split_loop): Update.
3404 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
3405 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
3406 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
3407 * tree-ssa-reassoc.c (branch_fixup): Update.
3408 * tree-ssa-tail-merge.c (replace_block_by): Update.
3409 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
3410 (compute_path_counts): Update.
3411 (update_profile): Update.
3412 (recompute_probabilities): Update.
3413 (update_joiner_offpath_counts): Update.
3414 (estimated_freqs_path): Update.
3415 (freqs_to_counts_path): Update.
3416 (clear_counts_path): Update.
3417 (ssa_fix_duplicate_block_edges): Update.
3418 (duplicate_thread_path): Update.
3419 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
3420 (case_bit_test_cmp): Update.
3421 (collect_switch_conv_info): Update.
3422 (gen_inbound_check): Update.
3423 (do_jump_if_equal): Update.
3424 (emit_cmp_and_jump_insns): Update.
3425 * tree-tailcall.c (decrease_profile): Update.
3426 (eliminate_tail_call): Update.
3427 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
3428 (vect_do_peeling): Update.
3429 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
3430 * ubsan.c (ubsan_expand_null_ifn): Update.
3431 (ubsan_expand_ptr_ifn): Update.
3432 * value-prof.c (gimple_divmod_fixed_value): Update.
3433 (gimple_mod_pow2): Update.
3434 (gimple_mod_subtract): Update.
3435 (gimple_ic): Update.
3436 (gimple_stringop_fixed_value): Update.
3438 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3441 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
3443 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
3445 PR rtl-optimization/82395
3446 * ira-color.c (allocno_priority_compare_func): Fix comparison step
3447 based on non_spilled_static_chain_regno_p.
3449 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
3451 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
3452 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
3453 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
3455 2017-10-19 Martin Sebor <msebor@redhat.com>
3457 PR tree-optimization/82596
3458 * tree.c (array_at_struct_end_p): Handle STRING_CST.
3460 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3462 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
3463 (get_mem_refs_of_builtin_call): Likewise.
3464 * builtins.c (expand_builtin_apply): Adjust call to
3465 allocate_dynamic_stack_space.
3466 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
3467 the third argument to allocate_dynamic_stack_space, otherwise -1.
3468 (expand_builtin): Deal with all alloca variants.
3469 (is_inexpensive_builtin): Likewise.
3470 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
3471 * calls.c (special_function_p): Deal with all alloca variants.
3472 (initialize_argument_information): Adjust call to
3473 allocate_dynamic_stack_space.
3474 (expand_call): Likewise.
3475 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
3476 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
3477 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
3478 use it for the stack usage computation.
3479 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
3480 * function.c (gimplify_parameters): Call build_alloca_call_expr.
3481 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
3482 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
3483 (in_loop_p): Remove first argument and useless check.
3484 (pass_walloca::execute): Remove useless test and adjust call to above.
3485 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
3486 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
3487 (gimplify_call_expr): Deal with all alloca variants.
3488 * hsa-gen.c (gen_hsa_alloca): Likewise.
3489 (gen_hsa_insns_for_call): Likewise.
3490 * ipa-pure-const.c (special_builtin_state): Likewise.
3491 * tree-chkp.c (chkp_build_returned_bound): Likewise.
3492 * tree-object-size.c (alloc_object_size): Likewise.
3493 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
3494 (call_may_clobber_ref_p_1): Likewise.
3495 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
3496 (ccp_fold_stmt): Likewise.
3497 (optimize_stack_restore): Likewise.
3498 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
3499 (mark_all_reaching_defs_necessary_1): Likewise.
3500 (propagate_necessity): Likewise.
3501 (eliminate_unnecessary_stmts): Likewise.
3502 * tree.c (build_common_builtin_nodes): Build
3503 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
3504 (build_alloca_call_expr): New function.
3505 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
3506 (CASE_BUILT_IN_ALLOCA): Likewise.
3507 (build_alloca_call_expr): Declare.
3508 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
3510 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
3513 * dwarf2out.c (new_die_raw): New static inline function.
3514 (new_die): Use it to create the DIE.
3515 (add_AT_external_die_ref): Likewise.
3516 (clone_die): Likewise.
3517 (clone_as_declaration): Likewise.
3518 (dwarf2out_vms_debug_main_pointer): Likewise.
3519 (base_type_die): Likewise. Remove early return for corner cases.
3520 Do not call add_pubtype on the DIE here.
3521 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
3522 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
3523 typedefs for base types with DW_AT_endianity. Make sure a DIE with
3524 native order exists for base types, attach the DIE manually and call
3525 add_pubtype on it. Do not equate a reverse order DIE to the type.
3527 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
3529 * config/arm/arm.c (align_ok_ldrd_strd): New function.
3530 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
3532 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
3534 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3536 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
3537 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
3538 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
3539 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
3540 * opts.c (sanitizer_opts): Add builtin.
3541 * ubsan.c (instrument_builtin): New function.
3542 (pass_ubsan::execute): Call it.
3543 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
3544 * doc/invoke.texi: Document -fsanitize=builtin.
3546 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
3547 builtins, store max (log2 (align), 0) into uchar field instead of
3548 align into uptr field.
3549 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
3550 store uchar 0 field instead of uptr 0 field.
3551 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
3552 instead of passing one address of struct with 2 locations pass
3553 two addresses of structs with 1 location each.
3554 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
3555 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
3556 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
3557 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
3558 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
3559 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
3560 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
3561 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
3563 2017-10-19 Martin Liska <mliska@suse.cz>
3566 * file-find.c (remove_prefix): Remove.
3567 * file-find.h (remove_prefix): Likewise.
3568 * gcc-ar.c: Remove smartness of lookup.
3570 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
3572 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
3573 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
3574 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
3576 2017-10-19 Jakub Jelinek <jakub@redhat.com>
3579 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
3580 (setcc + and to xor + setcc): New peephole2.
3582 2017-10-19 Tom de Vries <tom@codesourcery.com>
3584 * doc/sourcebuild.texi (Test Directives, Variants of
3585 dg-require-support): Add dg-require-stack-size.
3587 2017-10-19 Martin Liska <mliska@suse.cz>
3590 * gimplify.c (gimplify_decl_expr): Do not instrument variables
3591 that have a large alignment.
3592 (gimplify_target_expr): Likewise.
3594 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
3596 PR rtl-optimization/82602
3597 * ira.c (rtx_moveable_p): Return false for volatile asm.
3599 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
3602 * config/i386/i386-modes.def (CCGZ): New CC mode.
3603 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
3604 * config/i386/predicates.md (ix86_comparison_operator):
3606 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
3607 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
3608 with double-word subtraction.
3609 (put_condition_code): Handle CCGZmode.
3611 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
3613 * wide-int.cc (debug (const wide_int &)): New.
3614 (debug (const wide_int *)): New.
3615 (debug (const widest_int &)): New.
3616 (debug (const widest_int *)): New.
3618 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
3621 * lra-constraints.c (curr_insn_transform): Use non-input operand
3622 instead of output one for matched reload.
3624 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3626 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
3627 (tree-ssa-loop-ivopts.h): New header file.
3628 (struct builtin_info): New fields.
3629 (classify_builtin_1): Compute and record base and offset parts for
3630 memset builtin partition by calling strip_offset.
3631 (offset_cmp, fuse_memset_builtins): New functions.
3632 (finalize_partitions): Fuse adjacent memset partitions by calling
3634 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
3635 Expose the interface.
3636 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
3638 2017-10-18 Bin Cheng <bin.cheng@arm.com>
3640 PR tree-optimization/82574
3641 * tree-loop-distribution.c (find_single_drs): New parameter. Check
3642 that data reference must be executed exactly once per iteration
3643 against the outermost loop in nest.
3644 (classify_partition): Update call to above function.
3646 2017-10-18 Richard Biener <rguenther@suse.de>
3648 PR tree-optimization/82591
3649 * graphite.c (graphite_transform_loops): Move code gen message
3651 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3652 Here. Handle scop_to_isl_ast failing.
3653 (scop_to_isl_ast): Limit the number of ISL operations.
3655 2017-10-18 Richard Biener <rguenther@suse.de>
3657 * graphite-isl-ast-to-gimple.c
3658 (translate_isl_ast_to_gimple::set_rename): Simplify.
3659 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
3660 (graphite_copy_stmts_from_block): ... here.
3661 (copy_bb_and_scalar_dependences): Simplify.
3662 (add_parameters_to_ivs_params): Canonicalize.
3663 (generate_entry_out_of_ssa_copies): Simplify.
3664 * graphite-sese-to-poly.c (extract_affine_name): Simplify
3665 by passing in ISL dimension.
3666 (parameter_index_in_region_1): Rename to ...
3667 (parameter_index_in_region): ... this.
3668 (extract_affine): Adjust assert, pass down parameter index.
3669 (add_param_constraints): Use range-info when available.
3670 (build_scop_context): Adjust.
3671 * sese.c (new_sese_info): Adjust.
3672 (free_sese_info): Likewise.
3673 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
3674 Remove unused typedefs.
3675 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
3677 2017-10-18 Martin Liska <mliska@suse.cz>
3679 * combine.c (simplify_compare_const): Add gcc_fallthrough.
3681 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3683 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
3684 (s390_sched_init): Do not reset s390_sched_state if we entered the
3685 current basic block via a fallthru edge and all others are unlikely.
3687 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
3689 * config/s390/s390.c (NUM_SIDES): New variable.
3690 (LONGRUNNING_THRESHOLD): New variable.
3691 (LATENCY_FACTOR): New variable.
3692 (s390_sched_score): Decrease score for long-running instructions on
3694 (s390_sched_variable_issue): Perform bookkeeping for long-running
3697 2017-10-18 Richard Biener <rguenther@suse.de>
3699 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
3700 Simplify with removal of the parameter rename map.
3701 (set_rename): Likewise.
3702 (should_copy_to_new_region): Likewise.
3703 (graphite_copy_stmts_from_block): Likewise.
3704 (copy_bb_and_scalar_dependences): Remove initialization of
3705 unused copied_bb_map.
3707 (copy_internal_parameters): Likewise.
3708 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
3709 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
3710 Use INTEGRAL_TYPE_P.
3711 (parameter_index_in_region_1): Rename to ...
3712 (assign_parameter_index_in_region): ... this. Assert we have
3713 a parameter we handle.
3714 (scan_tree_for_params): Adjust.
3715 * sese.h (parameter_rename_map_t): Remove.
3716 (struct sese_info_t): Remove unused parameter_rename_map and
3717 copied_bb_map members.
3718 * sese.c (new_sese_info): Adjust.
3719 (free_sese_info): Likewise.
3721 2017-10-18 Martin Liska <mliska@suse.cz>
3724 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
3725 on an abnormal edge.
3727 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
3729 * doc/invoke.texi (ffunction-sections and fdata-sections):
3732 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3734 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
3735 the use statement can throw internally.
3737 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
3739 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
3740 any RTX present on the RHS of a SET.
3741 * compare-elim.c (try_eliminate_compare): Restore comment.
3743 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3745 * langhooks.h (struct lang_hooks): Document that tree_size langhook
3746 may be also called on tcc_type nodes.
3747 * langhooks.c (lhd_tree_size): Likewise.
3749 2017-10-17 David Malcolm <dmalcolm@redhat.com>
3751 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
3752 format_warning_at_substring.
3753 (maybe_warn): Convert source_range * param to a location_t. Pass
3754 UNKNOWN_LOCATION rather than NULL to fmtwarn.
3755 (format_directive): Remove code to extract source_ranges and
3756 source_range * in favor of just a location_t.
3757 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
3759 * substring-locations.c (format_warning_va): Convert
3760 source_range * param to a location_t.
3761 (format_warning_at_substring): Likewise.
3762 * substring-locations.h (format_warning_va): Likewise.
3763 (format_warning_at_substring): Likewise.
3765 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
3767 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
3769 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
3770 and scatter/gather ops.
3772 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
3773 vec_gather_load and vec_scatter_store.
3774 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
3775 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
3776 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
3777 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
3778 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
3779 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
3781 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
3783 * reg-stack.c (compare_for_stack_reg): Add bool argument.
3784 Detect FTST instruction and handle its register pops. Only pop
3785 second operand if can_pop_second_op is true.
3786 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
3787 set can_pop_second_op to false in the compare_for_stack_reg call.
3789 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
3790 output_fp_compare for stack register operands.
3791 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
3792 instructions here. Do not emit stack register pops here. Assert
3793 that FCOMPP pops next to top stack register. Rewrite function.
3795 2017-10-17 Nathan Sidwell <nathan@acm.org>
3798 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
3799 use DECL_ASSEMBLER_NAME_RAW.
3802 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
3805 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
3806 Wilco Dijkstra <wilco.dijkstra@arm.com>
3808 * builtins.c (expand_builtin_update_setjmp_buf): Add a
3809 converstion to Pmode from the buf_addr.
3811 2017-10-17 Richard Biener <rguenther@suse.de>
3813 * graphite-dependences.c (scop_get_reads_and_writes): Change
3814 output parameters to references.
3816 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
3818 PR 71026/tree-optimization
3819 * fold-const.c (distribute_real_division): Removed.
3820 (fold_binary_loc): Remove calls to distribute_real_divison.
3822 2017-10-17 Richard Biener <rguenther@suse.de>
3824 * graphite-scop-detection.c
3825 (scop_detection::stmt_has_simple_data_refs_p): Always use
3826 the full nest as region.
3827 (try_generate_gimple_bb): Likewise.
3828 * sese.c (scalar_evolution_in_region): Simplify now that
3829 SCEV can handle instantiation in regions.
3830 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
3831 in the non-loop part of a function if requested.
3833 2017-10-17 Richard Biener <rguenther@suse.de>
3835 PR tree-optimization/82563
3836 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
3838 (graphite_regenerate_ast_isl): Call it.
3839 * graphite-scop-detection.c (build_scops): Remove entry edge split.
3841 2017-10-17 Jakub Jelinek <jakub@redhat.com>
3843 PR tree-optimization/82549
3844 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
3845 Formatting fixes. Instead of calling make_bit_field_ref with negative
3848 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
3850 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
3851 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
3852 _mm_maskz_reduce_ss): New.
3853 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
3854 __builtin_ia32_reducess_mask): Ditto..
3855 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
3856 * config/i386/sse.md (reduces<mode>): Renamed to ...
3857 (reduces<mode><mask_scalar_name>): ... this.
3858 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
3860 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
3861 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
3863 2017-10-16 David Malcolm <dmalcolm@redhat.com>
3865 * Makefile.in (OBJS): Add unique-ptr-tests.o.
3866 * selftest-run-tests.c (selftest::run_tests): Call
3867 selftest::unique_ptr_tests_cc_tests.
3868 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
3869 * unique-ptr-tests.cc: New file.
3871 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
3874 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3876 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3877 (make_hard_regno_born, make_hard_regno_dead): Update
3878 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
3880 2017-10-16 Jeff Law <law@redhat.com>
3882 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3884 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
3886 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
3888 2017-10-16 Olivier Hainque <hainque@adacore.com>
3890 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
3891 with_cpu if we were configured for an e500v2 target cpu name.
3893 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
3895 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
3896 * doc/invoke.texi: Document +nodsp as a valid extension for
3899 2017-10-16 Martin Liska <mliska@suse.cz>
3901 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
3902 (test_set_range): Likewise.
3903 (test_range_functions): Rename to ...
3904 (test_bit_in_range): ... this.
3905 (sbitmap_c_tests): Add new test.
3907 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3909 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
3911 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
3912 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
3914 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3916 * config/aarch64/aarch64-builtins.c
3917 (aarch64_types_quadopu_lane_qualifiers): New.
3918 (TYPES_QUADOPU_LANE): New.
3919 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
3920 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
3921 (aarch64_<sur>dot_laneq<vsi2qi>): New.
3922 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
3923 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
3924 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
3925 (Vdottype, DOTPROD): New.
3926 (sur): Add SDOT and UDOT.
3928 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3930 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
3931 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
3932 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
3934 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
3935 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
3936 Enable TARGET_DOTPROD.
3937 (cortex-a75.cortex-a55): Likewise.
3938 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
3940 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3942 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
3943 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
3945 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
3947 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
3948 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
3949 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
3950 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
3951 * config/arm/types.md (neon_dot, neon_dot_q): New.
3952 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
3954 2017-10-16 Tamar Christina <tamar.christina@arm.com>
3956 * config/arm/arm.h (TARGET_DOTPROD): New.
3957 * config/arm/arm.c (arm_arch_dotprod): New.
3958 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
3959 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
3960 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
3961 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
3962 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
3963 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
3964 * doc/invoke.texi (armv8.2-a): Document dotprod
3966 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3968 * i386.c (ix86_vec_cost): New function.
3969 (ix86_rtx_costs): Handle vector operations better.
3970 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
3971 * x86-tune-costs.h: Add new costs to all tables.
3973 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
3975 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
3977 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
3978 divsd, sqrtss and sqrtsd
3979 * x86-tune-costs.h: Add new entries to all costs.
3980 (znver1_cost): Fix to match real instruction latencies.
3982 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3983 Michael Collison <michael.collison@arm.com>
3985 * compare-elim.c: Include emit-rtl.h.
3986 (can_merge_compare_into_arith): New function.
3987 (try_validate_parallel): Likewise.
3988 (try_merge_compare): Likewise.
3989 (try_eliminate_compare): Call the above when no previous clobber
3991 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
3994 2017-10-14 Jakub Jelinek <jakub@redhat.com>
3998 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
3999 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
4003 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
4004 to be any operand_equal_p operands. For & (B - 1) require
4005 B to be power of 2. Recognize
4006 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
4008 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4011 * optabs.c (expand_memory_blockage): Fix call of
4012 targetm.have_memory_blockage.
4014 2017-10-14 Jakub Jelinek <jakub@redhat.com>
4017 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
4018 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
4019 objects to extra_objs instead of overwriting it.
4021 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4023 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
4024 Use any_fp_register_operand as operand[3] predicate. Simplify
4025 equality test for operands[2] and operands[4] memory location.
4026 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
4027 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
4028 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
4029 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
4030 any_fp_register_operand as operand[1] predicate. Simplify
4031 equality test for operands[0] and operands[3] memory location.
4032 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
4033 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
4034 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
4036 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
4038 * target-insns.def: Add memory_blockage.
4039 * optabs.c (expand_memory_blockage): New function.
4040 (expand_asm_memory_barrier): Rename ...
4041 (expand_asm_memory_blockage): ... to this.
4042 (expand_mem_thread_fence): Call expand_memory_blockage
4043 instead of expand_asm_memory_barrier.
4044 (expand_mem_singnal_fence): Ditto.
4045 (expand_atomic_load): Ditto.
4046 (expand_atomic_store): Ditto.
4047 * doc/md.texi (Standard Pattern Names For Generation):
4048 Document memory_blockage instruction pattern.
4050 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
4052 * config/rl78/rl78.c (rl78_emit_libcall): New function.
4053 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
4054 * config/rl78/rl78.md: New define_expand "adddi3".
4056 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4058 * cfghooks.c (verify_flow_info): Disable check that all probabilities
4061 2017-10-13 Jeff Law <law@redhat.com>
4063 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
4065 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4068 * internal-fn.c (expand_mul_overflow): If both operands have
4069 the same highpart of -1 or 0 and the topmost bit of lowpart
4070 is different, overflow is if res <= 0 rather than res < 0.
4072 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
4074 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
4075 TARGET_P9_VECTOR code for unaligned_load case.
4077 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
4079 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
4081 2017-10-13 Nathan Sidwell <nathan@acm.org>
4083 * tree-core.h (tree_contains_struct): Make bool.
4084 * tree.c (tree_contains_struct): Likewise.
4085 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
4086 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
4087 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
4088 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
4089 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
4091 2017-10-13 Richard Biener <rguenther@suse.de>
4093 * graphite-isl-ast-to-gimple.c
4094 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
4095 parameters and dominance check.
4096 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
4097 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
4098 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
4099 Do not update SSA form here or do intermediate IL verification.
4100 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
4101 (graphite_initialize): Remove check on the number of loops in
4102 the function and inline into graphite_transform_loops.
4103 (graphite_finalize): Inline into graphite_transform_loops.
4104 (graphite_transform_loops): Perform SSA update and IL verification
4106 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
4108 2017-10-13 Richard Biener <rguenther@suse.de>
4110 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
4111 graphite_expression_type_precision): Avoid global constructor
4113 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
4114 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
4115 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
4116 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
4117 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
4118 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
4120 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
4123 * config/i386/i386.h (ix86_red_zone_size): New.
4124 * config/i386/i386.md (push peephole2s): Replace
4125 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
4127 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4128 Alan Hayward <alan.hayward@arm.com>
4129 David Sherwood <david.sherwood@arm.com>
4131 * combine.c (can_change_dest_mode): Reject changes in
4132 REGMODE_NATURAL_SIZE.
4134 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4135 Alan Hayward <alan.hayward@arm.com>
4136 David Sherwood <david.sherwood@arm.com>
4138 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
4139 (expand_debug_source_expr): Likewise.
4140 * combine.c (combine_simplify_rtx): Likewise.
4141 * cse.c (fold_rtx): Likewise.
4142 * fwprop.c (canonicalize_address): Likewise.
4143 * targhooks.c (default_shift_truncation_mask): Likewise.
4145 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4146 Alan Hayward <alan.hayward@arm.com>
4147 David Sherwood <david.sherwood@arm.com>
4149 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
4150 (widened_mode): Likewise.
4151 (expand_unop): Likewise.
4152 * ree.c (transform_ifelse): Likewise.
4153 (merge_def_and_ext): Likewise.
4154 (combine_reaching_defs): Likewise.
4155 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
4157 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4158 Alan Hayward <alan.hayward@arm.com>
4159 David Sherwood <david.sherwood@arm.com>
4161 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
4162 * combine.c (gen_lowpart_for_combine): Likewise.
4163 * dwarf2out.c (rtl_for_decl_location): Likewise.
4164 * final.c (alter_subreg): Likewise.
4165 * rtlhooks.c (gen_lowpart_general): Likewise.
4166 (gen_lowpart_if_possible): Likewise.
4168 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
4169 Alan Hayward <alan.hayward@arm.com>
4170 David Sherwood <david.sherwood@arm.com>
4172 * calls.c (expand_call): Use subreg_lowpart_offset.
4173 * cse.c (cse_insn): Likewise.
4174 * regcprop.c (copy_value): Likewise.
4175 (copyprop_hardreg_forward_1): Likewise.
4177 2017-10-13 Jakub Jelinek <jakub@redhat.com>
4180 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
4181 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
4182 =Q constraints to +Q and into insn condition add check
4183 that operands[0] and operands[1] are equal.
4184 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
4185 =Q constraints to +Q and into insn condition add check
4186 that operands[0] is equal to either operands[1] or operands[2].
4189 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
4190 instead of handling MINUS_EXPR twice (once for each argument),
4191 canonicalize operand order and handle just once, use rtype where
4192 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
4195 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
4196 any values of __C while still being pattern recognizable as a simple
4199 2017-10-13 Richard Biener <rguenther@suse.de>
4201 PR tree-optimization/82451
4203 2017-10-02 Richard Biener <rguenther@suse.de>
4205 PR tree-optimization/82355
4206 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4207 a mapping for the enclosing loop but avoid generating one for
4209 (copy_bb_and_scalar_dependences): Remove premature codegen
4210 error on PHIs in blocks duplicated into multiple places.
4211 * graphite-scop-detection.c
4212 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4213 in the region use it as loop and nest to analyze the DR in.
4214 (try_generate_gimple_bb): Likewise.
4215 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4216 (add_loop_constraints): For blocks in a loop not in the region
4217 create a dimension with a single iteration.
4218 * sese.h (gbb_loop_at_index): Remove assert.
4220 * cfgloop.c (loop_preheader_edge): For the loop tree root
4221 return the single successor of the entry block.
4222 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
4223 Reset the SCEV hashtable and niters.
4224 * graphite-scop-detection.c
4225 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
4226 assert that we only have POLYNOMIAL_CHREC that vary in loops
4227 contained in the region.
4228 (scop_detection::graphite_can_represent_expr): Adjust.
4229 (scop_detection::stmt_has_simple_data_refs_p): For loops
4230 not in the region set loop to NULL. The nest is now the
4231 entry edge to the region.
4232 (try_generate_gimple_bb): Likewise.
4233 * sese.c (scalar_evolution_in_region): Adjust for
4234 instantiate_scev change.
4235 * tree-data-ref.h (graphite_find_data_references_in_stmt):
4236 Make nest parameter the edge into the region.
4237 (create_data_ref): Likewise.
4238 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
4239 entry edge into a region and adjust instantiate_scev calls.
4240 (create_data_ref): Likewise.
4241 (graphite_find_data_references_in_stmt): Likewise.
4242 (find_data_references_in_stmt): Pass the loop preheader edge
4243 from the nest argument.
4244 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
4245 parameter the edge into the region.
4246 (instantiate_parameters): Use the loop preheader edge as entry.
4247 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
4249 (get_instantiated_value_entry): Make instantiate_below parameter
4250 the edge into the region.
4251 (instantiate_scev_name): Likewise. Adjust dominance checks,
4252 when we cannot use loop-based instantiation instantiate by
4253 walking use-def chains.
4254 (instantiate_scev_poly): Adjust.
4255 (instantiate_scev_binary): Likewise.
4256 (instantiate_scev_convert): Likewise.
4257 (instantiate_scev_not): Likewise.
4258 (instantiate_array_ref): Remove.
4259 (instantiate_scev_3): Likewise.
4260 (instantiate_scev_2): Likewise.
4261 (instantiate_scev_1): Likewise.
4262 (instantiate_scev_r): Do not blindly handle N-operand trees.
4263 Do not instantiate array-refs. Handle all constants and invariants.
4264 (instantiate_scev): Make instantiate_below parameter
4265 the edge into the region.
4266 (resolve_mixers): Use the loop preheader edge for the region
4267 parameter to instantiate_scev_r.
4268 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
4270 2017-10-13 Richard Biener <rguenther@suse.de>
4272 PR tree-optimization/82525
4273 * graphite-isl-ast-to-gimple.c
4274 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
4276 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
4277 Fail code generation when we cannot represent the isl integer.
4278 (binary_op_to_tree): Elide modulo operations that are no-ops
4279 in the type we code generate. Remove now superfluous code
4282 2017-10-13 Richard Biener <rguenther@suse.de>
4284 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
4285 (scop_detection::harmful_loop_in_region): Remove premature
4286 IV type restriction.
4287 (scop_detection::graphite_can_represent_scev): We can handle
4288 pointer IVs just fine.
4290 2017-10-13 Alan Modra <amodra@gmail.com>
4292 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
4293 "Clobbers and Scratch Registers". Add paragraph on
4294 alternative to clobbers for scratch registers and OpenBLAS
4297 2017-10-13 Alan Modra <amodra@gmail.com>
4299 * doc/extend.texi (Clobbers): Correct vax example. Delete old
4300 example of a memory input for a string of known length. Move
4301 commentary out of table. Add a number of new examples
4302 covering array memory inputs.
4304 2017-10-12 Martin Liska <mliska@suse.cz>
4306 PR tree-optimization/82493
4307 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
4308 (test_range_functions): New function.
4309 (sbitmap_c_tests): Likewise.
4310 * selftest-run-tests.c (selftest::run_tests): Run new tests.
4311 * selftest.h (sbitmap_c_tests): New function.
4313 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
4315 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
4317 * config/rs6000/amo.h: Fix spacing issue.
4319 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4322 * config/i386/i386.md (*ashl<mode>3_mask_1,
4323 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
4324 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
4327 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4329 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
4330 (profile_probability): Set max_probability
4331 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
4333 * profile-count.c (profile_probability::differs_from_p): Do not
4334 rely on max_probaiblity == 10000
4336 2017-10-12 Jeff Law <law@redhat.com>
4338 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
4341 2017-10-12 Martin Sebor <msebor@redhat.com>
4345 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
4346 (handle_alias_pairs): Call it.
4347 * common.opt (-Wattribute-alias): New option.
4348 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
4349 * doc/invoke.texi (-Wattribute-alias): Document.
4351 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
4354 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4356 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4358 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4359 (make_hard_regno_born, make_hard_regno_dead): Update
4360 bb_killed_pseudos and bb_gen_pseudos.
4362 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
4364 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
4366 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
4368 * config/alpha/alpha.c (alpha_split_conditional_move):
4369 Use std::swap instead of manually swapping.
4370 (alpha_stdarg_optimize_hook): Ditto.
4371 (alpha_canonicalize_comparison): Ditto.
4373 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4375 * tree-loop-distribution.c (struct builtin_info): New struct.
4376 (struct partition): Refactor fields into struct builtin_info.
4377 (partition_free): Free struct builtin_info.
4378 (build_size_arg_loc, build_addr_arg_loc): Delete.
4379 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
4380 information from struct builtin_info.
4381 (find_single_drs): New function refactored from classify_partition.
4382 Also moved builtin validity checks to this function.
4383 (compute_access_range, alloc_builtin): New functions.
4384 (classify_builtin_st, classify_builtin_ldst): New functions.
4385 (classify_partition): Refactor code into functions find_single_drs,
4386 classify_builtin_st and classify_builtin_ldst.
4387 (distribute_loop): Don't do runtime alias check when distributing
4389 (find_seed_stmts_for_distribution): New function.
4390 (pass_loop_distribution::execute): Refactor code finding seed
4391 stmts into above function. Support distribution for the innermost
4392 two-level loop nest. Adjust dump information.
4394 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4396 * tree-loop-distribution.c: Adjust the general comment.
4397 (NUM_PARTITION_THRESHOLD): New macro.
4398 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
4399 (classify_partition): Skip builtin pattern of loop nest's inner loop.
4400 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
4401 in call to build_partition_graph.
4402 (finalize_partitions): New parameter. Make loop distribution more
4403 conservative by fusing more partitions.
4404 (distribute_loop): Don't do runtime alias check in case of loop nest
4406 (find_seed_stmts_for_distribution): New function.
4407 (prepare_perfect_loop_nest): New function.
4408 (pass_loop_distribution::execute): Refactor code finding seed stmts
4409 and loop nest into above functions. Support loop nest distribution.
4410 Adjust dump information accordingly.
4412 2017-10-12 Bin Cheng <bin.cheng@arm.com>
4414 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
4415 and set PTYPE_SEQUENTIAL for merged partition.
4417 2017-10-12 Richard Biener <rguenther@suse.de>
4419 PR tree-optimization/69728
4421 2017-09-19 Richard Biener <rguenther@suse.de>
4423 PR tree-optimization/69728
4424 * graphite-sese-to-poly.c (schedule_error): New global.
4425 (add_loop_schedule): Handle empty domain by failing the
4427 (build_original_schedule): Handle schedule_error.
4429 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
4430 domain by returning an unchanged schedule.
4432 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4434 * genrecog.c (validate_pattern): For VEC_SELECT verify that
4435 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
4437 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
4439 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
4442 2017-10-12 Jakub Jelinek <jakub@redhat.com>
4445 * expr.c (store_field): Don't optimize away bitsize == 0 store
4446 from CALL_EXPR with addressable return type.
4448 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4450 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
4451 * config/rs6000/rs6000.md (sel): Delete mode attribute.
4452 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
4453 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
4454 TARGET_ISEL instead of TARGET_ISEL<sel>.
4456 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
4458 * config/rs6000/rs6000.c
4459 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
4461 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
4463 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
4465 (reg_or_cint_operand): Fix comment.
4466 (reg_or_zero_operand): New predicate.
4467 * config/rs6000/rs6000-protos.h (output_isel): Delete.
4468 * config/rs6000/rs6000.c (output_isel): Delete.
4469 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
4470 instead of reg_or_cint_operand. Output instruction directly (not via
4472 (isel_unsigned_<mode>): Ditto.
4473 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
4474 gpc_reg_operand. Add an instruction alternative for this. Output
4475 instruction directly.
4476 (*isel_reversed_unsigned_<mode>): Ditto.
4478 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4480 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
4481 (TARGET_CANONICALIZE_COMPARISON): Define.
4483 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
4486 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
4487 Check whether the dest is REG before adding REG_EQUIV note.
4489 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
4492 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
4494 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
4495 (make_hard_regno_born, make_hard_regno_dead): Update
4496 bb_killed_pseudos and bb_gen_pseudos.
4498 2017-10-11 Nathan Sidwell <nathan@acm.org>
4500 * incpath.h (enum incpath_kind): Name enum, prefix values.
4501 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
4502 * incpath.c (heads, tails): Use INC_MAX.
4503 (add_env_var_paths, add_standard_paths): Use incpath_kind.
4504 (merge_include_chains, split_quote_chain,
4505 register_include_chains): Update incpath_kind names.
4506 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
4507 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
4509 (add_framework_path, darwin_register_objc_includes): Likewise.
4510 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
4512 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4514 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
4515 Do not use float_operator operator predicate.
4516 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
4517 * config/i386/predicates.md (float_operator): Remove predicate.
4519 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
4521 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
4522 (*jccxf_i387): Ditto.
4523 (*jcc<mode>_i387): Ditto.
4524 (*jccu<mode>_i387): Ditto.
4525 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
4526 (*jcc_*_i387 splitters): Remove.
4527 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
4528 * config/i386/i386.c (ix86_split_fp_branch): Remove.
4529 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
4532 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4534 * profile-count.h (slow_safe_scale_64bit): New function.
4535 (safe_scale_64bit): New inline.
4536 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
4537 * profile-count.c: Include wide-int.h
4538 (slow_safe_scale_64bit): New.
4540 2017-10-11 Nathan Sidwell <nathan@acm.org>
4542 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
4543 HAS_DECL_ASSEMBLER_NAME_P.
4544 * gimple-expr.c (gimple_decl_printable_name: Check
4545 HAS_DECL_ASSEMBLER_NAME_P too.
4546 * ipa-utils.h (type_in_anonymous_namespace_p): Check
4547 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
4548 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
4549 * passes.c (rest_of_decl_compilation): Check
4550 HAS_DECL_ASSEMBLER_NAME_P too.
4551 * recog.c (verify_changes): Likewise.
4552 * tree-pretty-print.c (dump_decl_name): Likewise.
4553 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
4555 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
4556 (SET_DECL_ASSEMBLER_NAME): Use it.
4557 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
4558 (COPY_DECL_ASSEMBLER_NAME): Likewise.
4559 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
4561 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
4563 * config.gcc (i386, x86_64): Add extra objects.
4564 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
4565 (ix86_min_insn_size): Declare.
4566 (ix86_issue_rate): Declare.
4567 (ix86_adjust_cost): Declare.
4568 (ia32_multipass_dfa_lookahead): Declare.
4569 (ix86_macro_fusion_p): Declare.
4570 (ix86_macro_fusion_pair_p): Declare.
4571 (ix86_bd_has_dispatch): Declare.
4572 (ix86_bd_do_dispatch): Declare.
4573 (ix86_core2i7_init_hooks): Declare.
4574 (ix86_atom_sched_reorder): Declare.
4575 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
4576 (COSTS_N_BYTES): Move to x86-tune-costs.h.
4577 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
4578 (rip_relative_addr_p): Rename to ...
4579 (ix86_rip_relative_addr_p): ... this one; export.
4580 (memory_address_length): Update.
4581 (ix86_issue_rate): Move to x86-tune-sched.c.
4582 (ix86_flags_dependent): Move to x86-tune-sched.c.
4583 (ix86_agi_dependent): Move to x86-tune-sched.c.
4584 (exact_dependency_1): Move to x86-tune-sched.c.
4585 (exact_store_load_dependency): Move to x86-tune-sched.c.
4586 (ix86_adjust_cost): Move to x86-tune-sched.c.
4587 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
4588 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
4589 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
4590 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
4591 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
4592 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
4593 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
4594 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
4595 (min_insn_size): Rename to ...
4596 (ix86_min_insn_size): ... this one; export.
4597 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
4598 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
4599 (core2i7_first_cycle_multipass_backtrack): Move to
4600 x86-tune-sched-core.c.
4601 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
4602 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
4603 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
4604 (ix86_avoid_jump_mispredicts): Update.
4605 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
4606 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
4607 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
4608 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
4609 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
4610 (MAX_INSN): Move to ix86-tune-sched-bd.c.
4611 (MAX_IMM): Move to ix86-tune-sched-bd.c.
4612 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
4613 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
4614 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
4615 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
4616 (MAX_STORE): Move to ix86-tune-sched-bd.c.
4617 (BIG): Move to ix86-tune-sched-bd.c.
4618 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
4619 (enum insn_path): Move to ix86-tune-sched-bd.c.
4620 (get_mem_group): Move to ix86-tune-sched-bd.c.
4621 (is_cmp): Move to ix86-tune-sched-bd.c.
4622 (dispatch_violation): Move to ix86-tune-sched-bd.c.
4623 (is_branch): Move to ix86-tune-sched-bd.c.
4624 (is_prefetch): Move to ix86-tune-sched-bd.c.
4625 (init_window): Move to ix86-tune-sched-bd.c.
4626 (allocate_window): Move to ix86-tune-sched-bd.c.
4627 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
4628 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
4629 (process_end_window): Move to ix86-tune-sched-bd.c.
4630 (allocate_next_window): Move to ix86-tune-sched-bd.c.
4631 (find_constant): Move to ix86-tune-sched-bd.c.
4632 (get_num_immediates): Move to ix86-tune-sched-bd.c.
4633 (has_immediate): Move to ix86-tune-sched-bd.c.
4634 (get_insn_path): Move to ix86-tune-sched-bd.c.
4635 (get_insn_group): Move to ix86-tune-sched-bd.c.
4636 (count_num_restricted): Move to ix86-tune-sched-bd.c.
4637 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
4638 (add_insn_window): Move to ix86-tune-sched-bd.c.
4639 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
4640 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
4641 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
4642 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
4643 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
4644 (do_dispatch): Move to ix86-tune-sched-bd.c.
4645 (has_dispatch): Move to ix86-tune-sched-bd.c.
4646 * config/i386/t-i386: Add new object files.
4647 * config/i386/x86-tune-costs.h: New file.
4648 * config/i386/x86-tune-sched-atom.c: New file.
4649 * config/i386/x86-tune-sched-bd.c: New file.
4650 * config/i386/x86-tune-sched-core.c: New file.
4651 * config/i386/x86-tune-sched.c: New file.
4653 2017-10-11 Liu Hao <lh_mouse@126.com>
4655 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
4656 the generic version below instead.
4657 (should_colorize): Recognize Windows consoles as terminals
4659 * pretty-print.c [__MINGW32__] (write_all): New function.
4660 [__MINGW32__] (find_esc_head): Likewise.
4661 [__MINGW32__] (find_esc_terminator): Likewise.
4662 [__MINGW32__] (eat_esc_sequence): Likewise.
4663 [__MINGW32__] (mingw_ansi_fputs): New function that handles
4665 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
4668 2017-10-11 Richard Biener <rguenther@suse.de>
4670 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
4671 Properly call analyze_scalar_evolution with the loop of the stmt.
4673 2017-10-11 Richard Biener <rguenther@suse.de>
4675 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
4676 * tree-core.h (tree_base): Add chrec_var union member.
4677 * tree.h (CHREC_VAR): Remove.
4678 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
4679 * tree-chrec.h (build_polynomial_chrec): Adjust.
4680 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
4681 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
4683 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
4685 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
4686 * match.pd: ... here.
4687 ((T) X == (T) Y): Relax condition.
4689 2017-10-11 Bin Cheng <bin.cheng@arm.com>
4691 PR tree-optimization/82472
4692 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
4694 (break_alias_scc_partitions): Update postorder number.
4696 2017-10-11 Martin Liska <mliska@suse.cz>
4699 * opts.c (parse_no_sanitize_attribute): Do not use error_value
4701 * opts.h (parse_no_sanitize_attribute): Remove last argument.
4703 2017-10-11 Martin Liska <mliska@suse.cz>
4705 * print-rtl.c (print_insn): Move declaration of idbuf
4706 to same scope as name.
4708 2017-10-11 Martin Liska <mliska@suse.cz>
4713 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4716 2017-10-11 Martin Liska <mliska@suse.cz>
4719 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
4722 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
4724 * config.gcc (default_gnu_indirect_function): Default to yes for
4725 arm*-*-linux* with glibc.
4727 2017-10-11 Richard Biener <rguenther@suse.de>
4729 * tree-scalar-evolution.c (get_scalar_evolution): Handle
4730 default-defs and types we do not want to analyze.
4731 (interpret_loop_phi): Replace unreachable code with an assert.
4732 (compute_scalar_evolution_in_loop): Remove and inline ...
4733 (analyze_scalar_evolution_1): ... here, replacing condition with
4734 what makes the intent clearer. Remove handling of cases
4735 get_scalar_evolution now handles.
4737 2017-10-10 Jim Wilson <wilson@tuliptree.org>
4739 PR rtl-optimization/81434
4740 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
4741 comment for main loop. In sched_group_found if, also add checks for
4742 pass and min_cost_group.
4744 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
4746 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
4747 (rs6000_insn_cost): New function.
4748 * config/rs6000/rs6000.md (cost): New attribute.
4750 2017-10-10 Jakub Jelinek <jakub@redhat.com>
4751 H.J. Lu <hongjiu.lu@intel.com>
4755 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
4756 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
4757 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
4758 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
4759 to OPTION_MASK_ISA_AVX512VL - builtins that have both
4760 OPTION_MASK_ISA_MMX and some other bit set require both
4761 mmx and the ISAs without the mmx bit.
4762 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
4763 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
4764 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
4765 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
4766 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
4767 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
4768 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
4769 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
4770 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
4771 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
4772 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
4773 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
4774 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
4775 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
4776 Add OPTION_MASK_ISA_MMX.
4778 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
4780 * config.gcc (armv7*-*-freebsd*): New target.
4781 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
4783 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
4785 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
4786 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
4789 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
4791 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
4792 a template parameter.
4793 (WIDE_INT_REF_FOR): Update accordingly.
4794 * tree.h (wi::int_traits <const_tree>): Delete.
4795 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
4796 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
4797 (wi::tree_to_wide_ref): New typedef.
4798 (wi::to_wide): New function.
4799 * calls.c (get_size_range): Use wi::to_wide when operating on
4801 * cgraph.c (cgraph_node::create_thunk): Likewise.
4802 * config/i386/i386.c (ix86_data_alignment): Likewise.
4803 (ix86_local_alignment): Likewise.
4804 * dbxout.c (stabstr_O): Likewise.
4805 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
4806 * expr.c (const_vector_from_tree): Likewise.
4807 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
4808 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
4809 (fold_negate_expr_1, int_const_binop_1, const_binop)
4810 (fold_convert_const_int_from_real, optimize_bit_field_compare)
4811 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
4812 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
4813 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
4814 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
4815 (fold_not_const, round_up_loc): Likewise.
4816 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
4817 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
4818 (alloca_call_type): Likewise.
4819 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
4820 * godump.c (go_output_typedef): Likewise.
4821 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
4822 * internal-fn.c (get_min_precision): Likewise.
4823 * ipa-cp.c (ipcp_store_vr_results): Likewise.
4824 * ipa-polymorphic-call.c
4825 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
4826 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
4827 (ipa_modify_call_arguments): Likewise.
4828 * match.pd: Likewise.
4829 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
4830 * print-tree.c (print_node_brief, print_node): Likewise.
4831 * stmt.c (expand_case): Likewise.
4832 * stor-layout.c (layout_type): Likewise.
4833 * tree-affine.c (tree_to_aff_combination): Likewise.
4834 * tree-cfg.c (group_case_labels_stmt): Likewise.
4835 * tree-data-ref.c (dr_analyze_indices): Likewise.
4836 (prune_runtime_alias_test_list): Likewise.
4837 * tree-dump.c (dequeue_and_dump): Likewise.
4838 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
4839 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
4840 * tree-pretty-print.c (dump_generic_node): Likewise.
4841 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
4842 (simple_iv_with_niters): Likewise.
4843 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
4844 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
4845 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
4846 * tree-ssa-loop-niter.c (split_to_var_and_offset)
4847 (refine_value_range_using_guard, number_of_iterations_ne_max)
4848 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
4849 (get_cst_init_from_scev, record_nonwrapping_iv)
4850 (scev_var_range_cant_overflow): Likewise.
4851 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
4852 * tree-ssa-pre.c (compute_avail): Likewise.
4853 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
4854 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
4855 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
4856 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
4857 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
4858 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
4859 (dump_case_nodes, try_switch_expansion): Likewise.
4860 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
4861 (vect_do_peeling): Likewise.
4862 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
4863 * tree-vect-stmts.c (vectorizable_load): Likewise.
4864 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
4865 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
4866 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
4867 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
4868 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
4869 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
4870 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
4871 (evrp_dom_walker::before_dom_children): Likewise.
4872 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
4873 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
4874 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
4875 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
4876 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
4877 (get_range_pos_neg): Likewise.
4878 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
4879 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
4880 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
4881 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
4882 * config/avr/avr.c (avr_fold_builtin): Likewise.
4883 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
4884 * config/msp430/msp430.c (msp430_attr): Likewise.
4885 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
4886 * config/powerpcspe/powerpcspe-c.c
4887 (altivec_resolve_overloaded_builtin): Likewise.
4888 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
4889 (rs6000_expand_ternop_builtin): Likewise.
4890 * config/rs6000/rs6000-c.c
4891 (altivec_resolve_overloaded_builtin): Likewise.
4892 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
4893 (rs6000_expand_ternop_builtin): Likewise.
4894 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
4896 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4898 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
4899 when copying loop nest with only one inner loop.
4901 2017-10-10 Richard Biener <rguenther@suse.de>
4903 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
4904 blocks if SCEV is active.
4905 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
4907 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
4908 (scev_initialize): Assert we are not yet initialized.
4910 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4912 * tree-loop-distribution.c (generate_loops_for_partition): Remove
4913 inner loop's exit stmt by making it always exit the loop, otherwise
4914 we would generate an infinite empty loop.
4916 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4918 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
4919 renaming variables in new preheader if it's deleted.
4921 2017-10-10 Bin Cheng <bin.cheng@arm.com>
4923 * tree-loop-distribution.c (struct partition): Remove unused field
4924 loops of the structure.
4925 (partition_alloc, partition_free): Ditto.
4926 (build_rdg_partition_for_vertex): Ditto.
4928 2017-10-09 Jeff Law <law@redhat.com>
4930 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
4931 return type to match prototype and documentation.
4933 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4935 * config/rs6000/rs6000.c (processor_costs): Move to ...
4936 * config/rs6000/rs6000.h: ... here.
4937 (rs6000_cost): Declare.
4939 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
4941 * except.c (setjmp_fn): New global variable.
4942 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
4943 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
4944 if DONT_USE_BUILTIN_SETJMP is defined.
4946 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4948 * target.def (insn_cost): New hook.
4949 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
4950 * doc/tm.texi: Regenerate.
4951 * rtlanal.c (insn_cost): Use the new hook.
4953 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4955 * combine.c (combine_validate_cost): Compute the new insn_cost,
4956 not just pattern_cost.
4957 (try_combine): Adjust comment.
4959 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
4961 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
4963 * combine.c (uid_insn_cost): Adjust comment.
4964 (combine_validate_cost): Adjust comment. Use pattern_cost instead
4966 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
4967 * dse.c (find_shift_sequence): Ditto.
4968 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
4969 (bb_valid_for_noce_process_p): Use pattern_cost.
4970 * rtl.h (insn_rtx_cost): Delete.
4971 (pattern_cost): New prototype.
4972 (insn_cost): New prototype.
4973 * rtlanal.c (insn_rtx_cost): Rename to...
4974 (pattern_cost): ... this.
4977 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
4979 * config/i386/i386.md (*jcc_2): Remove insn pattern.
4980 (*jcc<mode>_0_r_i387): Ditto.
4981 (*jccxf_r_i387): Ditto.
4982 (*jcc<mode>_r_i387): Ditto.
4983 (*jccu<mode>_r_i387): Ditto.
4984 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
4985 (*jcc): Rename from *jcc_1.
4987 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4989 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
4990 deferred rescans after the lvx/stvx recombination pre-pass.
4992 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
4994 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
4995 memory operation instruction support.
4996 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
4997 (rs6000-ibm-aix[789]*): Likewise.
4998 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
4999 Document new functions.
5001 2017-10-09 Richard Biener <rguenther@suse.de>
5003 PR tree-optimization/82397
5004 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
5005 equality only for semantically equal trees.
5007 2017-10-09 Richard Biener <rguenther@suse.de>
5009 PR tree-optimization/82449
5010 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
5011 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
5012 allow constant addresses.
5013 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
5016 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5018 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
5021 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5024 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
5027 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5030 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
5032 2017-10-09 Jakub Jelinek <jakub@redhat.com>
5035 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
5036 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
5038 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
5040 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
5041 (WI_BINARY_PREDICATE_RESULT): Likewise.
5042 (wi::binary_traits::operator_result): New type.
5043 (wi::binary_traits::predicate_result): Likewise.
5044 (generic_wide_int::operator~, unary generic_wide_int::operator-)
5045 (generic_wide_int::operator==, generic_wide_int::operator!=)
5046 (generic_wide_int::operator&, generic_wide_int::and_not)
5047 (generic_wide_int::operator|, generic_wide_int::or_not)
5048 (generic_wide_int::operator^, generic_wide_int::operator+
5049 (binary generic_wide_int::operator-, generic_wide_int::operator*):
5051 (operator~, unary operator-, operator==, operator!=, operator&)
5052 (operator|, operator^, operator+, binary operator-, operator*): New
5054 * expr.c (get_inner_reference): Use wi::bit_and_not.
5055 * fold-const.c (fold_binary_loc): Likewise.
5056 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
5057 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
5058 (bit_value_binop): Likewise.
5059 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
5060 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
5061 (extract_range_from_binary_expr_1): Likewise.
5062 (masked_increment): Likewise.
5063 (simplify_bit_ops_using_ranges): Likewise.
5065 2017-10-09 Martin Jambor <mjambor@suse.cz>
5068 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
5069 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
5070 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
5072 (hsa_fixup_mov_insn_type): New function.
5073 (hsa_op_with_type::get_in_type): Use it.
5074 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
5075 immediates in an assert.
5076 (hsa_op_with_type::extend_int_to_32bit): New method.
5077 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
5078 types. Convert to dest type if necessary.
5079 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
5080 (reg_for_gimple_ssa): Pass false as min32int to
5081 hsa_type_for_scalar_tree_type.
5082 (gen_hsa_addr): Fixup type when creating addresable temporary.
5083 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
5084 (gen_hsa_unary_operation): Extend operands and convert to dest type if
5085 necessary. Call hsa_fixup_mov_insn_type.
5086 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
5087 extend operands and convert to dest type if necessary.
5088 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
5089 to dest type if necessary.
5090 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
5091 if conversion nt necessary and size matches.
5092 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
5093 to dest type if necessary.
5094 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
5095 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
5097 (gen_hsa_clrsb): Likewise.
5098 (gen_hsa_ffs): Likewise.
5099 (gen_hsa_divmod): Extend operands and convert to dest type if
5101 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
5103 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5105 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
5106 Remove empty default arguments. Use a brace block as output
5108 (conditional return): Ditto.
5110 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
5111 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
5113 (group_ending_nop): Ditto.
5114 (doloop_end): Ditto.
5115 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
5116 (splitters for those): Ditto.
5118 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
5120 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
5121 a conditional jump (and the compare for it) so that pc_rtx is the
5123 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
5124 for the deleted and renamed ctr<mode>_internal[234] patterns.
5125 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
5126 Delete second conditional return pattern.
5127 (ctr<mode>_internal2): Delete this second bdnz pattern.
5128 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
5129 (ctr<mode>_internal4): Delete this second bdz pattern.
5131 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5133 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
5134 (always_initialized_rtx_for_ssa_name_p): New predicate.
5135 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
5136 (finish_out_of_ssa): Free new field of SA.
5137 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
5138 * tree-ssa-coalesce.c: Include tree-ssa.h.
5139 (get_parm_default_def_partitions): Remove extern keyword.
5140 (get_undefined_value_partitions): New function.
5141 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
5142 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
5144 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
5146 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
5148 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
5150 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5152 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
5153 for targets that preffer 128bit.
5155 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
5157 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
5159 2017-10-08 Olivier Hainque <hainque@adacore.com>
5161 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
5162 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
5164 (thumb_set_return_address): Likewise.
5166 2017-10-08 Olivier Hainque <hainque@adacore.com>
5168 * common/config/arm/arm-common.c (arm_except_unwind_info):
5169 Handle DWARF2_UNWIND_INFO.
5171 2017-10-07 Michael Collison <michael.collison@arm.com>
5173 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
5176 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
5178 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
5179 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
5180 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
5181 defined, force the creation of a new block for a dispatch label.
5183 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
5185 * invoke.texi (Wsuggest-attribute=cold): Document.
5186 * common.opt (Wsuggest-attribute=cold): New
5187 * ipa-pure-const.c (warn_function_cold): New function.
5188 * predict.c (compute_function_frequency): Use it.
5189 * predict.h (warn_function_cold): Declare.
5191 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
5193 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
5196 2017-10-06 Martin Liska <mliska@suse.cz>
5198 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
5199 keyword for member functions.
5200 (struct sanopt_tree_couple): New struct.
5201 (struct sanopt_tree_couple_hash): New function.
5202 (struct sanopt_ctx): Add new hash_map.
5203 (has_dominating_ubsan_ptr_check): New function.
5204 (record_ubsan_ptr_check_stmt): Likewise.
5205 (maybe_optimize_ubsan_ptr_ifn): Likewise.
5206 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
5207 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
5209 2017-10-06 Sudakshina Das <sudi.das@arm.com>
5212 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
5213 aarch64_simd_valid_immediate on CONST_VECTORs.
5214 (aarch64_reg_or_bic_imm): Likewise.
5216 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
5218 PR rtl-optimization/82396
5219 * haifa-sched.c (ready_sort_real): Disable qsort checking.
5221 2017-10-06 Sebastian Pop <sebpop@gmail.com>
5223 * graphite-dependences.c (scop_get_reads): Move code to...
5224 (scop_get_must_writes): Move code to...
5225 (scop_get_may_writes): Move code to...
5226 (scop_get_reads_and_writes): ... here.
5227 (scop_get_dependences): Call scop_get_reads_and_writes.
5229 2017-10-06 Jakub Jelinek <jakub@redhat.com>
5231 PR tree-optimization/82434
5232 * fold-const.h (can_native_encode_type_p,
5233 can_native_encode_string_p): Remove.
5234 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
5235 don't encode anything, just return what would be otherwise returned.
5236 (native_encode_fixed, native_encode_complex, native_encode_vector):
5238 (native_encode_string): Likewise. Inline by hand
5239 can_native_encode_string_p.
5240 (can_native_encode_type_p): Remove.
5241 (can_native_encode_string_p): Remove.
5242 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
5243 STRING_CSTs using can_native_encode_string_p, test all
5244 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
5245 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
5246 argument from native_encode_expr.
5247 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
5248 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
5251 2017-10-06 Richard Biener <rguenther@suse.de>
5253 PR tree-optimization/82397
5254 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
5255 operand_equal_p but rely on data_ref_compare_tree for detecting
5257 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
5258 to match up with dr_group_sort_cmp.
5260 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5263 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
5265 * config/s390/s390-builtin-types.def: Regenerate.
5267 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5270 * config/s390/s390-builtin-types.def: Regenerate.
5271 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
5272 Change flag from B_VXE to B_VX.
5273 (s390_vec_min_dbl): Remove B_VXE flag.
5275 2017-10-06 Richard Biener <rguenther@suse.de>
5277 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
5278 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
5279 translate_isl_ast_to_gimple::is_valid_rename,
5280 translate_isl_ast_to_gimple::get_rename,
5281 translate_isl_ast_to_gimple::get_def_bb_for_const,
5282 translate_isl_ast_to_gimple::get_new_name,
5283 translate_isl_ast_to_gimple::collect_all_ssa_names,
5284 translate_isl_ast_to_gimple::copy_loop_phi_args,
5285 translate_isl_ast_to_gimple::collect_all_ssa_names,
5286 translate_isl_ast_to_gimple::copy_loop_phi_args,
5287 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
5288 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
5289 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
5290 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
5291 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
5292 translate_isl_ast_to_gimple::copy_cond_phi_args,
5293 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
5294 translate_isl_ast_to_gimple::edge_for_new_close_phis,
5295 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
5296 translate_isl_ast_to_gimple::rename_uses,
5297 translate_isl_ast_to_gimple::rename_all_uses): Remove.
5298 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
5299 (set_rename_for_each_def): Likewise.
5300 (graphite_copy_stmts_from_block): Handle debug stmt resetting
5301 here. Handle rewriting SCEV analyzable uses here.
5302 (copy_bb_and_scalar_dependences): Generate code for PHI
5304 (graphite_regenerate_ast_isl): Adjust.
5305 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
5306 (add_write, add_read): New functions.
5307 (build_cross_bb_scalars_def): Use it and simplify.
5308 (build_cross_bb_scalars_use): Likewise.
5309 (graphite_find_cross_bb_scalar_vars): Inline into...
5310 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
5311 simulating out-of-SSA. Compute liveout and add dependencies.
5312 (build_scops): Force an empty entry block.
5313 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
5315 (sese_build_liveouts): Declare.
5316 (sese_trivially_empty_bb_p): Likewise.
5317 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
5318 compute liveout and debug_liveout.
5319 (sese_bad_liveouts_use): Remove.
5320 (sese_reset_debug_liveouts_bb): Likewise.
5321 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
5322 (sese_build_liveouts): Build liveout and debug_liveout and store
5324 (new_sese_info): Adjust.
5325 (free_sese_info): Likewise.
5326 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
5327 do not build liveout here.
5328 (move_sese_in_condition): Adjust region entry.
5329 (scev_analyzable_p): Match up with chrec_apply requirements.
5330 (sese_trivially_empty_bb_p): New.
5331 * tree-into-ssa.c (get_reaching_def): Properly support generating
5332 default-defs for incremental rewrite of anonymous names.
5334 2017-10-06 Richard Biener <rguenther@suse.de>
5336 * graphite-sese-to-poly.c (extract_affine): For casts increasing
5337 precision do not perform modulo reduction.
5339 2017-10-06 Richard Biener <rguenther@suse.de>
5341 PR tree-optimization/82436
5342 * tree-vect-slp.c (vect_supported_load_permutation_p): More
5343 conservatively choose the vectorization factor when checking
5344 whether we can perform the required load permutation.
5345 (vect_transform_slp_perm_load): Assert when we may not fail.
5347 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
5349 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
5350 message for incompatible -msdata=* and -mcall-* options.
5352 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5354 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
5355 rate for post-reload scheduling.
5357 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5359 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
5361 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5363 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
5364 to improve monte carlo in scimark.
5366 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
5368 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
5369 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
5370 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
5371 pentium4_cost, nocona_cost): Set reassociation width to 1.
5372 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
5373 width to 2 for fp operations and 1 otherwise.
5374 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
5376 (atom_cost): Set reassociation width to 2.
5377 (slm_cost, generic_cost): Set fp reassociation width
5378 to 2 and 1 otherwise.
5379 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
5380 (core_cost): Set fp reassociation width to 4 and vector to 2.
5381 (ix86_reassociation_width): Rewrite using cost table; special case
5382 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
5383 and TARGET_AVX128_OPTIMAL.
5384 * config/i386/i386.h (processor_costs): Add
5385 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
5386 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
5387 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
5388 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
5389 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
5390 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
5392 2017-10-05 Nathan Sidwell <nathan@acm.org>
5394 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
5396 2017-10-05 Tamar Christina <tamar.christina@arm.com>
5398 * config/arm/arm.c (arm_test_fpu_data): New.
5399 (arm_run_selftests): Call arm_test_fpu_data.
5401 2017-10-04 Nathan Sidwell <nathan@acm.org>
5403 * toplev.c (toplev::main): Remove excess parens on pretty_printer
5405 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
5407 2017-10-04 Sudakshina Das <sudi.das@arm.com>
5409 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
5410 check type for aarch64_simd_valid_immediate.
5411 (aarch64_output_simd_mov_immediate): Update prototype.
5412 (aarch64_simd_valid_immediate): Update prototype.
5413 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
5414 support for ORR-immediate.
5415 (and<mode>3): modified pattern to add support for BIC-immediate.
5416 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
5417 now checks for valid immediate for BIC and ORR based on new enum
5419 (aarch64_output_simd_mov_immediate): Function now used to output
5420 BIC/ORR imm as well based on new enum argument.
5421 * config/aarch64/constraints.md (Do): New vector immediate constraint.
5423 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
5424 (aarch64_reg_or_bic_imm): Likewise.
5426 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5428 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
5429 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
5431 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5435 PR rtl-optimization/82396
5436 * haifa-sched.c (autopref_multipass_init): Simplify
5438 (autopref_rank_data): Simplify sort order.
5439 * sched-int.h (autopref_multipass_data_): Remove
5440 multi_mem_insn_p, min_offset and max_offset.
5442 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5444 * doc/sourcebuild.texi: Document vect_peeling_profitable.
5446 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5448 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
5451 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5453 * doc/sourcebuild.texi: Document vect_long_mult.
5455 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
5457 PR tree-optimization/82413
5458 * fold-const.c (build_range_check): Use widest_int when comparing
5459 the maximum ETYPE value with HIGH.
5461 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
5463 PR rtl-optimization/82396
5464 * haifa-sched.c (autopref_multipass_init): Simplify
5466 (autopref_rank_data): Simplify sort order.
5467 * sched-int.h (autopref_multipass_data_): Remove
5468 multi_mem_insn_p, min_offset and max_offset.
5470 2017-10-04 Jakub Jelinek <jakub@redhat.com>
5472 PR tree-optimization/82381
5473 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
5474 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
5477 PR tree-optimization/82374
5478 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
5479 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
5480 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
5481 current_function_decl to the new decl.
5483 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
5485 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
5486 helper macro for IEEE float128 hardware built-in functions.
5487 (SQRTF128_ODD): Add built-in functions with the round-to-odd
5489 (TRUNCF128_ODD): Likewise.
5490 (ADDF128_ODD): Likewise.
5491 (SUBF128_ODD): Likewise.
5492 (MULF128_ODD): Likewise.
5493 (DIVF128_ODD): Likewise.
5494 (FMAF128_ODD): Likewise.
5495 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
5496 UNSPEC_TRUNC_ROUND_TO_ODD.
5497 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
5498 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
5499 floating point round to odd instructions.
5500 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
5501 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
5502 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
5503 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
5504 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
5505 (trunc<mode>sf2_hw): Change the truncate with round to odd
5506 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
5507 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
5508 to odd hardware instructions.
5509 (sub<mode>3_odd): Likewise.
5510 (mul<mode>3_odd): Likewise.
5511 (div<mode>3_odd): Likewise.
5512 (sqrt<mode>2_odd): Likewise.
5513 (fma<mode>4_odd): Likewise.
5514 (fms<mode>4_odd): Likewise.
5515 (nfma<mode>4_odd): Likewise.
5516 (nfms<mode>4_odd): Likewise.
5517 (trunc<mode>df2_odd): Change the truncate with round to odd
5518 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
5520 * doc/extend.texi (PowerPC built-in functions): Update documentation
5521 for existing IEEE float128-bit built-in functions. Add built-in
5522 functions that generate the IEEE 128-bit floating point round to
5525 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
5527 PR rtl-optimization/77729
5528 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
5529 to (X&(C1&~C2))|C2 transformations.
5531 2017-10-03 Martin Jambor <mjambor@suse.cz>
5533 PR tree-optimization/82363
5534 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
5535 mismatch, mark lacc written regardless of racc.
5537 2017-10-03 Jakub Jelinek <jakub@redhat.com>
5539 PR tree-optimization/82381
5540 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
5541 stmt_to_insert nor wheather SSA_NAMEs are default defs.
5542 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
5543 fallthrough into reassoc_stmt_dominates_stmt_p.
5546 * combine.c (combine_instructions): Don't combine in unreachable
5549 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
5552 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
5553 function to not use the have_cpu variable. Do not set cpu_index,
5554 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
5556 (rs6000_valid_attribute_p): Remove duplicate initializations of
5557 old_optimize and func_optimize.
5558 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
5559 (rs6000_activate_target_options): Make global.
5560 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
5563 2017-10-02 Jakub Jelinek <jakub@redhat.com>
5565 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
5566 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
5567 Set *poffset to 0 and *psize and *pmax_size to -1 if
5568 *poffset + *psize overflows in HOST_WIDE_INT.
5570 PR tree-optimization/82387
5571 PR tree-optimization/82388
5572 PR tree-optimization/82389
5573 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
5574 instead of live_bytes non-NULL.
5576 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
5579 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
5582 2017-10-02 Richard Biener <rguenther@suse.de>
5584 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
5585 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
5586 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
5588 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5590 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
5591 requested precision matches the type's.
5592 * calls.c (alloc_max_size): Calculate the new candidate size as
5593 a widest_int and use wi::to_widest when comparing it with the
5594 current candidate size.
5595 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
5596 zero rather than integer_zero_node.
5597 * match.pd: Check for a no-op conversion before using wi::add
5598 rather than after. Use tree_to_uhwi when summing small shift
5599 counts into an unsigned int.
5601 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
5602 Alan Hayward <alan.hayward@arm.com>
5603 David Sherwood <david.sherwood@arm.com>
5606 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
5607 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
5608 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
5609 POINTER_AND_FP_REGS.
5611 2017-10-02 Richard Biener <rguenther@suse.de>
5613 PR tree-optimization/82355
5614 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
5615 a mapping for the enclosing loop but avoid generating one for
5617 (copy_bb_and_scalar_dependences): Remove premature codegen
5618 error on PHIs in blocks duplicated into multiple places.
5619 * graphite-scop-detection.c
5620 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
5621 in the region use it as loop and nest to analyze the DR in.
5622 (try_generate_gimple_bb): Likewise.
5623 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
5624 (add_loop_constraints): For blocks in a loop not in the region
5625 create a dimension with a single iteration.
5626 * sese.h (gbb_loop_at_index): Remove assert.
5628 2017-10-01 Kevin Buettner <kevinb@redhat.com>
5630 * omp-expand.c (adjust_context_scope): New function.
5631 (expand_parallel_call): Call adjust_context_scope.
5633 2017-10-01 Jeff Law <law@redhat.com>
5635 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
5636 dom_opt_dom_walker class with direct access to private members.
5637 Add comments. Call test_for_singularity.
5638 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
5639 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
5640 m_dummy_cond anymore.
5641 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
5643 (pass_dominator:execute): Build the dummy_cond here and pass it
5644 to the dom_opt_dom_walker ctor.
5645 (test_for_singularity): New function.
5647 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
5648 Maya Rashish <coypu@sdf.org>
5650 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
5651 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
5652 (alpha*-*-netbsd*) Use nbsd_tm_file.
5653 (arm*-*-netbsdelf*) Likewise.
5654 (i[34567]86-*-netbsdelf*) Likewise.
5655 (x86_64-*-netbsd*) Likewise.
5656 (mips*-*-netbsd*) Likewise.
5657 (powerpc-*-netbsd*) Likewise.
5658 (sh*-*-netbsd*) Likewise.
5659 (sparc-*-netbsdelf*) Likewise.
5660 (sparc64-*-netbsd*) Likewise.
5661 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
5663 (vax-*-netbsdelf*) Likewise.
5664 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
5665 (UINT_FAST8_TYPE) Likewise.
5666 (INT_FAST16_TYPE) Check CHAR_FAST16.
5667 (UINT_FAST16_TYPE) Likewise.
5669 2017-09-30 Jakub Jelinek <jakub@redhat.com>
5672 * config/i386/i386.md
5673 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
5674 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
5675 *divmodsi4_zext_2): New define_insn_and_split.
5676 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
5677 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
5678 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
5679 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
5680 New define_insn_and_split.
5681 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
5682 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
5683 operands[1] having DImode when mode is SImode.
5685 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
5686 always SImode for DIV and MOD in REG_EQUAL notes.
5688 2017-09-29 Yury Gribov <tetra2005@gmail.com>
5691 * match.pd: Fix handling of NaNs in pattern.
5693 2017-09-29 Jeff Law <law@redhat.com>
5695 * sbitmap.c (bitmap_bit_in_range_p): New function.
5696 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
5697 * tree-ssa-dse.c (live_bytes_read): New function.
5698 (dse_classify_store): Ignore reads of dead bytes.
5700 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
5701 typos and whitespace errors.
5702 * config/i386/predicates.md (address_no_seg_operand): Likewise.
5703 * config/s390/s390.c (s390_emit_prologue): Likewise.
5705 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5708 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
5709 with a symbol for LRA.
5711 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
5713 PR rtl-optimization/82338
5714 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
5716 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
5718 * genmodes.c (calc_wider_mode): Suppress qsort macro.
5719 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
5720 (qsort_chk): Declare.
5721 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
5722 (qsort_chk): New function.
5724 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5726 PR tree-optimization/82337
5727 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
5728 phi definition if the PHI result appears in an abnormal PHI.
5729 (find_basis_for_base_expr): Don't record a basis if the LHS of the
5730 basis appears in an abnormal PHI.
5732 2017-09-29 Richard Biener <rguenther@suse.de>
5734 * graphite-isl-ast-to-gimple.c
5735 (translate_isl_ast_to_gimple::set_codegen_error): New function.
5736 (binary_op_to_tree): Use it.
5737 (get_rename_from_scev): Likewise.
5738 (copy_loop_phi_nodes): Likewise.
5739 (copy_bb_and_scalar_dependences): Likewise.
5740 (translate_pending_phi_nodes): Likewise.
5742 2017-09-29 Jakub Jelinek <jakub@redhat.com>
5745 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
5746 for movabsq $(i32 << shift), r64.
5748 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
5750 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
5751 index when encoding %esp as %rsp to avoid 0x67 prefix.
5753 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
5755 * config/i386/i386.md (*movsf_internal, *movdf_internal):
5756 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
5758 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
5760 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
5761 Extensions with more than 16 double VFP registers.
5762 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
5763 to_clear_mask and all code related to it. Replace the remaining
5764 entry by a sbitmap and adapt code accordingly.
5766 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
5768 * brig-builtins.def: Change pure attributes to const.
5770 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5772 * config.gcc (default_gnu_indirect_function): Default to yes for
5773 sparc*-*-linux* with glibc.
5775 2017-09-28 Joseph Myers <joseph@codesourcery.com>
5777 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
5778 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
5779 when creating .init_array and .fini_array sections with priority
5782 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
5785 * config/aarch64/aarch64.c
5786 (aarch64_builtin_support_vector_misalignment): Always return false
5787 when misalignment is unknown.
5789 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
5791 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
5792 this function to return false if the definition used by the swap
5793 instruction is artificial, or if the memory address from which the
5794 constant value is loaded is not represented by a base address held
5795 in a register or if the base address register is a frame or stack
5796 pointer. Additionally, return false if the base address of the
5797 loaded constant is a SYMBOL_REF but is not considered to be a
5799 (replace_swapped_load_constant): New function.
5800 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
5801 loaded constant vector with a load of a swapped constant vector.
5803 2017-09-27 Carl Love <cel@us.ibm.com>
5805 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
5806 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
5807 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
5810 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
5812 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
5813 first, always call autopref_rank_data otherwise.
5815 2017-09-27 Richard Biener <rguenther@suse.de>
5817 * graphite-scop-detection.c (find_scop_parameters): Move
5818 loop bound handling ...
5819 (gather_bbs::before_dom_children): ... here, avoiding the need
5820 to build scop_info->loop_nest.
5821 (record_loop_in_sese): Remove.
5822 * sese.h (sese_info_t::loop_nest): Remove.
5823 * sese.c (new_sese_info): Do not allocate loop_nest.
5824 (free_sese_info): Do not free loop_nest.
5826 2017-09-27 Jakub Jelinek <jakub@redhat.com>
5829 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
5830 lhs from calls if the lhs has addressable type.
5832 2017-09-27 Richard Biener <rguenther@suse.de>
5834 * graphite.h (scop::max_alias_set): New member.
5835 * graphite-scop-detection.c: Remove references to non-existing
5836 --param in comments.
5837 (build_alias_sets): Record the maximum alias set used for drs.
5838 (build_scops): Support zero as unlimited for
5839 --param graphite-max-arrays-per-scop.
5840 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
5842 (build_poly_sr_1): ... here. Compute alias set based on the
5843 maximum alias set used for drs rather than
5844 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
5846 2017-09-27 Richard Biener <rguenther@suse.de>
5848 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
5849 --param loop-block-tile-size=0 to disable tiling.
5851 2017-09-27 Richard Biener <rguenther@suse.de>
5853 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
5854 (graphite-max-nb-scop-params): Document special value zero.
5855 * domwalk.h (dom_walker::STOP): New symbolical constant.
5856 (dom_walker::dom_walker): Add optional parameter for bb to
5858 (dom_walker::~dom_walker): Declare.
5859 (dom_walker::before_dom_children): Document STOP return value.
5860 (dom_walker::m_user_bb_to_rpo): New member.
5861 (dom_walker::m_bb_to_rpo): Likewise.
5862 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
5863 mapping here if not provided by the user.
5864 (dom_walker::~dom_walker): Free bb to RPO mapping if not
5865 provided by the user.
5866 (dom_walker::STOP): Define.
5867 (dom_walker::walk): Do not compute bb to RPO mapping here.
5868 Support STOP return value from before_dom_children to stop
5870 * graphite-optimize-isl.c (optimize_isl): If the schedule
5871 is the same still generate code if -fgraphite-identity
5872 or -floop-parallelize-all are given.
5873 * graphite-scop-detection.c: Include cfganal.h.
5874 (gather_bbs::gather_bbs): Get and pass through bb to RPO
5876 (gather_bbs::before_dom_children): Return STOP for BBs
5878 (build_scops): Compute bb to RPO mapping and pass it to
5879 the domwalk. Treat --param graphite-max-nb-scop-params=0
5880 as not limiting the number of params.
5881 * graphite.c (graphite_initialize): Remove limit on the
5882 number of basic-blocks in a function.
5883 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
5884 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
5885 default value of 10.
5887 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5889 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
5890 Adjust code to eliminate needing to do the shift right 32-bits
5891 operation after XSCVDPSPN.
5893 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5895 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
5896 ((X / Y) != 0 -> X >= Y): Likewise.
5898 2017-09-26 Carl Love <cel@us.ibm.com>
5900 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
5901 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
5902 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
5903 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
5904 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
5905 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
5906 definitions and overloading.
5907 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
5908 statement for P9V_BUILTIN_XST_LEN_R.
5909 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
5910 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
5911 define_expand and define_insn for the instructions and builtins.
5912 * doc/extend.texi: Update the built-in documentation file for the new
5914 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
5915 define_insn for the instructions
5917 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
5920 * gcc/config/netbsd-protos.h: New file.
5921 * gcc/config/netbsd.c: New file.
5922 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
5923 * gcc/config/t-netbsd: New file.
5924 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
5925 (tmake_file) Add t-netbsd.
5926 (extra_objs) Add netbsd.o.
5928 2017-09-26 Janus Weil <janus@gcc.gnu.org>
5932 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
5934 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
5936 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
5937 sign extension from a vector register to a GPR by doing a 32-bit
5938 direct move and then an EXTSW.
5939 (extendsi<mode>2 splitter): Likewise.
5940 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
5941 right or vector extract after doing XSCVDPSPN. Use
5942 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
5944 (movdi_from_sf_zero_ext): Likewise.
5945 (reload_gpr_from_vsxsf): Likewise.
5946 (p8_mfvsrd_4_disf): Delete, no longer used.
5947 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
5948 then needing to move the SFmode to a GPR to use the XSCVDPSP
5949 instruction instead of FRSP and XSCVDPSPN.
5950 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
5951 it is adjacent to the other XSCVSPDP insns.
5952 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
5953 SFmode to be in traditional Altivec registers.
5954 (vsx_xscvdpspn): Eliminate useless alternative constraint.
5955 (vsx_xscvspdpn): Likewise.
5956 (vsx_xscvspdpn_scalar): Likewise.
5958 2017-09-26 Martin Jambor <mjambor@suse.cz>
5960 * tree-sra.c (compare_access_positions): Put integral types first,
5961 stabilize sorting of integral types, remove conditions putting
5962 non-full-precision integers last.
5963 (sort_and_splice_var_accesses): Disable scalarization if a
5964 non-integert would be represented by a non-full-precision integer.
5966 2017-09-26 Joseph Myers <joseph@codesourcery.com>
5968 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
5969 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
5970 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
5971 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
5972 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
5973 conditionals inside the function instead of around it. Call
5974 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
5975 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
5977 2017-09-26 Richard Biener <rguenther@suse.de>
5979 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
5981 (scop_detection::build_scop_breadth): ... this. Removed.
5982 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
5983 (scop_detection::harmful_stmt_in_bb): Likewise.
5984 (scop_detection::graphite_can_represent_stmt): Likewise.
5985 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
5986 (scop_detection::can_represent_loop): Remove recursion, fold in ...
5987 (scop_detection::can_represent_loop_1): ... this. Removed.
5988 (scop_detection::harmful_loop_in_region): Simplify after inlining
5989 the above and remove more quadraticness.
5990 (build_scops): Adjust.
5991 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
5994 2017-09-26 Jakub Jelinek <jakub@redhat.com>
5997 * config/i386/i386.c (ix86_print_operand_address_as): Only test
5998 REGNO (base) == SP_REG if base is a REG.
6001 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
6002 if it is different SSA_NAME.
6003 (optimize_range_tests_cmp_bitwise): New function.
6004 (optimize_range_tests): Call it.
6006 2017-09-26 Richard Biener <rguenther@suse.de>
6008 PR tree-optimization/82321
6009 * graphite.c (canonicalize_loop_closed_ssa): Properly check
6010 for the def being inside the loop.
6012 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6014 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
6016 * config/s390/s390-builtins.def: Fix constraint on op4.
6018 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6020 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
6021 independent expanders.
6022 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
6023 ("vec_ordered", "vec_unordered"): New expanders.
6025 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6027 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
6030 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6032 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
6033 vec_unpacks_lo_v16qi.
6034 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
6036 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6038 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
6039 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
6040 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
6042 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6044 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
6046 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
6048 ("*vec_slb<mode>"): New insn pattern.
6049 ("vec_shr_<mode>"): New expander.
6050 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
6051 and force the shift count operand to V16QImode.
6052 ("vec_srb<mode>"): Set shift count mode to V16QI.
6054 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6056 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
6057 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
6058 ("vec_widen_smult_hi_<mode>"): New expander definitions.
6060 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
6063 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
6065 2017-09-26 Richard Biener <rguenther@suse.de>
6067 PR tree-optimization/82320
6068 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
6071 2017-09-25 Jeff Law <law@redhat.com>
6073 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
6074 prototype for new argument.
6075 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
6076 mostly extracted from rs6000_emit_allocate_stack.
6077 (rs6000_emit_probe_stack_range_stack_clash): New function.
6078 (rs6000_emit_allocate_stack): Call
6079 rs6000_emit_probe_stack_range_stack_clash as needed.
6080 (rs6000_emit_probe_stack_range): Add additional argument
6081 to call to gen_probe_stack_range{si,di}.
6082 (output_probe_stack_range): New.
6083 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
6084 (output_probe_stack_range_stack_clash): New.
6085 (rs6000_emit_prologue): Emit notes into dump file as requested.
6086 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
6087 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
6088 Add additional operand and pass it to output_probe_stack_range.
6090 2017-09-25 Bin Cheng <bin.cheng@arm.com>
6092 PR tree-optimization/82163
6093 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
6094 (checking_verify_loop_closed_ssa): New parameter.
6095 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
6096 (check_loop_closed_ssa_stmt): Delete.
6097 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
6098 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
6099 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
6102 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
6104 * brig-builtins.def: Treat HSAIL barrier builtins as
6105 setjmp/longjump style functions.
6107 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6109 * target.def (constant_alignment): New hook.
6110 * defaults.h (CONSTANT_ALIGNMENT): Delete.
6111 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
6112 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
6113 * doc/tm.texi: Regenerate.
6114 * targhooks.h (default_constant_alignment): Declare.
6115 (constant_alignment_word_strings): Likewise.
6116 * targhooks.c (default_constant_alignment): New function.
6117 (constant_alignment_word_strings): Likewise.
6118 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
6119 instead of CONSTANT_ALIGNMENT.
6120 * varasm.c (align_variable, get_variable_align, build_constant_desc)
6121 (force_const_mem): Likewise.
6122 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
6123 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
6124 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
6125 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6126 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
6128 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
6129 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6130 constant_alignment_word_strings.
6131 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
6132 (CONSTANT_ALIGNMENT): Likewise.
6133 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6134 (arm_constant_alignment): New function.
6135 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
6136 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6137 constant_alignment_word_strings.
6138 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
6139 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6140 constant_alignment_word_strings.
6141 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
6142 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6143 (cris_constant_alignment): New function.
6144 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
6145 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6146 (epiphany_constant_alignment): New function.
6147 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
6148 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6149 constant_alignment_word_strings.
6150 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
6151 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6152 constant_alignment_word_strings.
6153 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
6154 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6155 constant_alignment_word_strings.
6156 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
6157 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
6158 * config/i386/i386.c (ix86_constant_alignment): Make static.
6159 Use the same interface as the target hook.
6160 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6161 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
6162 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6163 constant_alignment_word_strings.
6164 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
6165 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
6166 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6167 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
6168 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6169 constant_alignment_word_strings.
6170 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
6171 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6172 constant_alignment_word_strings.
6173 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
6174 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6175 constant_alignment_word_strings.
6176 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
6177 * config/microblaze/microblaze.c (microblaze_constant_alignment):
6179 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6180 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
6181 * config/mips/mips.c (mips_constant_alignment): New function.
6182 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6183 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
6184 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
6185 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6186 (mmix_constant_alignment): Make static. Use the same interface
6188 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
6189 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6190 constant_alignment_word_strings.
6191 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
6192 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6193 constant_alignment_word_strings.
6194 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
6195 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6196 constant_alignment_word_strings.
6197 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
6198 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6199 (rs6000_constant_alignment): New function.
6200 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
6201 * config/riscv/riscv.c (riscv_constant_alignment): New function.
6202 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6203 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
6204 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6205 (rs6000_constant_alignment): New function.
6206 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
6207 * config/s390/s390.c (s390_constant_alignment): New function.
6208 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6209 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
6210 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6211 constant_alignment_word_strings.
6212 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
6213 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6214 (sparc_constant_alignment): New function.
6215 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
6216 * config/spu/spu.c (spu_constant_alignment): New function.
6217 (TARGET_CONSTANT_ALIGNMENT): Redefine.
6218 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
6219 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6220 constant_alignment_word_strings.
6221 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
6222 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6223 constant_alignment_word_strings.
6224 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
6225 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
6226 constant_alignment_word_strings.
6227 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
6228 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6229 (visium_constant_alignment): New function.
6230 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
6231 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
6232 (xtensa_constant_alignment): New function.
6233 * system.h (CONSTANT_ALIGNMENT): Poison.
6235 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
6237 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6238 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
6239 (rs6000_builtin_valid_without_lhs): New helper function.
6240 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6241 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
6243 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
6245 * target.h (vec_perm_indices): Use unsigned short rather than
6247 (auto_vec_perm_indices): Likewise.
6248 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
6249 Use unsigned int rather than unsigned char.
6250 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
6252 2017-09-25 Richard Biener <rguenther@suse.de>
6254 * cfgloop.h (sort_sibling_loops): Declare.
6255 * cfgloop.c (sort_sibling_loops_cmp): New helper.
6256 (sort_sibling_loops): New function sorting the sibling loop list
6258 * graphite.c (graphite_transform_loops): Sort sibling loops.
6260 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
6262 * target.def (vec_perm_const_ok): Change sel parameter to
6264 * optabs-query.c (can_vec_perm_p): Update accordingly.
6265 * doc/tm.texi: Regenerate.
6266 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
6267 auto_vec_perm_indices and remove separate nelt field.
6268 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
6269 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
6270 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
6271 (aarch64_expand_vec_perm_const): Update accordingly.
6272 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
6273 to vec_perm_indices.
6274 * config/arm/arm.c (expand_vec_perm_d): Change perm to
6275 auto_vec_perm_indices and remove separate nelt field.
6276 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
6277 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
6278 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
6280 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
6281 to vec_perm_indices.
6282 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
6283 sel to vec_perm_indices.
6284 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
6285 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
6286 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
6288 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
6291 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
6294 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
6295 on the FUNCTION_DECL function context if it has a DIE that is a
6298 2017-09-25 Richard Biener <rguenther@suse.de>
6300 PR tree-optimization/82285
6301 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
6304 2017-09-25 Tom de Vries <tom@codesourcery.com>
6308 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
6311 2017-09-25 Richard Biener <rguenther@suse.de>
6313 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
6314 ISL errors other than isl_error_quota happen. Dump if the
6315 schedule is the same.
6316 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
6317 errors instead of aborting inside ISL.
6319 2017-09-25 Iain Sandoe <iain@codesourcery.com>
6322 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
6323 of libgcc_eh for m64.
6324 * config/i386/darwin64.h: Likewise.
6326 2017-09-25 Richard Biener <rguenther@suse.de>
6329 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
6330 attribute for incomplete types nor twice for complete ones.
6332 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
6335 * config/i386/i386.c (ix86_print_operand_address_as): Encode
6336 %esp as %rsp to avoid 0x67 prefix if there is no index or base
6339 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
6342 * config/i386/i386.opt (mprefer-avx256): Use
6343 ix86_target_flags variable.
6344 * config/i386/i386.c (ix86_target_string): Move
6345 -mprefer-avx256 to flag2_opts.
6347 2017-09-22 Jakub Jelinek <jakub@redhat.com>
6350 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
6351 and x != -1 | y != -1 into (x & y) != -1.
6353 2017-09-22 Steve Ellcey <sellcey@cavium.com>
6355 * config.gcc: Add new case statement to set
6356 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
6357 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
6358 s390x-*-linux* case statements. Added aarch64 to the list of
6359 supported architectures.
6361 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6363 PR tree-optimization/82289
6364 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
6365 STMT_VINFO_RELEVANT_P.
6367 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6368 Alan Hayward <alan.hayward@arm.com>
6369 David Sherwood <david.sherwood@arm.com>
6371 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
6372 for VR_RANGE only; don't allow VR_ANTI_RANGE.
6373 (extract_range_from_binary_expr_1): Don't call
6374 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
6376 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6377 Alan Hayward <alan.hayward@arm.com>
6378 David Sherwood <david.sherwood@arm.com>
6380 * target.def (preferred_vector_alignment): New hook.
6381 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
6383 * doc/tm.texi: Regenerate.
6384 * targhooks.h (default_preferred_vector_alignment): Declare.
6385 * targhooks.c (default_preferred_vector_alignment): New function.
6386 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
6388 (DR_TARGET_ALIGNMENT): New macro.
6389 (aligned_access_p): Update commentary.
6390 (vect_known_alignment_in_bytes): New function.
6391 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
6393 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
6394 Calculate the misalignment based on the target alignment rather than
6396 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
6397 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
6398 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
6399 the target alignment, rather than masking the element misalignment
6400 with the number of elements in a vector. Also use the target
6401 alignment when calculating the maximum number of peels.
6402 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
6403 instead of TYPE_ALIGN_UNIT.
6404 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
6405 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
6406 (vect_create_addr_base_for_vector_ref): Update call accordingly.
6407 (vect_create_data_ref_ptr): Likewise.
6408 (vect_setup_realignment): Realign by ANDing with
6409 -DR_TARGET_MISALIGNMENT.
6410 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
6411 the number of peels based on DR_TARGET_ALIGNMENT.
6412 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
6413 with the guaranteed alignment boundary when deciding whether
6415 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
6416 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
6417 (ensure_base_align): Remove stmt_info parameter. Get the
6418 target base alignment from DR_TARGET_ALIGNMENT.
6419 (vectorizable_store): Update call accordingly. Interpret
6420 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
6422 (vectorizable_load): Likewise.
6424 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
6425 Alan Hayward <alan.hayward@arm.com>
6426 David Sherwood <david.sherwood@arm.com>
6428 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
6429 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
6430 (vect_enhance_data_refs_alignment): Likewise.
6432 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6434 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
6435 error. Only quit immediately if parsing is complete.
6436 (BEGIN): Initialize fatal_err and parse_done.
6437 (begin fpu, end fpu): Check number of arguments.
6438 (begin arch, end arch): Likewise.
6439 (begin cpu, end cpu): Likewise.
6440 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
6441 (optalias): Likewise.
6443 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
6445 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
6446 * config/arm/arm-isa.h: Delete. Move definitions to ...
6447 * arm-cpus.in: ... here. Use new feature and fgroup values.
6448 * config/arm/arm.c (arm_option_override): Use lower case for feature
6450 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
6451 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
6452 * config/arm/parsecpu.awk (END): Add new command 'isa'.
6454 (print_isa_bits_for): New function.
6455 (gen_isa): New function.
6456 (gen_comm_data): Use print_isa_bits_for.
6457 (define feature): New keyword.
6458 (define fgroup): New keyword.
6459 * config/arm/t-arm (TM_H): Remove.
6460 (GTM_H): Add arm-isa.h.
6461 (arm-isa.h): Add rule to generate file.
6462 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
6463 case for feature bit names.
6465 2017-09-22 Richard Biener <rguenther@suse.de>
6467 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
6469 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
6470 print of no dependency loops ...
6471 * graphite.c (graphite_transform_loops): ... here.
6472 (canonicalize_loop_closed_ssa_form): Work from inner to outer
6474 (same_close_phi_node, remove_duplicate_close_phi,
6475 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
6476 (canonicalize_loop_closed_ssa): ... here and simplify.
6477 * graphite-optimize-isl.c: Include tree-vectorizer.h.
6478 (optimize_isl): Use dump_printf_loc to tell when we stopped
6479 optimizing because of an ISL timeout.
6481 2017-09-22 Richard Biener <rguenther@suse.de>
6483 PR tree-optimization/82291
6484 * tree-if-conv.c (predicate_mem_writes): Make sure to
6485 remove writes in blocks predicated with false.
6487 2017-09-22 Richard Biener <rguenther@suse.de>
6489 * sese.c: Include cfganal.h.
6490 (if_region_set_false_region): Remove.
6491 (create_if_region_on_edge): Likewise.
6492 (move_sese_in_condition): Re-implement without destroying
6495 2017-09-22 Richard Biener <rguenther@suse.de>
6497 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
6498 Verify both BBs contain loop PHI nodes before dispatching to
6500 (graphite_regenerate_ast_isl): Do not recompute dominators,
6501 do not verify three times. Restructure for clarity.
6502 * graphite-scop-detection.c (same_close_phi_node,
6503 remove_duplicate_close_phi, make_close_phi_nodes_unique,
6504 defined_in_loop_p, canonicalize_loop_closed_ssa,
6505 canonicalize_loop_closed_ssa_form): Simplify, remove excess
6506 checking and SSA rewrite, move to ...
6507 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
6508 (graphite_initialize): Do not pass in ctx, do not reset the
6509 SCEV cache, compute only dominators.
6510 (graphite_transform_loops): Allocate ISL ctx after
6511 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
6512 Maintain post-dominators only around build_scops.
6513 * sese.c (if_region_set_false_region): Make static. Free
6514 and recompute dominators.
6515 (move_sese_in_condition): Assert we don't get called with
6516 post-dominators computed.
6517 * sese.h (if_region_set_false_region): Remove.
6519 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
6521 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
6522 mode attribute for TARGET_AVX512VL.
6524 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
6526 * config/i386/i386.opt (mprefer-avx256): New option.
6527 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
6529 (ix86_preferred_simd_mode): Return 256-bit AVX modes
6530 for TARGET_PREFER_AVX256.
6531 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
6533 2017-09-21 Jeff Law <law@redhat.com>
6535 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
6536 Fix dump output if the only stack space is for pushed registers.
6538 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6540 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
6543 2017-09-21 Martin Sebor <msebor@redhat.com>
6546 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
6547 code (in C++) or code that triggers warnings.
6549 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
6551 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
6553 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
6555 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
6556 * sched-rgn.c: Ditto.
6557 * sel-sched-ir.c: Ditto.
6559 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
6561 * toplev.h (set_random_seed): Adjust return type.
6562 * toplev.c (init_local_tick): Move eager initialization of random_seed
6563 to get_random_seed. Adjust comment.
6564 (init_random_seed): Inline to get_random_seed, delete.
6565 (get_random_seed): Initialize random_seed lazily.
6566 (set_random_seed): Do not return previous value.
6567 (print_switch_value): Do not call get_random_seed.
6569 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
6571 * cgraph.c (delete_function_version): New, broken out from...
6572 (cgraph_node::delete_function_version): ...here. Rename to
6573 cgraph_node::delete_function_version_by_decl. Update all uses.
6574 (cgraph_node::remove): Call delete_function_version.
6576 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6579 * tree-inline.c (expand_call_inline): Emit clobber stmts for
6580 VAR_DECLs to which addressable non-volatile parameters are mapped
6581 and for id->retvar after the return value assignment. Clear
6582 id->retval and id->retbnd after inlining.
6584 2017-09-21 Richard Biener <rguenther@suse.de>
6586 PR tree-optimization/82276
6587 PR tree-optimization/82244
6588 * tree-vrp.c (build_assert_expr_for): Set
6589 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
6591 (remove_range_assertions): Revert earlier change.
6593 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
6596 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
6598 2017-09-21 Richard Biener <rguenther@suse.de>
6600 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
6601 Restore valid IL after code generation errors.
6602 * graphite.c (graphite_transform_loops): Diagnose code
6603 generation issues as MSG_MISSED_OPTIMIZATION and continue
6604 with processing SCOPs.
6606 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6607 Alan Hayward <alan.hayward@arm.com>
6608 David Sherwood <david.sherwood@arm.com>
6610 * calls.c (compute_argument_addresses): Use simplify_gen_binary
6611 rather than choosing between plus_constant and gen_rtx_<CODE>.
6612 * expr.c (emit_push_insn): Likewise.
6613 (expand_expr_real_2): Likewise.
6615 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6616 Alan Hayward <alan.hayward@arm.com>
6617 David Sherwood <david.sherwood@arm.com>
6619 * loop-unroll.c (split_iv): Call copy_rtx on the step.
6621 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6622 Alan Hayward <alan.hayward@arm.com>
6623 David Sherwood <david.sherwood@arm.com>
6625 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
6626 calling tree_to_uhwi.
6628 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6629 Alan Hayward <alan.hayward@arm.com>
6630 David Sherwood <david.sherwood@arm.com>
6632 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
6633 INTEGER_CST rather than a negative test for ADDR_EXPR.
6635 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
6636 Alan Hayward <alan.hayward@arm.com>
6637 David Sherwood <david.sherwood@arm.com>
6639 * tree-vrp.c (extract_range_from_binary_expr_1): Check
6640 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
6642 2017-09-21 Richard Biener <rguenther@suse.de>
6644 PR tree-optimization/71351
6645 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
6646 graphite_create_new_loop_guard): Remove, fold remaining parts
6648 (translate_isl_ast_node_for): ... here and simplify.
6650 2017-09-21 Jakub Jelinek <jakub@redhat.com>
6653 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
6654 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
6655 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
6656 alternative always use QI mode, for -Os imov (=R,R) alternative
6657 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
6660 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6661 Jeff Law <law@redhat.com>
6663 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
6664 (allocate_stack_space): New function, partially extracted from
6666 (s390_emit_prologue): Track offset to most recent stack probe.
6667 Code to allocate space moved into allocate_stack_space.
6668 Dump actions when no stack is allocated.
6669 (s390_prologue_plus_offset): New function.
6670 (s390_emit_stack_probe): Likewise.
6672 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
6674 * common.opt (Wa, Wl, Wp, g, gz=): Add
6676 (gno-column-info): Remove.
6677 (gcolumn-info): Drop RejectNegative.
6679 (gno-record-gcc-switches): Remove.
6680 (grecord-gcc-switches): Drop RejectNegative.
6681 (gno-split-dwarf): Remove.
6682 (gsplit-dwarf): Drop RejectNegative.
6683 (gno-strict-dwarf): Remove.
6684 (gstrict-dwarf): Drop RejectNegative.
6685 * config/darwin.opt (gfull, gused): Add RejectNegative.
6686 * dwarf2out.c (gen_producer_string): Drop
6687 gno-record-gcc-switches handler.
6688 * optc-gen.awk: Add g to prefixes with negative forms.
6689 * opts-common.c (remapping_prefix_p): New.
6690 (find_opt): Check it.
6691 (generate_canonical_option): Test g prefix.
6692 (option_map): Add -gno- mapping.
6693 (add_misspelling_candidates): Check remapping_prefix_p.
6695 2017-09-20 Jeff Law <law@redhat.com>
6697 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
6698 thinko in stack clash protection support.
6700 * explow.c (compute_stack_clash_protection_loop_data): Use
6701 CONST_INT_P instead of explicit test. Verify object is a
6702 CONST_INT_P before looking at INTVAL.
6703 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
6704 instead of explicit test.
6706 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
6709 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
6710 address instead of to r1 and r11.
6712 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
6714 * config.gcc: Support "knm".
6715 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
6716 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
6718 * config/i386/i386.c (m_KNM): Define.
6719 (processor_target_table): Add "knm".
6721 (ix86_option_override_internal): Add "knm".
6722 (ix86_issue_rate): Add PROCESSOR_KNM.
6723 (ix86_adjust_cost): Ditto.
6724 (ia32_multipass_dfa_lookahead): Ditto.
6725 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
6726 (fold_builtin_cpu): Add M_INTEL_KNM.
6727 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
6728 (processor_type): Add PROCESSOR_KNM.
6729 * config/i386/x86-tune.def: Add m_KNM.
6730 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
6732 2017-09-20 Richard Biener <rguenther@suse.de>
6734 PR tree-optimization/80213
6735 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
6736 are allowed in empty BBs as well.
6737 (canonicalize_loop_closed_ssa): Also look for other complex
6739 (scop_detection::get_sese): Include the loop-closed PHI block
6741 (scop_detection::merge_sese): Remove code adding extra blocks.
6742 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
6743 (build_scops): Assert the final returned scop is invalid.
6745 2017-09-20 Richard Biener <rguenther@suse.de>
6747 PR tree-optimization/82264
6748 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
6750 (vn_phi_lookup): Likewise.
6751 (vn_phi_insert): Likewise.
6753 2017-09-20 Jakub Jelinek <jakub@redhat.com>
6755 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
6756 that fits into uhwi or shwi, add DW_AT_const_value regardless
6757 of early_dwarf without going through RTL, using add_AT_unsigned
6760 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
6761 (DEBUG_LTO_ABBREV_SECTION): Likewise.
6762 (DEBUG_LTO_MACINFO_SECTION): Likewise.
6763 (DEBUG_MACRO_SECTION): Likewise.
6764 (DEBUG_LTO_MACRO_SECTION): Likewise.
6765 (DEBUG_STR_DWO_SECTION): Likewise.
6766 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
6767 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
6768 (DEBUG_LTO_DWO_LINE_SECTION): Define.
6769 (DEBUG_LTO_LINE_STR_SECTION): Define.
6770 (init_sections_and_labels): Initialize debug_line_str_section
6771 variable. Initialize debug_loc_section for -gdwarf-5 to
6772 DEBUG_LOCLISTS_SECTION. Formatting fixes.
6774 2017-09-20 Richard Biener <rguenther@suse.de>
6776 * graphite-sese-to-poly.c (extract_affine): Properly handle
6777 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
6779 2017-09-20 Richard Biener <rguenther@suse.de>
6781 PR tree-optimization/81373
6782 * graphite-scop-detection.c (build_cross_bb_scalars_def):
6783 Force SESE live-out defs to be handled even if they are
6786 2017-09-19 Jeff Law <law@redhat.com>
6788 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
6789 nothing for stack adjustments with REG_STACK_CHECK.
6790 * sched-deps.c (parse_add_or_inc): Reject insns with
6791 REG_STACK_CHECK from dependency breaking.
6792 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
6793 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
6794 * reg-notes.def (STACK_CHECK): New note.
6796 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
6797 (ix86_expand_prologue): Dump stack clash info as needed.
6798 Call ix86_adjust_stack_and_probe_stack_clash as needed.
6800 * function.c (dump_stack_clash_frame_info): New function.
6801 * function.h (dump_stack_clash_frame_info): Prototype.
6802 (enum stack_clash_probes): New enum.
6804 * config/alpha/alpha.c (alpha_expand_prologue): Also check
6805 flag_stack_clash_protection.
6806 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
6807 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
6808 (arm_frame_pointer_required): Likewise.
6809 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
6810 (ia64_expand_prologue): Likewise.
6811 * config/mips/mips.c (mips_expand_prologue): Likewise.
6812 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
6813 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6814 (sparc_flat_expand_prologue): Likewise.
6815 * config/spu/spu.c (spu_expand_prologue): Likewise.
6817 * explow.c: Include "params.h".
6818 (anti_adjust_stack_and_probe_stack_clash): New function.
6819 (get_stack_check_protect): Likewise.
6820 (compute_stack_clash_protection_loop_data): Likewise.
6821 (emit_stack_clash_protection_loop_start): Likewise.
6822 (emit_stack_clash_protection_loop_end): Likewise.
6823 (allocate_dynamic_stack_space): Use get_stack_check_protect.
6824 Use anti_adjust_stack_and_probe_stack_clash.
6825 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
6826 (emit_stack_clash_protection_loop_start): Likewise.
6827 (emit_stack_clash_protection_loop_end): Likewise.
6828 * rtl.h (get_stack_check_protect): Prototype.
6829 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
6830 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
6831 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
6833 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
6835 * doc/tm.texi: Rebuilt.
6836 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
6837 get_stack_check_protect.
6838 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
6839 * config/arm/arm.c (arm_expand_prologue): Likewise.
6840 (arm_frame_pointer_required): Likewise.
6841 * config/i386/i386.c (ix86_expand_prologue): Likewise.
6842 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
6843 * config/mips/mips.c (mips_expand_prologue): Likewise.
6844 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
6845 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
6846 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
6847 (sparc_flat_expand_prologue): Likewise.
6849 * common.opt (-fstack-clash-protection): New option.
6850 * flag-types.h (enum stack_check_type): Note difference between
6851 -fstack-check= and -fstack-clash-protection.
6852 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
6853 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
6854 * toplev.c (process_options): Issue warnings/errors for cases
6855 not handled with -fstack-clash-protection.
6856 * doc/invoke.texi (-fstack-clash-protection): Document new option.
6857 (-fstack-check): Note additional problem with -fstack-check=generic.
6858 Note that -fstack-check is primarily for Ada and refer users
6859 to -fstack-clash-protection for stack-clash-protection.
6860 Document new params for stack clash protection.
6862 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
6864 * config/i386/i386.c (ix86_split_long_move): Do not handle
6865 address used for LEA in a special way.
6867 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
6869 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
6871 2017-09-19 Martin Sebor <msebor@redhat.com>
6874 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
6875 of incompatible types.
6877 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
6879 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
6880 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
6881 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
6882 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
6884 2017-09-19 Richard Biener <rguenther@suse.de>
6886 PR tree-optimization/82244
6887 * tree-vrp.c (remove_range_assertions): Do not propagate
6888 a constant to abnormals but replace the assert with a copy.
6890 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
6892 PR rtl-optimization/57878
6893 PR rtl-optimization/68988
6894 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
6895 avoidance test involving non_reload_pseudos. Move frequency test
6896 below the general fragmentation avoidance test.
6898 2017-09-19 Richard Biener <rguenther@suse.de>
6900 PR tree-optimization/69728
6901 * graphite-sese-to-poly.c (schedule_error): New global.
6902 (add_loop_schedule): Handle empty domain by failing the
6904 (build_original_schedule): Handle schedule_error.
6906 2017-09-19 Richard Biener <rguenther@suse.de>
6908 * graphite-scop-detection.c (scop_detection::can_represent_loop):
6909 Do not iterate to sibling loops but only to siblings of inner
6912 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
6915 * config/m68k/m68k.md (moveq feeding equality comparison): Check
6916 that the registers are different.
6918 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
6920 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
6921 to processor_model and "amdfam17h" to arch_names_table.
6922 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
6924 2017-09-18 Jakub Jelinek <jakub@redhat.com>
6927 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
6929 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6930 Alan Hayward <alan.hayward@arm.com>
6931 David Sherwood <david.sherwood@arm.com>
6933 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
6935 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
6937 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
6938 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
6939 vect_schedule_slp_instance.
6940 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
6941 Update call to vect_slp_analyze_node_operations. Simplify return
6943 (vect_slp_analyze_bb_1): Update call accordingly.
6944 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
6945 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
6946 (vect_schedule_slp): Update call accordingly.
6948 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6949 Alan Hayward <alan.hayward@arm.com>
6950 David Sherwood <david.sherwood@arm.com>
6952 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
6953 with types that aren't in fact scalar.
6955 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6957 * tree-vect-slp.c (vect_record_max_nunits): New function,
6959 (vect_build_slp_tree_1): ...here.
6960 (vect_build_slp_tree_2): Call it for phis too.
6962 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6964 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
6965 to vect_get_vec_def_for_operand when getting the mask operand.
6967 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6968 Alan Hayward <alan.hayward@arm.com>
6969 David Sherwood <david.sherwood@arm.com>
6971 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
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-loop.c (vectorizable_live_operation): Fix element size
6979 calculation for vector booleans.
6981 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
6982 Alan Hayward <alan.hayward@arm.com>
6983 David Sherwood <david.sherwood@arm.com>
6985 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
6987 (vect_transform_stmt): ...here.
6988 (vect_analyze_stmt): Use it instead of calling
6989 vectorizable_live_operation directly.
6991 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
6993 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
6994 non-SIMT targets in acc vector loops.
6996 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
6998 * configure.ac: Add arc and check if assembler supports gdwarf2.
6999 * configure: Regenerate.
7001 2017-09-18 Richard Biener <rguenther@suse.de>
7003 PR tree-optimization/82220
7004 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
7005 epilogue niters from the min_profitable_iters compute.
7007 2017-09-18 Jakub Jelinek <jakub@redhat.com>
7010 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
7011 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
7012 (ix86_init_pic_reg): Revert 2017-09-01 changes.
7014 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
7017 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
7018 switching to a new text section.
7020 2017-09-18 Richard Biener <rguenther@suse.de>
7022 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
7024 (build_alias_set): Reject aliases with no access function.
7026 2017-09-18 Richard Biener <rguenther@suse.de>
7028 PR tree-optimization/79622
7029 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
7031 (build_cross_bb_scalars_use): Likewise.
7033 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
7035 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
7037 2017-09-18 Alan Modra <amodra@gmail.com>
7040 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
7041 stack_pointer_rtx for count 0. Update comments. Break up
7042 large rtl expression.
7044 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
7046 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
7047 Increase to 20 bytes.
7048 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
7049 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
7050 or avx version of the stub.
7052 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
7055 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
7056 compute the minimum stack alignment. Also update preferred stack
7057 boundary for leaf functions.
7059 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
7061 PR tree-optimization/82228
7062 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
7065 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7067 * common/config/nds32/nds32-common.c
7068 (nds32_option_optimization_table): Refine formatting.
7069 (nds32_option_optimization_table): Use -fsched-pressure and
7070 -fomit-frame-pointer for specific optimization level.
7072 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
7074 * config/nds32/nds32.c: Refine formatting and comments.
7075 * config/nds32/nds32.h: Likewise.
7076 * config/nds32/nds32.md: Likewise.
7077 * config/nds32/nds32-cost.c: Likewise.
7078 * config/nds32/nds32-isr.c: Likewise.
7079 * config/nds32/nds32-md-auxiliary.c: Likewise.
7080 * config/nds32/nds32-multiple.md: Likewise.
7081 * config/nds32/nds32-predicates.c: Likewise.
7083 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
7084 Jakub Jelinek <jakub@redhat.com>
7086 Add support for -std=c++2a.
7087 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
7089 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
7091 2017-09-15 Steve Ellcey <sellcey@cavium.com>
7094 * doc/extend.texi (Common Function Attributes): Add
7095 references to ARM, AArch64, and S/390 specific attributes.
7096 (Function Specific Option Pragmas): Add AArch64 and S/390
7097 to list of back ends that support the target pragma.
7099 2017-09-15 Nathan Sidwell <nathan@acm.org>
7101 * doc/standards.texi: Fix C++17 description. Update URLs for
7104 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
7106 * common.opt (Wcast-align=strict): New warning option.
7107 * doc/invoke.texi: Document -Wcast-align=strict.
7109 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
7111 * cgraph.h (cgraph_thunk_info): Add comments.
7112 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
7113 assert for VIRTUAL_* arguments stricter.
7115 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
7117 PR tree-optimization/71026
7118 * match.pd: Move RDIV patterns from fold-const.c
7119 * fold-const.c (distribute_real_division): Removed.
7120 (fold_binary_loc): Remove calls to distribute_real_divison.
7122 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7124 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
7125 c++1z and gnu++1z as deprecated. Change other references to
7126 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
7127 Change -Wc++1z-compat to -Wc++17-compat.
7128 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
7129 * dwarf2out.c (highest_c_language): Handle C++17.
7130 (gen_compile_unit_die): Likewise.
7132 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7134 PR rtl-optimization/82192
7135 * combine.c (make_extraction): Don't look through non-paradoxical
7136 SUBREGs or TRUNCATE if pos + len is or might be bigger than
7139 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7140 Alan Hayward <alan.hayward@arm.com>
7141 David Sherwood <david.sherwood@arm.com>
7143 * target.def (function_arg_offset): New hook.
7144 * targhooks.h (default_function_arg_offset): Declare.
7145 * targhooks.c (default_function_arg_offset): New function.
7146 * function.c (locate_and_pad_parm): Use
7147 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
7148 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
7149 (TARGET_FUNCTION_ARG_OFFSET): ...this.
7150 * doc/tm.texi: Regenerate.
7151 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
7152 * config/spu/spu.c (spu_function_arg_offset): New function.
7153 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
7154 * system.h (FUNCTION_ARG_OFFSET): Poison.
7156 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7157 Alan Hayard <alan.hayward@arm.com>
7158 David Sherwood <david.sherwood@arm.com>
7160 * target.def (truly_noop_truncation): New hook.
7161 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
7162 than TRULY_NOOP_TRUNCATION.
7163 * hooks.h (hook_bool_uint_uint_true): Declare.
7164 * hooks.c (hook_bool_uint_uint_true): New function.
7165 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
7166 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
7167 * doc/tm.texi: Regenerate.
7168 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
7169 rather than TRULY_NOOP_TRUNCATION in comments.
7170 (simplify_comparison): Likewise.
7171 (record_truncated_value): Likewise.
7172 * expmed.c (extract_bit_field_1): Likewise.
7173 (extract_split_bit_field): Likewise.
7174 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
7175 instead of TRULY_NOOP_TRUNCATION.
7176 * function.c (assign_parm_setup_block): Likewise.
7177 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
7178 * rtlhooks.c: Include target.h.
7179 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
7180 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
7181 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
7182 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
7183 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
7184 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
7185 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
7186 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
7187 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
7188 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
7189 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
7190 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
7191 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
7192 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
7193 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
7194 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
7195 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
7196 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
7197 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
7198 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
7199 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
7200 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
7201 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
7202 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
7203 * config/mips/mips.c (mips_truly_noop_truncation): New function.
7204 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7205 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7206 rather than TRULY_NOOP_TRUNCATION in comments.
7207 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
7208 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
7209 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
7210 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
7211 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
7212 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
7213 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
7214 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
7215 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
7216 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
7217 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
7218 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7219 rather than TRULY_NOOP_TRUNCATION in comments.
7220 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
7221 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
7222 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
7223 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
7224 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
7225 TRULY_NOOP_TRUNCATION condition.
7226 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
7227 (TRULY_NOOP_TRUNCATION): Delete.
7228 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
7229 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
7230 * config/spu/spu.c (spu_truly_noop_truncation): New function.
7231 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7232 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
7233 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
7234 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
7235 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
7236 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
7237 rather than TRULY_NOOP_TRUNCATION in comments.
7238 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
7239 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
7240 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
7241 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
7242 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
7243 * system.h (TRULY_NOOP_TRUNCATION): Poison.
7245 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
7248 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
7249 (*cmp_ior): Likewise.
7250 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
7251 (*ior_scc_scc_cmp): Likewise.
7252 (*and_scc_scc): Likewise.
7253 (*and_scc_scc_cmp): Likewise.
7255 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
7256 Alan Hayard <alan.hayward@arm.com>
7257 David Sherwood <david.sherwood@arm.com>
7259 * target.def (can_change_mode_class): New hook.
7260 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
7261 (hard_regno_nregs): Likewise.
7262 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
7263 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
7264 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
7265 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
7266 (LOAD_EXTEND_OP): Update accordingly.
7267 * doc/tm.texi: Regenerate.
7268 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
7269 CANNOT_CHANGE_MODE_CLASS.
7270 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
7271 (REG_CAN_CHANGE_MODE_P): ...this new macro.
7272 * combine.c (simplify_set): Update accordingly.
7273 * emit-rtl.c (validate_subreg): Likewise.
7274 * recog.c (general_operand): Likewise.
7275 * regcprop.c (mode_change_ok): Likewise.
7276 * reload1.c (choose_reload_regs): Likewise.
7277 (inherit_piecemeal_p): Likewise.
7278 * rtlanal.c (simplify_subreg_regno): Likewise.
7279 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
7280 instead of CANNOT_CHANGE_MODE_CLASS.
7281 (reload_cse_simplify_operands): Likewise.
7282 * reload.c (push_reload): Use targetm.can_change_mode_class
7283 instead of CANNOT_CHANGE_MODE_CLASS.
7284 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
7285 REG_CANNOT_CHANGE_MODE_P.
7286 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7287 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
7288 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7289 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7290 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7291 (arm_can_change_mode_class): New function.
7292 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7293 than CANNOT_CHANGE_MODE_CLASS in comments.
7294 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7295 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
7296 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
7297 (ix86_can_change_mode_class): ...this new function, inverting the
7298 sense of the return value.
7299 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7300 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7301 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7302 (ia64_can_change_mode_class): New function.
7303 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7304 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
7305 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
7306 (m32c_can_change_mode_class): ...this new function, inverting the
7307 sense of the return value.
7308 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7309 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7310 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
7311 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
7312 (mips_can_change_mode_class): ...this new function, inverting the
7313 sense of the return value.
7314 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7315 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7316 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7317 (msp430_can_change_mode_class): New function.
7318 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7319 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
7320 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7321 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7322 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7323 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
7324 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7325 (pa_cannot_change_mode_class): Replace with...
7326 (pa_can_change_mode_class): ...this new function, inverting the
7327 sense of the return value.
7328 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
7329 than CANNOT_CHANGE_MODE_CLASS in comments.
7330 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7331 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
7332 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7333 (pdp11_cannot_change_mode_class): Replace with...
7334 (pdp11_can_change_mode_class): ...this new function, inverting the
7335 sense of the return value.
7336 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7337 * config/powerpcspe/powerpcspe-protos.h
7338 (rs6000_cannot_change_mode_class_ptr): Delete.
7339 * config/powerpcspe/powerpcspe.c
7340 (rs6000_cannot_change_mode_class_ptr): Delete.
7341 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7342 (rs6000_option_override_internal): Assign to
7343 targetm.can_change_mode_class instead of
7344 rs6000_cannot_change_mode_class_ptr.
7345 (rs6000_cannot_change_mode_class): Replace with...
7346 (rs6000_can_change_mode_class): ...this new function, inverting the
7347 sense of the return value.
7348 (rs6000_debug_cannot_change_mode_class): Replace with...
7349 (rs6000_debug_can_change_mode_class): ...this new function.
7350 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7351 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
7352 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7353 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7354 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
7356 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
7357 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7358 (rs6000_option_override_internal): Assign to
7359 targetm.can_change_mode_class instead of
7360 rs6000_cannot_change_mode_class_ptr.
7361 (rs6000_cannot_change_mode_class): Replace with...
7362 (rs6000_can_change_mode_class): ...this new function, inverting the
7363 sense of the return value.
7364 (rs6000_debug_cannot_change_mode_class): Replace with...
7365 (rs6000_debug_can_change_mode_class): ...this new function.
7366 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7367 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
7368 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
7369 (s390_can_change_mode_class): ...this new function, inverting the
7370 sense of the return value.
7371 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7372 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7373 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
7374 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7375 (sh_cannot_change_mode_class): Replace with...
7376 (sh_can_change_mode_class): ...this new function, inverting the
7377 sense of the return value.
7378 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7379 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7380 (sparc_can_change_mode_class): New function.
7381 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7382 * config/spu/spu.c (spu_can_change_mode_class): New function.
7383 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7384 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
7385 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
7386 (visium_can_change_mode_class): New function.
7387 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
7389 2017-09-15 Richard Biener <rguenther@suse.de>
7391 PR tree-optimization/82217
7392 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
7393 but not undefined case.
7395 2017-09-15 Jakub Jelinek <jakub@redhat.com>
7398 * postreload.c (reload_cse_simplify_operands): Skip
7399 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
7401 2017-09-15 Richard Biener <rguenther@suse.de>
7403 PR tree-optimization/68823
7404 * graphite-scop-detection.c (build_alias_set): If we have a
7405 possible dependence check whether we can handle them by just
7406 looking at the DRs DR_ACCESS_FNs.
7407 (build_scops): If build_alias_set fails, fail the SCOP.
7409 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
7411 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
7412 to support float128 built-in functions that require the ISA 3.0
7414 (BU_FLOAT128_3_HW): Likewise.
7415 (SQRTF128): Add support for the IEEE 128-bit square root and fma
7417 (FMAF128): Likewise.
7419 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7420 support for built-in functions that need the ISA 3.0 IEEE 128-bit
7421 floating point instructions.
7422 (rs6000_invalid_builtin): Likewise.
7423 (rs6000_builtin_mask_names): Likewise.
7424 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
7425 (RS6000_BTM_FLOAT128_HW): Likewise.
7426 (RS6000_BTM_COMMON): Likewise.
7427 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
7429 * doc/extend.texi (RS/6000 built-in functions): Document the
7430 IEEE 128-bit floating point square root and fused multiply-add
7433 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
7435 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
7436 reg (r2) isn't in the set of registers defined in the prologue.
7438 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7439 Alan Hayward <alan.hayward@arm.com>
7440 David Sherwood <david.sherwood@arm.com>
7442 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
7443 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
7444 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
7445 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
7446 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
7448 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
7449 max_vectorization_factor.
7450 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
7452 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7453 Alan Hayward <alan.hayward@arm.com>
7454 David Sherwood <david.sherwood@arm.com>
7456 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
7457 (vect_worthwhile_without_simd_p): Declare.
7458 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
7459 (vectorizable_reduction): Use it.
7460 * tree-vect-stmts.c (vectorizable_shift): Likewise.
7461 (vectorizable_operation): Likewise.
7463 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7464 Alan Hayward <alan.hayward@arm.com>
7465 David Sherwood <david.sherwood@arm.com>
7467 * tree-vectorizer.h (vect_get_num_copies): New function.
7468 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
7469 * tree-vect-loop.c (vectorizable_reduction): Likewise.
7470 (vectorizable_induction): Likewise.
7471 (vectorizable_live_operation): Likewise.
7472 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
7473 (vectorizable_bswap): Likewise.
7474 (vectorizable_call): Likewise.
7475 (vectorizable_conversion): Likewise.
7476 (vectorizable_assignment): Likewise.
7477 (vectorizable_shift): Likewise.
7478 (vectorizable_operation): Likewise.
7479 (vectorizable_store): Likewise.
7480 (vectorizable_load): Likewise.
7481 (vectorizable_condition): Likewise.
7482 (vectorizable_comparison): Likewise.
7483 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
7485 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7486 Alan Hayward <alan.hayward@arm.com>
7487 David Sherwood <david.sherwood@arm.com>
7489 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
7490 of vect_init_vector.
7492 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7493 Alan Hayward <alan.hayward@arm.com>
7494 David Sherwood <david.sherwood@arm.com>
7496 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
7497 an inline wrapper that provides a location.
7498 (gimple_build_vector): Likewise.
7499 * gimple-fold.c (gimple_build_vector_from_val): New function.
7500 (gimple_build_vector): Likewise.
7501 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
7502 functions to build the initial value. Always return a gimple value.
7503 (get_initial_defs_for_reduction): Likewise. Only compute
7505 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
7506 vect_init_vector on the results from get_initial_def(s)_for_reduction.
7507 (vectorizable_induction): Use gimple_build_vector rather than
7510 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7511 Alan Hayward <alan.hayward@arm.com>
7512 David Sherwood <david.sherwood@arm.com>
7514 * target.h (vec_perm_indices): New typedef.
7515 (auto_vec_perm_indices): Likewise.
7516 * optabs-query.h: Include target.h
7517 (can_vec_perm_p): Take a vec_perm_indices *.
7518 * optabs-query.c (can_vec_perm_p): Likewise.
7519 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
7520 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7521 * tree-vect-generic.c (lower_vec_perm): Likewise.
7522 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
7523 (vect_grouped_load_supported): Likewise.
7524 (vect_shift_permute_load_chain): Likewise.
7525 (vect_permute_store_chain): Use auto_vec_perm_indices.
7526 (vect_permute_load_chain): Likewise.
7527 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
7528 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
7529 Update uses of can_vec_perm_p.
7530 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
7531 mode with a number of elements. Take a vec_perm_indices *.
7532 (vect_create_epilog_for_reduction): Update accordingly.
7533 Use auto_vec_perm_indices.
7534 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
7535 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
7536 (vect_transform_slp_perm_load): Likewise.
7537 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
7538 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
7539 (vect_gen_perm_mask_checked): Likewise.
7540 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
7541 (vect_gen_perm_mask_checked): Likewise.
7542 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
7543 (vectorizable_store): Likewise.
7544 (vectorizable_load): Likewise.
7545 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
7546 (vectorizable_bswap): Likewise.
7548 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7549 Alan Hayward <alan.hayward@arm.com>
7550 David Sherwood <david.sherwood@arm.com>
7552 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
7553 * tree.c (build_vector): Likewise.
7554 (build_vector_from_ctor): Update accordingly.
7555 (build_vector_from_val): Likewise.
7556 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
7557 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
7558 * tree-vect-generic.c (add_rshift): Likewise.
7559 (expand_vector_divmod): Likewise.
7560 (optimize_vector_constructor): Likewise.
7561 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
7562 (vect_transform_slp_perm_load): Likewise.
7563 (vect_schedule_slp_instance): Likewise.
7564 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
7565 (vectorizable_call): Likewise.
7566 (vect_gen_perm_mask_any): Likewise. Add elements in order.
7567 * expmed.c (make_tree): Likewise.
7568 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
7569 a vector passed to build_vector.
7570 (fold_convert_const): Likewise.
7571 (exact_inverse): Likewise.
7572 (fold_ternary_loc): Likewise.
7573 (fold_relational_const): Likewise.
7574 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
7575 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
7576 (const_unop): Likewise. Store the reduction accumulator in a
7577 variable rather than an array.
7578 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
7579 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
7580 the new vector, rather than constructing it after the input arrays.
7581 (native_interpret_vector): Use auto_vec<tree> when building
7582 a vector passed to build_vector. Add elements in order.
7583 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
7584 auto_vec<tree> when building a vector passed to build_vector.
7585 (vect_create_epilog_for_reduction): Likewise.
7586 (vectorizable_induction): Likewise.
7587 (get_initial_def_for_reduction): Likewise. Fix indentation of
7589 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
7591 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
7592 passed to build_vector.
7594 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
7595 Alan Hayward <alan.hayward@arm.com>
7596 David Sherwood <david.sherwood@arm.com>
7598 * tree-core.h (tree_base::u): Add an "nelts" field.
7599 (tree_vector): Use VECTOR_CST_NELTS as the length.
7600 * tree.c (tree_size): Likewise.
7601 (make_vector): Initialize VECTOR_CST_NELTS.
7602 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
7603 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
7604 TYPE_VECTOR_SUBPARTS.
7605 * expr.c (const_vector_mask_from_tree): Consistently use "units"
7606 as the number of units, setting it from VECTOR_CST_NELTS.
7607 (const_vector_from_tree): Likewise.
7608 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
7609 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
7610 (fold_negate_expr_1): Likewise.
7611 (fold_convert_const): Likewise.
7612 (const_binop): Likewise. Differentiate the number of output and
7614 (const_unop): Likewise.
7615 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
7616 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
7617 in cases that did the opposite.
7619 2017-09-14 Richard Biener <rguenther@suse.de>
7621 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
7624 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
7626 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
7628 2017-09-14 Jakub Jelinek <jakub@redhat.com>
7631 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
7632 if and where to split a bb, except for splitting before debug insn
7633 sequences followed by non-label real insn. Delete debug insns
7634 in between basic blocks.
7636 * combine.c (make_compound_operation_int): Formatting fixes.
7638 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
7639 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7640 * config/netbsd.h (LINK_EH_SPEC): Likewise.
7641 * config/sol2.h (LINK_EH_SPEC): Likewise.
7642 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7643 * config/s390/linux.h (LINK_SPEC): Likewise.
7644 * config/freebsd.h (LINK_EH_SPEC): Likewise.
7645 * config/openbsd.h (LINK_EH_SPEC): Likewise.
7646 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7647 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
7648 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
7649 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7650 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
7651 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
7653 2017-09-13 Jakub Jelinek <jakub@redhat.com>
7655 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
7657 (ENDFILE_LINUX_SPEC): Likewise.
7658 (LINK_EH_SPEC): Likewise.
7659 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
7660 (LINK_OS_LINUX_SPEC32): Likewise.
7661 (LINK_OS_LINUX_SPEC64): Likewise.
7662 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
7663 (LINK_OS_LINUX_SPEC): Likewise.
7665 2017-09-13 Martin Liska <mliska@suse.cz>
7668 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
7669 CASE_HIGH is NULL_TREE.
7671 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7672 Alan Hayward <alan.hayward@arm.com>
7673 David Sherwood <david.sherwood@arm.com>
7675 * target.def (secondary_memory_needed): New hook.
7676 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
7677 instead of SECONDARY_MEMORY_NEEDED.
7678 (secondary_memory_needed_mode): Likewise.
7679 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
7680 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
7681 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
7682 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
7683 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
7684 * doc/tm.texi: Regenerate.
7685 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
7686 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
7687 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7688 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
7689 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
7690 * config/i386/i386.c (inline_secondary_memory_needed): Put the
7691 mode argument first and change the reg_class arguments to reg_class_t.
7692 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
7693 Make static. Update the call to inline_secondary_memory_needed.
7694 (ix86_register_move_cost): Update the call to
7695 inline_secondary_memory_needed.
7696 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7697 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
7699 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
7700 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7702 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
7703 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
7704 * config/mips/mips.c (mips_secondary_memory_needed): Make static
7705 and match hook interface. Add comment from mips.h.
7706 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7707 * config/mmix/mmix.md (truncdfsf2): Refer to
7708 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7710 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
7711 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
7712 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7713 (pa_secondary_memory_needed): New function.
7714 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
7715 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
7716 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7717 (pdp11_secondary_memory_needed): Make static and match hook interface.
7718 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
7719 * config/powerpcspe/powerpcspe-protos.h
7720 (rs6000_secondary_memory_needed_ptr): Delete.
7721 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
7723 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7724 (rs6000_option_override_internal): Assign to
7725 targetm.secondary_memory_needed rather than
7726 rs6000_secondary_memory_needed_ptr.
7727 (rs6000_secondary_memory_needed): Match hook interface.
7728 (rs6000_debug_secondary_memory_needed): Likewise.
7729 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
7730 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
7731 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
7732 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7733 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
7734 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
7736 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
7737 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7738 (rs6000_option_override_internal): Assign to
7739 targetm.secondary_memory_needed rather than
7740 rs6000_secondary_memory_needed_ptr.
7741 (rs6000_secondary_memory_needed): Match hook interface.
7742 (rs6000_debug_secondary_memory_needed): Likewise.
7743 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
7744 * config/s390/s390.c (s390_secondary_memory_needed): New function.
7745 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7746 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
7747 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
7748 (sparc_secondary_memory_needed): New function.
7749 * lra-constraints.c (check_and_process_move): Refer to
7750 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
7752 (curr_insn_transform): Likewise.
7753 (process_alt_operands): Use targetm.secondary_memory_needed
7754 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7755 (check_secondary_memory_needed_p): Likewise.
7756 (choose_split_class): Likewise.
7757 * reload.c: Unconditionally include code that was previously
7758 conditional on SECONDARY_MEMORY_NEEDED.
7759 (push_secondary_reload): Use targetm.secondary_memory_needed
7760 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7761 (push_reload): Likewise.
7762 * reload1.c: Unconditionally include code that was previously
7763 conditional on SECONDARY_MEMORY_NEEDED.
7764 (choose_reload_regs): Use targetm.secondary_memory_needed
7765 instead of TARGET_SECONDARY_MEMORY_NEEDED.
7766 (gen_reload): Likewise.
7767 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
7769 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
7770 Alan Hayward <alan.hayward@arm.com>
7771 David Sherwood <david.sherwood@arm.com>
7773 * target.def (secondary_memory_needed_mode): New hook:
7774 * targhooks.c (default_secondary_memory_needed_mode): Declare.
7775 * targhooks.h (default_secondary_memory_needed_mode): New function.
7776 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
7777 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
7778 * doc/tm.texi: Regenerate.
7779 * lra-constraints.c (check_and_process_move): Use
7780 targetm.secondary_memory_needed_mode instead of
7781 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
7782 (curr_insn_transform): Likewise.
7783 * reload.c (get_secondary_mem): Likewise.
7784 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7785 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
7787 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7788 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7789 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
7790 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7791 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
7793 * config/powerpcspe/powerpcspe-protos.h
7794 (rs6000_secondary_memory_needed_mode): Delete.
7795 * config/powerpcspe/powerpcspe.c
7796 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7797 (rs6000_secondary_memory_needed_mode): Make static.
7798 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7799 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
7801 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7803 (rs6000_secondary_memory_needed_mode): Make static.
7804 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7805 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
7806 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
7807 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
7808 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
7810 (sparc_secondary_memory_needed_mode): New function.
7811 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
7813 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
7815 * config/aarch64/constraints.md (Umq): New constraint.
7816 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
7818 (mov<mode>): Update condition.
7820 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7822 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
7823 when bitposition is the same.
7825 2017-09-13 Richard Biener <rguenther@suse.de>
7827 * dwarf2out.c (output_die_symbol): Remove.
7828 (output_die): Do not output a DIEs symbol.
7830 2017-09-13 Richard Biener <rguenther@suse.de>
7833 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
7834 default-def to avoid breaking iterator update with the weird
7835 interaction with cgraph_update_edges_for_call_stmt_node.
7837 2017-09-13 Richard Biener <rguenther@suse.de>
7839 * tree-cfg.c (verify_gimple_assign_binary): Add verification
7840 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
7841 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
7842 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
7844 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
7846 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
7847 Disable pc relative literal load irrespective of
7848 TARGET_FIX_ERR_A53_84341 for default.
7850 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
7852 * config/sparc/sparc.c (output_return): Output the source location of
7853 the insn in the delay slot, if any.
7854 (output_sibcall): Likewise.
7856 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
7859 * common.opt (-static-pie): New alias.
7860 (shared): Negate static-pie.
7861 (-no-pie): Update help text.
7863 (static-pie): New option.
7864 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
7865 -static-pie support.
7866 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
7867 (LINK_EH_SPEC): Likewise.
7868 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
7869 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7870 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
7871 * gcc.c (LINK_COMMAND_SPEC): Likewise.
7872 (init_gcc_specs): Likewise.
7873 (init_spec): Likewise.
7874 (display_help): Update help message for -pie.
7875 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
7878 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
7880 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
7881 (movdi_aarch64): Likewise.
7882 (movti_aarch64): Likewise.
7884 2017-09-12 Simon Wright <simon@pushface.org>
7887 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
7888 calculation of the minor version, always output as 0.
7890 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7893 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
7894 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
7895 on it early, rather than manual conversion late. For
7896 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
7897 instead of performing manual conversion.
7899 2017-09-12 Carl Love <cel@us.ibm.com>
7901 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
7902 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
7904 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
7905 VMULOSW): Add definitions.
7906 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7907 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
7908 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
7909 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
7910 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
7912 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7914 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
7916 (movti_aarch64): Likewise.
7917 (movdf_aarch64): Likewise.
7918 (movtf_aarch64): Likewise.
7919 (load_pairdi): Likewise.
7920 (store_pairdi): Likewise.
7921 (load_pairdf): Likewise.
7922 (store_pairdf): Likewise.
7923 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
7924 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
7925 (ldr_got_small_<mode>): Likewise.
7926 (ldr_got_small_28k_<mode>): Likewise.
7927 (ldr_got_tiny): Likewise.
7928 * config/aarch64/iterators.md (ldst_sz): New.
7929 (ldpstp_sz): Likewise.
7930 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
7932 (thunderx_load): Split load_8 to load_16.
7933 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
7935 (thunderx2t99_storepair_basic): Split store_8 to store_16.
7936 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
7937 (xgene1_store_pair): Split store_8 to store_16.
7938 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
7939 (falkor_st_0_st_sd): Split store_8 to store_16.
7941 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
7943 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
7944 and store1/2/3/4 to store_4/8/12/16.
7945 * config/aarch64/aarch64.md: Update for rename.
7946 * config/arm/arm.md: Likewise.: Likewise.
7947 * config/arm/arm.c: Likewise.
7948 * config/arm/thumb1.md: Likewise.
7949 * config/arm/thumb2.md: Likewise.
7950 * config/arm/vfp.md: Likewise.
7951 * config/arm/arm-generic.md: Likewise.
7952 * config/arm/arm1020e.md: Likewise.
7953 * config/arm/arm1026ejs.md: Likewise.
7954 * config/arm/arm1136jfs.md: Likewise.
7955 * config/arm/arm926ejs.md: Likewise.
7956 * config/arm/cortex-a15.md: Likewise.
7957 * config/arm/cortex-a17.md: Likewise.
7958 * config/arm/cortex-a5.md: Likewise.
7959 * config/arm/cortex-a53.md: Likewise.
7960 * config/arm/cortex-a57.md: Likewise.
7961 * config/arm/cortex-a7.md: Likewise.
7962 * config/arm/cortex-a8.md: Likewise.
7963 * config/arm/cortex-a9.md: Likewise.
7964 * config/arm/cortex-m4.md: Likewise.
7965 * config/arm/cortex-m7.md: Likewise.
7966 * config/arm/cortex-r4.md: Likewise.
7967 * config/arm/exynos-m1.md: Likewise.
7968 * config/arm/fa526.md: Likewise.
7969 * config/arm/fa606te.md: Likewise.
7970 * config/arm/fa626te.md: Likewise.
7971 * config/arm/fa726te.md: Likewise.
7972 * config/arm/fmp626.md: Likewise.
7973 * config/arm/iwmmxt.md: Likewise.
7974 * config/arm/ldmstm.md: Likewise.
7975 * config/arm/marvell-pj4.md: Likewise.
7976 * config/arm/xgene1.md: Likewise.
7977 * config/aarch64/thunderx.md: Likewise.
7978 * config/aarch64/thunderx2t99.md: Likewise.
7979 * config/aarch64/falkor.md: Likewise.
7981 2017-09-12 Martin Liska <mliska@suse.cz>
7983 * attribs.c (private_lookup_attribute): New function.
7984 * attribs.h (private_lookup_attribute): Declared here.
7985 (lookup_attribute): Called from this place.
7987 2017-09-12 Richard Biener <rguenther@suse.de>
7989 PR tree-optimization/82157
7990 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
7991 stmts with side-effects.
7993 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7994 Alan Hayward <alan.hayward@arm.com>
7995 David Sherwood <david.sherwood@arm.com>
7997 * target.def (hard_regno_nregs): New hook.
7998 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
7999 * targhooks.h (default_hard_regno_nregs): Declare.
8000 * targhooks.c (default_hard_regno_nregs): New function.
8001 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
8002 (TARGET_HARD_REGNO_NREGS): ...this hook.
8003 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
8004 (CLASS_MAX_NREGS): Likewise.
8005 * doc/tm.texi: Regenerate.
8006 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
8007 instead of HARD_REGNO_NREGS.
8008 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
8009 HARD_REGNO_NREGS in the comment.
8010 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
8011 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
8012 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
8013 Return an unsigned int.
8014 (TARGET_HARD_REGNO_NREGS): Redefine.
8015 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
8016 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
8017 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8018 (arc_hard_regno_nregs): New function.
8019 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
8020 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
8021 (arm_hard_regno_nregs): New function.
8022 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
8023 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
8024 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
8025 (TARGET_HARD_REGNO_NREGS): Redefine.
8026 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
8027 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
8028 (HARD_REGNO_NREGS): Delete.
8029 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
8030 (cr16_hard_regno_nregs): New function.
8031 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
8032 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
8033 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
8034 (cris_hard_regno_nregs): New function.
8035 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
8036 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
8037 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8038 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
8039 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
8040 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
8041 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
8042 (frv_hard_regno_nregs): Make static. Take and return an
8044 (frv_class_max_nregs): Remove outdated copy of documentation.
8045 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
8046 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
8047 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
8048 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
8049 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
8050 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
8051 (TARGET_HARD_REGNO_NREGS): Redefine.
8052 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
8053 (CLASS_MAX_NREGS): Update comment.
8054 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
8055 (ia64_hard_regno_nregs): New function.
8056 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
8057 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
8058 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
8059 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
8060 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
8062 (m32c_hard_regno_nregs): Likewise. Make static.
8063 (TARGET_HARD_REGNO_NREGS): Redefine.
8064 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
8065 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
8066 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
8067 (m68k_hard_regno_nregs): New function.
8068 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
8069 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
8070 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
8071 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
8072 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
8073 Take and return an unsigned int.
8074 (TARGET_HARD_REGNO_NREGS): Redefine.
8075 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
8076 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
8077 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
8078 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
8079 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
8080 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
8081 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
8082 (msp430_hard_regno_nregs): Make static. Take and return an
8084 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
8085 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
8086 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
8087 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
8088 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
8089 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
8090 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
8091 (TARGET_HARD_REGNO_NREGS): Redefine.
8092 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
8093 (PA_HARD_REGNO_NREGS): ...this.
8094 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
8095 (PA_HARD_REGNO_NREGS): ...this.
8096 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8097 (pa_hard_regno_nregs): New function.
8098 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
8099 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
8100 (pdp11_hard_regno_nregs): New function.
8101 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
8102 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
8103 (rs6000_hard_regno_nregs_hook): New function.
8104 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
8105 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
8106 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
8107 Take and return an unsigned int. Move earlier in file.
8108 (TARGET_HARD_REGNO_NREGS): Redefine.
8109 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
8110 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
8111 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
8112 (rl78_hard_regno_nregs): Make static. Take and return an
8114 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
8115 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
8116 (rs6000_hard_regno_nregs_hook): New function.
8117 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
8118 * config/rx/rx.c (rx_hard_regno_nregs): New function.
8119 (TARGET_HARD_REGNO_NREGS): Redefine.
8120 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
8121 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
8122 instead of HARD_REGNO_NREGS.
8123 (s390_hard_regno_nregs): New function.
8124 (s390_hard_regno_mode_ok): Add comment from s390.h.
8125 (TARGET_HARD_REGNO_NREGS): Redefine.
8126 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
8127 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
8128 (sh_hard_regno_nregs): New function.
8129 (sh_pass_in_reg_p): Use it.
8130 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
8131 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
8132 (sparc_hard_regno_nregs): New function.
8133 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
8134 * config/spu/spu.c (spu_hard_regno_nregs): New function.
8135 (spu_function_arg_advance): Use it, supplying a valid register number.
8136 (TARGET_HARD_REGNO_NREGS): Redefine.
8137 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
8138 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
8139 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
8140 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
8141 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
8142 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
8143 (CLASS_MAX_NREGS): Remove copy of old documentation.
8144 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
8145 (visium_hard_regno_nregs): New function.
8146 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
8147 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
8148 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
8149 xtensa_hard_regno_nregs): New function.
8150 * system.h (HARD_REGNO_NREGS): Poison.
8152 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8154 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
8155 hard_regno_nregs instead of HARD_REGNO_NREGS.
8156 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
8157 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
8158 (c6x_expand_epilogue): Likewise.
8159 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
8160 (frv_read_iacc_argument): Likewise.
8161 * config/sh/sh.c: Include regs.h.
8162 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8163 (regs_used): Likewise.
8164 (output_stack_adjust): Likewise.
8165 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
8166 * expmed.c: Include regs.h.
8167 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
8168 * ree.c: Include regs.h.
8169 (combine_reaching_defs): Use hard_regno_nregs instead of
8171 (add_removable_extension): Likewise.
8173 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8175 * regs.h (hard_regno_nregs): Turn into a function.
8176 (end_hard_regno): Update accordingly.
8177 * caller-save.c (setup_save_areas): Likewise.
8178 (save_call_clobbered_regs): Likewise.
8179 (replace_reg_with_saved_mem): Likewise.
8180 (insert_restore): Likewise.
8181 (insert_save): Likewise.
8182 * combine.c (can_change_dest_mode): Likewise.
8183 (move_deaths): Likewise.
8184 (distribute_notes): Likewise.
8185 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
8186 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
8187 (rs6000_split_multireg_move): Likewise.
8188 (rs6000_register_move_cost): Likewise.
8189 (rs6000_memory_move_cost): Likewise.
8190 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
8191 (rs6000_split_multireg_move): Likewise.
8192 (rs6000_register_move_cost): Likewise.
8193 (rs6000_memory_move_cost): Likewise.
8194 * cselib.c (cselib_reset_table): Likewise.
8195 (cselib_lookup_1): Likewise.
8196 * emit-rtl.c (set_mode_and_regno): Likewise.
8197 * function.c (aggregate_value_p): Likewise.
8198 * ira-color.c (setup_profitable_hard_regs): Likewise.
8199 (check_hard_reg_p): Likewise.
8200 (calculate_saved_nregs): Likewise.
8201 (assign_hard_reg): Likewise.
8202 (improve_allocation): Likewise.
8203 (calculate_spill_cost): Likewise.
8204 * ira-emit.c (modify_move_list): Likewise.
8205 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
8206 (ira_hard_reg_in_set_p): Likewise.
8207 * ira.c (setup_reg_mode_hard_regset): Likewise.
8208 (clarify_prohibited_class_mode_regs): Likewise.
8209 (check_allocation): Likewise.
8210 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8211 (lra_setup_reg_renumber): Likewise.
8212 (setup_try_hard_regno_pseudos): Likewise.
8213 (spill_for): Likewise.
8214 (assign_hard_regno): Likewise.
8215 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
8216 * lra-constraints.c (in_class_p): Likewise.
8217 (lra_constraint_offset): Likewise.
8218 (simplify_operand_subreg): Likewise.
8219 (lra_constraints): Likewise.
8220 (split_reg): Likewise.
8221 (split_if_necessary): Likewise.
8222 (invariant_p): Likewise.
8223 (inherit_in_ebb): Likewise.
8224 * lra-lives.c (process_bb_lives): Likewise.
8225 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
8226 (get_hard_regs): Likewise.
8227 (do_remat): Likewise.
8228 * lra-spills.c (assign_spill_hard_regs): Likewise.
8229 * mode-switching.c (create_pre_exit): Likewise.
8230 * postreload.c (reload_combine_recognize_pattern): Likewise.
8231 * recog.c (peep2_find_free_register): Likewise.
8232 * regcprop.c (kill_value_regno): Likewise.
8233 (set_value_regno): Likewise.
8234 (copy_value): Likewise.
8235 (maybe_mode_change): Likewise.
8236 (find_oldest_value_reg): Likewise.
8237 (copyprop_hardreg_forward_1): Likewise.
8238 * regrename.c (check_new_reg_p): Likewise.
8239 (regrename_do_replace): Likewise.
8240 * reload.c (push_reload): Likewise.
8241 (combine_reloads): Likewise.
8242 (find_dummy_reload): Likewise.
8243 (operands_match_p): Likewise.
8244 (find_reloads): Likewise.
8245 (find_equiv_reg): Likewise.
8246 (reload_adjust_reg_for_mode): Likewise.
8247 * reload1.c (count_pseudo): Likewise.
8248 (count_spilled_pseudo): Likewise.
8249 (find_reg): Likewise.
8250 (clear_reload_reg_in_use): Likewise.
8251 (free_for_value_p): Likewise.
8252 (allocate_reload_reg): Likewise.
8253 (choose_reload_regs): Likewise.
8254 (reload_adjust_reg_for_temp): Likewise.
8255 (emit_reload_insns): Likewise.
8256 (delete_output_reload): Likewise.
8257 * rtlanal.c (subreg_get_info): Likewise.
8258 * sched-deps.c (sched_analyze_reg): Likewise.
8259 * sel-sched.c (init_regs_for_mode): Likewise.
8260 (mark_unavailable_hard_regs): Likewise.
8261 (choose_best_reg_1): Likewise.
8262 (verify_target_availability): Likewise.
8263 * valtrack.c (dead_debug_insert_temp): Likewise.
8264 * var-tracking.c (track_loc_p): Likewise.
8265 (emit_note_insn_var_location): Likewise.
8266 * varasm.c (make_decl_rtl): Likewise.
8267 * reginfo.c (choose_hard_reg_mode): Likewise.
8268 (init_reg_modes_target): Refer directly to
8269 this_target_regs->x_hard_regno_nregs.
8271 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8273 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
8274 instead of hard_regno_nregs.
8276 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8278 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
8279 end_hard_regno instead of hard_regno_nregs.
8280 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
8281 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
8282 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
8283 * ira-color.c (improve_allocation): Likewise.
8284 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8285 * lra-lives.c (mark_regno_live): Likewise.
8286 (mark_regno_dead): Likewise.
8287 * lra-remat.c (operand_to_remat): Likewise.
8288 * lra.c (collect_non_operand_hard_regs): Likewise.
8289 * postreload.c (reload_combine_note_store): Likewise.
8290 (move2add_valid_value_p): Likewise.
8291 * reload.c (regno_clobbered_p): Likewise.
8293 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8295 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
8297 * config/v850/v850.c (v850_reorg): Likewise.
8298 * reload.c (refers_to_regno_for_reload_p): Likewise.
8299 (find_equiv_reg): Likewise.
8300 * reload1.c (reload_reg_reaches_end_p): Likewise.
8302 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8304 * caller-save.c (add_used_regs): Use REG_NREGS instead of
8306 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
8307 * config/arm/arm.c (output_move_neon): Likewise.
8308 (arm_attr_length_move_neon): Likewise.
8309 (neon_split_vcombine): Likewise.
8310 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
8311 (c6x_mark_reg_written): Likewise.
8312 (c6x_dwarf_register_span): Likewise.
8313 * config/i386/i386.c (ix86_save_reg): Likewise.
8314 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
8315 (rws_access_reg): Likewise.
8316 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
8317 * mode-switching.c (create_pre_exit): Likewise.
8318 * ree.c (combine_reaching_defs): Likewise.
8319 (add_removable_extension): Likewise.
8320 * regcprop.c (find_oldest_value_reg): Likewise.
8321 (copyprop_hardreg_forward_1): Likewise.
8322 * reload.c (reload_inner_reg_of_subreg): Likewise.
8323 (push_reload): Likewise.
8324 (combine_reloads): Likewise.
8325 (find_dummy_reload): Likewise.
8326 (reload_adjust_reg_for_mode): Likewise.
8327 * reload1.c (find_reload_regs): Likewise.
8328 (forget_old_reloads_1): Likewise.
8329 (reload_reg_free_for_value_p): Likewise.
8330 (reload_adjust_reg_for_temp): Likewise.
8331 (emit_reload_insns): Likewise.
8332 (delete_output_reload): Likewise.
8333 * sel-sched.c (choose_best_reg_1): Likewise.
8334 (choose_best_pseudo_reg): Likewise.
8336 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8337 Alan Hayward <alan.hayward@arm.com>
8338 David Sherwood <david.sherwood@arm.com>
8340 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
8341 * target.def (slow_unaligned_access): New hook.
8342 * targhooks.h (default_slow_unaligned_access): Declare.
8343 * targhooks.c (default_slow_unaligned_access): New function.
8344 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
8345 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
8346 * doc/tm.texi: Regenerate.
8347 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
8348 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
8349 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
8351 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
8352 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
8354 (rs6000_slow_unaligned_access): New function.
8355 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8356 (expand_block_compare): Likewise.
8357 (expand_strn_compare): Likewise.
8358 (rs6000_rtx_costs): Likewise.
8359 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
8360 (riscv_slow_unaligned_access): Likewise.
8361 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
8362 (riscv_slow_unaligned_access_p): ...this and make static.
8363 (riscv_option_override): Update accordingly.
8364 (riscv_slow_unaligned_access): New function.
8365 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8366 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
8367 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
8368 (rs6000_slow_unaligned_access): New function.
8369 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
8370 (rs6000_rtx_costs): Likewise.
8371 * config/rs6000/rs6000-string.c (expand_block_compare)
8372 (expand_strn_compare): Use targetm.slow_unaligned_access instead
8373 of SLOW_UNALIGNED_ACCESS.
8374 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
8375 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
8376 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
8377 of SLOW_UNALIGNED_ACCESS.
8378 * expmed.c (simple_mem_bitfield_p): Likewise.
8379 * expr.c (alignment_for_piecewise_move): Likewise.
8380 (emit_group_load_1): Likewise.
8381 (emit_group_store): Likewise.
8382 (copy_blkmode_from_reg): Likewise.
8383 (emit_push_insn): Likewise.
8384 (expand_assignment): Likewise.
8385 (store_field): Likewise.
8386 (expand_expr_real_1): Likewise.
8387 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
8388 * lra-constraints.c (simplify_operand_subreg): Likewise.
8389 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
8390 * gimple-ssa-store-merging.c: Likewise in block comment at start
8392 * tree-ssa-strlen.c: Include target.h.
8393 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
8394 of SLOW_UNALIGNED_ACCESS.
8395 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
8397 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
8399 PR rtl-optimization/82185
8400 * expmed.c (emit_store_flag_int): Only test tem if it has been
8403 2017-09-12 Richard Biener <rguenther@suse.de>
8406 * match.pd ((FTYPE) N CMP CST): Fix typo.
8408 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
8410 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
8412 (mips_near_type_p): Add 'short_call' attribute as a synonym
8414 * doc/extend.texi (short_call): Document new function attribute.
8416 2017-09-12 Jakub Jelinek <jakub@redhat.com>
8419 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
8420 assertion check that in the condition.
8421 (get_atomic_generic_size): Likewise. Before testing if parameter
8422 has pointer type, if it has array type, call for C++
8423 default_conversion to perform array-to-pointer conversion.
8425 2017-09-12 Richard Biener <rguenther@suse.de>
8427 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
8428 for operations we cannot scalarize.
8430 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
8432 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
8433 vectors heap vectors. Clean up comments.
8434 Make visited_bbs a reference.
8435 (profitable_jump_thread_path): Make GC
8436 vectors heap vectors. Clean up comments.
8438 (convert_and_register_jump_thread_path): Make GC vectors heap
8440 (check_subpath_and_update_thread_path): Same. Clean up comments.
8441 Make visited_bbs a reference.
8442 (handle_phi): Abstract common code to to
8443 register_jump_thread_path_if_profitable.
8444 Rename VAR_BB to DEF_BB.
8446 Make GC vectors heap vectors.
8447 Make visited_bbs a reference.
8448 (handle_assignment): Same.
8449 (register_jump_thread_path_if_profitable): New.
8450 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
8452 Make GC vectors heap vectors. Clean up comments.
8453 Make visited_bbs a reference.
8454 (find_jump_threads_backwards): Make visited_bbs live in the stack.
8455 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
8458 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
8461 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
8462 words of E_DImode object are reachable by xtensa_uimm8x4 access.
8464 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
8466 Revert r251800 and r251799.
8468 2017-09-11 Martin Jambor <mjambor@suse.cz>
8471 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
8472 arguments in advance.
8473 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
8474 use it to find predecessor edges.
8475 (naive_outof_ssa): Collect vector of predecessors.
8477 2017-09-08 Jason Merrill <jason@redhat.com>
8479 PR c++/70029 - ICE with ref-qualifier and -flto
8480 * langhooks.h (struct lang_hooks_for_types): Add
8481 copy_lang_qualifiers.
8482 * attribs.c (build_type_attribute_qual_variant): Use it.
8483 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
8485 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
8486 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
8488 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
8491 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
8492 (*mulsi3_sp64): New instruction.
8493 (mulsi3): New expander.
8495 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
8497 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
8499 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8501 * sancov.c: Include memmodel.h.
8503 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
8506 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
8509 2017-09-07 Carl Love <cel@us.ibm.com>
8511 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
8512 the sldi instruction.
8514 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
8516 * sancov.c: Include tm_p.h.
8518 2017-09-07 Jakub Jelinek <jakub@redhat.com>
8521 * output.h (switch_to_other_text_partition): New declaration.
8522 * varasm.c (switch_to_other_text_partition): New function.
8523 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
8524 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
8525 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
8526 to the other text partition before emitting LCL label and switch back
8527 after emitting the word after it.
8529 2017-09-07 Richard Biener <rguenther@suse.de>
8531 * passes.def (pass_split_crit_edges): Remove instance before PRE.
8532 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
8533 critical edges here, after loop init.
8534 (pass_data_pre): Remove PROP_no_crit_edges flags.
8535 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
8536 for valueization of call args to avoid leaking VN_TOP.
8537 (visit_use): Assert we do not visit default defs.
8538 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
8539 Use error_mark_node to more easily detect leaking VN_TOP.
8540 All default-defs are varying, not VN_TOP. Mark them visited.
8541 (run_scc_vn): Make code match comment.
8543 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
8545 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
8546 OPTION_MASK_FLOAT128_KEYWORD.
8547 (POWERPC_MASKS): Likewise.
8548 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
8549 support for the -mfloat128-type option, and make -mfloat128
8550 default on PowerPC Linux systems. Define or undefine
8551 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
8552 Define __float128 to be __ieee128 if IEEE 128-bit support is
8553 enabled, or undefine it.
8554 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
8555 Delete defining __FLOAT128_TYPE__.
8556 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
8557 -mfloat128-type option and make -mfloat128 default on PowerPC
8559 (TARGET_FLOAT128_TYPE): Likewise.
8560 (-mfloat128-type): Likewise.
8561 * config/rs6000/rs6000.c (rs6000_option_override_internal):
8562 Delete the -mfloat128-type option and make -mfloat128 default on
8563 PowerPC Linux systems. Always use __ieee128 to be the keyword for
8564 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
8565 128-bit floating point is enabled. Change tests from using
8566 -mfloat128-type to -mfloat128.
8567 (rs6000_mangle_type): Use the correct mangling for the __float128
8568 type even if normal long double is restricted to 64-bits.
8569 (floatn_mode): Enable the _Float128 type by default on VSX Linux
8571 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
8572 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
8573 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
8575 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
8576 documentation for -mfloat128.
8578 2017-09-06 Olivier Hainque <hainque@adacore.com>
8580 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
8582 2017-09-06 Wish Wu <wishwu007@gmail.com>
8583 Jakub Jelinek <jakub@redhat.com>
8585 * asan.c (initialize_sanitizer_builtins): Add
8586 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
8587 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
8588 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
8589 BT_FN_VOID_UINT64_PTR variables.
8590 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
8591 (BT_FN_VOID_UINT16_UINT16): Likewise.
8592 (BT_FN_VOID_UINT32_UINT32): Likewise.
8593 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
8594 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
8595 (BT_FN_VOID_UINT64_PTR): Likewise.
8596 * common.opt (flag_sanitize_coverage): New variable.
8597 (fsanitize-coverage=trace-pc): Remove.
8598 (fsanitize-coverage=): Add.
8599 * flag-types.h (enum sanitize_coverage_code): New enum.
8600 * fold-const.c (fold_range_test): Disable non-short-circuit
8601 optimization if flag_sanitize_coverage.
8602 (fold_truth_andor): Likewise.
8603 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
8604 * opts.c (COVERAGE_SANITIZER_OPT): Define.
8605 (coverage_sanitizer_opts): New array.
8606 (get_closest_sanitizer_option): Add OPTS argument, handle also
8607 OPT_fsanitize_coverage_.
8608 (parse_sanitizer_options): Adjusted to also handle
8609 OPT_fsanitize_coverage_.
8610 (common_handle_option): Add OPT_fsanitize_coverage_.
8611 * sancov.c (instrument_comparison, instrument_switch): New function.
8612 (sancov_pass): Add trace-cmp support.
8613 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
8614 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
8615 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
8616 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
8617 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
8618 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
8619 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
8620 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
8621 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
8622 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
8624 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8626 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
8627 error. Only quit immediately if parsing is complete.
8628 (BEGIN): Initialize fatal_err and parse_done.
8629 (begin fpu, end fpu): Check number of arguments.
8630 (begin arch, end arch): Likewise.
8631 (begin cpu, end cpu): Likewise.
8632 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
8633 (optalias): Likewise.
8635 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
8637 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
8638 * config/arm/arm-isa.h: Delete. Move definitions to ...
8639 * arm-cpus.in: ... here. Use new feature and fgroup values.
8640 * config/arm/arm.c (arm_option_override): Use lower case for feature
8642 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
8643 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
8644 * config/arm/parsecpu.awk (END): Add new command 'isa'.
8646 (print_isa_bits_for): New function.
8647 (gen_isa): New function.
8648 (gen_comm_data): Use print_isa_bits_for.
8649 (define feature): New keyword.
8650 (define fgroup): New keyword.
8651 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
8652 (arm-isa.h): Add rule to generate file.
8653 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
8654 case for feature bit names.
8656 2017-09-06 Richard Biener <rguenther@suse.de>
8658 * tree-ssa-pre.c (NECESSARY): Remove.
8659 (create_expression_by_pieces): Do not touch pass-local flags.
8660 (insert_into_preds_of_block): Likewise.
8661 (do_pre_regular_insertion): Likewise.
8662 (eliminate_insert): Likewise.
8663 (eliminate_dom_walker::before_dom_children): Likewise.
8664 (fini_eliminate): Do not look at inserted_exprs.
8665 (mark_operand_necessary): Remove.
8666 (remove_dead_inserted_code): Replace with simple work-list
8667 algorithm based on inserted_exprs and SSA uses.
8668 (pass_pre::execute): Re-order fini_eliminate and
8669 remove_dead_inserted_code.
8671 2017-09-06 Olivier Hainque <hainque@adacore.com>
8673 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
8674 for VxWorks 7. Adjust surrounding comments.
8676 2017-09-06 Richard Biener <rguenther@suse.de>
8678 * gimple-ssa-strength-reduction.c
8679 (find_candidates_dom_walker::before_dom_children): Also allow
8682 2017-09-06 Richard Biener <rguenther@suse.de>
8684 PR tree-optimization/82108
8685 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
8686 for gap in the non-permutation SLP case.
8688 2017-09-06 Martin Jambor <mjambor@suse.cz>
8690 PR tree-optimization/82078
8691 * tree-sra.c (sort_and_splice_var_accesses): Move call to
8692 add_access_to_work_queue...
8693 (build_accesses_from_assign): ...here.
8694 (propagate_all_subaccesses): Make sure racc is the group
8695 representative, if there is one.
8697 2017-09-06 Jakub Jelinek <jakub@redhat.com>
8700 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
8703 2017-09-06 Richard Biener <rguenther@suse.de>
8705 * gimple-ssa-strength-reduction.c
8706 (find_candidates_dom_walker::before_doom_children): Use a
8707 type and not a mode check.
8709 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8712 * config/arm/predicates.md (arm_general_adddi_operand): Create new
8714 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
8716 2017-09-05 Jeff Law <law@redhat.com>
8718 PR tree-optimization/64910
8719 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
8720 cases where we have 3 or more operands.
8722 2017-09-05 Jakub Jelinek <jakub@redhat.com>
8725 * omp-low.c (lower_omp_for): Recompute tree invariant if
8726 gimple_omp_for_initial/final is ADDR_EXPR.
8729 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
8730 into gimple val before gimplification fo the COND_EXPR.
8732 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
8734 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
8735 REGION_COPY argument.
8736 (thread_through_all_blocks): Remove unused argument to
8737 duplicate_thread_path.
8739 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8740 Alan Hayward <alan.hayward@arm.com>
8741 David Sherwood <david.sherwood@arm.com>
8743 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
8744 Take a scalar_mode rather than a machine_mode.
8745 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8746 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
8747 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
8748 (aarch64_gen_adjusted_ldpstp): Likewise.
8749 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
8751 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8752 Alan Hayward <alan.hayward@arm.com>
8753 David Sherwood <david.sherwood@arm.com>
8755 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
8756 Take a scalar_int_mode instead of a machine_mode.
8757 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8758 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8759 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8760 (aarch64_simd_attr_length_rglist): Delete.
8761 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
8762 a scalar_int_mode instead of a machine_mode.
8763 (aarch64_add_offset): Likewise.
8764 (aarch64_internal_mov_immediate): Likewise
8765 (aarch64_add_constant_internal): Likewise.
8766 (aarch64_add_constant): Likewise.
8767 (aarch64_movw_imm): Likewise.
8768 (aarch64_rtx_arith_op_extract_p): Likewise.
8769 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
8770 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
8771 Remove assert that the mode isn't a vector.
8772 (aarch64_output_scalar_simd_mov_immediate): Likewise.
8773 (aarch64_expand_mov_immediate): Update calls after above changes.
8774 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
8775 (aarch64_and_bitmask_imm): Check for scalar integer modes.
8776 (aarch64_move_imm): Likewise.
8777 (aarch64_can_const_movi_rtx_p): Likewise.
8778 (aarch64_strip_extend): Likewise.
8779 (aarch64_extr_rtx_p): Likewise.
8780 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
8781 a CONST_INT when the mode parameter is VOIDmode.
8782 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
8784 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8786 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
8787 * stor-layout.c (bitwise_mode_for_mode): Likewise.
8788 (bitwise_type_for_mode): Update accordingly.
8790 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8792 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
8793 * stor-layout.c (mode_for_size_tree): Likewise.
8794 (mode_for_array): Update accordingly.
8795 (layout_decl): Likewise.
8796 (compute_record_mode): Likewise. Only set the mode once.
8798 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8800 * target.def (get_mask_mode): Change return type to opt_mode.
8802 * doc/tm.texi: Regenerate.
8803 * targhooks.h (default_get_mask_mode): Return an opt_mode.
8804 * targhooks.c (default_get_mask_mode): Likewise.
8805 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
8806 * optabs-query.c (can_vec_mask_load_store_p): Update use of
8807 targetm.get_mask_mode.
8808 * tree.c (build_truth_vector_type): Likewise.
8810 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8812 * machmode.h (mode_for_vector): Return an opt_mode.
8813 * stor-layout.c (mode_for_vector): Likewise.
8814 (mode_for_int_vector): Update accordingly.
8815 (layout_type): Likewise.
8816 * config/i386/i386.c (emit_memmov): Likewise.
8817 (ix86_expand_set_or_movmem): Likewise.
8818 (ix86_expand_vector_init): Likewise.
8819 (ix86_get_mask_mode): Likewise.
8820 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
8822 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
8823 * expmed.c (extract_bit_field_1): Likewise.
8824 * expr.c (expand_expr_real_2): Likewise.
8825 * optabs-query.c (can_vec_perm_p): Likewise.
8826 (can_vec_mask_load_store_p): Likewise.
8827 * optabs.c (expand_vec_perm): Likewise.
8828 * targhooks.c (default_get_mask_mode): Likewise.
8829 * tree-vect-stmts.c (vectorizable_store): Likewise.
8830 (vectorizable_load): Likewise.
8831 (get_vectype_for_scalar_type_and_size): Likewise.
8833 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8835 * machmode.h (mode_for_int_vector): New function.
8836 * stor-layout.c (mode_for_int_vector): Likewise.
8837 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
8838 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
8839 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
8840 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
8841 (s390_expand_vcond): Likewise.
8843 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8845 * machmode.h (opt_machine_mode): New type.
8846 (opt_mode<T>): Allow construction from anything that can be
8848 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
8849 (mode_for_size): Return an opt_machine_mode.
8850 * stor-layout.c (mode_for_size): Likewise.
8851 (mode_for_size_tree): Update call accordingly.
8852 (bitwise_mode_for_mode): Likewise.
8853 (make_fract_type): Likewise.
8854 (make_accum_type): Likewise.
8855 * caller-save.c (replace_reg_with_saved_mem): Update call
8857 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8858 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8859 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8860 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
8861 * expmed.c (extract_bit_field_1): Likewise.
8862 * reload.c (get_secondary_mem): Likewise.
8863 * varasm.c (assemble_integer): Likewise.
8864 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
8867 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8869 * machmode.h (decimal_float_mode_for_size): New function.
8870 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
8871 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
8872 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
8873 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
8874 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
8875 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
8877 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8879 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
8880 (get_builtin_sync_mode): Likewise.
8881 (expand_ifn_atomic_compare_exchange): Likewise.
8882 (expand_builtin_atomic_clear): Likewise.
8883 (expand_builtin_atomic_test_and_set): Likewise.
8884 (fold_builtin_atomic_always_lock_free): Likewise.
8885 * calls.c (compute_argument_addresses): Likewise.
8886 (emit_library_call_value_1): Likewise.
8887 (store_one_arg): Likewise.
8888 * combine.c (combine_instructions): Likewise.
8889 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
8890 * config/arm/arm.c (arm_function_value): Likewise.
8891 (aapcs_allocate_return_reg): Likewise.
8892 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
8893 * config/i386/i386.c (construct_container): Likewise.
8894 (ix86_gimplify_va_arg): Likewise.
8895 (ix86_expand_sse_cmp): Likewise.
8896 (emit_memmov): Likewise.
8897 (emit_memset): Likewise.
8898 (expand_small_movmem_or_setmem): Likewise.
8899 (ix86_expand_pextr): Likewise.
8900 (ix86_expand_pinsr): Likewise.
8901 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
8902 * config/microblaze/microblaze.c (microblaze_block_move_straight):
8904 * config/mips/mips.c (mips_function_value_1) Likewise.
8905 (mips_block_move_straight): Likewise.
8906 (mips_expand_ins_as_unaligned_store): Likewise.
8907 * config/powerpcspe/powerpcspe.c
8908 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8909 (rs6000_darwin64_record_arg_flush): Likewise.
8910 * config/rs6000/rs6000.c
8911 (rs6000_darwin64_record_arg_advance_flush): Likewise.
8912 (rs6000_darwin64_record_arg_flush): Likewise.
8913 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
8914 (sparc_function_value_1): Likewise.
8915 * config/spu/spu.c (adjust_operand): Likewise.
8916 (spu_emit_branch_or_set): Likewise.
8917 (arith_immediate_p): Likewise.
8918 * emit-rtl.c (gen_lowpart_common): Likewise.
8919 * expr.c (expand_expr_real_1): Likewise.
8920 * function.c (assign_parm_setup_block): Likewise.
8921 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
8922 * reload1.c (alter_reg): Likewise.
8923 * stor-layout.c (mode_for_vector): Likewise.
8924 (layout_type): Likewise.
8926 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
8928 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
8929 (spu_convert_move): Likewise.
8930 * lower-subreg.c (resolve_simple_move): Likewise.
8932 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8935 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
8936 define_insn to a define_expand.
8937 (altivec_vsum2sws_direct): New define_insn.
8938 (altivec_vsumsws): Convert from a define_insn to a define_expand.
8940 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8942 * config/arm/arm.c (arm_option_params_internal): Improve setting of
8945 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
8950 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
8951 realign stack if stack alignment needed is less than incoming
8954 2017-09-05 Marek Polacek <polacek@redhat.com>
8957 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
8960 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
8962 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
8964 2017-09-05 Richard Biener <rguenther@suse.de>
8966 PR tree-optimization/82084
8967 * fold-const.c (can_native_encode_string_p): Handle wide characters.
8969 2017-09-05 Richard Biener <rguenther@suse.de>
8971 PR tree-optimization/82102
8972 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
8974 2017-09-05 Martin Liska <mliska@suse.cz>
8976 PR tree-optimization/82032
8977 * tree-cfg.c (generate_range_test): New function.
8978 * tree-cfg.h (generate_range_test): Declared here.
8979 * tree-cfgcleanup.c (convert_single_case_switch): New function.
8980 (cleanup_control_expr_graph): Use it.
8981 * tree-switch-conversion.c (try_switch_expansion): Remove
8983 (emit_case_nodes): Use generate_range_test.
8985 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
8988 * config/i386/i386.md (*<btsc><mode>_mask): Add
8989 TARGET_USE_BT to insn constraint.
8990 (*btr<mode>_mask): Ditto.
8992 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
8994 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
8995 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
8997 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
9000 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
9001 TARGET_NEON and TARGET_IWMMXT.
9002 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
9003 TARGET_NEON and TARGET_IWMMXT.
9004 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
9006 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
9008 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
9009 (ix86_rewrite_tls_address): Ditto.
9010 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
9011 (ix86_rewrite_tls_address_1): Ditto.
9012 (ix86_rewrite_tls_address): Ditto.
9013 * config/i386/predicates.md (tls_address_pattern): New predicate.
9014 * config/i386/i386.md (TLS address splitter): New splitter.
9016 2017-09-04 Richard Biener <rguenther@suse.de>
9018 PR tree-optimization/82084
9019 * fold-const.h (can_native_encode_string_p): Declare.
9020 * fold-const.c (can_native_encode_string_p): Factor out from ...
9021 (native_encode_string): ... here.
9022 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
9023 vectorizing stores from constants we later cannot handle.
9025 2017-09-04 Marek Polacek <polacek@redhat.com>
9028 * doc/invoke.texi: Update -Wtautological-compare documentation.
9030 2017-09-04 Jeff Law <law@redhat.com>
9032 PR tree-optimization/64910
9033 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
9034 swap the first and last operand if the last is a constant.
9036 2017-09-04 Marek Polacek <polacek@redhat.com>
9039 * convert.c (do_narrow): When sanitizing signed integer overflows,
9040 bail out for signed types.
9041 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
9043 2017-09-04 Richard Biener <rguenther@suse.de>
9045 PR tree-optimization/82060
9046 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
9047 Move devirtualization after stmt folding and before EH/AB/noreturn
9048 cleanup to get the stmt refs canonicalized. Use a bool instead
9049 of gimple_modified_p since that doesn't work for NOPs. Schedule
9050 NOPs generated by folding for removal.
9052 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9053 Alan Hayward <alan.hayward@arm.com>
9054 David Sherwood <david.sherwood@arm.com>
9056 * coretypes.h (pad_direction): New enum.
9057 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
9058 (FUNCTION_ARG_PADDING): Likewise.
9059 * target.def (function_arg_padding): New hook.
9060 * targhooks.h (default_function_arg_padding): Declare.
9061 * targhooks.c (default_function_arg_padding): New function.
9062 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
9063 (TARGET_FUNCTION_ARG_PADDING): ...this.
9064 * doc/tm.texi: Regenerate.
9065 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
9066 instead of direction.
9067 (compute_argument_addresses): Likewise.
9068 (load_register_parameters): Likewise.
9069 (emit_library_call_value_1): Likewise.
9070 (store_one_arg): Use targetm.calls.function_arg_padding instead
9071 of FUNCTION_ARG_PADDING.
9072 (must_pass_in_stack_var_size_or_pad): Likewise.
9073 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
9074 (emit_group_store): Likewise.
9075 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
9076 instead of FUNCTION_ARG_PADDING.
9077 (emit_push_insn): Likewise, and propagate enum change throughout
9079 * function.h (direction): Delete.
9080 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
9082 * function.c (assign_parm_find_stack_rtl): Likewise.
9083 (assign_parm_setup_block_p): Likewise.
9084 (assign_parm_setup_block): Likewise.
9085 (gimplify_parameters): Likewise.
9086 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
9087 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
9089 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
9090 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9091 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
9092 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
9093 (aarch64_function_arg_padding): ...this new function.
9094 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
9095 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9096 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
9097 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9098 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
9099 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9100 (arm_pad_arg_upward): Replace with...
9101 (arm_function_arg_padding): ...this new function.
9102 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
9104 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
9105 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
9106 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9107 (ia64_hpux_function_arg_padding): Replace with...
9108 (ia64_function_arg_padding): ...this new function. Use pad_direction
9109 instead of direction. Check for TARGET_HPUX.
9110 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
9111 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9112 (iq2000_function_arg_padding): New function.
9113 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
9114 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
9115 (mips_function_arg_padding): ...this new function.
9116 (mips_pad_reg_upward): Update accordingly.
9117 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9118 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
9119 targetm.calls.function_arg_padding.
9120 (FUNCTION_ARG_PADDING): Delete.
9121 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
9122 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
9123 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9124 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
9125 (nios2_block_reg_padding): Return pad_direction instead of direction.
9126 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
9127 instead of direction.
9128 (nios2_function_arg_padding): Likewise. Make static.
9129 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9130 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
9131 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
9132 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
9133 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9134 (pa_function_arg_padding): Make static. Return pad_direction instead
9136 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
9137 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9138 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
9139 instead of direction. Use targetm.calls.function_arg_padding.
9140 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
9141 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9142 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
9143 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
9144 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
9146 (function_arg_padding): Rename to...
9147 (rs6000_function_arg_padding): ...this. Make static. Return
9148 pad_direction instead of direction.
9149 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9150 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
9151 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
9152 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
9153 instead of direction. Use targetm.calls.function_arg_padding.
9154 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
9155 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
9156 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
9157 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
9158 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9159 (function_arg_padding): Rename to...
9160 (rs6000_function_arg_padding): ...this. Make static. Return
9161 pad_direction instead of direction.
9162 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
9163 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
9164 * config/s390/s390.c (s390_function_arg_padding): New function.
9165 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9166 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
9167 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
9168 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
9169 (function_arg_padding): Rename to...
9170 (sparc_function_arg_padding): ...this. Make static. Return
9171 pad_direction instead of direction.
9172 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
9173 * config/spu/spu.c (spu_function_arg_padding): New function.
9174 (TARGET_FUNCTION_ARG_PADDING): Redefine.
9175 * system.h (FUNCTION_ARG_PADDING): Poison.
9177 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9178 Alan Hayward <alan.hayward@arm.com>
9179 David Sherwood <david.sherwood@arm.com>
9181 * target.def (modes_tieable_p): New hook.
9182 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
9183 (TARGET_MODES_TIEABLE_P): ...this.
9184 * doc/tm.texi.in: Regenerate.
9185 * hooks.h (hook_bool_mode_mode_true): Declare.
9186 * hooks.c (hook_bool_mode_mode_true): New function.
9187 * combine.c (subst): Use targetm.modes_tieable_p instead of
9189 * dse.c (find_shift_sequence): Likewise.
9190 * expmed.c (extract_low_bits): Likewise.
9191 * lower-subreg.c: Include target.h.
9192 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
9194 * rtlanal.c (rtx_cost): Likewise.
9195 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
9196 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
9197 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
9198 (TARGET_MODES_TIEABLE_P): Redefine.
9199 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
9200 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
9201 (TARGET_MODES_TIEABLE_P): Redefine.
9202 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
9203 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
9204 (arc_modes_tieable_p): New function.
9205 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
9206 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
9207 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
9208 (arm_modes_tieable_p): Make static.
9209 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
9210 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
9211 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
9212 (TARGET_MODES_TIEABLE_P): Redefine.
9213 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
9214 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
9215 (TARGET_MODES_TIEABLE_P): Redefine.
9216 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
9217 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
9218 (cr16_modes_tieable_p): New function.
9219 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
9220 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
9221 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
9222 (TRULY_NOOP_TRUNCATION): Update comment.
9223 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
9224 (TRULY_NOOP_TRUNCATION): Update comment.
9225 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
9226 (frv_modes_tieable_p): New function.
9227 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
9228 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
9229 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
9230 (TARGET_MODES_TIEABLE_P): Redefine.
9231 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
9232 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
9233 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
9234 (TARGET_MODES_TIEABLE_P): Redefine.
9235 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
9236 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
9237 (ia64_modes_tieable_p): New function.
9238 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
9239 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
9240 (iq2000_modes_tieable_p): New function.
9241 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
9242 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
9243 (lm32_modes_tieable_p): New function.
9244 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
9245 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
9246 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
9247 (TARGET_MODES_TIEABLE_P): Redefine.
9248 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
9249 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
9250 (m32r_modes_tieable_p): New function.
9251 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
9252 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
9253 (m68k_modes_tieable_p): New function.
9254 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
9255 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
9256 (mcore_modes_tieable_p): New function.
9257 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
9258 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
9260 (TARGET_MODES_TIEABLE_P): Redefine.
9261 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
9262 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
9263 * config/mips/mips.c (mips_modes_tieable_p): Make static.
9264 (TARGET_MODES_TIEABLE_P): Redefine.
9265 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
9266 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
9267 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
9268 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
9269 (mn10300_modes_tieable_p): ...this and make static.
9270 (TARGET_MODES_TIEABLE_P): Redefine.
9271 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
9272 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
9273 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
9274 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
9275 (msp430_modes_tieable_p): Make static.
9276 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
9277 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
9278 (TARGET_MODES_TIEABLE_P): Redefine.
9279 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
9280 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
9281 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
9282 (TARGET_MODES_TIEABLE_P): Redefine.
9283 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
9284 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
9285 * config/pa/pa.c (pa_modes_tieable_p): Make static.
9286 (TARGET_MODES_TIEABLE_P): Redefine.
9287 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
9288 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
9289 (pdp11_modes_tieable_p): New function.
9290 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
9291 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
9292 (rs6000_modes_tieable_p): New function.
9293 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9294 * config/powerpcspe/powerpcspe.md: Update comment.
9295 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
9296 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
9297 (TARGET_MODES_TIEABLE_P): Redefine.
9298 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
9299 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
9300 (rl78_modes_tieable_p): New function.
9301 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
9302 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
9303 (rs6000_modes_tieable_p): New function.
9304 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
9305 * config/rs6000/rs6000.md: Update comment.
9306 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
9307 * config/rx/rx.c (rx_modes_tieable_p): New function.
9308 (TARGET_MODES_TIEABLE_P): Redefine.
9309 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
9310 * config/s390/s390.c (s390_modes_tieable_p): New function.
9311 (TARGET_MODES_TIEABLE_P): Redefine.
9312 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
9313 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
9314 (sh_modes_tieable_p): New function.
9315 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
9316 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
9317 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
9318 (sparc_modes_tieable_p): Make static.
9319 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
9320 * config/spu/spu.c (spu_modes_tieable_p): New function.
9321 (TARGET_MODES_TIEABLE_P): Redefine.
9322 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
9323 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
9324 (TARGET_MODES_TIEABLE_P): Redefine.
9325 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
9326 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
9327 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
9328 * config/v850/v850.c (v850_modes_tieable_p): New function.
9329 (TARGET_MODES_TIEABLE_P): Redefine.
9330 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
9331 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
9332 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
9333 (visium_modes_tieable_p): New function.
9334 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
9335 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
9336 (xtensa_modes_tieable_p): New function.
9337 * system.h (MODES_TIEABLE_P): Poison.
9339 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9340 Alan Hayward <alan.hayward@arm.com>
9341 David Sherwood <david.sherwood@arm.com>
9343 * target.def (hard_regno_mode_ok): New hook.
9344 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
9345 (TARGET_HARD_REGNO_MODE_OK): ...this.
9346 * doc/tm.texi.in: Regenerate.
9347 * hooks.h (hook_bool_uint_mode_true): Declare.
9348 * hooks.c (hook_bool_uint_mode_true): New function.
9349 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
9351 * genpreds.c (write_insn_preds_c): Add an include of target.h.
9352 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
9353 instead of HARD_REGNO_MODE_OK.
9354 * caller-save.c: Include target.h.
9355 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
9357 * combine.c (can_combine_p): Likewise.
9358 (combinable_i3pat): Likewise.
9359 (can_change_dest_mode): Likewise.
9360 * expr.c (init_expr_target): Likewise.
9361 (convert_move): Likewise.
9362 (convert_modes): Likewise.
9363 * ira.c (setup_prohibited_class_mode_regs): Likewise.
9364 (setup_prohibited_mode_move_regs): Likewise.
9365 * ira.h (target_ira): Likewise.
9366 * lra-assigns.c (find_hard_regno_for_1): Likewise.
9367 * lra-constraints.c (process_alt_operands): Likewise.
9368 (split_reg): Likewise.
9369 * recog.c (peep2_find_free_register): Likewise.
9370 * ree.c (combine_reaching_defs): Likewise.
9371 * regcprop.c (maybe_mode_change): Likewise.
9372 * reginfo.c (init_reg_sets_1): Likewise.
9373 (choose_hard_reg_mode): Likewise.
9374 (simplifiable_subregs): Likewise.
9375 * regrename.c (check_new_reg_p): Likewise.
9376 * reload.c (find_valid_class): Likewise.
9377 (find_valid_class_1): Likewise.
9378 (reload_inner_reg_of_subreg): Likewise.
9379 (push_reload): Likewise.
9380 (combine_reloads): Likewise.
9381 (find_dummy_reload): Likewise.
9382 (find_reloads): Likewise.
9383 * reload1.c (find_reg): Likewise.
9384 (set_reload_reg): Likewise.
9385 (allocate_reload_reg): Likewise.
9386 (choose_reload_regs): Likewise.
9387 (reload_adjust_reg_for_temp): Likewise.
9388 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
9389 (simplify_subreg_regno): Likewise.
9390 * sel-sched.c (init_regs_for_mode): Likewise.
9391 * varasm.c (make_decl_rtl): Likewise.
9392 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
9393 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
9395 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
9396 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
9397 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9398 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
9399 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
9400 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9401 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
9402 (arc_mode_class): Delete.
9403 (HARD_REGNO_MODE_OK): Delete.
9404 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9405 (arc_hard_regno_mode_ok): Rename old array to...
9406 (arc_hard_regno_mode_ok_modes): ...this.
9407 (arc_conditional_register_usage): Update accordingly.
9408 (arc_mode_class): Make static.
9409 (arc_hard_regno_mode_ok): New function.
9410 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
9411 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
9412 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9413 (arm_hard_regno_mode_ok): Make static.
9414 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
9416 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
9417 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
9418 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
9420 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9421 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
9422 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
9423 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
9424 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
9425 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9426 * config/bfin/predicates.md (valid_reg_operand): Use
9427 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
9428 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
9429 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
9430 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9431 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
9432 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
9433 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9434 (cr16_hard_regno_mode_ok): Make static and return a bool.
9435 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
9436 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9437 (cris_hard_regno_mode_ok): New function.
9438 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
9439 (epiphany_mode_class): Delete.
9440 (HARD_REGNO_MODE_OK): Delete.
9441 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
9442 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9443 (hard_regno_mode_ok): Rename to...
9444 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
9445 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
9446 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
9448 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
9449 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
9450 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9451 (frv_hard_regno_mode_ok): Make static and return a bool.
9452 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
9454 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
9455 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
9456 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
9457 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
9459 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9460 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
9461 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
9462 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
9464 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9465 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
9466 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9467 (ia64_hard_regno_mode_ok): New function.
9468 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
9469 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9470 (iq2000_hard_regno_mode_ok): New function.
9471 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
9472 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9473 (lm32_hard_regno_mode_ok): New function.
9474 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
9475 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
9476 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
9477 instead of HARD_REGNO_MODE_OK.
9478 (m32c_hard_regno_ok): Rename to...
9479 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
9480 (m32c_cannot_change_mode_class): Update accordingly.
9481 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9482 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
9483 (m32r_mode_class): Delete.
9484 (HARD_REGNO_MODE_OK): Delete.
9485 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9486 (m32r_hard_regno_mode_ok): Rename to...
9487 (m32r_hard_regno_modes): ...this.
9488 (m32r_mode_class): Make static.
9489 (m32r_hard_regno_mode_ok): New function.
9490 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
9491 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
9492 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9493 (m68k_hard_regno_mode_ok): Make static.
9494 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
9495 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9496 (mcore_hard_regno_mode_ok): New function.
9497 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
9498 (HARD_REGNO_MODE_OK): Delete.
9499 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
9501 (microblaze_hard_regno_mode_ok_p): ...this and make static.
9502 (microblaze_hard_regno_mode_ok): New function.
9503 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9504 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
9505 (mips_hard_regno_mode_ok): Delete.
9506 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
9507 (mips_hard_regno_mode_ok_p): ...this and make static.
9508 (mips_hard_regno_mode_ok_p): Rename to...
9509 (mips_hard_regno_mode_ok_uncached): ...this.
9510 (mips_hard_regno_mode_ok): New function.
9511 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
9512 of HARD_REGNO_MODE_OK.
9513 (mips_option_override): Update after above name changes.
9514 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9515 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
9516 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
9517 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
9518 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
9519 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9520 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
9521 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
9522 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
9523 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9524 (msp430_hard_regno_mode_ok): Make static and return a bool.
9525 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
9526 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
9527 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
9529 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9530 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
9531 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
9532 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
9533 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
9534 (PA_HARD_REGNO_MODE_OK): ...this
9535 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
9536 (PA_HARD_REGNO_MODE_OK): ...this.
9537 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9538 (pa_hard_regno_mode_ok): New function.
9539 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
9540 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9541 (pdp11_hard_regno_mode_ok): New function.
9542 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
9543 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
9545 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
9547 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9548 (rs6000_hard_regno_mode_ok): Rename to...
9549 (rs6000_hard_regno_mode_ok_uncached): ...this.
9550 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9551 (rs6000_hard_regno_mode_ok): New function.
9552 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
9553 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
9554 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
9555 (riscv_hard_regno_mode_ok): ...this and make static.
9556 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9557 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
9558 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
9559 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9560 (rl78_hard_regno_mode_ok): Make static and return bool.
9561 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
9562 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
9564 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
9565 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9566 (rs6000_hard_regno_mode_ok): Rename to...
9567 (rs6000_hard_regno_mode_ok_uncached): ...this.
9568 (rs6000_init_hard_regno_mode_ok): Update accordingly.
9569 (rs6000_hard_regno_mode_ok): New function.
9570 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
9571 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
9572 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9573 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
9574 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
9575 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
9576 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9577 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
9578 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
9579 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9580 (sh_hard_regno_mode_ok): Make static.
9581 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
9582 instead of HARD_REGNO_MODE_OK.
9583 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
9584 (sparc_mode_class): Delete.
9585 (HARD_REGNO_MODE_OK): Delete.
9586 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9587 (hard_regno_mode_classes): Make static.
9588 (sparc_mode_class): Likewise.
9589 (sparc_hard_regno_mode_ok): New function.
9590 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
9591 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
9592 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
9594 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9595 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
9596 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
9597 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
9598 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
9599 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9600 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
9601 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
9602 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
9603 (visium_hard_regno_mode_ok): New function.
9604 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
9605 instead of HARD_REGNO_MODE_OK.
9606 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
9607 (HARD_REGNO_MODE_OK): Delete.
9608 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
9609 (xtensa_hard_regno_mode_ok_p): ...this and make static.
9610 (xtensa_option_override): Update accordingly.
9611 (TARGET_HARD_REGNO_MODE_OK): Redefine.
9612 (xtensa_hard_regno_mode_ok): New function.
9613 * system.h (HARD_REGNO_MODE_OK): Poison.
9615 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9616 Alan Hayward <alan.hayward@arm.com>
9617 David Sherwood <david.sherwood@arm.com>
9619 * target.def (hard_regno_call_part_clobbered): New hook.
9620 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
9621 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
9622 * doc/tm.texi: Regenerate.
9623 * hooks.h (hook_bool_uint_mode_false): Declare.
9624 * hooks.c (hook_bool_uint_mode_false): New function.
9625 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9626 * cselib.c (cselib_process_insn): Use
9627 targetm.hard_regno_call_part_clobbered instead of
9628 HARD_REGNO_CALL_PART_CLOBBERED.
9629 * ira-conflicts.c (ira_build_conflicts): Likewise.
9630 * ira-costs.c (ira_tune_allocno_costs): Likewise.
9631 * lra-constraints.c (need_for_call_save_p): Likewise.
9632 * lra-lives.c: Include target.h.
9633 (check_pseudos_live_through_calls): Use
9634 targetm.hard_regno_call_part_clobbered instead of
9635 HARD_REGNO_CALL_PART_CLOBBERED.
9636 * regcprop.c: Include target.h.
9637 (copyprop_hardreg_forward_1): Use
9638 targetm.hard_regno_call_part_clobbered instead of
9639 HARD_REGNO_CALL_PART_CLOBBERED.
9640 * reginfo.c (choose_hard_reg_mode): Likewise.
9641 * regrename.c (check_new_reg_p): Likewise.
9642 * reload.c (find_equiv_reg): Likewise.
9643 * reload1.c (emit_reload_insns): Likewise.
9644 * sched-deps.c (deps_analyze_insn): Likewise.
9645 * sel-sched.c (init_regs_for_mode): Likewise.
9646 (mark_unavailable_hard_regs): Likewise.
9647 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
9648 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9649 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
9651 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9652 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9653 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
9655 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
9657 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9658 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9659 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
9661 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9662 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9663 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
9665 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9666 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
9668 * config/powerpcspe/powerpcspe.c
9669 (rs6000_hard_regno_call_part_clobbered): New function.
9670 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9671 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9672 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
9674 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9675 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9676 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
9678 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
9679 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
9680 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
9682 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9683 Alan Hayward <alan.hayward@arm.com>
9684 David Sherwood <david.sherwood@arm.com>
9686 * rtl.h (subreg_memory_offset): Declare.
9687 * emit-rtl.c (subreg_memory_offset): New function.
9688 * expmed.c (store_bit_field_1): Use it.
9689 * expr.c (undefined_operand_subword_p): Likewise.
9690 * simplify-rtx.c (simplify_subreg): Likewise.
9692 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
9694 PR rtl-optimization/57448
9697 * optabs.c (expand_atomic_load): Place compiler memory barriers if
9698 using atomic_load pattern.
9699 (expand_atomic_store): Likewise.
9701 2017-09-04 Jakub Jelinek <jakub@redhat.com>
9704 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
9705 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
9706 handling. Use replace_call_with_value with NULL instead of
9707 gsi_replace, unlink_stmt_vdef and release_defs.
9709 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
9712 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
9714 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
9717 * rtl.h (emit_library_call_value_1): Declare.
9718 (emit_library_call): Replace declaration with a series of overloads.
9719 Remove the parameter count argument.
9720 (emit_library_call_value): Likewise.
9721 * calls.c (emit_library_call_value_1): Make global. Replace varargs
9722 with an "rtx_mode_t *".
9723 (emit_library_call_value): Delete.
9724 (emit_library_call): Likewise.
9725 * asan.c (asan_emit_stack_protection): Update calls accordingly.
9726 (asan_emit_allocas_unpoison): Likewise.
9727 * builtins.c (expand_builtin_powi): Likewise.
9728 (expand_asan_emit_allocas_unpoison): Likewise.
9729 * cfgexpand.c (expand_main_function): Likewise.
9730 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
9731 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
9732 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
9733 * config/arm/arm.c (arm_trampoline_init): Likewise.
9734 (arm_call_tls_get_addr): Likewise.
9735 (arm_expand_divmod_libfunc): Likewise.
9736 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
9737 (smulsi3_highpart): Likewise.
9738 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
9739 (c6x_expand_compare): Likewise.
9740 (c6x_expand_movmem): Likewise.
9741 * config/frv/frv.c (frv_trampoline_init): Likewise.
9742 * config/i386/i386.c (ix86_trampoline_init): Likewise.
9743 (ix86_expand_divmod_libfunc): Likewise.
9744 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
9745 (ia64_expand_compare): Likewise.
9746 (ia64_profile_hook): Likewise.
9747 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
9748 (nonlocal_goto): Likewise.
9749 (restore_stack_nonlocal): Likewise.
9750 * config/m32r/m32r.c (block_move_call): Likewise.
9751 (m32r_trampoline_init): Likewise.
9752 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
9753 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
9754 (m68k_call_m68k_read_tp): Likewise.
9755 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
9756 (microblaze_expand_divide): Likewise.
9757 * config/mips/mips.h (mips_args): Likewise.
9758 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
9759 (MIPS_ICACHE_SYNC): Likewise.
9760 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
9761 (nios2_trampoline_init): Likewise.
9762 * config/pa/pa.c (hppa_tls_call): Likewise.
9763 (pa_trampoline_init): Likewise.
9764 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
9765 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
9766 (expand_strn_compare): Likewise.
9767 (rs6000_generate_compare): Likewise.
9768 (rs6000_expand_float128_convert): Likewise.
9769 (output_profile_hook): Likewise.
9770 (rs6000_trampoline_init): Likewise.
9771 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
9772 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
9773 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
9774 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
9775 (rs6000_generate_compare): Likewise.
9776 (rs6000_expand_float128_convert): Likewise.
9777 (output_profile_hook): Likewise.
9778 (rs6000_trampoline_init): Likewise.
9779 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
9780 * config/sh/sh.c (sh_trampoline_init): Likewise.
9781 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
9782 (sparc_emit_float_lib_cmp): Likewise.
9783 (sparc32_initialize_trampoline): Likewise.
9784 (sparc64_initialize_trampoline): Likewise.
9785 (sparc_profile_hook): Likewise.
9786 * config/spu/spu.c (ea_load_store): Likewise.
9787 * config/spu/spu.md (floatunssidf2): Likewise.
9788 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
9789 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
9790 * config/visium/visium.c (expand_block_move_4): Likewise.
9791 (expand_block_move_2): Likewise.
9792 (expand_block_move_1): Likewise.
9793 (expand_block_set_4): Likewise.
9794 (expand_block_set_2): Likewise.
9795 (expand_block_set_1): Likewise.
9796 (visium_trampoline_init): Likewise.
9797 (visium_profile_hook): Likewise.
9798 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
9799 (xtensa_setup_frame_addresses): Likewise.
9800 (xtensa_trampoline_init): Likewise.
9801 * except.c (sjlj_emit_function_enter): Likewise.
9802 (sjlj_emit_function_exit): Likewise.
9803 * explow.c (allocate_dynamic_stack_space): Likewise.
9804 (probe_stack_range): Likewise.
9805 * expr.c (convert_mode_scalar): Likewise.
9806 * optabs.c (expand_binop): Likewise.
9807 (expand_twoval_binop_libfunc): Likewise.
9808 (expand_unop): Likewise.
9809 (prepare_cmp_insn): Likewise.
9810 (prepare_float_lib_cmp): Likewise.
9811 (expand_float): Likewise.
9812 (expand_fix): Likewise.
9813 (expand_fixed_convert): Likewise.
9814 (maybe_emit_sync_lock_test_and_set): Likewise.
9815 (expand_atomic_compare_and_swap): Likewise.
9816 (expand_mem_thread_fence): Likewise.
9817 (expand_atomic_fetch_op): Likewise.
9819 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
9821 * doc/generic.texi (OpenACC): Adjust URL.
9822 * doc/invoke.texi (C Dialect Options): Ditto.
9824 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
9826 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
9827 predicate for operand 1. Add (m,<S>) constraint.
9828 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
9829 Prevent memory operand 1 with register operand 2.
9831 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
9833 PR rtl-optimization/82024
9834 * combine.c (try_combine): If the combination result is a PARALLEL,
9835 and we only need to retain the SET in there that would be placed
9836 at I2, check that we can place that at I3 instead, before doing so.
9838 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9841 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
9843 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
9844 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
9847 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
9848 Jeff Law <law@redhat.com>
9850 * varasm.c (bss_initializer_p): Do not put constants into .bss
9851 (categorize_decl_for_section): Handle bss_initializer_p returning
9852 false when DECL_INITIAL is NULL.
9854 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9857 * config/s390/s390.c (s390_can_inline_p): New function.
9859 2017-09-01 Jeff Law <law@redhat.com>
9861 PR tree-optimization/82052
9862 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
9863 Always initialize the returned slot after a hash table miss
9864 when INSERT is true.
9866 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
9868 * config/s390/s390.md (mem_signal_fence): Remove.
9869 * doc/md.texi (mem_signal_fence): Remove.
9870 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
9872 * target-insns.def (mem_signal_fence): Remove.
9874 2017-09-01 Jakub Jelinek <jakub@redhat.com>
9877 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
9880 * asan.c (create_odr_indicator): Strip name encoding from assembler
9881 name before appending it after __odr_asan_.
9883 2017-09-01 Martin Liska <mliska@suse.cz>
9885 PR tree-optimization/82059
9886 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
9887 frequency only when an edge is redirected.
9889 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9891 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
9892 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
9893 (arc_conditional_register_usage): Remove ARC600 lp_count
9895 (arc_file_start): Emit Tag_ARC_CPU_variation.
9896 (arc_can_use_doloop_p): New conditions to use ZOLs.
9897 (hwloop_fail): New function.
9898 (hwloop_optimize): Likewise.
9899 (hwloop_pattern_reg): Likewise.
9900 (arc_doloop_hooks): New struct, to be used with reorg_loops.
9901 (arc_reorg_loops): New function, calls reorg_loops.
9902 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
9903 (arc600_corereg_hazard): Remove ZOL checking, case handled by
9905 (arc_loop_hazard): Remove function, functionality moved into
9907 (arc_hazard): Remove arc_loop_hazard call.
9908 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
9909 into hwloop_optimize.
9910 (arc_label_align): Remove ZOL handling.
9911 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
9912 * config/arc/arc.md (doloop_begin): Remove pattern.
9913 (doloop_begin_i): Likewise.
9914 (doloop_end_i): Likewise.
9915 (doloop_fallback): Likewise.
9916 (doloop_fallback_m): Likewise.
9917 (doloop_end): Reimplement expand.
9918 (arc_lp): New pattern for LP instruction.
9919 (loop_end): New pattern.
9920 (loop_fail): Likewise.
9921 (decrement_and_branch_until_zero): Likewise.
9922 * config/arc/arc.opt (mlpc-width): New option.
9923 * doc/invoke.texi (mlpc-width): Document option.
9925 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9927 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
9928 (arc_ccfsm_advance): Fix checking for delay slots.
9929 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
9931 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9933 * config/arc/arc.md (movqi_insn): Add stores to save constant long
9935 (movhi_insn): Update store instruction constraint which are saving
9936 6-bit short immediates.
9937 (movsi_insn): Consider also short scaled load operations.
9938 (zero_extendhisi2_i): Use Usd constraint instead of T.
9939 (extendhisi2_i): Add q constraint.
9940 (arc_clzsi2): Add type and length attributes.
9941 (arc_ctzsi2): Likewise.
9942 * config/arc/constraints.md (Usc): Update constraint, the
9943 assembler can parse two relocations for a single instruction.
9945 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
9947 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
9948 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
9950 2017-08-31 Olivier Hainque <hainque@adacore.com>
9952 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
9953 match as powerpc-wrs-vxworks*.
9955 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
9957 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
9958 register constraint for by-element operand.
9959 (aarch64_mls_elt_merge<mode>): Likewise.
9961 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9963 * config/arc/arc.c (arc_can_follow_jump): Check for short
9966 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9968 * config.gcc: Use g.opt for arc.
9969 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
9970 functionality moved to ...
9971 (legitimate_scaled_address_p): New function, ...here.
9972 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
9973 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
9974 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
9976 (arc_override_options): Handle G option.
9977 (arc_output_pic_addr_const): Correct function definition.
9978 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
9979 (arc_decl_anon_ns_mem_p): Delete.
9980 (arc_in_small_data_p): Overhaul this function to take into
9981 consideration the value given via G option.
9982 (arc_rewrite_small_data_1): Renamed and corrected old
9983 arc_rewrite_small_data function.
9984 (arc_rewrite_small_data): New function.
9985 (small_data_pattern): Don't use pic_offset_table_rtx.
9986 * config/arc/arc.h (CC1_SPEC): Recognize G option.
9987 * config/arc/simdext.md (movmisalignv2hi): Use
9988 prepare_move_operands function.
9990 (movmisalign*): Likewise.
9991 * doc/invoke.texi (ARC options): Document -G option.
9993 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
9995 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
9997 * config/arc/arc.c (arc_print_operand): Output scalled address for
9998 sdata whenever is possible.
9999 (arc_in_small_data_p): Allow sdata for 64bit datum when double
10000 load/stores are available.
10001 (compact_sda_memory_operand): Check for the alignment required by
10002 code density instructions.
10003 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
10005 * config/arc/constraints.md (Usd): Update constraint.
10006 (Us0): New constraint.
10007 (Usc): Update constraint.
10009 2017-08-31 Richard Biener <rguenther@suse.de>
10011 PR middle-end/82054
10012 * dwarf2out.c (dwarf2out_early_global_decl): Process each
10013 function only once.
10015 2017-08-31 Tamar Christina <tamar.christina@arm.com>
10017 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
10018 Resize type_signature.
10020 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10021 Alan Hayward <alan.hayward@arm.com>
10022 David Sherwood <david.sherwood@arm.com>
10024 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
10025 subregs whose inner modes can be stored in GPRs.
10026 (aarch64_classify_index): Likewise.
10028 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
10029 Alan Hayward <alan.hayward@arm.com>
10030 David Sherwood <david.sherwood@arm.com>
10032 * config/aarch64/iterators.md (V_cmp_result): Rename to...
10033 (V_INT_EQUIV): ...this.
10034 (v_cmp_result): Rename to...
10035 (v_int_equiv): ...this.
10036 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
10037 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
10038 (copysign<mode>3): Likewise.
10039 (aarch64_simd_bsl<mode>_internal): Likewise.
10040 (aarch64_simd_bsl<mode>): Likewise.
10041 (vec_cmp<mode><mode>): Likewise.
10042 (vcond<mode><mode>): Likewise.
10043 (vcond<v_cmp_mixed><mode>): Likewise.
10044 (vcondu<mode><v_cmp_mixed>): Likewise.
10045 (aarch64_cm<optab><mode>): Likewise.
10046 (aarch64_cmtst<mode>): Likewise.
10047 (aarch64_fac<optab><mode>): Likewise.
10048 (vec_perm_const<mode>): Likewise.
10049 (vcond_mask_<mode><v_cmp_result>): Rename to...
10050 (vcond_mask_<mode><v_int_equiv>): ...this.
10051 (vec_cmp<mode><v_cmp_result>): Rename to...
10052 (vec_cmp<mode><v_int_equiv>): ...this.
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/aarch64-modes.def: Remove 32-, 48- and 64-byte
10060 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
10061 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
10062 (UNSPEC_LD4_DREG): New unspecs.
10063 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
10064 (aarch64_ld2<mode>_dreg_be): Replace with...
10065 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
10067 (aarch64_ld3<mode>_dreg_le)
10068 (aarch64_ld3<mode>_dreg_be): Replace with...
10069 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
10071 (aarch64_ld4<mode>_dreg_le)
10072 (aarch64_ld4<mode>_dreg_be): Replace with...
10073 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
10076 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10078 PR tree-optimization/81987
10079 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
10080 insert an initializer in a location not dominated by the stride
10083 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10085 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
10086 on the entire header of the finally block in the fallthru case.
10088 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
10090 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
10092 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
10094 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
10095 rs6000_emit_move_from_cr and call renamed function.
10096 (rs6000_emit_prologue): Call renamed functions.
10097 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
10098 movesi_from_cr, remove volatile CRs.
10100 2017-08-30 Jon Beniston <jon@beniston.com>
10101 Richard Biener <rguenther@suse.de>
10103 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
10104 of VECTOR_MODE_P check.
10105 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
10106 element vector types.
10108 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10110 * df.h (df_read_modify_subreg_p): Remove in favor of...
10111 * rtl.h (read_modify_subreg_p): ...this new function. Take a
10112 const_rtx instead of an rtx.
10113 * cprop.c (local_cprop_find_used_regs): Update accordingly.
10114 * df-problems.c (df_word_lr_mark_ref): Likewise.
10115 * ira-lives.c (mark_pseudo_reg_live): Likewise.
10116 (mark_pseudo_reg_dead): Likewise.
10117 (mark_ref_dead): Likewise.
10118 * reginfo.c (init_subregs_of_mode): Likewise.
10119 * sched-deps.c (sched_analyze_1): Likewise.
10120 * df-scan.c (df_def_record_1): Likewise.
10121 (df_uses_record): Likewise.
10122 (df_read_modify_subreg_p): Remove in favor of...
10123 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
10124 const_rtx instead of an rtx.
10126 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10127 Alan Hayward <alan.hayward@arm.com>
10128 David Sherwood <david.sherwood@arm.com>
10130 * rtl.h (partial_subreg_p): New function.
10131 * caller-save.c (save_call_clobbered_regs): Use it.
10132 * calls.c (expand_call): Likewise.
10133 * combine.c (combinable_i3pat): Likewise.
10134 (simplify_set): Likewise.
10135 (make_extraction): Likewise.
10136 (make_compound_operation_int): Likewise.
10137 (gen_lowpart_or_truncate): Likewise.
10138 (force_to_mode): Likewise.
10139 (make_field_assignment): Likewise.
10140 (reg_truncated_to_mode): Likewise.
10141 (record_truncated_value): Likewise.
10142 (move_deaths): Likewise.
10143 * cse.c (record_jump_cond): Likewise.
10144 (cse_insn): Likewise.
10145 * cselib.c (cselib_lookup_1): Likewise.
10146 * expmed.c (extract_bit_field_using_extv): Likewise.
10147 * function.c (assign_parm_setup_reg): Likewise.
10148 * ifcvt.c (noce_convert_multiple_sets): Likewise.
10149 * ira-build.c (create_insn_allocnos): Likewise.
10150 * lra-coalesce.c (merge_pseudos): Likewise.
10151 * lra-constraints.c (match_reload): Likewise.
10152 (simplify_operand_subreg): Likewise.
10153 (curr_insn_transform): Likewise.
10154 * lra-lives.c (process_bb_lives): Likewise.
10155 * lra.c (new_insn_reg): Likewise.
10156 (lra_substitute_pseudo): Likewise.
10157 * regcprop.c (mode_change_ok): Likewise.
10158 (maybe_mode_change): Likewise.
10159 (copyprop_hardreg_forward_1): Likewise.
10160 * reload.c (push_reload): Likewise.
10161 (find_reloads): Likewise.
10162 (find_reloads_subreg_address): Likewise.
10163 * reload1.c (alter_reg): Likewise.
10164 (eliminate_regs_1): Likewise.
10165 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10167 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
10169 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
10170 back to if statements, including unpack.
10172 2017-08-30 Martin Liska <mliska@suse.cz>
10174 PR inline-asm/82001
10175 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
10177 (func_checker::compare_asm_inputs_outputs): ... this function.
10178 (func_checker::compare_gimple_asm): Use the function to compare
10179 also ASM constrains.
10180 * ipa-icf-gimple.h: Rename the function.
10182 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10183 Alan Hayward <alan.hayward@arm.com>
10184 David Sherwood <david.sherwood@arm.com>
10186 * coretypes.h (complex_mode): New type.
10187 * gdbhooks.py (build_pretty_printer): Handle it.
10188 * machmode.h (complex_mode): New class.
10189 (complex_mode::includes_p): New function.
10190 (is_complex_int_mode): Likewise.
10191 (is_complex_float_mode): Likewise.
10192 * genmodes.c (get_mode_class): Handle complex mode classes.
10193 * function.c (expand_function_end): Use is_complex_int_mode.
10195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10196 Alan Hayward <alan.hayward@arm.com>
10197 David Sherwood <david.sherwood@arm.com>
10199 * coretypes.h (scalar_mode_pod): New typedef.
10200 * gdbhooks.py (build_pretty_printer): Handle it.
10201 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
10202 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
10203 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
10204 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
10205 as_a <scalar_mode>.
10207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10208 Alan Hayward <alan.hayward@arm.com>
10209 David Sherwood <david.sherwood@arm.com>
10211 * machmode.h (mode_for_vector): Take a scalar_mode instead
10213 * stor-layout.c (mode_for_vector): Likewise.
10214 * explow.c (promote_mode): Use as_a <scalar_mode>.
10215 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
10217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10218 Alan Hayward <alan.hayward@arm.com>
10219 David Sherwood <david.sherwood@arm.com>
10221 * target.def (preferred_simd_mode): Take a scalar_mode
10222 instead of a machine_mode.
10223 * targhooks.h (default_preferred_simd_mode): Likewise.
10224 * targhooks.c (default_preferred_simd_mode): Likewise.
10225 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
10226 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
10227 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
10228 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
10229 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
10230 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
10231 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
10232 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
10234 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
10235 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
10236 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
10237 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
10238 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
10239 * doc/tm.texi: Regenerate.
10240 * optabs-query.c (can_vec_mask_load_store_p): Return false for
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 (scalar_mode_supported_p): Take a scalar_mode
10248 instead of a machine_mode.
10249 * targhooks.h (default_scalar_mode_supported_p): Likewise.
10250 * targhooks.c (default_scalar_mode_supported_p): Likewise.
10251 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
10252 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10253 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
10254 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
10255 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
10256 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
10257 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
10258 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
10259 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
10260 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
10261 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
10262 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
10264 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
10265 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
10266 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10267 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10268 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
10270 * doc/tm.texi: Regenerate.
10272 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10273 Alan Hayward <alan.hayward@arm.com>
10274 David Sherwood <david.sherwood@arm.com>
10276 * coretypes.h (opt_scalar_mode): New typedef.
10277 * gdbhooks.py (build_pretty_printers): Handle it.
10278 * machmode.h (mode_iterator::get_2xwider): Add overload for
10280 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
10282 * expr.c (convert_mode_scalar): Likewise.
10283 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10284 * optabs.c (expand_float): Likewise.
10285 (expand_fix): Likewise.
10286 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10288 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10289 Alan Hayward <alan.hayward@arm.com>
10290 David Sherwood <david.sherwood@arm.com>
10292 * optabs.c (expand_float): Explicitly check for scalars before
10293 using a branching expansion.
10294 (expand_fix): Likewise.
10296 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10297 Alan Hayward <alan.hayward@arm.com>
10298 David Sherwood <david.sherwood@arm.com>
10300 * expr.c (convert_mode): Split scalar handling out into...
10301 (convert_mode_scalar): ...this new function. Treat the modes
10304 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10305 Alan Hayward <alan.hayward@arm.com>
10306 David Sherwood <david.sherwood@arm.com>
10308 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
10310 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
10312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10313 Alan Hayward <alan.hayward@arm.com>
10314 David Sherwood <david.sherwood@arm.com>
10316 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
10317 rather than a machine_mode.
10318 (fixed_from_string): Likewise.
10319 (fixed_convert): Likewise.
10320 (fixed_convert_from_int): Likewise.
10321 (fixed_convert_from_real): Likewise.
10322 (real_convert_from_fixed): Likewise.
10323 * fixed-value.c (fixed_from_double_int): Likewise.
10324 (fixed_from_string): Likewise.
10325 (fixed_convert): Likewise.
10326 (fixed_convert_from_int): Likewise.
10327 (fixed_convert_from_real): Likewise.
10328 (real_convert_from_fixed): Likewise.
10329 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
10331 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10332 Alan Hayward <alan.hayward@arm.com>
10333 David Sherwood <david.sherwood@arm.com>
10335 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
10336 of separate mode class checks. Do not allow vector modes here.
10337 (immed_wide_int_const): Use as_a <scalar_mode>.
10338 * explow.c (trunc_int_for_mode): Likewise.
10339 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
10340 (wi::shwi): Likewise.
10341 (wi::min_value): Likewise.
10342 (wi::max_value): Likewise.
10343 * dwarf2out.c (loc_descriptor): Likewise.
10344 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
10345 for CONST_WIDE_INT.
10347 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10348 Alan Hayward <alan.hayward@arm.com>
10349 David Sherwood <david.sherwood@arm.com>
10351 * tree.h (SCALAR_TYPE_MODE): New macro.
10352 * expr.c (expand_expr_addr_expr_1): Use it.
10353 (expand_expr_real_2): Likewise.
10354 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
10355 (fold_convert_const_fixed_from_int): Likewise.
10356 (fold_convert_const_fixed_from_real): Likewise.
10357 (native_encode_fixed): Likewise
10358 (native_encode_complex): Likewise
10359 (native_encode_vector): Likewise.
10360 (native_interpret_fixed): Likewise.
10361 (native_interpret_real): Likewise.
10362 (native_interpret_complex): Likewise.
10363 (native_interpret_vector): Likewise.
10364 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
10365 (simd_clone_adjust_argument_types): Likewise.
10366 (simd_clone_init_simd_arrays): Likewise.
10367 (simd_clone_adjust): Likewise.
10368 * stor-layout.c (layout_type): Likewise.
10369 * tree.c (build_minus_one_cst): Likewise.
10370 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
10371 * tree-inline.c (estimate_move_cost): Likewise.
10372 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
10373 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
10374 (vectorizable_reduction): Likewise.
10375 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
10376 (vect_recog_mixed_size_cond_pattern): Likewise.
10377 (check_bool_pattern): Likewise.
10378 (adjust_bool_pattern): Likewise.
10379 (search_type_for_mask_1): Likewise.
10380 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
10381 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10382 (vectorizable_load): Likewise.
10383 (vectorizable_store): Likewise.
10384 * ubsan.c (ubsan_encode_value): Likewise.
10385 * varasm.c (output_constant): Likewise.
10387 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10388 Alan Hayward <alan.hayward@arm.com>
10389 David Sherwood <david.sherwood@arm.com>
10391 * coretypes.h (scalar_mode): New class.
10392 * machmode.h (scalar_mode): Likewise.
10393 (scalar_mode::includes_p): New function.
10394 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
10395 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
10396 * genmodes.c (get_mode_class): Handle remaining scalar modes.
10397 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
10398 * expmed.c (store_bit_field_1): Likewise.
10399 (extract_bit_field_1): Likewise.
10400 * expr.c (write_complex_part): Likewise.
10401 (read_complex_part): Likewise.
10402 (emit_move_complex_push): Likewise.
10403 (expand_expr_real_2): Likewise.
10404 * function.c (assign_parm_setup_reg): Likewise.
10405 (assign_parms_unsplit_complex): Likewise.
10406 * optabs.c (expand_binop): Likewise.
10407 * rtlanal.c (subreg_get_info): Likewise.
10408 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10409 * varasm.c (output_constant_pool_2): Likewise.
10411 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10412 Alan Hayward <alan.hayward@arm.com>
10413 David Sherwood <david.sherwood@arm.com>
10415 * expmed.c (extract_high_half): Use scalar_int_mode and remove
10417 (expmed_mult_highpart_optab): Likewise.
10418 (expmed_mult_highpart): Likewise.
10420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10421 Alan Hayward <alan.hayward@arm.com>
10422 David Sherwood <david.sherwood@arm.com>
10424 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
10425 instead of a machine_mode.
10426 (builtin_memset_read_str): Likewise.
10427 * builtins.c (c_readstr): Likewise.
10428 (builtin_memcpy_read_str): Likewise.
10429 (builtin_strncpy_read_str): Likewise.
10430 (builtin_memset_read_str): Likewise.
10431 (builtin_memset_gen_str): Likewise.
10432 (expand_builtin_signbit): Use scalar_int_mode for local variables.
10433 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
10434 instead of a machine_mode.
10435 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
10437 (make_extraction): Likewise.
10438 (try_widen_shift_mode): Take and return scalar_int_modes instead
10440 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
10441 a scalar_int_mode instead of a machine_mode.
10442 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
10443 (avr_addr_space_pointer_mode): Likewise.
10444 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
10445 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
10446 (msp430_unwind_word_mode): Likewise.
10447 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
10448 (spu_addr_space_pointer_mode): Likewise.
10449 (spu_addr_space_address_mode): Likewise.
10450 (spu_libgcc_cmp_return_mode): Likewise.
10451 (spu_libgcc_shift_count_mode): Likewise.
10452 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
10453 (rl78_addr_space_pointer_mode): Likewise.
10454 (fl78_unwind_word_mode): Likewise.
10455 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
10457 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
10458 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
10459 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
10460 (mips_valid_pointer_mode): Likewise.
10461 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
10462 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
10463 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
10465 (ft32_addr_space_address_mode): Likewise.
10466 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
10467 scalar_int_mode instead of a machine_mode.
10468 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
10470 (m32c_addr_space_address_mode): Likewise.
10471 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
10472 (rs6000_eh_return_filter_mode): Likewise.
10473 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
10474 (rs6000_eh_return_filter_mode): Likewise.
10475 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
10476 (s390_libgcc_shift_count_mode): Likewise.
10477 (s390_unwind_word_mode): Likewise.
10478 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
10480 * target.def (mode_rep_extended): Likewise.
10481 (valid_pointer_mode): Likewise.
10482 (addr_space.valid_pointer_mode): Likewise.
10483 (eh_return_filter_mode): Return a scalar_int_mode rather than
10485 (libgcc_cmp_return_mode): Likewise.
10486 (libgcc_shift_count_mode): Likewise.
10487 (unwind_word_mode): Likewise.
10488 (addr_space.pointer_mode): Likewise.
10489 (addr_space.address_mode): Likewise.
10490 * doc/tm.texi: Regenerate.
10491 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
10493 (do_jump): Use scalar_int_mode for local variables.
10494 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
10495 rather than a machine_mode.
10496 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
10497 (scompare_loc_descriptor_wide): Likewise.
10498 (scompare_loc_descriptor_narrow): Likewise.
10499 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
10501 * except.c (sjlj_emit_dispatch_table): Likewise.
10502 (expand_builtin_eh_copy_values): Likewise.
10503 * explow.c (convert_memory_address_addr_space_1): Likewise.
10504 Take a scalar_int_mode rather than a machine_mode.
10505 (convert_memory_address_addr_space): Take a scalar_int_mode rather
10506 than a machine_mode.
10507 (memory_address_addr_space): Use scalar_int_mode for local variables.
10508 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
10509 rather than a machine_mode.
10510 * expmed.c (mask_rtx): Likewise.
10511 (init_expmed_one_conv): Likewise.
10512 (expand_mult_highpart_adjust): Likewise.
10513 (extract_high_half): Likewise.
10514 (expmed_mult_highpart_optab): Likewise.
10515 (expmed_mult_highpart): Likewise.
10516 (expand_smod_pow2): Likewise.
10517 (expand_sdiv_pow2): Likewise.
10518 (emit_store_flag_int): Likewise.
10519 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
10521 (extract_low_bits): Likewise.
10522 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
10524 * expr.c (pieces_addr::adjust): Likewise.
10525 (can_store_by_pieces): Likewise.
10526 (store_by_pieces): Likewise.
10527 (clear_by_pieces_1): Likewise.
10528 (expand_expr_addr_expr_1): Likewise.
10529 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
10530 (expand_expr_real_1): Likewise.
10531 (try_casesi): Likewise.
10532 * final.c (shorten_branches): Likewise.
10533 * fold-const.c (fold_convert_const_int_from_fixed): Change the
10534 type of "mode" to machine_mode.
10535 * internal-fn.c (expand_arith_overflow_result_store): Take a
10536 scalar_int_mode rather than a machine_mode.
10537 (expand_mul_overflow): Use scalar_int_mode for local variables.
10538 * loop-doloop.c (doloop_modify): Likewise.
10539 (doloop_optimize): Likewise.
10540 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
10541 than a machine_mode.
10542 (expand_doubleword_shift_condmove): Likewise.
10543 (expand_doubleword_shift): Likewise.
10544 (expand_doubleword_clz): Likewise.
10545 (expand_doubleword_popcount): Likewise.
10546 (expand_doubleword_parity): Likewise.
10547 (expand_absneg_bit): Use scalar_int_mode for local variables.
10548 (prepare_float_lib_cmp): Likewise.
10549 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
10550 rather than a machine_mode.
10551 (convert_memory_address_addr_space): Likewise.
10552 (get_mode_bounds): Likewise.
10553 (get_address_mode): Return a scalar_int_mode rather than a
10555 * rtlanal.c (get_address_mode): Likewise.
10556 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
10557 than a machine_mode.
10558 * targhooks.c (default_mode_rep_extended): Likewise.
10559 (default_valid_pointer_mode): Likewise.
10560 (default_addr_space_valid_pointer_mode): Likewise.
10561 (default_eh_return_filter_mode): Return a scalar_int_mode rather
10562 than a machine_mode.
10563 (default_libgcc_cmp_return_mode): Likewise.
10564 (default_libgcc_shift_count_mode): Likewise.
10565 (default_unwind_word_mode): Likewise.
10566 (default_addr_space_pointer_mode): Likewise.
10567 (default_addr_space_address_mode): Likewise.
10568 * targhooks.h (default_eh_return_filter_mode): Likewise.
10569 (default_libgcc_cmp_return_mode): Likewise.
10570 (default_libgcc_shift_count_mode): Likewise.
10571 (default_unwind_word_mode): Likewise.
10572 (default_addr_space_pointer_mode): Likewise.
10573 (default_addr_space_address_mode): Likewise.
10574 (default_mode_rep_extended): Take a scalar_int_mode rather than
10576 (default_valid_pointer_mode): Likewise.
10577 (default_addr_space_valid_pointer_mode): Likewise.
10578 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
10580 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
10581 rather than a machine_mode.
10582 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
10583 for local variables.
10584 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10585 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
10586 than a machine_mode.
10588 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10589 Alan Hayward <alan.hayward@arm.com>
10590 David Sherwood <david.sherwood@arm.com>
10592 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
10593 the mode argument to scalar_int_mode.
10594 (do_jump_by_parts_zero_rtx): Likewise.
10595 (do_jump_by_parts_equality_rtx): Likewise.
10596 (do_jump_by_parts_greater): Take a mode argument.
10597 (do_jump_by_parts_equality): Likewise.
10598 (do_jump_1): Update calls accordingly.
10600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10601 Alan Hayward <alan.hayward@arm.com>
10602 David Sherwood <david.sherwood@arm.com>
10604 * is-a.h (safe_dyn_cast): New function.
10605 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
10606 (jump_table_for_label): Likewise.
10607 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
10608 instead of an rtx_insn *.
10609 (shorten_branches): Use dyn_cast instead of LABEL_P and
10610 JUMP_TABLE_DATA_P. Use jump_table_for_label and
10611 rtx_jump_table_data::get_data_mode.
10612 (final_scan_insn): Likewise.
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 * combine.c (try_combine): Use is_a <scalar_int_mode> when
10619 trying to combine a full-register integer set with a subreg
10622 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10623 Alan Hayward <alan.hayward@arm.com>
10624 David Sherwood <david.sherwood@arm.com>
10626 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
10627 that is always either address_mode or pointer_mode.
10629 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10630 Alan Hayward <alan.hayward@arm.com>
10631 David Sherwood <david.sherwood@arm.com>
10633 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
10634 when the two are known to be equal.
10636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10637 Alan Hayward <alan.hayward@arm.com>
10638 David Sherwood <david.sherwood@arm.com>
10640 * simplify-rtx.c (simplify_const_unary_operation): Use
10641 is_a <scalar_int_mode> instead of checking for a nonzero
10642 precision. Forcibly convert op_mode to a scalar_int_mode
10643 in that case. More clearly differentiate the operand and
10644 result modes and use the former when deciding what the value
10645 of a count-bits operation should be. Use is_int_mode instead
10646 of checking for a MODE_INT. Remove redundant check for whether
10647 this mode has a zero precision.
10649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10650 Alan Hayward <alan.hayward@arm.com>
10651 David Sherwood <david.sherwood@arm.com>
10653 * optabs.c (widen_leading): Change the type of the mode argument
10654 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
10655 (widen_bswap): Likewise.
10656 (expand_parity): Likewise.
10657 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
10658 (expand_ffs): Likewise.
10659 (epand_unop): Check for scalar integer modes before calling the
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 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
10668 (expand_expr_real_1): Update call accordingly.
10670 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10671 Alan Hayward <alan.hayward@arm.com>
10672 David Sherwood <david.sherwood@arm.com>
10674 * expmed.c (store_bit_field_using_insv): Add op0_mode and
10675 value_mode arguments. Use scalar_int_mode internally.
10676 (store_bit_field_1): Rename the new integer mode from imode
10677 to op0_mode and use it instead of GET_MODE (op0). Update calls
10678 to store_split_bit_field, store_bit_field_using_insv and
10679 store_fixed_bit_field.
10680 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
10681 Use scalar_int_mode internally. Use a bit count rather than a mode
10682 when calculating the largest bit size for get_best_mode.
10683 Update calls to store_split_bit_field and store_fixed_bit_field_1.
10684 (store_fixed_bit_field_1): Add mode and value_mode arguments.
10685 Remove assertion that OP0 has a scalar integer mode.
10686 (store_split_bit_field): Add op0_mode and value_mode arguments.
10687 Update calls to extract_fixed_bit_field.
10688 (extract_bit_field_using_extv): Add an op0_mode argument.
10689 Use scalar_int_mode internally.
10690 (extract_bit_field_1): Rename the new integer mode from imode to
10691 op0_mode and use it instead of GET_MODE (op0). Update calls to
10692 extract_split_bit_field, extract_bit_field_using_extv and
10693 extract_fixed_bit_field.
10694 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
10695 to extract_split_bit_field and extract_fixed_bit_field_1.
10696 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
10697 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
10698 on the target mode.
10699 (extract_split_bit_field): Add an op0_mode argument. Update call
10700 to extract_fixed_bit_field.
10702 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10703 Alan Hayward <alan.hayward@arm.com>
10704 David Sherwood <david.sherwood@arm.com>
10706 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
10707 * explow.c (hard_function_value): Likewise.
10708 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
10709 convert_to_mode call outside the loop.
10710 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
10711 for the mode iterator. Require the mode specified by max_pieces
10713 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
10715 (copy_blkmode_to_reg): Likewise.
10716 (set_storage_via_setmem): Likewise.
10717 * optabs.c (prepare_cmp_insn): Likewise.
10718 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10719 * stor-layout.c (finish_bitfield_representative): Likewise.
10721 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10722 Alan Hayward <alan.hayward@arm.com>
10723 David Sherwood <david.sherwood@arm.com>
10725 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
10726 * expr.c (convert_move): Use them.
10727 (convert_modes): Likewise.
10728 (store_expr_with_bounds): Likewise.
10730 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10731 Alan Hayward <alan.hayward@arm.com>
10732 David Sherwood <david.sherwood@arm.com>
10734 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
10735 parameter for the mode of "x". Remove the "known_x", "known_mode"
10736 and "known_ret" arguments. Change the type of the mode argument
10737 to scalar_int_mode.
10738 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
10739 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
10740 (reg_num_sign_bit_copies_for_combine): Likewise.
10741 * rtlanal.c (nonzero_bits1): Likewise.
10742 (num_sign_bit_copies1): Likewise.
10743 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
10744 (reg_num_sign_bit_copies_general): Likewise.
10745 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
10746 (reg_nonzero_bits_general): Likewise.
10748 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10749 Alan Hayward <alan.hayward@arm.com>
10750 David Sherwood <david.sherwood@arm.com>
10752 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
10753 than in subroutines. Return 1 for non-integer modes.
10754 (cached_num_sign_bit_copies): Change the type of the mode parameter
10755 to scalar_int_mode.
10756 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
10757 classes. Handle CONST_INT_P first and then check whether X also
10758 has a scalar integer mode. Check the same thing for inner registers
10759 of a SUBREG and for values that are being extended or truncated.
10761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10762 Alan Hayward <alan.hayward@arm.com>
10763 David Sherwood <david.sherwood@arm.com>
10765 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
10766 in subroutines. Return the mode mask for non-integer modes.
10767 (cached_nonzero_bits): Change the type of the mode parameter
10768 to scalar_int_mode.
10769 (nonzero_bits1): Likewise. Remove early exit for other mode
10770 classes. Handle CONST_INT_P first and then check whether X
10771 also has a scalar integer mode.
10773 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10774 Alan Hayward <alan.hayward@arm.com>
10775 David Sherwood <david.sherwood@arm.com>
10777 * expr.c (widest_int_mode_for_size): Make the comment match the code.
10778 Return a scalar_int_mode and assert that the size is greater than
10780 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
10781 (op_by_pieces_d::op_by_pieces_d): Likewise.
10782 (op_by_pieces_d::run): Likewise.
10783 (can_store_by_pieces): Likewise.
10785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10786 Alan Hayward <alan.hayward@arm.com>
10787 David Sherwood <david.sherwood@arm.com>
10789 * combine.c (extract_left_shift): Add a mode argument and update
10791 (make_compound_operation_int): Change the type of the mode parameter
10792 to scalar_int_mode and update the call to extract_left_shift.
10794 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10795 Alan Hayward <alan.hayward@arm.com>
10796 David Sherwood <david.sherwood@arm.com>
10798 * combine.c (simplify_and_const_int): Change the type of the mode
10799 parameter to scalar_int_mode.
10800 (simplify_and_const_int_1): Likewise. Update recursive call.
10802 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10803 Alan Hayward <alan.hayward@arm.com>
10804 David Sherwood <david.sherwood@arm.com>
10806 * combine.c (simplify_compare_const): Check that the mode is a
10807 scalar_int_mode (rather than VOIDmode) before testing its
10809 (simplify_comparison): Move COMPARISON_P handling out of the
10810 loop and restrict the latter part of the loop to scalar_int_modes.
10811 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
10812 and when considering SUBREG_REGs. Use is_int_mode instead of
10813 checking GET_MODE_CLASS against MODE_INT.
10815 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10816 Alan Hayward <alan.hayward@arm.com>
10817 David Sherwood <david.sherwood@arm.com>
10819 * combine.c (try_widen_shift_mode): Move check for equal modes to...
10820 (simplify_shift_const_1): ...here. Use scalar_int_mode for
10821 shift_unit_mode and for modes involved in scalar shifts.
10823 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10824 Alan Hayward <alan.hayward@arm.com>
10825 David Sherwood <david.sherwood@arm.com>
10827 * combine.c (force_int_to_mode): New function, split out from...
10828 (force_to_mode): ...here. Keep xmode up-to-date and use it
10829 instead of GET_MODE (x).
10831 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10832 Alan Hayward <alan.hayward@arm.com>
10833 David Sherwood <david.sherwood@arm.com>
10835 * optabs-query.h (extraction_insn::struct_mode): Change type to
10836 opt_scalar_int_mode and update comment.
10837 (extraction_insn::field_mode): Change type to scalar_int_mode.
10838 (extraction_insn::pos_mode): Likewise.
10839 * combine.c (make_extraction): Update accordingly.
10840 * optabs-query.c (get_traditional_extraction_insn): Likewise.
10841 (get_optab_extraction_insn): Likewise.
10842 * recog.c (simplify_while_replacing): Likewise.
10843 * expmed.c (narrow_bit_field_mem): Change the type of the mode
10844 parameter to opt_scalar_int_mode.
10846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10847 Alan Hayward <alan.hayward@arm.com>
10848 David Sherwood <david.sherwood@arm.com>
10850 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
10851 to a scalar_int_mode instead of a machine_mode.
10852 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
10853 (get_best_mode): Return a boolean and use a pointer argument to store
10854 the selected mode. Replace the limit mode parameter with a bit limit.
10855 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
10856 for the values returned by bit_field_mode_iterator::next_mode.
10857 (store_bit_field): Update call to get_best_mode.
10858 (store_fixed_bit_field): Likewise.
10859 (extract_fixed_bit_field): Likewise.
10860 * expr.c (optimize_bitfield_assignment_op): Likewise.
10861 * fold-const.c (optimize_bit_field_compare): Likewise.
10862 (fold_truth_andor_1): Likewise.
10863 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
10864 Update for new type of m_mode.
10865 (get_best_mode): As above.
10867 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10868 Alan Hayward <alan.hayward@arm.com>
10869 David Sherwood <david.sherwood@arm.com>
10871 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
10872 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
10873 (store_bit_field): Check is_a <scalar_int_mode> before calling
10874 strict_volatile_bitfield_p.
10875 (extract_bit_field): Likewise.
10877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10878 Alan Hayward <alan.hayward@arm.com>
10879 David Sherwood <david.sherwood@arm.com>
10881 * target.def (cstore_mode): Return a scalar_int_mode.
10882 * doc/tm.texi: Regenerate.
10883 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
10884 * targhooks.h (default_cstore_mode): Likewise.
10885 * targhooks.c (default_cstore_mode): Likewise, using a forced
10887 * expmed.c (emit_cstore): Expect the target of the cstore to be
10890 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10891 Alan Hayward <alan.hayward@arm.com>
10892 David Sherwood <david.sherwood@arm.com>
10894 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
10896 (niter_desc): Likewise mode.
10897 (iv_analyze): Add a mode parameter.
10899 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
10900 and change its type to scalar_int_mode.
10901 * loop-iv.c: Update commentary at head of file.
10902 (iv_constant): Pass the mode paraeter before the rtx it describes
10903 and change its type to scalar_int_mode. Remove VOIDmode handling.
10904 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
10905 (iv_extend): Likewise.
10906 (shorten_into_mode): Likewise.
10907 (iv_add): Use scalar_int_mode.
10908 (iv_mult): Likewise.
10909 (iv_shift): Likewise.
10910 (canonicalize_iv_subregs): Likewise.
10911 (get_biv_step_1): Pass the outer_mode parameter before the rtx
10912 it describes and change its mode to scalar_int_mode. Also change
10913 the type of the returned inner_mode to scalar_int_mode.
10914 (get_biv_step): Likewise, turning outer_mode from a pointer
10915 into a direct parameter. Update call to get_biv_step_1.
10916 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
10917 iv_constant and get_biv_step.
10918 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
10919 and change its type to scalar_int_mode. Don't initialise iv->mode
10920 to VOIDmode and remove later checks for its still being VOIDmode.
10921 Update calls to iv_analyze_op and iv_analyze_expr. Check
10922 is_a <scalar_int_mode> when changing the mode under consideration.
10923 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
10924 Update call to iv_analyze_expr.
10925 (iv_analyze_op): Add a mode parameter. Reject subregs whose
10926 inner register is not also a scalar_int_mode. Update call to
10928 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
10929 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
10930 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
10931 separate mode class checks. Update calls to iv_analyze. Remove
10932 fix-up of VOIDmodes after iv_analyze_biv.
10933 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
10934 don't have a scalar_int_mode. Update call to biv_p.
10936 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10937 Alan Hayward <alan.hayward@arm.com>
10938 David Sherwood <david.sherwood@arm.com>
10940 * cfgexpand.c (convert_debug_memory_address): Use
10941 as_a <scalar_int_mode>.
10942 * combine.c (expand_compound_operation): Likewise.
10943 (make_extraction): Likewise.
10944 (change_zero_ext): Likewise.
10945 (simplify_comparison): Likewise.
10946 * cse.c (cse_insn): Likewise.
10947 * dwarf2out.c (minmax_loc_descriptor): Likewise.
10948 (mem_loc_descriptor): Likewise.
10949 (loc_descriptor): Likewise.
10950 * expmed.c (init_expmed_one_mode): Likewise.
10951 (synth_mult): Likewise.
10952 (emit_store_flag_1): Likewise.
10953 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
10954 of a comparison with size.
10955 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
10956 (reduce_to_bit_field_precision): Likewise.
10957 * function.c (expand_function_end): Likewise.
10958 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
10959 * loop-doloop.c (doloop_modify): Likewise.
10960 * optabs.c (expand_binop): Likewise.
10961 (expand_unop): Likewise.
10962 (expand_copysign_absneg): Likewise.
10963 (prepare_cmp_insn): Likewise.
10964 (maybe_legitimize_operand): Likewise.
10965 * recog.c (const_scalar_int_operand): Likewise.
10966 * rtlanal.c (get_address_mode): Likewise.
10967 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10968 (simplify_cond_clz_ctz): Likewise.
10969 * tree-nested.c (get_nl_goto_field): Likewise.
10970 * tree.c (build_vector_type_for_mode): Likewise.
10971 * var-tracking.c (use_narrower_mode): Likewise.
10973 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10974 Alan Hayward <alan.hayward@arm.com>
10975 David Sherwood <david.sherwood@arm.com>
10977 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
10978 * builtins.c (expand_builtin_signbit): Use it.
10979 * cfgexpand.c (expand_debug_expr): Likewise.
10980 * dojump.c (do_jump): Likewise.
10981 (do_compare_and_jump): Likewise.
10982 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
10983 * expmed.c (make_tree): Likewise.
10984 * expr.c (expand_expr_real_2): Likewise.
10985 (expand_expr_real_1): Likewise.
10986 (try_casesi): Likewise.
10987 * fold-const-call.c (fold_const_call_ss): Likewise.
10988 * fold-const.c (unextend): Likewise.
10989 (extract_muldiv_1): Likewise.
10990 (fold_single_bit_test): Likewise.
10991 (native_encode_int): Likewise.
10992 (native_encode_string): Likewise.
10993 (native_interpret_int): Likewise.
10994 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
10995 * internal-fn.c (expand_addsub_overflow): Likewise.
10996 (expand_neg_overflow): Likewise.
10997 (expand_mul_overflow): Likewise.
10998 (expand_arith_overflow): Likewise.
10999 * match.pd: Likewise.
11000 * stor-layout.c (layout_type): Likewise.
11001 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
11002 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
11003 * tree-ssanames.c (get_range_info): Likewise.
11004 * tree-switch-conversion.c (array_value_type) Likewise.
11005 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
11006 (vect_recog_divmod_pattern): Likewise.
11007 (vect_recog_mixed_size_cond_pattern): Likewise.
11008 * tree-vrp.c (extract_range_basic): Likewise.
11009 (simplify_float_conversion_using_ranges): Likewise.
11010 * tree.c (int_fits_type_p): Likewise.
11011 * ubsan.c (instrument_bool_enum_load): Likewise.
11012 * varasm.c (mergeable_string_section): Likewise.
11013 (narrowing_initializer_constant_valid_p): Likewise.
11014 (output_constant): Likewise.
11016 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11017 Alan Hayward <alan.hayward@arm.com>
11018 David Sherwood <david.sherwood@arm.com>
11020 * machmode.h (NARROWEST_INT_MODE): New macro.
11021 * expr.c (alignment_for_piecewise_move): Use it instead of
11022 GET_CLASS_NARROWEST_MODE (MODE_INT).
11023 (push_block): Likewise.
11024 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
11026 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11028 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11029 Alan Hayward <alan.hayward@arm.com>
11030 David Sherwood <david.sherwood@arm.com>
11032 * postreload.c (move2add_valid_value_p): Change the type of the
11033 mode parameter to scalar_int_mode.
11034 (move2add_use_add2_insn): Add a mode parameter and use it instead
11036 (move2add_use_add3_insn): Likewise.
11037 (reload_cse_move2add): Update accordingly.
11039 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11040 Alan Hayward <alan.hayward@arm.com>
11041 David Sherwood <david.sherwood@arm.com>
11043 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
11045 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
11046 * optabs.c (expand_unop): Likewise.
11048 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11049 Alan Hayward <alan.hayward@arm.com>
11050 David Sherwood <david.sherwood@arm.com>
11052 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
11053 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
11054 (popcount_loc_descriptor): Likewise.
11055 (bswap_loc_descriptor): Likewise.
11056 (rotate_loc_descriptor): Likewise.
11057 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
11058 calling the functions above.
11060 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11061 Alan Hayward <alan.hayward@arm.com>
11062 David Sherwood <david.sherwood@arm.com>
11064 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
11066 (try_combine): Likewise.
11067 (simplify_if_then_else): Likewise.
11068 * cse.c (cse_insn): Likewise.
11069 * dwarf2out.c (mem_loc_descriptor): Likewise.
11070 * emit-rtl.c (gen_lowpart_common): Likewise.
11071 * simplify-rtx.c (simplify_truncation): Likewise.
11072 (simplify_binary_operation_1): Likewise.
11073 (simplify_const_relational_operation): Likewise.
11074 (simplify_ternary_operation): Likewise.
11075 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
11077 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11078 Alan Hayward <alan.hayward@arm.com>
11079 David Sherwood <david.sherwood@arm.com>
11081 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
11082 * reload.c (push_reload): Likewise.
11083 (find_reloads): Likewise.
11085 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11086 Alan Hayward <alan.hayward@arm.com>
11087 David Sherwood <david.sherwood@arm.com>
11089 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
11090 (make_compound_operation_int): Likewise.
11091 (change_zero_ext): Likewise.
11092 * expr.c (convert_move): Likewise.
11093 (convert_modes): Likewise.
11094 * fwprop.c (forward_propagate_subreg): Likewise.
11095 * loop-iv.c (get_biv_step_1): Likewise.
11096 * optabs.c (widen_operand): Likewise.
11097 * postreload.c (move2add_valid_value_p): Likewise.
11098 * recog.c (simplify_while_replacing): Likewise.
11099 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11100 (simplify_binary_operation_1): Likewise. Remove redundant
11101 mode equality check.
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 * combine.c (combine_simplify_rtx): Add checks for
11108 is_a <scalar_int_mode>.
11109 (simplify_if_then_else): Likewise.
11110 (make_field_assignment): Likewise.
11111 (simplify_comparison): Likewise.
11112 * ifcvt.c (noce_try_bitop): Likewise.
11113 * loop-invariant.c (canonicalize_address_mult): Likewise.
11114 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11117 Alan Hayward <alan.hayward@arm.com>
11118 David Sherwood <david.sherwood@arm.com>
11120 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
11121 is_a <scalar_int_mode> instead of != BLKmode.
11123 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11124 Alan Hayward <alan.hayward@arm.com>
11125 David Sherwood <david.sherwood@arm.com>
11127 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
11128 instead of != VOIDmode.
11129 * combine.c (if_then_else_cond): Likewise.
11130 (change_zero_ext): Likewise.
11131 * dwarf2out.c (mem_loc_descriptor): Likewise.
11132 (loc_descriptor): Likewise.
11133 * rtlanal.c (canonicalize_condition): Likewise.
11134 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
11136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11137 Alan Hayward <alan.hayward@arm.com>
11138 David Sherwood <david.sherwood@arm.com>
11140 * simplify-rtx.c (simplify_binary_operation_1): Use
11141 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
11143 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11144 Alan Hayward <alan.hayward@arm.com>
11145 David Sherwood <david.sherwood@arm.com>
11147 * wide-int.h (int_traits<unsigned char>) New class.
11148 (int_traits<unsigned short>) Likewise.
11149 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
11150 Use GET_MODE_UNIT_PRECISION and remove redundant test for
11152 * combine.c (set_nonzero_bits_and_sign_copies): Use
11153 is_a <scalar_int_mode>.
11154 (find_split_point): Likewise.
11155 (combine_simplify_rtx): Likewise.
11156 (simplify_logical): Likewise.
11157 (expand_compound_operation): Likewise.
11158 (expand_field_assignment): Likewise.
11159 (make_compound_operation): Likewise.
11160 (extended_count): Likewise.
11161 (change_zero_ext): Likewise.
11162 (simplify_comparison): Likewise.
11163 * dwarf2out.c (scompare_loc_descriptor): Likewise.
11164 (ucompare_loc_descriptor): Likewise.
11165 (minmax_loc_descriptor): Likewise.
11166 (mem_loc_descriptor): Likewise.
11167 (loc_descriptor): Likewise.
11168 * expmed.c (init_expmed_one_mode): Likewise.
11169 * lra-constraints.c (lra_constraint_offset): Likewise.
11170 * optabs.c (prepare_libcall_arg): Likewise.
11171 * postreload.c (move2add_note_store): Likewise.
11172 * reload.c (operands_match_p): Likewise.
11173 * rtl.h (load_extend_op): Likewise.
11174 * rtlhooks.c (gen_lowpart_general): Likewise.
11175 * simplify-rtx.c (simplify_truncation): Likewise.
11176 (simplify_unary_operation_1): Likewise.
11177 (simplify_binary_operation_1): Likewise.
11178 (simplify_const_binary_operation): Likewise.
11179 (simplify_const_relational_operation): Likewise.
11180 (simplify_subreg): Likewise.
11181 * stor-layout.c (bitwise_mode_for_mode): Likewise.
11182 * var-tracking.c (adjust_mems): Likewise.
11183 (prepare_call_arguments): Likewise.
11185 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11186 Alan Hayward <alan.hayward@arm.com>
11187 David Sherwood <david.sherwood@arm.com>
11189 * machmode.h (is_int_mode): New fuction.
11190 * combine.c (find_split_point): Use it.
11191 (combine_simplify_rtx): Likewise.
11192 (simplify_if_then_else): Likewise.
11193 (simplify_set): Likewise.
11194 (simplify_shift_const_1): Likewise.
11195 (simplify_comparison): Likewise.
11196 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
11197 * cse.c (notreg_cost): Likewise.
11198 (cse_insn): Likewise.
11199 * cselib.c (cselib_lookup_1): Likewise.
11200 * dojump.c (do_jump_1): Likewise.
11201 (do_compare_rtx_and_jump): Likewise.
11202 * dse.c (get_call_args): Likewise.
11203 * dwarf2out.c (rtl_for_decl_init): Likewise.
11204 (native_encode_initializer): Likewise.
11205 * expmed.c (emit_store_flag_1): Likewise.
11206 (emit_store_flag): Likewise.
11207 * expr.c (convert_modes): Likewise.
11208 (store_field): Likewise.
11209 (expand_expr_real_1): Likewise.
11210 * fold-const.c (fold_read_from_constant_string): Likewise.
11211 * gimple-ssa-sprintf.c (get_format_string): Likewise.
11212 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
11213 * optabs.c (expand_binop): Likewise.
11214 (expand_unop): Likewise.
11215 (expand_abs_nojump): Likewise.
11216 (expand_one_cmpl_abs_nojump): Likewise.
11217 * simplify-rtx.c (mode_signbit_p): Likewise.
11218 (val_signbit_p): Likewise.
11219 (val_signbit_known_set_p): Likewise.
11220 (val_signbit_known_clear_p): Likewise.
11221 (simplify_relational_operation_1): Likewise.
11222 * tree.c (vector_type_mode): Likewise.
11224 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11225 Alan Hayward <alan.hayward@arm.com>
11226 David Sherwood <david.sherwood@arm.com>
11228 * machmode.h (smallest_mode_for_size): Fix formatting.
11229 (smallest_int_mode_for_size): New function.
11230 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
11231 instead of smallest_mode_for_size.
11232 * combine.c (make_extraction): Likewise.
11233 * config/arc/arc.c (arc_expand_movmem): Likewise.
11234 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
11235 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
11236 * config/s390/s390.c (s390_expand_insv): Likewise.
11237 * config/sparc/sparc.c (assign_int_registers): Likewise.
11238 * config/spu/spu.c (spu_function_value): Likewise.
11239 (spu_function_arg): Likewise.
11240 * coverage.c (get_gcov_type): Likewise.
11241 (get_gcov_unsigned_t): Likewise.
11242 * dse.c (find_shift_sequence): Likewise.
11243 * expmed.c (store_bit_field_1): Likewise.
11244 * expr.c (convert_move): Likewise.
11245 (store_field): Likewise.
11246 * internal-fn.c (expand_arith_overflow): Likewise.
11247 * optabs-query.c (get_best_extraction_insn): Likewise.
11248 * optabs.c (expand_twoval_binop_libfunc): Likewise.
11249 * stor-layout.c (layout_type): Likewise.
11250 (initialize_sizetypes): Likewise.
11251 * targhooks.c (default_get_mask_mode): Likewise.
11252 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
11254 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11255 Alan Hayward <alan.hayward@arm.com>
11256 David Sherwood <david.sherwood@arm.com>
11258 * machmode.h (opt_mode::else_blk): New function.
11259 (int_mode_for_mode): Declare.
11260 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
11261 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
11263 * cfgexpand.c (expand_debug_expr): Likewise.
11264 * combine.c (gen_lowpart_or_truncate): Likewise.
11265 (gen_lowpart_for_combine): Likewise.
11266 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
11267 * config/avr/avr.c (avr_to_int_mode): Likewise.
11268 (avr_out_plus_1): Likewise.
11269 (avr_out_plus): Likewise.
11270 (avr_out_round): Likewise.
11271 * config/i386/i386.c (ix86_split_to_parts): Likewise.
11272 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
11273 (s390_expand_vcond): Likewise.
11274 * config/spu/spu.c (spu_split_immediate): Likewise.
11275 (spu_expand_mov): Likewise.
11276 * dse.c (get_stored_val): Likewise.
11277 * expmed.c (store_bit_field_1): Likewise.
11278 (convert_extracted_bit_field): Use int_mode_for_mode instead of
11280 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
11281 (extract_low_bits): Likewise.
11282 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
11283 handling rather than repeating the check.
11284 (emit_group_store): Likewise.
11285 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
11286 * optabs.c (expand_absneg_bit): Likewise.
11287 (expand_copysign_absneg): Likewise.
11288 (expand_copysign_bit): Likewise.
11289 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
11290 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
11291 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
11292 * var-tracking.c (prepare_call_arguments): Likewise.
11293 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
11294 int_mode_for_mode instead of mode_for_size.
11295 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
11297 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11298 Alan Hayward <alan.hayward@arm.com>
11299 David Sherwood <david.sherwood@arm.com>
11301 * machmode.h (int_mode_for_size): New function.
11302 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
11303 instead of mode_for_size.
11304 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
11306 * combine.c (expand_field_assignment): Use int_mode_for_size
11307 instead of mode_for_size.
11308 (make_extraction): Likewise.
11309 (simplify_shift_const_1): Likewise.
11310 (simplify_comparison): Likewise.
11311 * dojump.c (do_jump): Likewise.
11312 * dwarf2out.c (mem_loc_descriptor): Likewise.
11313 * emit-rtl.c (init_derived_machine_modes): Likewise.
11314 * expmed.c (flip_storage_order): Likewise.
11315 (convert_extracted_bit_field): Likewise.
11316 * expr.c (copy_blkmode_from_reg): Likewise.
11317 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
11318 * internal-fn.c (expand_mul_overflow): Likewise.
11319 * lower-subreg.c (simple_move): Likewise.
11320 * optabs-libfuncs.c (init_optabs): Likewise.
11321 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11322 * tree.c (vector_type_mode): Likewise.
11323 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
11324 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
11325 * tree-vect-generic.c (expand_vector_parallel): Likewise.
11326 * tree-vect-stmts.c (vectorizable_load): Likewise.
11327 (vectorizable_store): Likewise.
11329 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11330 Alan Hayward <alan.hayward@arm.com>
11331 David Sherwood <david.sherwood@arm.com>
11333 * coretypes.h (pod_mode): New type.
11334 (scalar_int_mode_pod): New typedef.
11335 * machmode.h (pod_mode): New class.
11336 (int_n_data_t::m): Change type to scalar_int_mode_pod.
11337 * genmodes.c (emit_mode_int_n): Update accordingly.
11338 * lower-subreg.h (target_lower_subreg): Change type to
11339 scalar_int_mode_pod.
11340 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
11341 scalar_int_mode_pod.
11343 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11344 Alan Hayward <alan.hayward@arm.com>
11345 David Sherwood <david.sherwood@arm.com>
11347 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
11348 machine_mode to scalar_int_mode.
11349 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
11350 (rs6000_option_override_internal): Remove cast to int.
11351 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
11352 machine_mode to scalar_int_mode.
11353 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
11354 (rs6000_option_override_internal): Remove cast to int.
11355 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
11356 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
11358 * config/s390/s390.c (s390_expand_builtin): Likewise.
11359 * coretypes.h (scalar_int_mode): New type.
11360 (opt_scalar_int_mode): New typedef.
11361 * machmode.h (scalar_int_mode): New class.
11362 (scalar_int_mode::includes_p): New function.
11363 (byte_mode): Change type to scalar_int_mode.
11364 (word_mode): Likewise.
11365 (ptr_mode): Likewise.
11366 * emit-rtl.c (byte_mode): Likewise.
11367 (word_mode): Likewise.
11368 (ptr_mode): Likewise.
11369 (init_derived_machine_modes): Update accordingly.
11370 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
11371 and MODE_PARTIAL_INT.
11372 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
11373 opt_scalar_int_mode.
11375 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11376 Alan Hayward <alan.hayward@arm.com>
11377 David Sherwood <david.sherwood@arm.com>
11379 * target.def (libgcc_floating_mode_supported_p): Take a
11381 * doc/tm.texi: Regenerate.
11382 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
11384 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11385 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
11388 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11389 Alan Hayward <alan.hayward@arm.com>
11390 David Sherwood <david.sherwood@arm.com>
11392 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
11393 * doc/tm.texi: Regenerate.
11394 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
11395 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
11396 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
11397 * targhooks.h (default_floatn_mode): Likewise.
11398 * targhooks.c (default_floatn_mode): Likewise.
11399 * tree.c (build_common_tree_nodes): Update accordingly.
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 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
11406 (mode_iterator::iterate_p): Likewise.
11407 (mode_iterator::get_wider): Likewise.
11408 * expr.c (init_expr_target): Use opt_scalar_float_mode.
11410 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11411 Alan Hayward <alan.hayward@arm.com>
11412 David Sherwood <david.sherwood@arm.com>
11414 * coretypes.h (opt_scalar_float_mode): New typedef.
11415 * machmode.h (float_mode_for_size): New function.
11416 * emit-rtl.c (double_mode): Delete.
11417 (init_emit_once): Use float_mode_for_size.
11418 * stor-layout.c (layout_type): Likewise.
11419 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
11421 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11422 Alan Hayward <alan.hayward@arm.com>
11423 David Sherwood <david.sherwood@arm.com>
11425 * output.h (assemble_real): Take a scalar_float_mode.
11426 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
11427 * config/arm/arm.md (consttable_4): Likewise.
11428 (consttable_8): Likewise.
11429 (consttable_16): Likewise.
11430 * config/mips/mips.md (consttable_float): Likewise.
11431 * config/s390/s390.c (s390_output_pool_entry): Likewise.
11432 * varasm.c (assemble_real): Take a scalar_float_mode.
11433 (output_constant_pool_2): Update accordingly.
11434 (output_constant): Likewise.
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 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
11441 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
11442 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
11443 (native_encode_real): Likewise.
11444 (native_interpret_real): Likewise.
11445 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
11446 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11448 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11449 Alan Hayward <alan.hayward@arm.com>
11450 David Sherwood <david.sherwood@arm.com>
11452 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
11453 <scalar_float_mode>. Simplify.
11454 (gen_extend_conv_libfunc): Likewise.
11456 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11457 Alan Hayward <alan.hayward@arm.com>
11458 David Sherwood <david.sherwood@arm.com>
11460 * coretypes.h (scalar_float_mode): New type.
11461 * machmode.h (mode_traits::from_int): Use machine_mode if
11462 USE_ENUM_MODES is defined.
11463 (is_a): New function.
11465 (dyn_cast): Likewise.
11466 (scalar_float_mode): New class.
11467 (scalar_float_mode::includes_p): New function.
11468 (is_float_mode): Likewise.
11469 * gdbhooks.py (MachineModePrinter): New class.
11470 (build_pretty_printer): Use it for scalar_float_mode.
11471 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
11472 (format_helper::format_helper): Turn into a template.
11473 * genmodes.c (get_mode_class): New function.
11474 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
11475 or machine_mode if none.
11476 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
11477 as_a <scalar_float_mode>.
11478 * dwarf2out.c (mem_loc_descriptor): Likewise.
11479 (insert_float): Likewise.
11480 (add_const_value_attribute): Likewise.
11481 * simplify-rtx.c (simplify_immed_subreg): Likewise.
11482 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
11483 (expand_unop): Update accordingly.
11484 (expand_abs_nojump): Likewise.
11485 (expand_copysign_absneg): Take a scalar_float_mode.
11486 (expand_copysign_bit): Likewise.
11487 (expand_copysign): Update accordingly.
11489 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11490 Alan Hayward <alan.hayward@arm.com>
11491 David Sherwood <david.sherwood@arm.com>
11493 * coretypes.h (opt_mode): New class.
11494 * machmode.h (opt_mode): Likewise.
11495 (opt_mode::else_void): New function.
11496 (opt_mode::require): Likewise.
11497 (opt_mode::exists): Likewise.
11498 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
11499 (GET_MODE_2XWIDER_MODE): Likewise.
11500 (mode_iterator::get_wider): Update accordingly.
11501 (mode_iterator::get_2xwider): Likewise.
11502 (mode_iterator::get_known_wider): Likewise, turning into a template.
11503 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
11504 forcing a wider mode to exist.
11505 * config/cr16/cr16.h (LONG_REG_P): Likewise.
11506 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11507 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
11508 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
11509 * lower-subreg.c (init_lower_subreg): Likewise.
11510 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
11511 on the final iteration.
11512 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
11513 a wider mode exists before asking for a move pattern.
11514 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
11515 forcing a wider mode to exist.
11516 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
11517 returning false if no such mode exists.
11518 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
11519 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
11520 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
11521 Avoid checking for a MODE_INT if we already know the mode is not a
11523 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
11524 forcing a wider mode to exist.
11525 (expmed_mult_highpart_optab): Likewise.
11526 (expmed_mult_highpart): Likewise.
11527 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
11529 * lto-streamer-in.c (lto_input_mode_table): Likewise.
11530 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
11531 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
11532 * internal-fn.c (expand_mul_overflow): Update use of
11533 GET_MODE_2XWIDER_MODE.
11534 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11535 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
11536 GET_MODE_WIDER_MODE.
11537 (convert_plusminus_to_widen): Likewise.
11538 * tree-switch-conversion.c (array_value_type): Likewise.
11539 * var-tracking.c (emit_note_insn_var_location): Likewise.
11540 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
11541 Return false inside rather than outside the loop if no wider mode
11543 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
11544 and GET_MODE_2XWIDER_MODE
11545 (can_compare_p): Use else_void.
11546 * gdbhooks.py (OptMachineModePrinter): New class.
11547 (build_pretty_printer): Use it for opt_mode.
11549 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11550 Alan Hayward <alan.hayward@arm.com>
11551 David Sherwood <david.sherwood@arm.com>
11553 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
11554 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
11556 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11557 Alan Hayward <alan.hayward@arm.com>
11558 David Sherwood <david.sherwood@arm.com>
11560 * machmode.h (mode_traits): New structure.
11561 (get_narrowest_mode): New function.
11562 (mode_iterator::start): Likewise.
11563 (mode_iterator::iterate_p): Likewise.
11564 (mode_iterator::get_wider): Likewise.
11565 (mode_iterator::get_known_wider): Likewise.
11566 (mode_iterator::get_2xwider): Likewise.
11567 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
11568 (FOR_EACH_MODE): Likewise.
11569 (FOR_EACH_MODE_FROM): Likewise.
11570 (FOR_EACH_MODE_UNTIL): Likewise.
11571 (FOR_EACH_WIDER_MODE): Likewise.
11572 (FOR_EACH_2XWIDER_MODE): Likewise.
11573 * builtins.c (expand_builtin_strlen): Use new mode iterators.
11574 * combine.c (simplify_comparison): Likewise
11575 * config/i386/i386.c (type_natural_mode): Likewise.
11576 * cse.c (cse_insn): Likewise.
11577 * dse.c (find_shift_sequence): Likewise.
11578 * emit-rtl.c (init_derived_machine_modes): Likewise.
11579 (init_emit_once): Likewise.
11580 * explow.c (hard_function_value): Likewise.
11581 * expmed.c (extract_fixed_bit_field_1): Likewise.
11582 (extract_bit_field_1): Likewise.
11583 (expand_divmod): Likewise.
11584 (emit_store_flag_1): Likewise.
11585 * expr.c (init_expr_target): Likewise.
11586 (convert_move): Likewise.
11587 (alignment_for_piecewise_move): Likewise.
11588 (widest_int_mode_for_size): Likewise.
11589 (emit_block_move_via_movmem): Likewise.
11590 (copy_blkmode_to_reg): Likewise.
11591 (set_storage_via_setmem): Likewise.
11592 (compress_float_constant): Likewise.
11593 * omp-low.c (omp_clause_aligned_alignment): Likewise.
11594 * optabs-query.c (get_best_extraction_insn): Likewise.
11595 * optabs.c (expand_binop): Likewise.
11596 (expand_twoval_unop): Likewise.
11597 (expand_twoval_binop): Likewise.
11598 (widen_leading): Likewise.
11599 (widen_bswap): Likewise.
11600 (expand_parity): Likewise.
11601 (expand_unop): Likewise.
11602 (prepare_cmp_insn): Likewise.
11603 (prepare_float_lib_cmp): Likewise.
11604 (expand_float): Likewise.
11605 (expand_fix): Likewise.
11606 (expand_sfix_optab): Likewise.
11607 * postreload.c (move2add_use_add2_insn): Likewise.
11608 * reg-stack.c (reg_to_stack): Likewise.
11609 * reginfo.c (choose_hard_reg_mode): Likewise.
11610 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
11611 * stor-layout.c (mode_for_size): Likewise.
11612 (smallest_mode_for_size): Likewise.
11613 (mode_for_vector): Likewise.
11614 (finish_bitfield_representative): Likewise.
11615 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
11616 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
11617 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11618 * var-tracking.c (prepare_call_arguments): Likewise.
11620 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11621 Alan Hayward <alan.hayward@arm.com>
11622 David Sherwood <david.sherwood@arm.com>
11624 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
11625 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
11626 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
11627 * machmode.h (mode_size): Move earlier in file.
11628 (mode_precision): Likewise.
11629 (mode_inner): Likewise.
11630 (mode_nunits): Likewise.
11631 (mode_unit_size): Likewise.
11632 (unit_unit_precision): Likewise.
11633 (mode_wider): Likewise.
11634 (mode_2xwider): Likewise.
11635 (machine_mode): New class.
11636 (mode_to_bytes): New function.
11637 (mode_to_bits): Likewise.
11638 (mode_to_precision): Likewise.
11639 (mode_to_inner): Likewise.
11640 (mode_to_unit_size): Likewise.
11641 (mode_to_unit_precision): Likewise.
11642 (mode_to_nunits): Likewise.
11643 (GET_MODE_SIZE): Use mode_to_bytes.
11644 (GET_MODE_BITSIZE): Use mode_to_bits.
11645 (GET_MODE_PRECISION): Use mode_to_precision.
11646 (GET_MODE_INNER): Use mode_to_inner.
11647 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
11648 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
11649 (GET_MODE_NUNITS): Use mode_to_nunits.
11650 * system.h (ALWAYS_INLINE): New macro.
11651 * config/powerpcspe/powerpcspe-c.c
11652 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
11653 int for arg1_mode and arg2_mode.
11655 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11656 Alan Hayward <alan.hayward@arm.com>
11657 David Sherwood <david.sherwood@arm.com>
11659 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
11660 Prefix mode names with E_ in case statements.
11661 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11662 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
11663 (aarch64_split_simd_move): Likewise.
11664 (aarch64_gen_storewb_pair): Likewise.
11665 (aarch64_gen_loadwb_pair): Likewise.
11666 (aarch64_gen_store_pair): Likewise.
11667 (aarch64_gen_load_pair): Likewise.
11668 (aarch64_get_condition_code_1): Likewise.
11669 (aarch64_constant_pool_reload_icode): Likewise.
11670 (get_rsqrte_type): Likewise.
11671 (get_rsqrts_type): Likewise.
11672 (get_recpe_type): Likewise.
11673 (get_recps_type): Likewise.
11674 (aarch64_gimplify_va_arg_expr): Likewise.
11675 (aarch64_simd_container_mode): Likewise.
11676 (aarch64_emit_load_exclusive): Likewise.
11677 (aarch64_emit_store_exclusive): Likewise.
11678 (aarch64_expand_compare_and_swap): Likewise.
11679 (aarch64_gen_atomic_cas): Likewise.
11680 (aarch64_emit_bic): Likewise.
11681 (aarch64_emit_atomic_swap): Likewise.
11682 (aarch64_emit_atomic_load_op): Likewise.
11683 (aarch64_evpc_trn): Likewise.
11684 (aarch64_evpc_uzp): Likewise.
11685 (aarch64_evpc_zip): Likewise.
11686 (aarch64_evpc_ext): Likewise.
11687 (aarch64_evpc_rev): Likewise.
11688 (aarch64_evpc_dup): Likewise.
11689 (aarch64_gen_ccmp_first): Likewise.
11690 (aarch64_gen_ccmp_next): Likewise.
11691 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
11692 (alpha_emit_xfloating_libcall): Likewise.
11693 (emit_insxl): Likewise.
11694 (alpha_arg_type): Likewise.
11695 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
11696 (arc_preferred_simd_mode): Likewise.
11697 (arc_secondary_reload): Likewise.
11698 (get_arc_condition_code): Likewise.
11699 (arc_print_operand): Likewise.
11700 (arc_legitimate_constant_p): Likewise.
11701 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11702 * config/arc/arc.md (casesi_load): Likewise.
11703 (casesi_compact_jump): Likewise.
11704 * config/arc/predicates.md (proper_comparison_operator): Likewise.
11705 (cc_use_register): Likewise.
11706 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11707 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
11708 (arm_init_iwmmxt_builtins): Likewise.
11709 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
11710 (neon_expand_vector_init): Likewise.
11711 (arm_attr_length_move_neon): Likewise.
11712 (maybe_get_arm_condition_code): Likewise.
11713 (arm_emit_vector_const): Likewise.
11714 (arm_preferred_simd_mode): Likewise.
11715 (arm_output_iwmmxt_tinsr): Likewise.
11716 (thumb1_output_casesi): Likewise.
11717 (thumb2_output_casesi): Likewise.
11718 (arm_emit_load_exclusive): Likewise.
11719 (arm_emit_store_exclusive): Likewise.
11720 (arm_expand_compare_and_swap): Likewise.
11721 (arm_evpc_neon_vuzp): Likewise.
11722 (arm_evpc_neon_vzip): Likewise.
11723 (arm_evpc_neon_vrev): Likewise.
11724 (arm_evpc_neon_vtrn): Likewise.
11725 (arm_evpc_neon_vext): Likewise.
11726 (arm_validize_comparison): Likewise.
11727 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
11728 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
11729 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
11730 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
11731 (c6x_preferred_simd_mode): Likewise.
11732 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
11733 (epiphany_rtx_costs): Likewise.
11734 * config/epiphany/predicates.md (proper_comparison_operator):
11736 * config/frv/frv.c (condexec_memory_operand): Likewise.
11737 (frv_emit_move): Likewise.
11738 (output_move_single): Likewise.
11739 (output_condmove_single): Likewise.
11740 (frv_hard_regno_mode_ok): Likewise.
11741 (frv_matching_accg_mode): Likewise.
11742 * config/h8300/h8300.c (split_adds_subs): Likewise.
11743 (h8300_rtx_costs): Likewise.
11744 (h8300_print_operand): Likewise.
11745 (compute_mov_length): Likewise.
11746 (output_logical_op): Likewise.
11747 (compute_logical_op_length): Likewise.
11748 (compute_logical_op_cc): Likewise.
11749 (h8300_shift_needs_scratch_p): Likewise.
11750 (output_a_shift): Likewise.
11751 (compute_a_shift_length): Likewise.
11752 (compute_a_shift_cc): Likewise.
11753 (expand_a_rotate): Likewise.
11754 (output_a_rotate): Likewise.
11755 * config/i386/i386.c (classify_argument): Likewise.
11756 (function_arg_advance_32): Likewise.
11757 (function_arg_32): Likewise.
11758 (function_arg_64): Likewise.
11759 (function_value_64): Likewise.
11760 (ix86_gimplify_va_arg): Likewise.
11761 (ix86_legitimate_constant_p): Likewise.
11762 (put_condition_code): Likewise.
11763 (split_double_mode): Likewise.
11764 (ix86_avx256_split_vector_move_misalign): Likewise.
11765 (ix86_expand_vector_logical_operator): Likewise.
11766 (ix86_split_idivmod): Likewise.
11767 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
11768 (ix86_build_const_vector): Likewise.
11769 (ix86_build_signbit_mask): Likewise.
11770 (ix86_match_ccmode): Likewise.
11771 (ix86_cc_modes_compatible): Likewise.
11772 (ix86_expand_branch): Likewise.
11773 (ix86_expand_sse_cmp): Likewise.
11774 (ix86_expand_sse_movcc): Likewise.
11775 (ix86_expand_int_sse_cmp): Likewise.
11776 (ix86_expand_vec_perm_vpermi2): Likewise.
11777 (ix86_expand_vec_perm): Likewise.
11778 (ix86_expand_sse_unpack): Likewise.
11779 (ix86_expand_int_addcc): Likewise.
11780 (ix86_split_to_parts): Likewise.
11781 (ix86_vectorize_builtin_gather): Likewise.
11782 (ix86_vectorize_builtin_scatter): Likewise.
11783 (avx_vpermilp_parallel): Likewise.
11784 (inline_memory_move_cost): Likewise.
11785 (ix86_tieable_integer_mode_p): Likewise.
11786 (x86_maybe_negate_const_int): Likewise.
11787 (ix86_expand_vector_init_duplicate): Likewise.
11788 (ix86_expand_vector_init_one_nonzero): Likewise.
11789 (ix86_expand_vector_init_one_var): Likewise.
11790 (ix86_expand_vector_init_concat): Likewise.
11791 (ix86_expand_vector_init_interleave): Likewise.
11792 (ix86_expand_vector_init_general): Likewise.
11793 (ix86_expand_vector_set): Likewise.
11794 (ix86_expand_vector_extract): Likewise.
11795 (emit_reduc_half): Likewise.
11796 (ix86_emit_i387_round): Likewise.
11797 (ix86_mangle_type): Likewise.
11798 (ix86_expand_round_sse4): Likewise.
11799 (expand_vec_perm_blend): Likewise.
11800 (canonicalize_vector_int_perm): Likewise.
11801 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
11802 (expand_vec_perm_1): Likewise.
11803 (expand_vec_perm_interleave3): Likewise.
11804 (expand_vec_perm_even_odd_pack): Likewise.
11805 (expand_vec_perm_even_odd_1): Likewise.
11806 (expand_vec_perm_broadcast_1): Likewise.
11807 (ix86_vectorize_vec_perm_const_ok): Likewise.
11808 (ix86_expand_vecop_qihi): Likewise.
11809 (ix86_expand_mul_widen_hilo): Likewise.
11810 (ix86_expand_sse2_abs): Likewise.
11811 (ix86_expand_pextr): Likewise.
11812 (ix86_expand_pinsr): Likewise.
11813 (ix86_preferred_simd_mode): Likewise.
11814 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
11815 * config/i386/sse.md (*andnot<mode>3): Likewise.
11816 (<mask_codefor><code><mode>3<mask_name>): Likewise.
11817 (*<code><mode>3): Likewise.
11818 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
11819 (ia64_expand_atomic_op): Likewise.
11820 (ia64_arg_type): Likewise.
11821 (ia64_mode_to_int): Likewise.
11822 (ia64_scalar_mode_supported_p): Likewise.
11823 (ia64_vector_mode_supported_p): Likewise.
11824 (expand_vec_perm_broadcast): Likewise.
11825 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
11826 (iq2000_function_arg_advance): Likewise.
11827 (iq2000_function_arg): Likewise.
11828 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
11829 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
11830 (m68k_libcall_value): Likewise.
11831 (m68k_function_value): Likewise.
11832 (sched_attr_op_type): Likewise.
11833 * config/mcore/mcore.c (mcore_output_move): Likewise.
11834 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
11836 (microblaze_function_arg): Likewise.
11837 * config/mips/mips.c (mips16_build_call_stub): Likewise.
11838 (mips_print_operand): Likewise.
11839 (mips_mode_ok_for_mov_fmt_p): Likewise.
11840 (mips_vector_mode_supported_p): Likewise.
11841 (mips_preferred_simd_mode): Likewise.
11842 (mips_expand_vpc_loongson_even_odd): Likewise.
11843 (mips_expand_vec_unpack): Likewise.
11844 (mips_expand_vi_broadcast): Likewise.
11845 (mips_expand_vector_init): Likewise.
11846 (mips_expand_vec_reduc): Likewise.
11847 (mips_expand_msa_cmp): Likewise.
11848 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
11849 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
11850 (cc_flags_for_mode): Likewise.
11851 * config/msp430/msp430.c (msp430_print_operand): Likewise.
11852 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
11853 (nds32_output_casesi_pc_relative): Likewise.
11854 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11855 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
11856 (nvptx_gen_unpack): Likewise.
11857 (nvptx_gen_pack): Likewise.
11858 (nvptx_gen_shuffle): Likewise.
11859 (nvptx_gen_wcast): Likewise.
11860 (nvptx_preferred_simd_mode): Likewise.
11861 * config/pa/pa.c (pa_secondary_reload): Likewise.
11862 * config/pa/predicates.md (base14_operand): Likewise.
11863 * config/powerpcspe/powerpcspe-c.c
11864 (altivec_resolve_overloaded_builtin): Likewise.
11865 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
11867 (rs6000_preferred_simd_mode): Likewise.
11868 (output_vec_const_move): Likewise.
11869 (rs6000_expand_vector_extract): Likewise.
11870 (rs6000_split_vec_extract_var): Likewise.
11871 (reg_offset_addressing_ok_p): Likewise.
11872 (rs6000_legitimate_offset_address_p): Likewise.
11873 (rs6000_legitimize_address): Likewise.
11874 (rs6000_emit_set_const): Likewise.
11875 (rs6000_const_vec): Likewise.
11876 (rs6000_emit_move): Likewise.
11877 (spe_build_register_parallel): Likewise.
11878 (rs6000_darwin64_record_arg_recurse): Likewise.
11879 (swap_selector_for_mode): Likewise.
11880 (spe_init_builtins): Likewise.
11881 (paired_init_builtins): Likewise.
11882 (altivec_init_builtins): Likewise.
11883 (do_load_for_compare): Likewise.
11884 (rs6000_generate_compare): Likewise.
11885 (rs6000_expand_float128_convert): Likewise.
11886 (emit_load_locked): Likewise.
11887 (emit_store_conditional): Likewise.
11888 (rs6000_output_function_epilogue): Likewise.
11889 (rs6000_handle_altivec_attribute): Likewise.
11890 (rs6000_function_value): Likewise.
11891 (emit_fusion_gpr_load): Likewise.
11892 (emit_fusion_p9_load): Likewise.
11893 (emit_fusion_p9_store): Likewise.
11894 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
11895 (fusion_gpr_mem_load): Likewise.
11896 (fusion_addis_mem_combo_load): Likewise.
11897 (fusion_addis_mem_combo_store): Likewise.
11898 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
11899 (fusion_gpr_mem_load): Likewise.
11900 (fusion_addis_mem_combo_load): Likewise.
11901 (fusion_addis_mem_combo_store): Likewise.
11902 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11904 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
11905 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
11906 (rs6000_preferred_simd_mode): Likewise.
11907 (output_vec_const_move): Likewise.
11908 (rs6000_expand_vector_extract): Likewise.
11909 (rs6000_split_vec_extract_var): Likewise.
11910 (reg_offset_addressing_ok_p): Likewise.
11911 (rs6000_legitimate_offset_address_p): Likewise.
11912 (rs6000_legitimize_address): Likewise.
11913 (rs6000_emit_set_const): Likewise.
11914 (rs6000_const_vec): Likewise.
11915 (rs6000_emit_move): Likewise.
11916 (rs6000_darwin64_record_arg_recurse): Likewise.
11917 (swap_selector_for_mode): Likewise.
11918 (paired_init_builtins): Likewise.
11919 (altivec_init_builtins): Likewise.
11920 (rs6000_expand_float128_convert): Likewise.
11921 (emit_load_locked): Likewise.
11922 (emit_store_conditional): Likewise.
11923 (rs6000_output_function_epilogue): Likewise.
11924 (rs6000_handle_altivec_attribute): Likewise.
11925 (rs6000_function_value): Likewise.
11926 (emit_fusion_gpr_load): Likewise.
11927 (emit_fusion_p9_load): Likewise.
11928 (emit_fusion_p9_store): Likewise.
11929 * config/rx/rx.c (rx_gen_move_template): Likewise.
11930 (flags_from_mode): Likewise.
11931 * config/s390/predicates.md (s390_alc_comparison): Likewise.
11932 (s390_slb_comparison): Likewise.
11933 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
11934 (s390_vector_mode_supported_p): Likewise.
11935 (s390_cc_modes_compatible): Likewise.
11936 (s390_match_ccmode_set): Likewise.
11937 (s390_canonicalize_comparison): Likewise.
11938 (s390_emit_compare_and_swap): Likewise.
11939 (s390_branch_condition_mask): Likewise.
11940 (s390_rtx_costs): Likewise.
11941 (s390_secondary_reload): Likewise.
11942 (__SECONDARY_RELOAD_CASE): Likewise.
11943 (s390_expand_cs): Likewise.
11944 (s390_preferred_simd_mode): Likewise.
11945 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
11946 * config/sh/sh.c (sh_print_operand): Likewise.
11947 (dump_table): Likewise.
11948 (sh_secondary_reload): Likewise.
11949 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11950 * config/sh/sh.md (casesi_worker_1): Likewise.
11951 (casesi_worker_2): Likewise.
11952 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
11953 (fcc_comparison_operator): Likewise.
11954 * config/sparc/sparc.c (sparc_expand_move): Likewise.
11955 (emit_soft_tfmode_cvt): Likewise.
11956 (sparc_preferred_simd_mode): Likewise.
11957 (output_cbranch): Likewise.
11958 (sparc_print_operand): Likewise.
11959 (sparc_expand_vec_perm_bmask): Likewise.
11960 (vector_init_bshuffle): Likewise.
11961 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
11962 (spu_vector_mode_supported_p): Likewise.
11963 (spu_expand_insv): Likewise.
11964 (spu_emit_branch_or_set): Likewise.
11965 (spu_handle_vector_attribute): Likewise.
11966 (spu_builtin_splats): Likewise.
11967 (spu_builtin_extract): Likewise.
11968 (spu_builtin_promote): Likewise.
11969 (spu_expand_sign_extend): Likewise.
11970 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
11971 (tilegx_simd_int): Likewise.
11972 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
11973 (tilepro_simd_int): Likewise.
11974 * config/v850/v850.c (const_double_split): Likewise.
11975 (v850_print_operand): Likewise.
11976 (ep_memory_offset): Likewise.
11977 * config/vax/vax.c (vax_rtx_costs): Likewise.
11978 (vax_output_int_move): Likewise.
11979 (vax_output_int_add): Likewise.
11980 (vax_output_int_subtract): Likewise.
11981 * config/visium/predicates.md (visium_branch_operator): Likewise.
11982 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
11983 (visium_print_operand_address): Likewise.
11984 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
11985 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
11986 (xtensa_expand_conditional_branch): Likewise.
11987 (xtensa_copy_incoming_a7): Likewise.
11988 (xtensa_output_literal): Likewise.
11989 * dfp.c (decimal_real_maxval): Likewise.
11990 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
11992 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
11993 Alan Hayward <alan.hayward@arm.com>
11994 David Sherwood <david.sherwood@arm.com>
11996 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
11997 (mode_nunits_inline): Likewise.
11998 (mode_inner_inline): Likewise.
11999 (mode_unit_size_inline): Likewise.
12000 (mode_unit_precision_inline): Likewise.
12001 (emit_insn_modes_h): Likewise. Also emit a #define of the
12003 (emit_mode_wider): Add an E_ prefix to mode names.
12004 (emit_mode_complex): Likewise.
12005 (emit_mode_inner): Likewise.
12006 (emit_mode_adjustments): Likewise.
12007 (emit_mode_int_n): Likewise.
12008 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
12009 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
12010 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
12011 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
12012 (CRC32_BUILTIN, ENTRY): Likewise.
12013 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
12014 (aarch64_pop_regs): Likewise.
12015 (aarch64_process_components): Likewise.
12016 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
12017 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
12018 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
12019 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
12020 * config/arm/arm.c (arm_init_libfuncs): Likewise.
12021 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
12023 * config/i386/i386-builtin.def (pcmpestr): Likewise.
12024 (pcmpistr): Likewise.
12025 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
12026 * config/mmix/mmix.c (mmix_output_condition): Likewise.
12027 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
12029 * config/rl78/rl78.c (mduc_regs): Likewise.
12030 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12031 (htm_expand_builtin): Likewise.
12032 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
12033 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
12034 * config/xtensa/xtensa.c (print_operand): Likewise.
12035 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
12036 (NUM_MODE_VECTOR_INT): Likewise.
12037 * genoutput.c (null_operand): Likewise.
12038 (output_operand_data): Likewise.
12039 * genrecog.c (print_parameter_value): Likewise.
12040 * lra.c (debug_operand_data): Likewise.
12042 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12044 * dwarf2out.c (scompare_loc_descriptor_wide)
12045 (scompare_loc_descriptor_narrow): New functions, split out from...
12046 (scompare_loc_descriptor): ...here.
12047 * expmed.c (emit_store_flag_int): New function, split out from...
12048 (emit_store_flag): ...here.
12050 2017-08-30 Richard Biener <rguenther@suse.de>
12052 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
12053 (dwarf2out_early_finish): Move setting of AT_pubnames from
12054 early debug output to early finish.
12056 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12058 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
12059 and -mdata-region to the assembler.
12061 2017-08-30 Richard Biener <rguenther@suse.de>
12063 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
12065 (gen_subprogram_die): Add DW_AT_object_pointer only early.
12066 (dwarf2out_early_global_decl): Only generate a DIE for the
12067 abstract origin if it doesn't already exist or is a declaration DIE.
12068 (resolve_addr): Do not add the linkage name twice when
12069 generating a stub DIE for the DW_TAG_GNU_call_site target.
12071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
12073 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12074 Use machine_mode rather than int for arg1_mode.
12076 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
12079 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
12080 that the second argument of the built-in functions to unpack
12081 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
12082 switch statement instead a lot of if statements.
12083 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
12084 Allow 64-bit values to be in Altivec registers as well as
12085 traditional floating point registers.
12086 (pack<mode>, FMOVE128_VSX iterator): Likewise.
12088 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
12090 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
12091 MAX_REGS_PER_ADDRESS == 1.
12093 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
12095 * config/i386/i386.opt (flag_fentry): Do not init to -1.
12096 * config/i386/i386.c (ix86_option_override_internal): Simplify
12097 setting of opts->x_flag_entry.
12099 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12100 Jakub Jelinek <jakub@redhat.com>
12101 Richard Biener <rguenther@suse.de>
12103 PR tree-optimization/81503
12104 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
12105 folded constant fits in the target type; reorder tests for clarity.
12107 2017-08-29 Martin Liska <mliska@suse.cz>
12109 * passes.def: Include pass_lower_switch.
12110 * stmt.c (dump_case_nodes): Remove and move to
12111 tree-switch-conversion.
12112 (case_values_threshold): Likewise.
12113 (expand_switch_as_decision_tree_p): Likewise.
12114 (emit_case_decision_tree): Likewise.
12115 (expand_case): Likewise.
12116 (balance_case_nodes): Likewise.
12117 (node_has_low_bound): Likewise.
12118 (node_has_high_bound): Likewise.
12119 (node_is_bounded): Likewise.
12120 (emit_case_nodes): Likewise.
12121 (struct simple_case_node): New struct.
12122 (add_case_node): Remove.
12123 (emit_case_dispatch_table): Use vector instead of case_list.
12124 (reset_out_edges_aux): Remove.
12125 (compute_cases_per_edge): Likewise.
12126 (expand_case): Build list of simple_case_node.
12127 (expand_sjlj_dispatch_table): Use it.
12128 * tree-switch-conversion.c (struct case_node): Moved from
12129 stmt.c and adjusted.
12130 (emit_case_nodes): Likewise.
12131 (node_has_low_bound): Likewise.
12132 (node_has_high_bound): Likewise.
12133 (node_is_bounded): Likewise.
12134 (case_values_threshold): Likewise.
12135 (reset_out_edges_aux): Likewise.
12136 (compute_cases_per_edge): Likewise.
12137 (add_case_node): Likewise.
12138 (dump_case_nodes): Likewise.
12139 (balance_case_nodes): Likewise.
12140 (expand_switch_as_decision_tree_p): Likewise.
12141 (emit_jump): Likewise.
12142 (emit_case_decision_tree): Likewise.
12143 (try_switch_expansion): Likewise.
12144 (do_jump_if_equal): Likewise.
12145 (emit_cmp_and_jump_insns): Likewise.
12146 (fix_phi_operands_for_edge): New function.
12147 (record_phi_operand_mapping): Likewise.
12148 (class pass_lower_switch): New pass.
12149 (pass_lower_switch::execute): New function.
12150 (make_pass_lower_switch): Likewise.
12151 (conditional_probability):
12152 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
12153 * tree-pass.h: Add make_pass_lower_switch.
12155 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
12158 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
12161 2017-08-29 Richard Biener <rguenther@suse.de>
12163 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
12164 we do not add a DW_AT_inline attribute twice.
12165 (gen_subprogram_die): Remove code setting DW_AT_inline on
12166 DECL_ABSTRACT_P nodes.
12168 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
12170 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
12171 calls to internal functions.
12172 (gimplify_modify_expr): Likewise.
12173 * tree-call-cdce.c (use_internal_fn): Likewise.
12174 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
12175 (convert_to_divmod): Set the nothrow flag.
12176 * tree-if-conv.c (predicate_mem_writes): Likewise.
12177 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
12178 (vectorizable_call): Likewise.
12179 (vectorizable_store): Likewise.
12180 (vectorizable_load): Likewise.
12181 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
12182 (vect_recog_mask_conversion_pattern): Likewise.
12184 2017-08-29 Martin Liska <mliska@suse.cz>
12187 * gcc.c (driver_handle_option): Add new argument.
12188 * opts-common.c (handle_option): Pass
12189 target_option_override_hook.
12190 * opts-global.c (lang_handle_option): Add new option.
12191 (set_default_handlers): Add new argument.
12192 (decode_options): Likewise.
12193 * opts.c (target_handle_option): Likewise.
12194 (common_handle_option): Call target_option_override_hook.
12195 * opts.h (struct cl_option_handler_func): Add hook for
12196 target option override.
12197 (struct cl_option_handlers): Likewise.
12198 (set_default_handlers): Add new argument.
12199 (decode_options): Likewise.
12200 (common_handle_option): Likewise.
12201 (target_handle_option): Likewise.
12202 * toplev.c (toplev::main): Pass targetm.target_option.override
12205 2017-08-29 Richard Biener <rguenther@suse.de>
12206 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
12208 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
12209 life to the active subtree.
12211 2017-08-28 Jeff Law <law@redhat.com>
12213 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
12214 derive_equivalences.
12215 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
12216 Code moved into....
12217 (edge_info::derive_equivalences): New private member function
12219 * tree-ssa-dom.c (class edge_info): Changed from a struct
12220 to a class. Add ctor/dtor, methods and data members.
12221 (edge_info::edge_info): Renamed from allocate_edge_info.
12222 Initialize additional members.
12223 (edge_info::~edge_info): New.
12224 (free_dom_edge_info): Delete the edge info.
12225 (record_edge_info): Use new class & associated member functions.
12226 Tighten forms for testing for edge equivalences.
12227 (record_temporary_equivalences): Iterate over the simple
12228 equivalences rather than assuming there's only one per edge.
12229 (cprop_into_successor_phis): Iterate over the simple
12230 equivalences rather than assuming there's only one per edge.
12231 (optimize_stmt): Use operand_equal_p rather than pointer
12232 equality for mini-DSE code.
12234 2017-08-28 Nathan Sidwell <nathan@acm.org>
12236 * gcc.c (execute): Fold SIGPIPE handling into switch
12237 statement. Adjust internal error message.
12239 2017-08-28 Richard Biener <rguenther@suse.de>
12242 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
12243 Do nothing for removed DIEs.
12245 2017-08-28 Richard Biener <rguenther@suse.de>
12247 PR tree-optimization/81977
12248 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
12251 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
12254 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
12255 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
12256 using targetm.gen_mem_thread_fence.
12258 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
12261 * config/i386/i386.md (*<btsc><mode>): Change operand 2
12262 predicate to register_operand. Reorder operands.
12263 (*btr<mode>): Ditto.
12264 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
12265 (*btr<mode>_mask): Ditto.
12267 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
12269 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
12270 * config/rs6000/xmmintrin.h: New file.
12271 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
12273 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12276 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
12277 parameter and_insn and return it.
12278 (recombine_lvx_pattern): Insert a copy to ensure availability of
12279 the base register of the copied masking operation at the point of
12280 the instruction replacement.
12281 (recombine_stvx_pattern): Likewise.
12283 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
12285 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
12286 undocumented switches.
12287 (-mpower9-dform-vector): Likewise.
12288 (-mpower9-dform): Likewise.
12289 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
12290 comments to delete references to -mpower9-dform* switches.
12291 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
12292 Delete reference to -mpower9-dform* switches, test for
12293 -mpower9-vector instead.
12294 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
12295 (OTHER_P9_VECTOR_MASKS): Likewise.
12296 (POWERPC_MASKS): Likewise.
12297 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
12298 tests against -mpower9-dform* to -mpower9-vector. Delete code
12299 that checked for -mpower9-dform* consistancy with other options.
12300 Add test for -mpower9-misc to enable other power9 switches.
12301 (rs6000_init_hard_regno_mode_ok): Likewise.
12302 (rs6000_option_override_internal): Likewise.
12303 (rs6000_emit_prologue): Likewise.
12304 (rs6000_emit_epilogue): Likewise.
12305 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
12306 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
12307 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
12309 (emit_fusion_p9_store): Likewise.
12310 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
12311 resetting these macros if the assembler does not support ISA 3.0
12313 (TARGET_P9_DFORM_VECTOR): Likewise.
12314 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
12315 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
12317 2017-08-25 Alan Modra <amodra@gmail.com>
12319 PR rtl-optimization/81747
12320 * cse.c (cse_extended_basic_block): Don't attempt to record
12321 equivalences for degenerate conditional branches that branch
12322 to their fall-through.
12324 2017-08-24 Martin Sebor <msebor@redhat.com>
12326 PR middle-end/81908
12327 * gimple-fold.c (size_must_be_zero_p): New function.
12328 (gimple_fold_builtin_memory_op): Call it.
12330 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
12332 * config/rs6000/mm_malloc.h: New file.
12334 2017-08-24 Bin Cheng <bin.cheng@arm.com>
12336 PR tree-optimization/81913
12337 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
12338 analysis when either IVs in condition can wrap.
12340 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
12342 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
12343 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
12345 2017-08-24 Richard Biener <rguenther@suse.de>
12348 * targhooks.c (default_target_can_inline_p): Properly
12349 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
12350 is present and always compare.
12351 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
12352 infer -mfpmath=sse from TARGET_SSE_P.
12353 (ix86_can_inline_p): Properly use target_option_default_node when
12354 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
12356 2017-08-24 Richard Biener <rguenther@suse.de>
12359 * dwarf2out.c (output_die): Handle flag_generate_offload like
12361 (output_comp_unit): Likewise.
12362 (gen_array_type_die): Likewise.
12363 (dwarf2out_early_finish): Likewise.
12364 (note_variable_value_in_expr): Likewise.
12365 (dwarf2out_finish): Likewise. Adjust assert.
12366 * cgraphunit.c (symbol_table::compile): Move setting of
12367 flag_generate_offload earlier ...
12368 (symbol_table::finalize_compilation_unit): ... here, before
12369 early debug finalization.
12371 2017-08-24 Richard Biener <rguenther@suse.de>
12373 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
12374 and ipa-fnsummary.h.
12375 (ix86_can_inline_p): When ix86_fpmath flags do not match
12376 check whether the callee uses FP math at all.
12378 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
12380 PR middle-end/81931
12381 * tree-ssanames.c (get_nonzero_bits): Use element_precision
12382 instead of TYPE_PRECISION.
12384 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
12385 Alan Hayward <alan.hayward@arm.com>
12386 David Sherwood <david.sherwood@arm.com>
12388 * combine.c (make_extraction): Use subreg_offset_from_lsb.
12390 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
12392 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
12394 (ix86_frame::stack_realign_allocate): New field.
12395 (struct machine_frame_state): Modify comments.
12396 (machine_frame_state::sp_realigned_fp_end): New field.
12397 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
12398 layout calculation.
12399 (sp_valid_at): Add assertion to assure no attempt to access invalid
12400 offset of a realigned stack.
12401 (fp_valid_at): Likewise.
12402 (choose_baseaddr): Modify comments.
12403 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
12404 ix86_expand_prologue.
12405 (ix86_expand_prologue): Modify stack realignment and allocation.
12406 (ix86_expand_epilogue): Modify comments.
12407 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
12408 avx2_runtime, avx512f, and avx512f_runtime.
12410 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
12412 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
12413 (mstackrealign): Do not init to -1.
12414 * config/i386/i386.c (ix86_option_override_internal):
12415 Check opts_set, not opts when setting default value of
12416 opts->x_ix86_force_align_arg_pointer.
12418 2017-08-23 Richard Biener <rguenther@suse.de>
12420 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
12421 lang_hooks.decl_printable_name.
12422 * print-rtl-function.c (print_rtx_function): Likewise.
12423 * tree-pretty-print.c (dump_function_header): Likewise.
12425 2017-08-23 Richard Biener <rguenther@suse.de>
12428 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
12429 -g0 at compile-time.
12431 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12433 PR middle-end/19706
12434 * doc/sourcebuild.texi (Other hardware attributes):
12437 2017-08-23 Tamar Christina <tamar.christina@arm.com>
12439 PR middle-end/19706
12440 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
12441 Fix single-use check.
12443 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
12445 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
12447 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12449 * doc/install.texi: Modify to add more details on running selected
12452 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
12454 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
12455 is combined with -mabi=ms.
12456 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
12459 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12461 PR tree-optimization/81488
12462 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
12463 and cached_basis fields.
12464 (MAX_SPREAD): New constant.
12465 (alloc_cand_and_find_basis): Initialize new fields.
12466 (clear_visited): New function.
12467 (create_phi_basis_1): Rename from create_phi_basis, set visited
12468 and cached_basis fields.
12469 (create_phi_basis): New wrapper function.
12470 (phi_add_costs_1): Rename from phi_add_costs, add spread
12471 parameter, set visited field, short-circuit when limits reached.
12472 (phi_add_costs): New wrapper function.
12473 (record_phi_increments_1): Rename from record_phi_increments, set
12475 (record_phi_increments): New wrapper function.
12476 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
12477 (phi_incr_cost): New wrapper function.
12478 (all_phi_incrs_profitable_1): Rename from
12479 all_phi_incrs_profitable, set visited field.
12480 (all_phi_incrs_profitable): New wrapper function.
12482 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12483 Alan Hayward <alan.hayward@arm.com>
12484 David Sherwood <david.sherwood@arm.com>
12486 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
12487 that takes the outer and inner modes.
12488 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
12489 comparison as the canonical test for a paradoxical subreg.
12490 * combine.c (simplify_set): Use paradoxical_subreg_p.
12491 (make_extraction): Likewise.
12492 (force_to_mode): Likewise.
12493 (rtx_equal_for_field_assignment_p): Likewise.
12494 (gen_lowpart_for_combine): Likewise.
12495 (simplify_comparison): Likewise.
12496 * cse.c (equiv_constant): Likewise.
12497 * expmed.c (store_bit_field_1): Likewise.
12498 * final.c (alter_subreg): Likewise.
12499 * fwprop.c (propagate_rtx): Likewise.
12500 (forward_propagate_subreg): Likewise.
12501 * ira-conflicts.c (ira_build_conflicts): Likewise.
12502 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
12503 * lra-constraints.c (curr_insn_transform): Likewise.
12504 (split_reg): Likewise.
12505 * lra-eliminations.c (move_plus_up): Likewise.
12506 (lra_eliminate_regs_1): Likewise.
12507 * recog.c (general_operand): Likewise.
12508 * ree.c (combine_reaching_defs): Likewise.
12509 * reload.c (push_reload): Likewise.
12510 (find_reloads): Likewise.
12511 * reload1.c (elimination_effects): Likewise.
12512 (compute_reload_subreg_offset): Likewise.
12513 (choose_reload_regs): Likewise.
12514 * rtlanal.c (subreg_lsb_1): Likewise.
12515 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
12516 (simplify_subreg): Likewise.
12517 * var-tracking.c (track_loc_p): Likewise.
12518 * emit-rtl.c (byte_lowpart_offset): Likewise.
12519 (paradoxical_subreg_p): Delete out-of-line definition.
12521 2017-08-22 Jeff Law <law@redhat.com>
12523 PR tree-optimization/81741
12524 PR tree-optimization/71947
12525 * tree-ssa-dom.c: Include tree-inline.h.
12526 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
12527 equivalences if one is more expensive to compute than the other.
12528 * tree-ssa-scopedtables.h (class const_or_copies): Make
12529 record_const_or_copy_raw method private.
12530 (class avail_exprs_stack): New method simplify_binary_operation.
12531 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
12532 avail_exprs_stack::simplify_binary_operation as needed.
12533 (avail_exprs_stack::simplify_binary_operation): New function.
12535 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
12537 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
12538 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
12539 (DOT_SYMBOLS): Likewise.
12540 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
12541 (RELOCATABLE_NEEDS_FIXUP): Likewise.
12542 (RS6000_ABI_NAME): Likewise.
12543 (TARGET_CMODEL): Likewise.
12544 (TOC_SECTION_ASM_OP): Likewise.
12545 (SET_CMODEL): New macro.
12546 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
12548 2017-08-22 Richard Biener <rguenther@suse.de>
12550 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
12551 to allow for free-lang-data replacements similar to verify_type_variant.
12553 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
12554 Alan Hayward <alan.hayward@arm.com>
12555 David Sherwood <david.sherwood@arm.com>
12557 * config/aarch64/aarch64.md (casesi): Use DImode rather than
12558 VOIDmode for the LABEL_REF.
12560 2017-08-22 Richard Biener <rguenther@suse.de>
12562 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
12564 2017-08-22 Richard Biener <rguenther@suse.de>
12566 * common.opt (feliminate-dwarf2-dups): Ignore.
12567 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
12568 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
12569 same_die_p_wrap, compute_section_prefix,
12570 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
12571 (comdat_symbol_id, comdat_symbol_number): Likewise.
12572 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
12574 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
12575 (output_die): Mark unreachable path unreachable.
12576 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
12577 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
12578 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
12579 (dwarf2out_early_finish): Likewise.
12581 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
12583 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
12585 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
12588 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
12589 not VAR_P. Filter attribute warnings with OPT_Wattributes.
12590 (avr_attribute_table) <io, io_low, address>: Initialize
12591 .decl_required with true.
12593 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
12595 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
12596 undocumented debugging options.
12597 (-mvsx-scalar-double): Likewise.
12598 (-mallow-df-permute): Likewise.
12599 (-mvectorize-builtins): Likewise.
12600 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
12601 (rs6000_builtin_vectorized_function): Likewise.
12602 (rs6000_builtin_md_vectorized_function): Likewise.
12603 (rs6000_opt_vars): Likewise.
12605 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
12608 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
12609 (*btrq_imm): Rename from *btrq.
12610 (*btcq_imm): Rename from *btcq.
12611 (btsc): New code attribute.
12612 (*<btsc><mode>): New insn pattern.
12613 (*btr<mode>): Ditto.
12614 (*<btsc><mode>_mask): New insn_and_split pattern.
12615 (*btr<mode>_mask): Ditto.
12617 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12618 Alan Hayward <alan.hayward@arm.com>
12619 David Sherwood <david.sherwood@arm.com>
12621 * function.c (pad_below): Simplify padding calculation.
12623 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12624 Alan Hayward <alan.hayward@arm.com>
12625 David Sherwood <david.sherwood@arm.com>
12627 * target.def (function_prologue): Remove frame size argument.
12628 (function_epilogue): Likewise.
12629 * doc/tm.texi: Regenerate.
12630 * final.c (final_start_function): Update call to function_prologue.
12631 (final_end_function): Update call to function_epilogue.
12632 (default_function_pro_epilogue): Remove frame size argument.
12633 * output.h (default_function_pro_epilogue): Likewise.
12634 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
12635 (arm_output_function_prologue): Likewise.
12636 * config/frv/frv.c (frv_function_prologue): Likewise.
12637 (frv_function_epilogue): Likewise.
12638 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
12639 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
12640 (ia64_output_function_epilogue): Likewise.
12641 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
12642 (m32r_output_function_epilogue): Likewise.
12643 * config/microblaze/microblaze.c (microblaze_function_prologue)
12644 (microblaze_function_epilogue): Likewise.
12645 * config/mips/mips.c (mips_output_function_prologue): Likewise.
12646 (mips_output_function_epilogue): Likewise.
12647 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
12648 (mmix_target_asm_function_epilogue): Likewise.
12649 * config/msp430/msp430.c (msp430_start_function): Likewise.
12650 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
12651 (nds32_asm_function_epilogue): Likewise.
12652 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
12653 * config/pa/pa.c (pa_output_function_prologue): Likewise.
12654 (pa_output_function_epilogue): Likewise.
12655 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
12656 (rs6000_output_function_epilogue): Likewise.
12657 * config/rl78/rl78.c (rl78_start_function): Likewise.
12658 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
12659 (rs6000_output_function_epilogue): Likewise.
12660 * config/rx/rx.c (rx_output_function_prologue): Likewise.
12661 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
12662 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
12663 (sparc_asm_function_epilogue): Likewise.
12665 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12667 * tree.h (type_has_mode_precision_p): New function.
12668 * convert.c (convert_to_integer_1): Use it.
12669 * expr.c (expand_expr_real_2): Likewise.
12670 (expand_expr_real_1): Likewise.
12671 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
12672 * match.pd: Likewise.
12673 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
12674 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
12675 * tree-tailcall.c (process_assignment): Likewise.
12676 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12677 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
12678 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
12679 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
12680 (vectorizable_assignment): Likewise.
12681 (vectorizable_shift): Likewise.
12682 (vectorizable_operation): Likewise.
12683 * tree-vrp.c (register_edge_assert_for_2): Likewise.
12685 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
12687 * match.pd: Add pow (C, x) simplification.
12689 2017-08-21 Richard Biener <rguenther@suse.de>
12691 PR tree-optimization/81900
12692 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
12693 for blocks with abnormal predecessors.
12694 (compute_antic): Do not set visited flag prematurely.
12696 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
12699 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
12701 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12703 * stor-layout.h (vector_type_mode): Move to...
12704 * tree.h (vector_type_mode): ...here.
12705 * stor-layout.c (vector_type_mode): Move to...
12706 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
12708 2017-08-21 Richard Biener <rguenther@suse.de>
12710 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
12711 register_external_die hooks.
12712 (debug_false_tree_charstarstar_uhwistar): Declare.
12713 (debug_nothing_tree_charstar_uhwi): Likewise.
12714 * debug.c (do_nothing_debug_hooks): Adjust.
12715 (debug_false_tree_charstarstar_uhwistar): New do nothing.
12716 (debug_nothing_tree_charstar_uhwi): Likewise.
12717 * dbxout.c (dbx_debug_hooks): Adjust.
12718 (xcoff_debug_hooks): Likewise.
12719 * sdbout.c (sdb_debug_hooks): Likewise.
12720 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
12721 * dwarf2out.c (macinfo_label_base): New global.
12722 (dwarf2out_register_external_die): New function for the
12723 register_external_die hook.
12724 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
12725 (dwarf2_debug_hooks): Use them.
12726 (dwarf2_lineno_debug_hooks): Adjust.
12727 (struct die_struct): Add with_offset flag.
12728 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
12729 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
12730 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
12731 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
12732 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
12733 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
12734 defining section names for the early LTO debug variants.
12735 (reset_indirect_string): New helper.
12736 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
12737 (print_dw_val): Add support for offsetted symbol references.
12738 (get_ultimate_context): Split out from is_cxx.
12739 (is_cxx): Use get_ultimate_context.
12740 (is_fortran): Add decl overload.
12741 (compute_comp_unit_symbol): Split out worker from
12742 compute_section_prefix.
12743 (compute_section_prefix): Call compute_comp_unit_symbol and
12744 set comdat_type_p here.
12745 (output_die): Skip DIE symbol output for the LTO added one.
12746 Handle DIE symbol references with offset.
12747 (output_comp_unit): Guard section name mangling properly.
12748 For LTO debug sections emit a symbol at the section beginning
12749 which we use to refer to its DIEs.
12750 (add_abstract_origin_attribute): For DIEs registered via
12751 dwarf2out_register_external_die directly refer to the early
12752 DIE rather than indirectly through the shadow one we created.
12753 Remove obsolete call to dwarf2out_abstract_function for
12754 non-function/block origins.
12755 (gen_array_type_die): When generating early LTO debug do
12756 not emit DW_AT_string_length.
12757 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
12758 late when in LTO. As suggested place a gcc_unreachable for
12759 the DECL_ABSTRACT_P case.
12760 (gen_subprogram_die): Avoid another specification DIE
12761 for early built declarations/definitions for the late LTO case.
12762 (gen_variable_die): Add type references for late duplicated VLA dies
12764 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
12765 we have the abstract instance already.
12766 (process_scope_var): Adjust decl DIE contexts in LTO which
12767 first puts them in limbo.
12768 (gen_decl_die): Do not generate type DIEs late apart from
12769 types for VLAs or for decls we do not yet have a DIE. Do not
12770 call dwarf2out_abstract_function late.
12771 (dwarf2out_early_global_decl): Make sure to create DIEs
12772 for abstract instances of a decl first.
12773 (dwarf2out_late_global_decl): Adjust comment.
12774 (output_macinfo_op): With multiple macro sections use
12775 macinfo_label_base to distinguish labels.
12776 (output_macinfo): Likewise. Update macinfo_label_base.
12777 Pass in the line info label.
12778 (note_variable_value_in_expr): When generating LTO resolve
12779 all variable values here by generating DIEs as needed.
12780 (init_sections_and_labels): Add early LTO debug flag parameter
12781 and generate different sections and names if set. Add generation
12782 counter for the labels so we can have multiple of them.
12783 (reset_dies): Helper to allow DIEs to be output multiple times.
12784 (dwarf2out_finish): When outputting DIEs to the fat part of an
12785 LTO object first reset DIEs.
12786 (dwarf2out_early_finish): Output early DIEs when generating LTO.
12787 (modified_type_die): Check for decl_ultimate_origin being self
12789 (gen_type_die_with_usage): Likewise.
12790 (gen_typedef_die): Allow decl_ultimate_origin being self.
12791 (set_decl_abstract_flags): Remove.
12792 (set_block_abstract_flags): Likewise.
12793 (dwarf2out_abstract_function): Treat the early generated DIEs
12794 as the abstract copy and only add DW_AT_inline and
12795 DW_AT_artificial here and call set_decl_origin_self.
12796 If the DIE has an abstract origin don't do anything.
12797 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
12798 if we have none yet (Go fails to build one, PR78628).
12799 (variably_modified_type_p): Prevent endless recursion for Ada
12800 cyclic pointer types.
12801 * lto-streamer-in.c: Include debug.h.
12802 (dref_queue): New global.
12803 (lto_read_tree_1): Stream in DIE references.
12804 (lto_input_tree): Register DIE references.
12805 (input_function): Stream DECL_DEBUG_ARGS.
12806 * lto-streamer-out.c: Include debug.h.
12807 (lto_write_tree_1): Output DIE references.
12808 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
12809 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
12810 (output_function): Stream DECL_DEBUG_ARGS.
12811 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
12812 Stream DECL_ABSTRACT_ORIGIN.
12813 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
12814 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
12815 DECL_CONTEXT for file-scope decls.
12816 * lto-streamer.h (struct dref_entry): Declare.
12817 (dref_queue): Likewise.
12818 * cfgexpand.c (pass_expand::execute): Do not call the
12819 outlining_inline_function hook here.
12820 * lto-wrapper.c (debug_obj): New global.
12821 (tool_cleanup): Unlink it if required.
12822 (debug_objcopy): New function.
12823 (run_gcc): Handle early debug sections in the IL files by
12824 extracting them to separate files, partially linkin them and
12825 feeding the result back as result to the linker.
12826 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
12827 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
12828 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
12829 sections into a separate segment.
12830 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
12832 (darwin_asm_dwarf_section): Likewise.
12833 (darwin_asm_output_dwarf_offset): Likewise.
12834 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
12836 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12837 Alan Hayward <alan.hayward@arm.com>
12838 David Sherwood <david.sherwood@arm.com>
12840 * read-md.h (md_reader::record_potential_iterator_use): Replace
12841 pointer argument with an rtx and an index.
12842 * read-rtl.c (iterator_group::apply_iterator): Likewise.
12843 (apply_mode_iterator): Likewise.
12844 (apply_code_iterator): Likewise.
12845 (apply_int_iterator): Likewise.
12846 (apply_subst_iterator): Likewise.
12847 (record_iterator_use): Likewise.
12848 (record_attribute_use): Likewise.
12849 (md_reader::record_potential_iterator_use): Likewise. Update calls
12850 to record_iterator_use and apply_iterator.
12851 (iterator_use): Replace ptr with x and index.
12852 (attribute_use): Likewise.
12853 (apply_attribute_uses): Update calls to apply_iterator.
12854 (apply_iterators): Likewise. Update initialization of iterator_use.
12855 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
12856 and record_potential_iterator_use.
12857 (rtx_reader::read_rtx_operand): Likewise.
12859 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
12860 Alan Hayward <alan.hayward@arm.com>
12861 David Sherwood <david.sherwood@arm.com>
12863 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
12866 2017-08-21 Richard Biener <rguenther@suse.de>
12868 PR middle-end/81884
12869 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
12870 at struct end conservatively when comparing common bases.
12872 2017-08-21 Richard Biener <rguenther@suse.de>
12874 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
12875 (mem_ref_in_stmt): Remove.
12876 (determine_max_movement): Use ref index to get at the reference.
12877 (invariantness_dom_walker::before_dom_children): Deal with
12878 lim data already initialized.
12879 (gather_mem_refs_stmt): Initialize lim data and record ref index.
12881 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12883 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
12884 (TARGET_ISA_ROUND): Ditto.
12885 (TARGET_ROUND): Ditto.
12886 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
12887 * config/i386/i386.md: Ditto.
12888 * config/i386/sse.md: Ditto.
12889 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
12890 with OPTION_MASK_ISA_SSE4_1.
12892 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
12895 * doc/extend.texi (x86 Built-in Functions): Correct the name of
12896 __builtin_ia32_lzcnt_u16.
12898 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
12901 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
12902 (rs6000_set_current_function): Rewrite function to use it.
12904 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
12907 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
12908 and TYPE_WARN_IF_NOT_ALIGN.
12909 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
12910 (handle_warn_if_not_align): New.
12911 (place_union_field): Call handle_warn_if_not_align.
12912 (place_field): Call handle_warn_if_not_align.
12913 Copy TYPE_WARN_IF_NOT_ALIGN.
12914 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
12915 (layout_type): Likewise.
12916 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
12918 (tree_decl_common): Add warn_if_not_align.
12919 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
12920 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
12921 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
12922 (DECL_WARN_IF_NOT_ALIGN): Likewise.
12923 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
12924 * doc/extend.texi: Document warn_if_not_aligned attribute.
12925 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
12927 2017-08-17 Martin Liska <mliska@suse.cz>
12930 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
12931 (get_data_dependence): Use it as pointer type.
12932 (distribute_loop): Likewise.
12934 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12936 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
12937 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
12938 (p8_vmrgow_<mode>_direct): New define_insn.
12939 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
12940 handle endianness for vmrgew and vmrgow permute patterns.
12942 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12944 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
12945 * config/rs6000/rs6000-cpus.def: Remove comment.
12946 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
12947 (POWERPC_MASKS): Likewise.
12948 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
12949 use of TARGET_VSX_TIMODE.
12950 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
12951 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
12952 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
12953 (rs6000_option_override_internal): Remove dead code.
12954 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
12955 (rs6000_legitimize_reload_address): Likewise.
12956 (rs6000_legitimate_address_p): Likewise.
12957 (rs6000_opt_masks): Delete "vsx-timode".
12958 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
12959 from function comment.
12960 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
12961 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
12962 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
12964 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
12965 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
12966 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
12967 (VSX_TI): Likewise.
12969 (define_peephole2): Likewise.
12971 2017-08-17 Martin Sebor <msebor@redhat.com>
12974 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
12975 past the end of an array.
12976 (test_pp_format): Add test cases.
12978 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
12980 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
12981 missing ECF_NOTHROW flags.
12983 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
12986 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
12987 operands residing in integer registers.
12988 (*vsx_le_perm_load_<mode>): Likewise.
12989 (*vsx_le_perm_store_<mode>): Likewise.
12990 (define_peephole2): Add peepholes to optimize the above.
12992 2017-08-17 Marek Polacek <polacek@redhat.com>
12994 PR middle-end/81814
12995 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
12996 to mimic what shorten_compare did. Change the return type to bool.
12997 (fold_cond_expr_with_comparison): Update call to
12998 operand_equal_for_comparison_p.
12999 (fold_ternary_loc): Likewise.
13001 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
13003 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
13005 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
13007 2017-08-17 Richard Biener <rguenther@suse.de>
13009 * tree-ssa-structalias.c (solve_graph): When propagating
13010 to successors update the graphs succ edges and avoid duplicate work.
13012 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
13015 * config/i386/i386.c (ix86_option_override_internal): Save target
13016 specific options after ix86_stack_protector_guard_reg was changed.
13018 2017-08-17 Richard Biener <rguenther@suse.de>
13020 PR tree-optimization/81827
13021 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
13022 (new_var_info): Initialize it conservatively.
13023 (get_call_vi): Mark register vars.
13024 (new_scalar_tmp_constraint_exp): Likewise.
13025 (handle_rhs_call): Likewise.
13026 (handle_const_call): Likewise.
13027 (create_function_info_for): Likewise.
13028 (solve_constraints): Sort varinfos to separate register from
13029 non-register vars to pack points-to solution bitmaps during
13032 2017-08-17 Marek Polacek <polacek@redhat.com>
13034 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
13036 2017-08-17 Richard Biener <rguenther@suse.de>
13038 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
13039 to true when overflow is undefined and we saturated the result.
13041 2017-08-17 Alan Modra <amodra@gmail.com>
13044 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
13045 Don't use store multiple if only one reg needs saving.
13046 (interesting_frame_related_regno): New function.
13047 (rs6000_frame_related): Don't emit frame info for regs that
13049 (rs6000_emit_epilogue): Likewise.
13051 2017-08-16 Nathan Sidwell <nathan@acm.org>
13053 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
13054 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
13055 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
13056 * tree.c (free_lang_data_in_type): Use else-if chain. Always
13057 clear TYPE_LANG_1. Remove obsolete member-function stripping.
13058 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
13059 (verify_type): Adjust for TYPE_BINFO move.
13060 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
13061 process TYPE_BINFO directly.
13062 (hash_tree): Likewise.
13063 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
13065 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
13068 2017-08-16 David Malcolm <dmalcolm@redhat.com>
13070 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
13072 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
13075 * config/i386/i386.md (*anddi_1_btr): Change predicates of
13076 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
13077 Add ix86_binary_operator_ok to insn constraint.
13078 (*iordi_1_bts): Ditto.
13079 (*xordi_1_btc): Ditto.
13080 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
13081 Update corresponding peephole2 pattern.
13085 2017-08-16 Bin Cheng <bin.cheng@arm.com>
13087 PR tree-optimization/81832
13088 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
13089 copy loop header which has IFN_LOOP_DIST_ALIAS call.
13091 2017-08-16 Marek Polacek <polacek@redhat.com>
13094 * fold-const.c (fold_indirect_ref_1): Restore original behavior
13095 regarding size_zero_node.
13097 2017-08-16 Martin Liska <mliska@suse.cz>
13100 * config.gcc: Respect previously set extra_objs in case
13103 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
13105 PR tree-optimization/81835
13106 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
13107 the phi SSA_NAME. Check that the condition in a COND_EXPR does
13108 not depend on the phi.
13110 2017-08-16 Alan Modra <amodra@gmail.com>
13112 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
13115 2017-08-16 Alan Modra <amodra@gmail.com>
13117 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
13118 (save_reg_p): ..into this. Update all callers.
13119 (first_reg_to_save): Simplify.
13121 2017-08-16 Alan Modra <amodra@gmail.com>
13123 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
13126 2017-08-15 Joseph Myers <joseph@codesourcery.com>
13130 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
13131 constant count if that count is less than 32.
13133 2017-08-15 Nathan Sidwell <nathan@acm.org>
13135 * gcc.c (execute): Emit friendlier message if inferior is killed
13136 by an external cause.
13138 2017-08-15 Richard Biener <rguenther@suse.de>
13140 PR tree-optimization/81790
13141 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
13142 CONSTRUCTORs from simplifying and VN.
13144 2017-08-14 Martin Sebor <msebor@redhat.com>
13146 * builtin-attrs.def: Add comments.
13148 2017-08-14 Martin Sebor <msebor@redhat.com>
13151 * doc/extend.texi (attribute nonstring): Document new attribute.
13153 2017-08-14 Martin Sebor <msebor@redhat.com>
13156 * tree-diagnostic.c (default_tree_printer): Handle %G.
13157 * gimple-pretty-print.h (percent_G_format): Declare new function.
13158 * gimple-pretty-print.c (percent_G_format): Define.
13159 * tree-pretty-print.c (percent_K_format): Add argument.
13161 2017-08-14 Martin Sebor <msebor@redhat.com>
13163 PR translation/79998
13164 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
13165 Remove a stray space.
13167 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
13170 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
13171 (*iordi_1_bts): Ditto.
13172 (*xordi_1_btc): Ditto.
13174 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13177 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
13178 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13180 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
13181 quoted strings, and make more translator-friendly.
13182 (darwin_rs6000_override_options): Likewise.
13183 (rs6000_option_override_internal): Likewise.
13184 (rs6000_return_in_memory): Fix overlong line.
13185 (init_cmulative_args): Use quoted strings, and make more
13186 translator-friendly.
13187 (rs6000_pass_by_reference): Fix overlong line.
13188 (def_builtin): Use quoted strings.
13189 (altivec_expand_predicate_builtin): Use quoted strings, and make
13190 more translator-friendly.
13191 (htm_expand_builtin): Use quoted strings.
13192 (cpu_expand_builtin): Use quoted strings, and make more
13193 translator-friendly.
13194 (altivec_expand_builtin): Likewise.
13195 (paired_expand_predicate_builtin): Likewise.
13196 (rs6000_invalid_builtin): Likewise.
13197 (builtin_function_type): Use quoted strings.
13198 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
13199 more translator-friendly.
13200 (rs6000_trampoline_init): Likewise.
13201 (rs6000_handle_altivec_attribute): Likewise.
13202 (rs6000_inner_target_options): Use quoted strings.
13203 (rs6000_disable_incompatible_switches): Likewise.
13204 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
13205 strings, and make more translator-friendly.
13206 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
13208 2017-08-14 Bin Cheng <bin.cheng@arm.com>
13210 PR tree-optimization/81799
13211 * tree-loop-distribution.c (version_loop_by_alias_check): Force
13212 cond_expr to simple gimple operand.
13214 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
13216 PR middle-end/46932
13217 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
13219 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
13223 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
13224 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
13225 TARGET_GASISR_PROLOGUES.
13226 * config/avr/avr.c (avr_option_override): Same.
13227 (avr_pass_pre_proep::execute): Same.
13229 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
13232 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
13233 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
13235 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
13237 * config/i386/i386.md (*load_tp_<mode>): Redefine as
13238 define_insn_and_split. Split to a memory load from 0 in
13239 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
13240 using PTR mode iterator.
13241 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
13242 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
13243 (*add_tp_<mode>): Redefine as define_insn_and_split.
13244 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
13245 address space. Merge with *add_tp_x32 using PTR mode iterator.
13246 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
13247 Split to an add with a memory load from 0 in
13248 DEFAULT_TLS_SEG_REG address space.
13250 2017-08-12 Andrew Pinski <apinski@cavium.com>
13252 * config/aarch64/aarch64-option-extensions.def (rdma):
13253 Fix feature string to what Linux prints out in /proc/cpuinfo.
13255 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
13258 * dwarf2out.c (modified_type_die): For C typedef types that have
13259 an ultimate origin, process the ultimate origin instead of the
13261 (gen_typedef_die): Assert that input DECLs have no ultimate
13263 (gen_type_die_with_usage): For typedef variants that have an
13264 ultimate origin, just call gen_decl_die on the original DECL.
13265 (process_scope_var): Avoid creating DIEs for local typedefs and
13266 concrete static variables.
13268 2017-08-12 Alan Modra <amodra@gmail.com>
13272 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
13273 match gnu-user.h startfile.
13274 (ENDFILE_LINUX_SPEC): Similarly.
13276 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
13279 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
13281 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
13283 2017-08-11 Tamar Christina <tamar.christina@arm.com>
13284 * config/aarch64/aarch64.md (mov<mode>): Change.
13285 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13286 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
13287 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
13289 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
13291 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
13292 for storage order barriers.
13294 2017-08-11 Martin Liska <mliska@suse.cz>
13297 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
13298 variables of void type.
13300 2017-08-11 Martin Liska <mliska@suse.cz>
13302 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
13303 TARGET_SUPPORTS_ALIASES.
13304 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
13305 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
13306 (optimize_weakref): Likewise.
13307 * symtab.c (symtab_node::noninterposable_alias): Likewise.
13308 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
13309 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
13311 2017-08-11 Martin Liska <mliska@suse.cz>
13314 * config/i386/i386.c (make_resolver_func): Do complete
13315 refactoring of the function.
13317 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13320 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
13321 * config/i386/i386.c (ix86_stack_protect_guard): Use
13322 ix86_stack_protect_guard_symbol_str to generate varible declaration.
13323 * doc/invoke.texi (x86 Options): Document
13324 -mstack-protector-guard-symbol= option.
13326 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
13328 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
13329 * config/i386/i386.c (ix86_split_stack_guard): New function.
13330 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
13331 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
13332 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
13333 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
13334 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
13335 (split_stack_space_check): Call ix86_split_stack_guard.
13337 2017-08-10 Martin Sebor <msebor@redhat.com>
13339 * print-tree.c (print_node): Print location using the established
13341 Replace spaces with colons.
13342 (debug_raw, debug): Ditto.
13344 2017-08-10 Martin Sebor <msebor@redhat.com>
13347 * pretty-print.c (pp_format): Correct the handling of %s precision.
13349 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
13352 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13354 (ix86_finalize_stack_frame_flags): This. Also clear
13355 frame_pointer_needed if -fno-omit-frame-pointer is used without
13357 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13358 with ix86_finalize_stack_frame_flags.
13359 (ix86_expand_epilogue): Likewise.
13360 (ix86_expand_split_stack_prologue): Likewise.
13361 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
13363 2017-08-10 Martin Liska <mliska@suse.cz>
13366 * c-attribs.c (handle_target_attribute):
13367 Report warning for an empty string argument of target attribute.
13369 2017-08-09 Jakub Jelinek <jakub@redhat.com>
13372 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
13374 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
13375 or DECL_NONLOCAL labels.
13376 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
13377 or DECL_NONLOCAL labels here.
13379 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
13381 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
13382 to indicate when early gimple folding has been disabled.
13383 (rs6000_gimple_fold_builtin): Add debug content.
13384 (rs6000_invalid_builtin): Fix whitespace.
13385 (rs6000_expand_builtin): Fix whitespace.
13386 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
13388 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
13391 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
13392 SAVE_MULTIPLE if not all the registers that saves, should be saved.
13394 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
13396 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
13397 (qdf24xx): Likewise.
13398 * config/aarch64/aarch64.md: Include falkor.md.
13399 * config/aarch64/falkor.md: New.
13401 2017-08-09 Marek Polacek <polacek@redhat.com>
13404 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
13405 * diagnostic.c (emit_diagnostic): Add a comment.
13406 (emit_diagnostic_valist): New function.
13408 2017-08-09 Marek Polacek <polacek@redhat.com>
13411 * input.c (make_location): New overload.
13412 * input.h (make_location): Declare.
13414 2017-08-08 Alan Modra <amodra@gmail.com>
13415 H.J. Lu <hongjiu.lu@intel.com>
13418 * gcc.c (NO_PIE_SPEC): Delete.
13419 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
13421 (LINK_PIE_SPEC): ..to here.
13422 (LINK_COMMAND_SPEC): Support -no-pie.
13423 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
13424 chain of crtbegin*.o selection, update for PIE_SPEC changes and
13426 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
13427 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
13428 (ENDFILE_CRTEND_SPEC): Similarly.
13430 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
13433 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
13434 (mstack-protector-guard-offset=): Ditto.
13435 * config/i386/i386.c (ix86_option_override): Handle
13436 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
13438 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
13439 ix86_stack_protect_guard_offset variables.
13440 (TARGET_STACK_PROTECT_GUARD): Always define.
13441 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
13442 and -mstack-protector-guard-offset= options.
13444 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13446 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
13447 boundary case for the last candidate.
13449 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13451 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
13452 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
13454 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13456 PR middle-end/19706
13457 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
13458 * config/aarch64/aarch64-builtins.c
13459 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
13460 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
13461 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
13463 2017-08-08 Tamar Christina <tamar.christina@arm.com>
13464 Andrew Pinski <pinskia@gmail.com>
13466 PR middle-end/19706
13467 * internal-fn.def (XORSIGN): New.
13468 * optabs.def (xorsign_optab): New.
13469 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
13470 (convert_expand_mult_copysign): New.
13471 (pass_optimize_widening_mul::execute): Call
13472 convert_expand_mult_copysign.
13474 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13476 PR tree-optimization/81354
13477 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
13478 Insert on edges rather than explicitly creating landing pads.
13479 (analyze_candidates_and_replace): Commit edge inserts.
13481 2017-08-08 Richard Biener <rguenther@suse.de>
13483 PR middle-end/81719
13484 * tree-ssa-loop-niter.c: Include tree-dfa.h.
13485 (expand_simple_operations): Also look through ADDR_EXPRs with
13486 MEM_REF bases treating them as POINTER_PLUS_EXPR.
13488 2017-08-08 Richard Biener <rguenther@suse.de>
13490 PR tree-optimization/81723
13491 * tree-vect-slp.c (struct bst_traits): New hash traits.
13492 (bst_fail): New global.
13493 (vect_build_slp_tree_2): New worker, split out from ...
13494 (vect_build_slp_tree): ... this now wrapping it with using
13495 bst_fail set to cache SLP tree build fails. Properly handle
13497 (vect_analyze_slp_instance): Allocate and free bst_fail.
13499 2017-08-08 Martin Liska <mliska@suse.cz>
13502 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
13503 LABEL_DECLs that can be from a different function.
13505 2017-08-08 Bin Cheng <bin.cheng@arm.com>
13507 PR tree-optimization/81744
13508 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
13509 loop's number of iterations.
13511 2017-08-08 Martin Liska <mliska@suse.cz>
13513 * asan.c: Include header files.
13514 * attribs.c (build_decl_attribute_variant): New function moved
13516 (build_type_attribute_qual_variant): Likewise.
13517 (cmp_attrib_identifiers): Likewise.
13518 (simple_cst_list_equal): Likewise.
13519 (omp_declare_simd_clauses_equal): Likewise.
13520 (attribute_value_equal): Likewise.
13521 (comp_type_attributes): Likewise.
13522 (build_type_attribute_variant): Likewise.
13523 (lookup_ident_attribute): Likewise.
13524 (remove_attribute): Likewise.
13525 (merge_attributes): Likewise.
13526 (merge_type_attributes): Likewise.
13527 (merge_decl_attributes): Likewise.
13528 (merge_dllimport_decl_attributes): Likewise.
13529 (handle_dll_attribute): Likewise.
13530 (attribute_list_equal): Likewise.
13531 (attribute_list_contained): Likewise.
13532 * attribs.h (lookup_attribute): New function moved from tree.[ch].
13533 (lookup_attribute_by_prefix): Likewise.
13534 * bb-reorder.c: Include header files.
13535 * builtins.c: Likewise.
13536 * calls.c: Likewise.
13537 * cfgexpand.c: Likewise.
13538 * cgraph.c: Likewise.
13539 * cgraphunit.c: Likewise.
13540 * convert.c: Likewise.
13541 * dwarf2out.c: Likewise.
13542 * final.c: Likewise.
13543 * fold-const.c: Likewise.
13544 * function.c: Likewise.
13545 * gimple-expr.c: Likewise.
13546 * gimple-fold.c: Likewise.
13547 * gimple-pretty-print.c: Likewise.
13548 * gimple.c: Likewise.
13549 * gimplify.c: Likewise.
13550 * hsa-common.c: Likewise.
13551 * hsa-gen.c: Likewise.
13552 * internal-fn.c: Likewise.
13553 * ipa-chkp.c: Likewise.
13554 * ipa-cp.c: Likewise.
13555 * ipa-devirt.c: Likewise.
13556 * ipa-fnsummary.c: Likewise.
13557 * ipa-inline.c: Likewise.
13558 * ipa-visibility.c: Likewise.
13560 * lto-cgraph.c: Likewise.
13561 * omp-expand.c: Likewise.
13562 * omp-general.c: Likewise.
13563 * omp-low.c: Likewise.
13564 * omp-offload.c: Likewise.
13565 * omp-simd-clone.c: Likewise.
13566 * opts-global.c: Likewise.
13567 * passes.c: Likewise.
13568 * predict.c: Likewise.
13569 * sancov.c: Likewise.
13570 * sanopt.c: Likewise.
13571 * symtab.c: Likewise.
13572 * toplev.c: Likewise.
13573 * trans-mem.c: Likewise.
13574 * tree-chkp.c: Likewise.
13575 * tree-eh.c: Likewise.
13576 * tree-into-ssa.c: Likewise.
13577 * tree-object-size.c: Likewise.
13578 * tree-parloops.c: Likewise.
13579 * tree-profile.c: Likewise.
13580 * tree-ssa-ccp.c: Likewise.
13581 * tree-ssa-live.c: Likewise.
13582 * tree-ssa-loop.c: Likewise.
13583 * tree-ssa-sccvn.c: Likewise.
13584 * tree-ssa-structalias.c: Likewise.
13585 * tree-ssa.c: Likewise.
13586 * tree-streamer-in.c: Likewise.
13587 * tree-vectorizer.c: Likewise.
13588 * tree-vrp.c: Likewise.
13589 * tsan.c: Likewise.
13590 * ubsan.c: Likewise.
13591 * varasm.c: Likewise.
13592 * varpool.c: Likewise.
13593 * tree.c: Remove functions moved to attribs.[ch].
13594 * tree.h: Likewise.
13595 * config/aarch64/aarch64.c: Add attrs.h header file.
13596 * config/alpha/alpha.c: Likewise.
13597 * config/arc/arc.c: Likewise.
13598 * config/arm/arm.c: Likewise.
13599 * config/avr/avr.c: Likewise.
13600 * config/bfin/bfin.c: Likewise.
13601 * config/c6x/c6x.c: Likewise.
13602 * config/cr16/cr16.c: Likewise.
13603 * config/cris/cris.c: Likewise.
13604 * config/darwin.c: Likewise.
13605 * config/epiphany/epiphany.c: Likewise.
13606 * config/fr30/fr30.c: Likewise.
13607 * config/frv/frv.c: Likewise.
13608 * config/ft32/ft32.c: Likewise.
13609 * config/h8300/h8300.c: Likewise.
13610 * config/i386/winnt.c: Likewise.
13611 * config/ia64/ia64.c: Likewise.
13612 * config/iq2000/iq2000.c: Likewise.
13613 * config/lm32/lm32.c: Likewise.
13614 * config/m32c/m32c.c: Likewise.
13615 * config/m32r/m32r.c: Likewise.
13616 * config/m68k/m68k.c: Likewise.
13617 * config/mcore/mcore.c: Likewise.
13618 * config/microblaze/microblaze.c: Likewise.
13619 * config/mips/mips.c: Likewise.
13620 * config/mmix/mmix.c: Likewise.
13621 * config/mn10300/mn10300.c: Likewise.
13622 * config/moxie/moxie.c: Likewise.
13623 * config/msp430/msp430.c: Likewise.
13624 * config/nds32/nds32-isr.c: Likewise.
13625 * config/nds32/nds32.c: Likewise.
13626 * config/nios2/nios2.c: Likewise.
13627 * config/nvptx/nvptx.c: Likewise.
13628 * config/pa/pa.c: Likewise.
13629 * config/pdp11/pdp11.c: Likewise.
13630 * config/powerpcspe/powerpcspe.c: Likewise.
13631 * config/riscv/riscv.c: Likewise.
13632 * config/rl78/rl78.c: Likewise.
13633 * config/rx/rx.c: Likewise.
13634 * config/s390/s390.c: Likewise.
13635 * config/sh/sh.c: Likewise.
13636 * config/sol2.c: Likewise.
13637 * config/sparc/sparc.c: Likewise.
13638 * config/spu/spu.c: Likewise.
13639 * config/stormy16/stormy16.c: Likewise.
13640 * config/tilegx/tilegx.c: Likewise.
13641 * config/tilepro/tilepro.c: Likewise.
13642 * config/v850/v850.c: Likewise.
13643 * config/vax/vax.c: Likewise.
13644 * config/visium/visium.c: Likewise.
13645 * config/xtensa/xtensa.c: Likewise.
13647 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
13650 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
13651 constraints since the -mupper-regs-* switches have been
13653 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
13654 into a vector from a double word element that was extracted from
13655 another vector, and eliminate extra XXPERMDI instructions.
13656 (vsx_concat_<mode>_2): Likewise.
13657 (vsx_concat_<mode>_3): Likewise.
13658 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
13659 concat to allow optimizing inserts from previous extracts.
13661 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
13663 * config/i386/i386.c (ix86_stack_protect_guard): Generate
13664 memory reference to a SSP offset in TLS address space.
13665 (ix86_print_operand) <case '@'>: Remove.
13666 (ix86_print_operand_punct_valid_p): Remove '@' code.
13667 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
13668 UNSPEC_SP_TLS_TEST.
13669 (stack_tls_protect_set_<mode>): Remove.
13670 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
13671 (stack_tls_protect_test_<mode>): Remove.
13672 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
13674 2017-08-07 Olivier Hainque <hainque@adacore.com>
13677 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
13679 2017-08-07 Douglas Rupp <rupp@adacore.com>
13681 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
13682 variable was referenced as multidir in command.
13684 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13687 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
13690 2017-08-07 Martin Liska <mliska@suse.cz>
13692 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
13693 * config/rl78/rl78.c: Add include of attribs.h.
13694 * config/sh/sh.c: Likewise.
13695 * config/v850/v850.c: Likewise.
13697 2017-08-07 Tom de Vries <tom@codesourcery.com>
13699 PR middle-end/78266
13700 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
13702 2017-08-07 Martin Liska <mliska@suse.cz>
13704 * config/mips/mips.c: Include attribs.h.
13706 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
13709 * doc/invoke.texi: Document change in behvaior for -Ofast for
13712 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
13714 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
13716 (aarch64_pop_regs): Likewise.
13717 (aarch64_gen_load_pair): Likewise.
13718 (aarch64_save_callee_saves): Likewise.
13719 (aarch64_restore_callee_saves): Likewise.
13721 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13723 * config/i386/i386.c: Revert the last change.
13725 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13728 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
13730 (ix86_finalize_stack_frame_flags): This. Also clear
13731 frame_pointer_needed if -fno-omit-frame-pointer is used without
13733 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
13734 with ix86_finalize_stack_frame_flags.
13735 (ix86_expand_epilogue): Likewise.
13736 (ix86_expand_split_stack_prologue): Likewise.
13738 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
13741 * config/i386/i386.c (get_builtin_code_for_version): Set priority
13742 to P_AES for Westmere.
13744 2017-08-07 Jonathan Yong <10walls@gmail.com>
13746 * config/i386/mingw.opt (fset-stack-executable): Removed.
13747 * config/i386/cygming.opt (fset-stack-executable): Moved
13749 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
13751 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
13753 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
13755 2017-08-07 Marek Polacek <polacek@redhat.com>
13757 PR middle-end/81737
13758 * fold-const.c (fold_indirect_ref_1): Check type_domain.
13760 2017-08-07 Martin Liska <mliska@suse.cz>
13762 * attribs.h (canonicalize_attr_name): New function.
13763 (cmp_attribs): Move from c-format.c and adjusted.
13764 (is_attribute_p): Moved from tree.h.
13765 * tree-inline.c: Add new includes.
13766 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
13767 (private_is_attribute_p): Remove.
13768 (private_lookup_attribute): Likewise.
13769 (private_lookup_attribute_by_prefix): Simplify.
13770 (remove_attribute): Use is_attribute_p.
13771 * tree.h: Remove removed declarations.
13773 2017-08-07 Jakub Jelinek <jakub@redhat.com>
13775 PR middle-end/81698
13776 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
13777 instead of computing it in the function. Formatting fix.
13778 (expand_case): Don't rely on default_edge being the first edge,
13779 clear it if removing it, pass default_edge to
13780 emit_case_dispatch_table.
13781 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
13784 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
13786 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
13787 insn in the function, emit NOP after the insn.
13789 2017-08-06 Tom de Vries <tom@codesourcery.com>
13791 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
13794 2017-08-06 Tom de Vries <tom@codesourcery.com>
13796 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
13799 2017-08-04 Yury Gribov <tetra2005@gmail.com>
13801 PR tree-optimization/57371
13802 * match.pd: New pattern.
13804 2017-08-04 Marek Polacek <polacek@redhat.com>
13806 PR middle-end/81695
13807 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
13808 perform the computation in offset_int.
13810 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13812 PR tree-optimization/81136
13813 * tree-vectorizer.h: Include tree-hash-traits.h.
13814 (vec_base_alignments): New typedef.
13815 (vec_info): Add a base_alignments field.
13816 (vect_record_base_alignments): Declare.
13817 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
13819 (DR_IS_CONDITIONAL_IN_STMT): New macro.
13820 (create_data_ref): Add an is_conditional_in_stmt argument.
13821 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
13822 the is_conditional_in_stmt field.
13823 (data_ref_loc): Add an is_conditional_in_stmt field.
13824 (get_references_in_stmt): Set the is_conditional_in_stmt field.
13825 (find_data_references_in_stmt): Update call to create_data_ref.
13826 (graphite_find_data_references_in_stmt): Likewise.
13827 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
13828 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
13829 (vect_record_base_alignment): New function.
13830 (vect_record_base_alignments): Likewise.
13831 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
13832 for nested statements even if we fail to compute a misalignment.
13833 Use pooled base alignments for unconditional references.
13834 (vect_find_same_alignment_drs): Compare base addresses instead
13836 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
13837 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
13839 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13841 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
13842 Add an explicit name for the enum. Use auto_vec for slp_instances
13843 and grouped_stores.
13844 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
13846 (_bb_vec_info): Add a constructor and destructor.
13847 (vinfo_for_stmt): Return NULL for uids of -1 as well.
13848 (destroy_loop_vec_info): Delete.
13849 (vect_destroy_datarefs): Likewise.
13850 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
13851 (vec_info::vec_info): New function.
13852 (vec_info::~vec_info): Likewise.
13853 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
13854 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
13855 destroy_loop_vec_info.
13856 * tree-vect-loop.c (new_loop_vec_info): Replace with...
13857 (_loop_vec_info::_loop_vec_info): ...this.
13858 (destroy_loop_vec_info): Replace with...
13859 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
13860 the stmt_vec_infos. Leave handling of vec_info information to its
13861 destructor. Remove explicit vector releases.
13862 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
13863 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
13864 * tree-vect-slp.c (new_bb_vec_info): Replace with...
13865 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
13866 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
13867 (destroy_bb_vec_info): Replace with...
13868 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
13869 information to its destructor.
13870 (vect_slp_analyze_bb_1): Use new and delete instead of
13871 new_bb_vec_info and destroy_bb_vec_info.
13872 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
13875 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
13877 * tree-data-ref.h (subscript): Add access_fn field.
13878 (data_dependence_relation): Add could_be_independent_p.
13879 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
13880 (same_access_functions): Move to tree-data-ref.c.
13881 * tree-data-ref.c (ref_contains_union_access_p): New function.
13882 (access_fn_component_p): Likewise.
13883 (access_fn_components_comparable_p): Likewise.
13884 (dr_analyze_indices): Add a reference to access_fn_component_p.
13885 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
13887 (constant_access_functions): Likewise.
13888 (add_other_self_distances): Likewise.
13889 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
13890 (initialize_data_dependence_relation): Use XCNEW and remove
13891 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
13892 of access functions that have the same type. Allow the
13893 subsequence to end with different bases in some circumstances.
13894 Record the chosen access functions in SUB_ACCESS_FN.
13895 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
13896 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
13897 (subscript_dependence_tester_1): Likewise dra and drb.
13898 (build_classic_dist_vector): Update calls accordingly.
13899 (subscript_dependence_tester): Likewise.
13900 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
13901 DDR_COULD_BE_INDEPENDENT_P.
13902 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
13903 comp_alias_ddrs instead of may_alias_ddrs.
13904 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
13906 (vect_analyze_data_ref_dependence): Use it if
13907 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
13908 distance vectors if that fails.
13909 (dependence_distance_ge_vf): New function.
13910 (vect_prune_runtime_alias_test_list): Use it. Don't clear
13911 LOOP_VINFO_MAY_ALIAS_DDRS.
13913 2017-08-04 Richard Biener <rguenther@suse.de>
13915 PR middle-end/81705
13916 * fold-const.c (fold_binary_loc): Properly restrict
13917 minus_var0 && minus_var1 case when associating undefined overflow
13920 2017-08-04 Tom de Vries <tom@codesourcery.com>
13922 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
13924 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13926 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
13927 Don't start diagnostic messages with a capital letter.
13928 * config/rs6000/rs6000.c (rs6000_option_override_internal):
13930 (rs6000_invalid_builtin): Likewise.
13931 (rs6000_trampoline_init): Likewise.
13933 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13936 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
13937 after setting changeable df flags.
13939 2017-08-03 Richard Biener <rguenther@suse.de>
13941 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
13942 up if the use is in USE - X.
13944 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
13946 * toplev.c (dumpfile.h): New include.
13947 (internal_error_reentered): New static function. Use it...
13948 (internal_error_function): ...here to handle reentered internal_error.
13950 2017-08-03 Richard Biener <rguenther@suse.de>
13952 PR middle-end/81148
13953 * fold-const.c (split_tree): Add minus_var and minus_con
13954 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
13955 here but always use minus_*.
13956 (associate_trees): Assert we never associate with MINUS_EXPR
13957 and NULL first operand. Do not recurse for PLUS_EXPR operands
13958 when associating as MINUS_EXPR either.
13959 (fold_binary_loc): Track minus_var and minus_con.
13961 2017-08-03 Tom de Vries <tom@codesourcery.com>
13964 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
13965 ACCEL_COMPILER, apply finish_options on
13966 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
13968 2017-08-03 Tom de Vries <tom@codesourcery.com>
13971 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
13972 function_entry_patch_area_size > 0.
13974 2017-08-03 Jakub Jelinek <jakub@redhat.com>
13977 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
13978 instead of 10??LU, perform unit multiplication in wide_int,
13979 don't change alloc_object_size_limit if the limit is larger
13982 PR tree-optimization/81655
13983 PR tree-optimization/81588
13984 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
13985 the case when ranges[i].low and high are 1 for unsigned type with
13988 PR middle-end/81052
13989 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
13990 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
13992 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13994 * tree-vrp.h: Add include guard.
13996 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
13999 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
14000 (ud2): New insn pattern.
14001 * config/i386/i386.c (ix86_expand_epilogue):
14002 For naked functions, generate ud2 instead of trap insn.
14004 2017-08-02 Marek Polacek <polacek@redhat.com>
14007 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
14009 2017-08-02 Tom de Vries <tom@codesourcery.com>
14010 Cesar Philippidis <cesar@codesourcery.com>
14012 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
14013 Add missing edge probabilities.
14015 2017-08-02 Tamar Christina <tamar.christina@arm.com>
14017 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
14018 Correct endianness.
14020 2017-08-02 Jakub Jelinek <jakub@redhat.com>
14022 PR middle-end/79499
14023 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
14024 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
14025 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
14026 prologue_seq sequences - if any.
14028 2017-08-02 Richard Biener <rguenther@suse.de>
14030 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
14031 via vectors if supported, integer extracts via punning if supported
14032 or otherwise vector extracts.
14034 2017-08-02 Richard Biener <rguenther@suse.de>
14036 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
14038 (bitmap_insert_into_set): ... this.
14040 2017-08-02 Richard Biener <rguenther@suse.de>
14042 PR tree-optimization/81633
14044 2015-08-17 Alan Hayward <alan.hayward@arm.com>
14046 PR tree-optimization/71752
14047 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
14049 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
14051 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
14052 (machine_function::call_ms2sysv_pad_out): Remove field.
14053 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
14054 (ix86_compute_frame_layout): Likewise.
14056 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
14059 * config/i386/i386.c (ix86_set_func_type): Disallow naked
14060 attribute with interrupt attribute.
14062 2017-08-01 Andrew Pinski <apinski@cavium.com>
14064 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
14065 BIT_INSERT_EXPR's operand 1
14066 to see if the types precision matches.
14068 2017-08-01 Martin Liska <mliska@suse.cz>
14070 PR middle-end/70140
14071 * builtins.c (expand_builtin_memcpy_args): Remove.
14072 (expand_builtin_memcpy): Call newly added function
14073 expand_builtin_memory_copy_args.
14074 (expand_builtin_memcpy_with_bounds): Likewise.
14075 (expand_builtin_mempcpy): Remove last argument.
14076 (expand_builtin_mempcpy_with_bounds): Likewise.
14077 (expand_builtin_memory_copy_args): New function created from
14078 expand_builtin_mempcpy_args with small modifications.
14079 (expand_builtin_mempcpy_args): Remove.
14080 (expand_builtin_stpcpy): Remove unused argument.
14081 (expand_builtin): Likewise.
14082 (expand_builtin_with_bounds): Likewise.
14084 2017-08-01 Martin Liska <mliska@suse.cz>
14087 Make mempcpy more optimal (PR middle-end/70140).
14089 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14092 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
14093 __builtin_vec_cmpne verify both arguments are compatible vectors
14094 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
14095 verify arg1_type is a pointer or array type. For __builtin_vec_st,
14096 move computation of aligned to after checking the argument types.
14100 * config/rs6000/vsx.md (vextract_fp_from_shorth,
14101 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
14104 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
14105 Doug Rupp <rupp@adacore.com>
14106 Olivier Hainque <hainque@adacore.com>
14108 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
14109 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
14111 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
14112 for TARGET_OS_CPP_BUILTIN.
14113 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
14114 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
14116 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
14117 passing required abi options to the assembler for EABI configurations.
14118 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
14119 of .text.hot and .text.unlikely sections for kernel modules when
14120 using ARM style exceptions.
14121 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
14122 options. Add EXTRA_CC1_SPEC.
14123 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
14125 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
14127 (ARM_TARGET2_DWARF_FORMAT): Define.
14128 * config/arm/t-vxworks: Adjust multilib control to removal of the
14129 Diab command line options.
14131 2017-08-01 Martin Liska <mliska@suse.cz>
14133 PR gcov-profile/81561
14134 * gcov.c (unblock): Make unblocking safe as we need to preserve
14135 index correspondence of blocks and block_lists.
14137 2017-08-01 Richard Biener <rguenther@suse.de>
14139 PR tree-optimization/81181
14140 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
14141 (compute_antic): ... end of iteration here.
14143 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
14145 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
14146 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
14147 (ftree-slp-vectorize): Likewise.
14148 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
14149 can no longer be set independent of flag_tree_loop_vectorize.
14150 * omp-general.c (emp_max_vf): Likewise.
14151 * opts.c (enable_fdo_optimizations): Remove references to
14152 flag_tree_vectorize, these are now implicit.
14153 (common_handle_option): Remove handling for OPT_ftree_vectorize,
14154 and leave it for the options machinery.
14156 2017-08-01 Martin Liska <mliska@suse.cz>
14158 PR middle-end/70140
14159 * builtins.c (expand_builtin_memcpy_args): Remove.
14160 (expand_builtin_memcpy): Call newly added function
14161 expand_builtin_memory_copy_args.
14162 (expand_builtin_memcpy_with_bounds): Likewise.
14163 (expand_builtin_mempcpy): Remove last argument.
14164 (expand_builtin_mempcpy_with_bounds): Likewise.
14165 (expand_builtin_memory_copy_args): New function created from
14166 expand_builtin_mempcpy_args with small modifications.
14167 (expand_builtin_mempcpy_args): Remove.
14168 (expand_builtin_stpcpy): Remove unused argument.
14169 (expand_builtin): Likewise.
14170 (expand_builtin_with_bounds): Likewise.
14172 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14175 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
14176 print "ds:" only for immediates in generic address space.
14178 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
14181 * config/i386/i386.c (ix86_funciton_naked): New prototype.
14182 (ix86_function_ok_for_sibcall): Return false for naked functions.
14184 2017-08-01 Richard Biener <rguenther@suse.de>
14186 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
14187 (compute_antic): Seed worklist with exit block predecessors.
14188 * cfganal.c (dfs_find_deadend): For a cycle return the source
14189 of the edge closing it.
14191 2017-08-01 Tamar Christina <tamar.christina@arm.com>
14193 * config/aarch64/aarch64.c
14194 (aarch64_can_const_movi_rtx_p): Move 0 check.
14196 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14198 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
14199 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
14201 * match.pd: Ditto in address comparison pattern.
14203 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14205 PR tree-optimization/81627
14206 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
14207 closed ssa form for store-store chain.
14209 2017-08-01 Bin Cheng <bin.cheng@arm.com>
14211 PR tree-optimization/81620
14212 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
14213 for store-store chain.
14215 2017-08-01 Jakub Jelinek <jakub@redhat.com>
14217 PR tree-optimization/81588
14218 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
14219 ranges[i].in_p, invert comparison code ccode. For >/>=,
14220 swap rhs1 and rhs2 and comparison code unconditionally,
14221 for </<= don't do that. Don't swap rhs1/rhs2 again if
14222 ranges[i].in_p, instead invert comparison code ccode if
14223 opcode or oe->rank is BIT_IOR_EXPR.
14226 * optabs.def (vec_extract_optab, vec_init_optab): Change from
14227 a direct optab to conversion optab.
14228 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
14229 with GET_MODE_INNER as last argument instead of optab_handler.
14230 * expmed.c (extract_bit_field_1): Likewise. Use vector from
14231 vector extraction if possible and optab is available.
14232 * expr.c (store_constructor): Use convert_optab_handler instead
14233 of optab_handler. Use vector initialization from smaller
14234 vectors if possible and optab is available.
14235 * tree-vect-stmts.c (vectorizable_load): Likewise.
14236 * doc/md.texi (vec_extract, vec_init): Document that the optabs
14237 now have two modes.
14238 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
14239 of vec_init from half-sized vectors with the same element mode.
14240 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
14241 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
14242 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
14243 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
14244 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
14245 after mode in gen_vec_extract* calls.
14246 (vec_extract<mode>): Renamed to ...
14247 (vec_extract<mode><ssescalarmodelower>): ... this.
14248 (vec_extract<mode><ssehalfvecmodelower>): New expander.
14249 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
14250 element mode after mode in gen_vec_init* calls.
14251 (VEC_INIT_HALF_MODE): New mode iterator.
14252 (vec_init<mode>): Renamed to ...
14253 (vec_init<mode><ssescalarmodelower>): ... this.
14254 (vec_init<mode><ssehalfvecmodelower>): New expander.
14255 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
14256 (vec_extractv2sfsf): ... this.
14257 (vec_initv2sf): Renamed to ...
14258 (vec_initv2sfsf): ... this.
14259 (vec_extractv2si): Renamed to ...
14260 (vec_extractv2sisi): ... this.
14261 (vec_initv2si): Renamed to ...
14262 (vec_initv2sisi): ... this.
14263 (vec_extractv4hi): Renamed to ...
14264 (vec_extractv4hihi): ... this.
14265 (vec_initv4hi): Renamed to ...
14266 (vec_initv4hihi): ... this.
14267 (vec_extractv8qi): Renamed to ...
14268 (vec_extractv8qiqi): ... this.
14269 (vec_initv8qi): Renamed to ...
14270 (vec_initv8qiqi): ... this.
14271 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
14272 (vec_init<mode>): Renamed to ...
14273 (vec_init<mode><VEC_base_l>): ... this.
14274 (vec_extract<mode>): Renamed to ...
14275 (vec_extract<mode><VEC_base_l>): ... this.
14276 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
14277 (vec_initv2sfsf): ... this.
14278 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
14279 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14280 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
14281 element mode after mode in gen_vec_init* calls.
14282 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
14283 (vec_init<mode><Vel>): ... this.
14284 (vec_extract<mode>): Renamed to ...
14285 (vec_extract<mode><Vel>): ... this.
14286 * config/aarch64/iterators.md (Vel): New mode attribute.
14287 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
14288 Add element mode after mode in gen_vec_extract* calls.
14289 * config/s390/vector.md (non_vec_l): New mode attribute.
14290 (vec_extract<mode>): Renamed to ...
14291 (vec_extract<mode><non_vec_l>): ... this.
14292 (vec_init<mode>): Renamed to ...
14293 (vec_init<mode><non_vec_l>): ... this.
14294 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
14295 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
14297 * config/arm/iterators.md (V_elem_l): New mode attribute.
14298 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
14299 (vec_extract<mode><V_elem_l>): ... this.
14300 (vec_extractv2di): Renamed to ...
14301 (vec_extractv2didi): ... this.
14302 (vec_init<mode>): Renamed to ...
14303 (vec_init<mode><V_elem_l>): ... this.
14304 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
14305 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
14306 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
14307 Add element mode after gen_vec_extract* calls.
14308 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
14309 (vec_init<mode><unitmode>): ... this.
14310 (vec_extract<mode>): Renamed to ...
14311 (vec_extract<mode><unitmode>): ... this.
14312 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
14313 (vec_init<mode><unitmode>): ... this.
14314 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
14315 (vec_initv2sfsf): ... this.
14316 (vec_extractv2sf): Renamed to ...
14317 (vec_extractv2sfsf): ... this.
14318 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
14319 Add element mode after gen_vec_extract* calls.
14320 * config/mips/mips.md (unitmode): New mode iterator.
14321 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
14322 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
14323 * config/spu/spu.md (inner_l): New mode attribute.
14324 (vec_init<mode>): Renamed to ...
14325 (vec_init<mode><inner_l>): ... this.
14326 (vec_extract<mode>): Renamed to ...
14327 (vec_extract<mode><inner_l>): ... this.
14328 * config/sparc/sparc.md (veltmode): New mode iterator.
14329 (vec_init<VMALL:mode>): Renamed to ...
14330 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
14331 * config/ia64/vect.md (vec_initv2si): Renamed to ...
14332 (vec_initv2sisi): ... this.
14333 (vec_initv2sf): Renamed to ...
14334 (vec_initv2sfsf): ... this.
14335 (vec_extractv2sf): Renamed to ...
14336 (vec_extractv2sfsf): ... this.
14337 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
14338 (vec_init<mode>): Renamed to ...
14339 (vec_init<mode><VEC_base_l>): ... this.
14340 (vec_extract<mode>): Renamed to ...
14341 (vec_extract<mode><VEC_base_l>): ... this.
14342 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
14343 (vec_initv2sfsf): ... this.
14344 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
14345 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
14346 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
14347 gen_vec_init* calls.
14349 2017-08-01 Richard Biener <rguenther@suse.de>
14351 PR tree-optimization/81297
14352 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
14353 TREE_OVERFLOW from INTEGER_CSTs.
14355 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
14357 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
14359 2017-07-31 Carl Love <cel@us.ibm.com>
14361 * config/rs6000/rs6000-c: Add support for built-in functions
14362 vector signed char vec_xl_be (signed long long, signed char *);
14363 vector unsigned char vec_xl_be (signed long long, unsigned char *);
14364 vector signed int vec_xl_be (signed long long, signed int *);
14365 vector unsigned int vec_xl_be (signed long long, unsigned int *);
14366 vector signed long long vec_xl_be (signed long long, signed long long *);
14367 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
14368 vector signed short vec_xl_be (signed long long, signed short *);
14369 vector unsigned short vec_xl_be (signed long long, unsigned short *);
14370 vector double vec_xl_be (signed long long, double *);
14371 vector float vec_xl_be (signed long long, float *);
14372 * config/rs6000/altivec.h (vec_xl_be): Add #define.
14373 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
14374 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
14376 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
14377 (altivec_expand_builtin): Add switch statement to call
14378 altivec_expand_xl_be for each builtin.
14379 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
14380 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
14381 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
14382 __builtin_vsx_le_be_v16qi.
14383 * doc/extend.texi: Update the built-in documentation file for the
14384 new built-in functions.
14386 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14389 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
14391 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14393 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14395 * config.gcc: Add z14.
14396 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
14397 CPU model numbers for z13s and z14.
14398 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
14400 * config/s390/s390-opts.h (enum processor_type): Rename
14401 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14402 * config/s390/s390.c (processor_table): Add field for CPU name to
14403 be passed to Binutils.
14404 (s390_asm_output_machine_for_arch): Use the new field in
14405 processor_table for Binutils.
14406 (s390_expand_builtin): Replace arch12 with z14.
14407 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
14408 (s390_get_sched_attrmask): Likewise.
14409 (s390_get_unit_mask): Likewise.
14410 * config/s390/s390.opt: Add z14 to processor_type enum.
14412 2017-07-31 Martin Jambor <mjambor@suse.cz>
14415 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
14416 regardless of optimization level.
14418 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
14419 Martin Liska <mliska@suse.cz>
14421 * predict.def: Remove old comment and adjust probability.
14422 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
14423 PREDICT statements.
14425 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
14428 * config/i386/i386.c (ix86_function_naked): New function.
14429 (ix86_can_use_return_insn_p): Return false for naked functions.
14430 (ix86_expand_prologue): Skip prologue for naked functions.
14431 (ix86_expand_epilogue): Skip epilogue for naked functions
14432 and emit trap instruction.
14433 (ix86_warn_func_return): New function.
14434 (ix86_attribute_table): Add "naked" attribute specification.
14435 (TARGET_WARN_FUNC_RETURN): Define.
14436 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
14438 2017-07-31 Martin Liska <mliska@suse.cz>
14440 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
14441 (dump_gimple_bb_header): Always dump BB info.
14442 (pp_cfg_jump): Do not append info about BB when dumping a jump.
14444 2017-07-31 Martin Liska <mliska@suse.cz>
14447 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
14448 also with current_function_decl non-null equality.
14450 2017-07-31 Jakub Jelinek <jakub@redhat.com>
14453 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
14454 change type to the element type, instead add eltype variable and
14455 use it where we are interested in the element type.
14457 PR tree-optimization/81603
14458 * ipa-polymorphic-call.c
14459 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
14460 offset arithmetic in offset_int, bail out if the resulting bit offset
14461 doesn't fit into shwi.
14463 2017-07-31 Martin Liska <mliska@suse.cz>
14465 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
14466 (gimplify_save_expr): Fix comment.
14468 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
14471 * config/i386/i386.c (ix86_function_arg): Update arguments for
14473 (ix86_compute_frame_layout): Set the initial stack offset to
14474 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
14475 INCOMING_FRAME_SP_OFFSET.
14476 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
14477 stack before exception handler returns.
14478 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
14479 the 'ERROR_CODE' for exception handler.
14481 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
14483 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
14484 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
14485 (ASM_OUTPUT_REG_POP): Ditto.
14486 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
14487 instead of asm_fprintf to output pure string.
14489 2017-07-29 Jakub Jelinek <jakub@redhat.com>
14491 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
14492 to imported_module_or_decl hook.
14493 (debug_nothing_tree_tree_tree_bool): Remove.
14494 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
14495 * debug.c (do_nothing_debug_hooks): Use
14496 debug_nothing_tree_tree_tree_bool_bool instead of
14497 debug_nothing_tree_tree_tree_bool.
14498 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
14499 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
14500 * sdbout.c (sdb_debug_hooks): Likewise.
14501 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
14502 (gen_namespace_die): Add DW_AT_export_symbols attribute if
14504 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
14505 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
14506 attribute, don't add anything.
14508 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14510 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
14511 (fold_build2_stat_loc): Likewise.
14512 (fold_build3_stat_loc): Likewise.
14513 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
14514 (fold_build1_loc): Remove macro.
14515 (fold_build2_loc): Likewise.
14516 (fold_build3_loc): Likewise.
14518 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14520 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
14521 (gimple_build_debug_bind_source_stat): Likewise.
14522 * gimple.h (gimple_build_debug_bind): Remove macro.
14523 (gimple_build_debug_bind_source): Likewise.
14525 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14527 * bitmap.c (bitmap_alloc): Adjust.
14528 (bitmap_gc_alloc): Likewise.
14529 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
14531 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14533 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
14534 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
14535 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
14536 (bitmap_gc_alloc_stat): Likewise.
14537 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
14539 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14541 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
14542 * rtl.h (shallow_copy_rtx): Remove macro.
14544 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14546 * emit-rtl.c (gen_raw_REG): Adjust.
14547 * gengenrtl.c (gendef): Likewise.
14548 * rtl.c (rtx_alloc_stat): Remove _stat from name.
14549 * rtl.h (rtx_alloc): Remove macro.
14551 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14553 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
14554 (build_tree_list_stat): Likewise.
14555 * tree.h (build_tree_list): Remove macro.
14556 (build_tree_list_vec): Likewise.
14558 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14560 * tree.c (make_vector_stat): Remove _stat from name.
14561 (build_vector_stat): Likewise.
14562 * tree.h (make_vector_stat): Remove macro.
14563 (build_vector_stat): Likewise.
14565 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14567 * tree.h (build_var_debug_value): Remove prototype.
14569 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14571 * tree.c (tree_cons_stat): Remove _stat from name.
14572 * tree.h (tree_cons): Remove macro.
14574 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14576 * tree.c (build_vl_exp_stat): Remove _stat from name.
14577 * tree.h (build_vl_exp): Remove macro.
14579 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14581 * tree.c (build_decl_stat): Remove _stat from name.
14582 * tree.h (build_decl): Remove macro.
14584 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14586 * gimple.c (gimple_build_with_ops_stat): Adjust.
14587 (gimple_alloc_stat): Remove _stat from name.
14588 * gimple.h (gimple_alloc): Remove macro.
14590 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14592 * tree.c (make_tree_vec_stat): Remove _stat from name.
14593 (grow_tree_vec_stat): Likewise.
14594 * tree.h (make_tree_vec_stat): Adjust prototype.
14595 (grow_tree_vec_stat): Likewise.
14596 (make_tree_vec): Remove macro.
14597 (grow_tree_vec): Likewise.
14599 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14601 * fold-const.c (fold_build1_stat_loc): Adjust.
14602 (fold_build2_stat_loc): Likewise.
14603 (fold_build3_stat_loc): Likewise.
14604 * tree.c (build0_stat): Remove _stat from name.
14605 (build1_stat): Likewise.
14606 (build2_stat): Likewise.
14607 (build3_stat): Likewise.
14608 (build4_stat): Likewise.
14609 (build5_stat): Likewise.
14610 * tree.h (build1_loc): Remove macro, and rename _stat function
14612 (build2_loc): Likewise.
14613 (build3_loc): Likewise.
14614 (build4_loc): Likewise.
14615 (build5_loc): Likewise.
14617 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14619 * tree.c (make_int_cst_stat): Remove _stat from name.
14620 * tree.h (make_int_cst_stat): Adjust prototype.
14621 (make_int_cst): Remove macro.
14623 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14625 * tree.c (make_tre_binfo_stat): Remove _stat from name.
14626 * tree.h (make_tree_binfo_stat): Adjust prototype.
14627 (make_tree_binfo): Remove.
14629 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14631 * tree.c (copy_node_stat): Rename to copy_node.
14632 (build_distinct_type_copy): Adjust.
14633 * tree.h (copy_node_stat): Adjust prototype.
14634 (copy_node): Remove macro.
14636 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
14638 * tree.c (make_node_stat): rename to make_node.
14639 (build_tree_list_stat): Adjust.
14640 (build0_stat): Likewise.
14641 (build2_stat): Likewise.
14642 (build3_stat): Likewise.
14643 (build4_stat): Likewise.
14644 (build5_stat): Likewise.
14645 (build_decl_stat): Likewise.
14646 * tree.h (make_node_stat): Adjust prototype.
14647 (make_node): remove macro.
14649 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
14651 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
14652 (PPC_FEATURE2_SCV): Likewise.
14653 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
14655 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14657 * config/aarch64/aarch64.c
14658 (aarch64_internal_mov_immediate): Add new special pattern.
14659 * config/aarch64/aarch64.md (*movdi_aarch64):
14660 Add reg/32bit const mov case.
14662 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14663 Richard Sandiford <richard.sandiford@linaro.org>
14665 * config/aarch64/aarch64.md (mov<mode>): Generalize.
14666 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
14667 Add integer and movi cases.
14668 (movi-split-hf-df-sf split, fp16): New.
14669 (enabled): Added TARGET_FP_F16INST.
14670 * config/aarch64/iterators.md (GPF_HF): New.
14671 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
14673 2017-07-28 Tamar Christina <tamar.christina@arm.com>
14675 * config/aarch64/aarch64.c
14676 (aarch64_simd_container_mode): Add prototype.
14677 (aarch64_expand_mov_immediate): Add HI support.
14678 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
14679 (aarch64_can_const_movi_rtx_p): New.
14680 (aarch64_preferred_reload_class):
14681 Remove restrictions of using FP registers for certain SIMD operations.
14682 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
14683 (aarch64_valid_floating_const): Add integer move validation.
14684 (aarch64_simd_imm_scalar_p): Remove.
14685 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
14686 (aarch64_legitimate_constant_p): Expand list of supported cases.
14687 * config/aarch64/aarch64-protos.h
14688 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
14689 (aarch64_reinterpret_float_as_int): New.
14690 (aarch64_simd_imm_scalar_p): Remove.
14691 * config/aarch64/constraints.md (Uvi): New.
14692 (Dd): Split into Ds and new Dd.
14693 * config/aarch64/aarch64.md (*movsi_aarch64):
14695 (*movdi_aarch64): Add SIMD mov case.
14697 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14699 * tree-predcom.c: (struct chain): Handle store-store chain in which
14700 stores for elimination only store loop invariant values.
14701 (execute_pred_commoning_chain): Ditto.
14702 (prepare_initializers_chain_store_elim): Ditto.
14703 (prepare_finalizers): Ditto.
14704 (is_inv_store_elimination_chain): New function.
14705 (initialize_root_vars_store_elim_1): New function.
14707 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14709 * tree-predcom.c: Revise general description of the pass.
14710 (enum chain_type): New enum type for store elimination.
14711 (struct chain): New field supporting store elimination.
14712 (struct component): Ditto.
14713 (dump_chain): Dump store-stores chain.
14714 (release_chain): Release resources.
14715 (split_data_refs_to_components): Compute and create component
14716 contains only stores for elimination.
14717 (get_chain_last_ref_at): New function.
14718 (make_invariant_chain): Initialization.
14719 (make_rooted_chain): Specify chain type in parameter and record it.
14720 (add_looparound_copies): Skip for store-stores chain.
14721 (determine_roots_comp): Compute type of chain and pass it to
14723 (initialize_root_vars_store_elim_2): New function.
14724 (finalize_eliminated_stores): New function.
14725 (remove_stmt): Handle store for elimination.
14726 (execute_pred_commoning_chain): Execute predictive commoning on
14727 store-store chains.
14728 (determine_unroll_factor): Skip unroll for store-stores chain.
14729 (prepare_initializers_chain_store_elim): New function.
14730 (prepare_initializers_chain): Hanlde store-store chain.
14731 (prepare_finalizers_chain, prepare_finalizers): New function.
14732 (tree_predictive_commoning_loop): Return integer value indicating
14733 if loop is unrolled or lcssa form is corrupted.
14734 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
14736 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14738 * tree-predcom.c (initialize_root): Delete.
14739 (execute_pred_commoning_chain): Initialize root vars and replace
14740 reference of non-combined chain directly, rather than call above
14743 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14745 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
14746 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
14748 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14750 * tree-predcom.c (struct chain): New field init_seq.
14751 (release_chain): Release init_seq.
14752 (prepare_initializers_chain): Record intialization stmts in above
14754 (insert_init_seqs): New function.
14755 (tree_predictive_commoning_loop): Call insert_init_seqs.
14757 2017-07-28 Bin Cheng <bin.cheng@arm.com>
14759 * tree-predcom.c (determine_roots_comp): Skip trivial components.
14761 2017-07-28 Richard Biener <rguenther@suse.de>
14763 * match.pd: Remove superfluous :c.
14764 * genmatch.c (simplify::id): Add member.
14765 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
14767 (current_id): New global.
14768 (dt_node::parent): Move from ...
14769 (dt_operand::parent): ... here. Add for_id member.
14770 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
14771 (decision_tree::find_node): Relax order requirement when
14772 merging DT_TRUE nodes to ones inbetween the current simplify
14773 and the one we try to merge with. Add diagnostic whenever
14774 we need to enforce pattern order by not merging.
14775 (decision_tree::insert): Set current_id.
14776 (decision_tree::print_node): Dump parent node and for_id.
14777 (parser::last_id): Add member.
14778 (parser::push_simplify): Assign unique id.
14779 (parser::parser): Initialize last_id.
14781 2017-07-28 Martin Liska <mliska@suse.cz>
14784 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
14785 gimple_build_debug_bind.
14787 2017-07-28 Richard Biener <rguenther@suse.de>
14789 PR tree-optimization/81502
14790 * match.pd: Add pattern combining BIT_INSERT_EXPR with
14792 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
14794 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
14795 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
14796 for BIT_FIELD_REF args.
14797 * fold-const.c (make_bit_field_ref): Likewise.
14798 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
14800 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14803 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
14804 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
14805 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
14806 Or it into SANITIZER_UNDEFINED.
14807 * ubsan.c: Include gimple-fold.h and varasm.h.
14808 (ubsan_expand_ptr_ifn): New function.
14809 (instrument_pointer_overflow): New function.
14810 (maybe_instrument_pointer_overflow): New function.
14811 (instrument_object_size): Formatting fix.
14812 (pass_ubsan::execute): Call instrument_pointer_overflow
14813 and maybe_instrument_pointer_overflow.
14814 * internal-fn.c (expand_UBSAN_PTR): New function.
14815 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
14816 * sanitizer.def (__ubsan_handle_pointer_overflow,
14817 __ubsan_handle_pointer_overflow_abort): New builtins.
14818 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
14819 * internal-fn.def (UBSAN_PTR): New internal function.
14820 * opts.c (sanitizer_opts): Add pointer-overflow.
14821 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
14822 * fold-const.c (build_range_check): Compute pointer range check in
14823 integral type if pointer arithmetics would be needed. Formatting
14826 2017-07-28 Martin Liska <mliska@suse.cz>
14829 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
14830 parameters that are of a variable-length.
14832 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
14834 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
14836 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
14837 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
14838 (CRT_CALL_STATIC_FUNCTION): Likewise.
14839 (ASM_DEFAULT_SPEC): New define.
14840 (ASM_SPEC32): Likewise.
14841 (ASM_SPEC64): Likewise.
14842 (ASM_SPEC_COMMON): Likewise.
14843 (ASM_SPEC): Likewise.
14844 (INVALID_64BIT): Likewise.
14845 (LINK_OS_DEFAULT_SPEC): Likewise.
14846 (LINK_OS_SPEC32): Likewise.
14847 (LINK_OS_SPEC64): Likewise.
14848 (POWERPC_LINUX): Likewise.
14849 (PTRDIFF_TYPE): Likewise.
14850 (RESTORE_FP_PREFIX): Likewise.
14851 (RESTORE_FP_SUFFIX): Likewise.
14852 (SAVE_FP_PREFIX): Likewise.
14853 (SAVE_FP_SUFFIX): Likewise.
14854 (SIZE_TYPE): Likewise.
14855 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
14856 (TARGET_64BIT): Likewise.
14857 (TARGET_64BIT): Likewise.
14858 (TARGET_AIX): Likewise.
14859 (WCHAR_TYPE_SIZE): Likewise.
14860 (WCHAR_TYPE): Undefine.
14861 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
14862 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
14863 (CPP_OS_RTEMS_SPEC): Delete.
14864 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
14865 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
14867 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
14869 2017-07-28 Jakub Jelinek <jakub@redhat.com>
14871 PR tree-optimization/81578
14872 * tree-parloops.c (build_new_reduction): Bail out if
14873 reduction_code isn't one of the standard OpenMP reductions.
14874 Move the details printing after that decision.
14876 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14878 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
14879 related to reload_in_progress.
14880 (splat_input_operand): Likewise.
14881 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
14883 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
14885 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
14886 (TARGET_INSTANTIATE_DECLS): Likewise.
14887 (legitimate_indexed_address_p): Delete reload_in_progress code.
14888 (rs6000_debug_legitimate_address_p): Likewise.
14889 (rs6000_eliminate_indexed_memrefs): Likewise.
14890 (rs6000_emit_le_vsx_store): Likewise.
14891 (rs6000_emit_move_si_sf_subreg): Likewise.
14892 (rs6000_emit_move): Likewise.
14893 (register_to_reg_type): Likewise.
14894 (rs6000_pre_atomic_barrier): Likewise.
14895 (rs6000_machopic_legitimize_pic_address): Likewise.
14896 (rs6000_allocate_stack_temp): Likewise.
14897 (rs6000_address_for_fpconvert): Likewise.
14898 (rs6000_address_for_altivec): Likewise.
14899 (rs6000_secondary_memory_needed_rtx): Delete function.
14900 (rs6000_check_sdmode): Likewise.
14901 (rs6000_alloc_sdmode_stack_slot): Likewise.
14902 (rs6000_instantiate_decls): Likewise.
14903 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
14904 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
14905 Delete reload_in_progress.
14906 (*vec_reload_and_plus_<mptrsize>): Likewise.
14907 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
14908 (vsx_div_v2di): Likewise.
14909 (vsx_udiv_v2di): Likewise.
14911 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14913 * config/rs6000/rs6000.opt (mlra): Replace with stub.
14914 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
14915 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
14916 (rs6000_debug_reg_global): Delete print of LRA status.
14917 (rs6000_option_override_internal): Delete dead LRA related code.
14918 (rs6000_lra_p): Delete function.
14919 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
14921 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
14923 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
14924 * config/riscv/rtems.h: New file.
14926 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14927 Sudakshina Das <sudi.das@arm.com>
14929 * config/aarch64/aarch64.md
14930 (define_split for and<mode>3nr_compare): Move
14931 non aarch64_logical_operand to a register.
14932 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
14933 register immediate operand to a register.
14934 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
14936 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
14938 PR middle-end/81564
14939 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
14941 2017-07-27 Richard Biener <rguenther@suse.de>
14943 PR tree-optimization/81573
14944 PR tree-optimization/81494
14945 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
14946 multi defuse cycle case.
14948 2017-07-27 Richard Biener <rguenther@suse.de>
14950 PR tree-optimization/81571
14951 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
14954 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
14956 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
14957 earlier and only if MASK_FPU is set. Adjust formatting.
14959 2017-07-27 Martin Liska <mliska@suse.cz>
14961 * opt-functions.awk: Add validation of value of Init.
14962 * optc-gen.awk: Pass new argument.
14964 2017-07-27 Martin Liska <mliska@suse.cz>
14966 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
14967 Fix wrong condition.
14969 2017-07-27 Martin Liska <mliska@suse.cz>
14971 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
14972 BBs and edges seen by autoFDO.
14974 2017-07-27 Richard Biener <rguenther@suse.de>
14976 PR tree-optimization/81502
14977 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
14978 with incompatible but same sized type.
14979 (execute_update_addresses_taken): Likewise.
14981 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
14983 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
14984 flag_tree_loop_vectorize rather than flag_tree_vectorize.
14986 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14989 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
14990 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
14991 Change s_operand to memory_operand.
14993 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
14995 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
14996 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
14997 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
14998 Emit instructions rather than returning an expression. Handle TFmode
14999 and KFmode by casting to TImode.
15000 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
15001 (rs6000_emit_le_vsx_store): Likewise.
15002 * config/rs6000/vsx.md (VSX_TI): New iterator.
15003 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
15004 (*vsx_le_undo_permute_<mode>): Likewise.
15005 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
15006 emit the split sequence.
15007 (*vsx_le_perm_store_<mode>): Likewise.
15009 2017-07-27 Jakub Jelinek <jakub@redhat.com>
15011 PR tree-optimization/81555
15012 PR tree-optimization/81556
15013 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
15014 if true, force CHANGED for the recursive invocation.
15015 (reassociate_bb): Remember original length of ops array, pass
15016 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
15018 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
15019 attributes for noipa attribute. For naked attribute use
15020 lookup_attribute first before lookup_attribute_spec.
15021 * final.c (rest_of_handle_final): Disable IPA RA for functions with
15023 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
15024 for functions with noipa attribute.
15025 (cgraph_externally_visible_p): Return true for functions with noipa
15027 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
15028 for functions with noipa attribute.
15029 * doc/extend.texi: Document noipa function attribute.
15030 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
15031 also for functions with noipa attribute.
15032 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
15034 2017-07-26 Andrew Pinski <apinski@cavium.com>
15036 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
15037 vec_unalign_load_cost and vec_unalign_store_cost.
15039 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
15041 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
15042 -mvsx-small-integer option.
15043 (ISA_3_0_MASKS_IEEE): Likewise.
15044 (OTHER_VSX_VECTOR_MASKS): Likewise.
15045 (POWERPC_MASKS): Likewise.
15046 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
15047 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
15048 code, only testing for DImode being allowed in non-VSX floating
15050 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
15051 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
15053 (rs6000_option_override_internal): Delete -mvsx-small-integer.
15054 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
15055 TARGET_P8_VECTOR test.
15056 (rs6000_secondary_reload_simple_move): Likewise.
15057 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
15058 since TARGET_P9_VECTOR was already tested.
15059 (rs6000_opt_masks): Remove -mvsx-small-integer.
15060 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
15061 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15063 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
15064 test for TARGET_VEXTRACTUB was used, and that uses
15066 (p9 extract splitter): Likewise.
15067 (vsx_extract_<mode>_di_p9): Likewise.
15068 (vsx_extract_<mode>_store_p9): Likewise.
15069 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
15070 for TARGET_P9_VECTOR was used. Delete code that is now dead with
15071 the elimination of TARGET_VSX_SMALL_INTEGER.
15072 (vsx_extract_<mode>_p8): Likewise.
15073 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
15074 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
15075 (vsx_set_<mode>_p9): Likewise.
15076 (vsx_set_v4sf_p9): Likewise.
15077 (vsx_set_v4sf_p9_zero): Likewise.
15078 (vsx_insert_extract_v4sf_p9): Likewise.
15079 (vsx_insert_extract_v4sf_p9_2): Likewise.
15080 * config/rs6000/rs6000.md (sign extend splitter): Change
15081 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
15082 (floatsi<mode>2_lfiwax_mem): Likewise.
15083 (floatunssi<mode>2_lfiwzx_mem): Likewise.
15084 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
15085 since a test for TARGET_P9_VECTOR was used.
15086 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15087 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
15088 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15089 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15090 TARGET_P8_VECTOR test.
15091 (fix_trunc<mode>si2_stfiwx): Likewise.
15092 (fix_trunc<mode>si2_internal): Likewise.
15093 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
15094 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15096 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15097 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
15098 TARGET_P8_VECTOR test.
15099 (fixuns_trunc<mode>si2_stfiwx): Likewise.
15100 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
15101 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
15103 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
15104 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
15105 since a test for TARGET_P9_VECTOR was used.
15106 (splitter for loading small constants): Likewise.
15108 2017-07-26 Andrew Pinski <apinski@cavium.com>
15110 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
15113 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
15116 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
15117 (fp_valid_at): Likewise.
15119 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15121 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
15122 (qdf24xx_addrcost_table): Likewise.
15123 (cortexa57_tunings): Update to use generic_branch_cost.
15124 (cortexa72_tunings): Likewise.
15125 (cortexa73_tunings): Likewise.
15126 (qdf24xx_tunings): Likewise.
15128 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
15130 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
15131 (thunderx2t99_branch_cost): Likewise.
15132 (cortexa35_tunings): Update to use generic_branch_cost.
15133 (cortexa53_tunings): Likewise.
15134 (cortexa57_tunings): Likewise.
15135 (cortexa72_tunings): Likewise.
15136 (cortexa73_tunings): Likewise.
15137 (thunderx2t99_tunings): Likewise.
15139 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15141 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
15142 (sparc_option_override): Honour MASK_FSMULD.
15143 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
15144 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
15145 * config/sparc/sparc.opt (mfsmuld): New option.
15146 * doc/invoke.texi (mfsmuld): Document option.
15148 2017-07-26 Marek Polacek <polacek@redhat.com>
15150 PR middle-end/70992
15151 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
15153 2017-07-26 Richard Biener <rguenther@suse.de>
15155 * gimple-match-head.c (do_valueize): Return OP if valueize
15157 (get_def): New helper to get at the def stmt of a SSA name
15158 if valueize allows.
15159 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
15160 do_valueize to get at the def stmt.
15161 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
15163 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
15165 PR middle-end/46932
15166 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
15168 2017-07-26 Martin Liska <mliska@suse.cz>
15171 * function.c (expand_function_start): Make expansion of
15172 nonlocal_goto_save_area after parm_birth_insn.
15174 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
15176 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
15177 from all CPU target flags enable members.
15179 2017-07-26 Richard Biener <rguenther@suse.de>
15181 * genmatch.c (dt_simplify::gen): Make iterator vars const.
15182 (decision_tree::gen): Make 'type' const.
15183 (write_predicate): Likewise.
15185 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15187 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
15188 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
15189 (rs6000_option_override_internal): Likewise.
15190 (rs6000_expand_vector_set): Likewise.
15191 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
15192 (TARGET_UPPER_REGS_SF): Likewise.
15193 (TARGET_UPPER_REGS_DI): Likewise.
15194 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
15195 (TARGET_DIRECT_MOVE_64BIT): Likewise.
15196 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
15197 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
15198 (Splitters for DI constants in Altivec registers): Likewise.
15199 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
15200 (vsx_set_v4sf_p9): Likewise.
15201 (vsx_set_v4sf_p9_zero): Likewise.
15202 (vsx_insert_extract_v4sf_p9): Likewise.
15203 (vsx_insert_extract_v4sf_p9_2): Likewise.
15205 2017-07-25 Carl Love <cel@us.ibm.com>
15207 * doc/extend.texi: Update the built-in documentation file for the
15208 existing built-in functions
15209 vector signed char vec_cnttz (vector signed char);
15210 vector unsigned char vec_cnttz (vector unsigned char);
15211 vector signed short vec_cnttz (vector signed short);
15212 vector unsigned short vec_cnttz (vector unsigned short);
15213 vector signed int vec_cnttz (vector signed int);
15214 vector unsigned int vec_cnttz (vector unsigned int);
15215 vector signed long long vec_cnttz (vector signed long long);
15216 vector unsigned long long vec_cnttz (vector unsigned long long);
15218 2017-07-25 Andrew Pinski <apinski@cavium.com>
15220 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
15221 accesses where the use is for the first operand of a BIT_INSERT.
15223 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15226 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15227 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15229 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
15231 * config/i386/gstabs.h: Delete.
15232 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
15234 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
15236 * config/i386/i386.c (ix86_decompose_address): Do not check for
15237 register RTX when looking at index_reg or base_reg.
15238 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
15240 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
15242 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
15243 to update EH info here.
15245 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15247 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
15249 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15251 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
15253 2017-07-25 Torsten Duwe <duwe@suse.de>
15255 * common.opt: Introduce -fpatchable-function-entry
15256 command line option, and its variables function_entry_patch_area_size
15257 and function_entry_patch_area_start.
15258 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
15259 including a two-value parser.
15260 * target.def (print_patchable_function_entry): New target hook.
15261 * targhooks.h (default_print_patchable_function_entry): New function.
15262 * targhooks.c (default_print_patchable_function_entry): Likewise.
15263 * toplev.c (process_options): Switch off IPA-RA if
15264 patchable function entries are being generated.
15265 * varasm.c (assemble_start_function): Look at the
15266 patchable-function-entry command line switch and current
15267 function attributes and maybe generate NOP instructions by
15268 calling the print_patchable_function_entry hook.
15269 * doc/extend.texi: Document patchable_function_entry attribute.
15270 * doc/invoke.texi: Document -fpatchable_function_entry
15271 command line option.
15272 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
15274 * doc/tm.texi: Re-generate.
15276 2017-07-25 Jakub Jelinek <jakub@redhat.com>
15279 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
15280 TARGET_AVX512DQ rather than TARGET_AVX512BW.
15282 2017-07-25 Tamar Christina <tamar.christina@arm.com>
15284 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
15286 2017-07-25 Richard Biener <rguenther@suse.de>
15288 PR tree-optimization/81455
15289 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
15290 not walk in cycles when looking for guards.
15292 2017-07-25 Richard Biener <rguenther@suse.de>
15294 PR tree-optimization/81529
15295 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
15296 when optimizing backedge uses.
15298 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
15300 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
15302 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
15303 to dl_section_ref. On AIX, append an expression to subtract
15304 the size of the section length to dl_section_ref.
15306 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
15308 * configure.ac: If any of the config.* scripts fail, exit 1.
15309 * configure: Regenerate.
15311 2017-07-25 Richard Biener <rguenther@suse.de>
15313 PR middle-end/81546
15314 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
15315 of immediate uses, be more verbose on errors.
15317 2017-07-25 Richard Biener <rguenther@suse.de>
15319 PR tree-optimization/81510
15320 * tree-vect-loop.c (vect_is_simple_reduction): When the
15321 reduction stmt is not inside the loop bail out.
15323 2017-07-25 Richard Biener <rguenther@suse.de>
15325 PR tree-optimization/81303
15326 * tree-vect-loop-manip.c (vect_loop_versioning): Build
15327 profitability check against LOOP_VINFO_NITERSM1.
15329 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
15331 * domwalk.c (cmp_bb_postorder): Simplify.
15332 (sort_bbs_postorder): New function. Use it...
15333 (dom_walker::walk): ...here to optimize common cases.
15335 2017-07-25 Martin Liska <mliska@suse.cz>
15338 * ipa-visibility.c (function_and_variable_visibility): Make the
15339 redirection just on target that supports aliasing.
15340 Fix GNU coding style.
15342 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
15345 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
15347 * config/arm/rtems.h: Likewise.
15348 * config/bfin/rtems.h: Likewise.
15349 * config/i386/rtemself.h: Likewise.
15350 * config/lm32/rtems.h: Likewise.
15351 * config/m32c/rtems.h: Likewise.
15352 * config/m68k/rtemself.h: Likewise.
15353 * config/microblaze/rtems.h: Likewise.
15354 * config/mips/rtems.h: Likewise.
15355 * config/moxie/rtems.h: Likewise.
15356 * config/nios2/rtems.h: Likewise.
15357 * config/powerpcspe/rtems.h: Likewise.
15358 * config/rs6000/rtems.h: Likewise.
15359 * config/rtems.h: Likewise.
15360 * config/sh/rtems.h: Likewise.
15361 * config/sh/rtemself.h: Likewise.
15362 * config/sparc/rtemself.h: Likewise.
15364 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15367 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
15368 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
15369 * tree-ssa-structalias.c (alias_get_name): Same.
15371 2017-07-25 Bin Cheng <bin.cheng@arm.com>
15374 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
15375 instructions if no du chain is found.
15377 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
15379 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
15381 2017-07-25 Richard Biener <rguenther@suse.de>
15383 PR middle-end/81505
15384 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
15387 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
15389 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
15390 upper-regs options.
15391 (ISA_2_7_MASKS_SERVER): Likewise.
15392 (ISA_3_0_MASKS_IEEE): Likewise.
15393 (OTHER_P8_VECTOR_MASKS): Likewise.
15394 (OTHER_VSX_VECTOR_MASKS): Likewise.
15395 (POWERPC_MASKS): Likewise.
15396 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
15397 duplicate list of options.
15398 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
15399 explicit -mupper-regs options.
15400 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
15401 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
15402 alias for -mupper-regs-df.
15403 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
15404 (rs6000_init_hard_regno_mode_ok): Likewise.
15405 (rs6000_option_override_internal): Likewise.
15406 (rs6000_opt_masks): Likewise.
15407 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
15408 options in terms of whether -mvsx or -mpower8-vector was used.
15409 (TARGET_UPPER_REGS_DI): Likewise.
15410 (TARGET_UPPER_REGS_SF): Likewise.
15411 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
15412 -mupper-regs-* options.
15414 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15416 * passes.c (emergency_dump_function): Print some empty lines and a
15417 header before the RTL dump.
15419 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
15421 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
15423 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
15426 * config/aarch64/aarch64.c (aarch64_classify_symbol):
15427 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
15429 2017-07-24 Carl Love <cel@us.ibm.com>
15431 * config/rs6000/rs6000-c.c: Add support for built-in functions
15432 vector float vec_extract_fp32_from_shorth (vector unsigned short);
15433 vector float vec_extract_fp32_from_shortl (vector unsigned short);
15434 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
15435 vec_extract_fp_from_shortl): Add defines for the two builtins.
15436 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
15437 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
15439 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
15440 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
15441 * doc/extend.texi: Update the built-in documentation file for the
15442 new built-in function.
15444 2017-07-24 Jakub Jelinek <jakub@redhat.com>
15447 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
15449 * doc/generic.texi: Likewise.
15450 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
15451 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
15453 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
15455 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
15456 (aarch64_mls_elt_merge<mode>): Likewise.
15458 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
15460 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15461 having __cxa_atexit.
15463 2017-07-23 Michael Collison <michael.collison@arm.com>
15465 * config/arm/arm.c (arm_option_override): Deprecate
15466 use of -mstructure-size-boundary.
15467 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
15468 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
15470 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15473 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
15474 Reduce cost estimate for direct moves.
15476 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
15479 * config/i386/i386.c (ix86_option_override_internal): Disable
15480 BMI, BMI2 and TBM instructions for -m16.
15482 2017-07-21 Carl Love <cel@us.ibm.com>
15484 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15485 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15486 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15487 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15488 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15489 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15490 VMULOSW): New enum "unspec" values.
15491 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15492 altivec_vmulosw): New patterns.
15493 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15494 VMULOSW): Add definitions.
15496 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
15498 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
15499 (qdf24xx): Likewise.
15500 * config/aarch64/aarch64-options-extensions.def (rdma); New.
15501 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
15502 (AARCH64_FL_V8_1): Renumber.
15503 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
15504 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
15505 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
15506 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
15507 rdma to feature modifiers list.
15509 2017-07-21 Yury Gribov <tetra2005@gmail.com>
15511 PR middle-end/56727
15512 * ipa-visibility (function_and_variable_visibility): Convert
15513 recursive PLT call to direct call if appropriate.
15515 2017-07-21 Andrew Pinski <apinski@cavium.com>
15517 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
15518 operand 1 to see if the types precision matches.
15519 * fold-const.c (operand_equal_p): Likewise.
15521 2017-07-21 Richard Biener <rguenther@suse.de>
15523 PR tree-optimization/81303
15524 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
15525 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
15526 (vect_peeling_hash_get_lowest_cost): Adjust.
15527 (vect_enhance_data_refs_alignment): Likewise. Use
15528 vect_get_peeling_costs_all_drs to compute the penalty for no
15529 peeling to match up costs.
15531 2017-07-21 Richard Biener <rguenther@suse.de>
15533 PR tree-optimization/81500
15534 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
15535 we didn't identify a reduction path.
15537 2017-07-21 Tom de Vries <tom@codesourcery.com>
15538 Cesar Philippidis <cesar@codesourcery.com>
15540 PR gcov-profile/81442
15541 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
15544 2017-07-21 Tom de Vries <tom@codesourcery.com>
15547 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
15549 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
15550 nvptx_override_options_after_change.
15552 2017-07-21 Ulrich Drepper <drepper@redhat.com>
15554 * dwarf2out.c (output_file_names): Avoid double testing for
15555 dwarf_version >= 5.
15557 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
15559 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
15561 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15563 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
15565 (try_crossjump_to_edge): Do not punt on partitioned functions.
15567 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
15569 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
15570 Put all BBs reachable only via paths crossing cold region to cold
15572 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
15574 2016-07-21 Richard Biener <rguenther@suse.de>
15576 PR tree-optimization/81303
15577 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
15578 into account prologue and epilogue iterations when raising
15579 min_profitable_iters to sth at least covering one vector iteration.
15581 2017-07-21 Tamar Christina <tamar.christina@arm.com>
15583 * config/arm/arm.c (arm_test_cpu_arch_dat):
15586 2017-07-20 Nathan Sidwell <nathan@acm.org>
15588 Remove TYPE_METHODS.
15589 * tree.h (TYPE_METHODS): Delete.
15590 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
15591 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
15592 (dbxout_type_methods): Scan TYPE_FIELDS.
15593 (dbxout_type): Don't check TYPE_METHODS here.
15594 * function.c (use_register_for_decl): Always ignore register for
15595 class types when not optimizing.
15596 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
15597 * tree.c (free_lang_data_in_type): Stitch out member functions and
15598 templates from TYPE_FIELDS.
15599 (build_distinct_type_copy, verify_type_variant,
15600 verify_type): Member fns are on TYPE_FIELDS.
15601 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
15602 * tree-pretty-print.c (dump_generic_node): Likewise.
15604 2017-07-20 Jakub Jelinek <jakub@redhat.com>
15607 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
15608 V2TImode and V4TImode.
15609 (ix86_expand_vector_extract): Likewise.
15610 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
15611 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
15612 (ssescalarmode): Handle V4TImode and V2TImode.
15613 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
15614 (*vec_extractv2ti, *vec_extractv4ti): New insns.
15615 (VEXTRACTI128_MODE): New mode iterator.
15616 (splitter for *vec_extractv?ti first element): New.
15617 (VEC_INIT_MODE): New mode iterator.
15618 (vec_init<mode>): Consolidate 3 expanders into one using
15619 VEC_INIT_MODE mode iterator.
15621 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15623 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
15624 non_spilled_static_chain_regno_p.
15626 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
15628 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
15630 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
15632 * bb-reorder.c (connect_traces): Allow copying of blocks within
15635 2017-07-20 Richard Biener <rguenther@suse.de>
15637 * gimple.h (gimple_phi_result): Add gphi * overload.
15638 (gimple_phi_result_ptr): Likewise.
15639 (gimple_phi_arg): Likewise. Adjust index assert to only
15640 allow actual argument accesses rather than all slots available
15642 (gimple_phi_arg_def): Add gphi * overload.
15643 * tree-phinodes.c (make_phi_node): Initialize only actual
15645 (resize_phi_node): Clear memory not covered by old node,
15646 do not initialize excess argument slots.
15647 (reserve_phi_args_for_new_edge): Initialize new argument slot
15650 2017-07-20 Bin Cheng <bin.cheng@arm.com>
15652 PR tree-optimization/81388
15654 2016-07-21 Bin Cheng <bin.cheng@arm.com>
15656 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
15657 by removing computation of may_be_zero.
15659 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15660 Tom de Vries <tom@codesourcery.com>
15662 PR middle-end/81030
15663 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
15664 when gimple level profile disagrees with what RTL expander did.
15666 2017-07-20 Richard Biener <rguenther@suse.de>
15668 PR tree-optimization/61171
15669 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
15670 (vect_analyze_stmt): Add slp instance parameter.
15671 (vectorizable_reduction): Likewise.
15672 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
15673 (vect_is_simple_reduction): Deal with chains not detected
15674 as SLP reduction chain, specifically not properly associated
15675 chains containing a mix of plus/minus.
15676 (get_reduction_op): Remove.
15677 (get_initial_defs_for_reduction): Simplify, pass in whether
15678 this is a reduction chain, pass in the SLP node for the PHIs.
15679 (vect_create_epilog_for_reduction): Get the SLP instance as
15681 (vectorizable_reduction): Get the SLP instance as arg.
15682 During analysis remember the SLP node with the PHIs in the
15683 instance. Simplify getting at the vectorized reduction PHIs.
15684 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
15685 through SLP instance.
15686 (vect_slp_analyze_operations): Likewise.
15687 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
15688 (vect_transform_stmt): Likewise.
15690 2017-07-20 Tom de Vries <tom@codesourcery.com>
15692 PR tree-optimization/81489
15693 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
15694 read of phi arg location to before loop that modifies phi.
15696 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
15698 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
15701 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15703 PR middle-end/81331
15704 * except.c (execute): Fix ordering issue.
15706 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15708 PR rtl-optimization/81423
15709 * combine.c (make_compound_operation_int): Don't try to optimize
15710 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
15712 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
15714 PR rtl-optimization/81423
15715 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
15716 with a constant that is -1 in the truncated to mode.
15718 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15720 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
15721 (determine_unlikely_bbs): ... here.
15722 * predict.h (propagate_unlikely_bbs_forward): Declare.
15723 * cfgexpand.c (pass_expand::execute): Use it.
15724 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
15726 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
15727 propagate_unlikely_bbs_forward.
15729 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
15731 PR middle-end/81331
15732 * except.c (maybe_add_nop_after_section_switch): New function.
15735 2017-07-19 Tom de Vries <tom@codesourcery.com>
15737 * gimple.h (gimple_phi_set_arg): Make assert more strict.
15739 2017-07-19 Tom de Vries <tom@codesourcery.com>
15741 * gimple.h (gimple_phi_arg): Make assert more strict.
15743 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
15745 * config.gcc (powerpc*-*-*): Add mmintrin.h.
15746 * config/rs6000/mmintrin.h: New file.
15747 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
15749 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15751 PR tree-optimization/81346
15752 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
15754 2017-07-19 Tom de Vries <tom@codesourcery.com>
15756 * config/nvptx/nvptx.md (VECIM): Add V2DI.
15758 2017-07-19 Tom de Vries <tom@codesourcery.com>
15760 * config/nvptx/nvptx-modes.def: Add V2DImode.
15761 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
15762 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
15763 (nvptx_output_mov_insn): Handle lack of mov.b128.
15764 (nvptx_print_operand): Handle 'H' and 'L' codes.
15765 (nvptx_vector_mode_supported): Allow V2DImode.
15766 (nvptx_preferred_simd_mode): New function.
15767 (nvptx_data_alignment): New function.
15768 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
15769 nvptx_preferred_simd_mode.
15770 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
15772 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
15774 2017-07-19 Tom de Vries <tom@codesourcery.com>
15776 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
15777 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
15778 (nvptx_vector_mode_supported): New function. Allow V2SImode.
15779 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
15780 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
15781 (mov<VECIM>_insn): New define_insn.
15782 (define_expand "mov<VECIM>): New define_expand.
15784 2017-07-19 Tom de Vries <tom@codesourcery.com>
15786 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
15788 2017-07-19 Jakub Jelinek <jakub@redhat.com>
15790 PR tree-optimization/81346
15791 * fold-const.h (fold_div_compare, range_check_type): Declare.
15792 * fold-const.c (range_check_type): New function.
15793 (build_range_check): Use range_check_type.
15794 (fold_div_compare): No longer static, rewritten into
15795 a match.pd helper function.
15796 (fold_comparison): Don't call fold_div_compare here.
15797 * match.pd (X / C1 op C2): New optimization using fold_div_compare
15798 as helper function.
15800 2017-07-19 Nathan Sidwell <nathan@acm.org>
15802 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
15803 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
15804 * tree.c (find_decls_types_r, verify_type): Use
15805 TYPE_{MIN,MAX}_VALUE_RAW.
15806 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
15807 (hash_tree): Likewise.
15808 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
15810 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
15813 2017-07-18 Tom de Vries <tom@codesourcery.com>
15815 PR middle-end/81464
15816 * omp-expand.c (expand_omp_for_static_chunk): Handle
15817 equal-argument loop exit phi.
15819 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
15822 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
15823 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
15824 operand 2 predicate.
15825 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
15826 operand 2 predicate.
15827 (ror,rol -> rorx splitters): Use const_int_operand as
15828 operand 2 predicate.
15830 2017-06-18 Richard Biener <rguenther@suse.de>
15832 PR tree-optimization/81410
15833 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
15834 the gap in the ! slp_perm SLP case after each group.
15836 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15838 PR middle-end/81463
15839 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
15842 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
15844 PR middle-end/81462
15845 * predict.c (set_even_probabilities): Cleanup; do not affect
15846 probabilities that are already known.
15847 (combine_predictions_for_bb): Call even when count is set.
15849 2017-07-18 Nathan Sidwell <nathan@acm.org>
15851 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
15854 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15857 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
15858 optimization for loop niter analysis.
15860 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
15863 * config/avr/avr.c (avr_optimize_casesi): Don't use
15864 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
15866 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
15868 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
15869 body_cost_vec from _vect_peel_extended_info.
15870 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
15871 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
15874 2017-07-18 Bin Cheng <bin.cheng@arm.com>
15876 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
15878 2017-07-18 Richard Biener <rguenther@suse.de>
15880 PR tree-optimization/80620
15881 PR tree-optimization/81403
15882 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
15883 info when re-using a VN table entry.
15885 2017-07-18 Richard Biener <rguenther@suse.de>
15887 PR tree-optimization/81418
15888 * tree-vect-loop.c (vectorizable_reduction): Properly compute
15889 vectype_in. Verify that with lane-reducing reduction operations
15890 we have a single def-use cycle.
15892 2017-07-17 Carl Love <cel@us.ibm.com>
15894 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
15896 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15897 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
15898 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
15899 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
15900 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
15901 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
15902 VMULOSW): New enum "unspec" values.
15903 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
15904 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
15905 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
15906 altivec_vmulosw): New patterns.
15907 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
15908 VMULOSW): Add definitions.
15909 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
15911 * config/alpha/alpha.c: Include predict.h.
15913 2017-07-17 Yury Gribov <tetra2005@gmail.com>
15915 * tree-vrp.c (compare_assert_loc): Fix comparison function
15916 to return predictable results.
15918 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15920 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
15922 (subdi3): Likewise.
15923 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
15924 * doc/invoke.texi (mexpand-adddi): Update text.
15926 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15928 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
15929 that also clobbers the CC register. The old expand code is moved
15931 (*arc_clzsi2): ... here.
15932 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
15933 the CC register. The old expand code is moved to ...
15934 (arc_ctzsi2): ... here.
15936 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15938 * config/arc/arc.opt (mindexed-loads): Use initial value
15939 TARGET_INDEXED_LOADS_DEFAULT.
15940 (mauto-modify-reg): Use initial value
15941 TARGET_AUTO_MODIFY_REG_DEFAULT.
15942 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15943 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15944 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
15945 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
15947 2017-07-17 Martin Liska <mliska@suse.cz>
15950 * opts.c (finish_options): Do not allow -fgnu-tm
15951 w/ -fsanitize={kernel-,}address. Say sorry.
15953 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15956 * tree-loop-distribution.c (classify_partition): Only assert on
15957 numer of iterations.
15958 (merge_dep_scc_partitions): Delete prameter. Update function call.
15959 (distribute_loop): Remove code handling loop with unknown niters.
15960 (pass_loop_distribution::execute): Skip loop with unknown niters.
15962 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15965 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
15966 function sort_partitions_by_post_order.
15968 2017-07-17 Bin Cheng <bin.cheng@arm.com>
15970 PR tree-optimization/81374
15971 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
15972 the max index of basic blocks, rather than number of basic blocks.
15974 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
15976 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
15978 (arc_legitimate_pic_operand_p): Likewise.
15979 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
15981 (arc_needs_pcl_p): Likewise.
15982 (arc_legitimate_pc_offset_p): Likewise.
15983 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
15984 function is also used in constrains.md.
15985 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
15986 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
15987 PLUS. Only return true/false in known cases, otherwise assert.
15988 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
15989 is already called in arc_legitimate_constant_p.
15990 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
15992 (LEGITIMATE_PIC_OPERAND_P): Use
15993 arc_raw_symbolic_reference_mentioned_p function.
15994 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
15999 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
16000 Andrew Burgess <andrew.burgess@embecosm.com>
16002 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
16003 (arc_return_address_register): New function.
16004 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
16005 (arc_handle_fndecl_attribute): Add naked attribute.
16006 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
16007 (TARGET_WARN_FUNC_RETURN): Likewise.
16008 (arc_allocate_stack_slots_for_args): New function.
16009 (arc_warn_func_return): Likewise.
16010 (machine_function): Change type fn_type.
16011 (arc_compute_function_type): Consider new naked function type,
16012 change function return type.
16013 (arc_must_save_register): Adapt to handle new
16014 arc_compute_function_type's return type.
16015 (arc_expand_prologue): Likewise.
16016 (arc_expand_epilogue): Likewise.
16017 (arc_return_address_regs): Delete.
16018 (arc_return_address_register): New function.
16019 (arc_epilogue_uses): Use above function.
16020 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
16021 (arc_function_type): Change encoding, add naked type.
16022 (ARC_INTERRUPT_P): Change to handle the new encoding.
16023 (ARC_FAST_INTERRUPT_P): Likewise.
16024 (ARC_NORMAL_P): Define.
16025 (ARC_NAKED_P): Likewise.
16026 (arc_compute_function_type): Delete prototype.
16027 * config/arc/arc.md (in_ret_delay_slot): Use
16028 arc_return_address_register function.
16029 (simple_return): Likewise.
16030 (p_return_i): Likewise.
16032 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16034 PR tree-optimization/81428
16035 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
16036 can't be built for those types.
16038 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16040 Remove stuff dead since r239246.
16042 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
16043 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
16044 (avr_inform_devices): Remove dead stuff.
16046 2017-07-17 Tamar Christina <tamar.christina@arm.com>
16048 * config/arm/arm_neon.h: Fix softp typo.
16050 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16052 PR tree-optimization/81365
16053 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
16054 aggregate moves onto bb predecessor edges, make sure there are no
16055 loads that could alias the lhs in between the start of bb and the
16058 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
16061 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
16062 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
16063 [LSHIFTRT, outer_code = TRUNCATE]: Same.
16065 2017-07-17 Jakub Jelinek <jakub@redhat.com>
16067 PR tree-optimization/81396
16068 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
16069 (init_symbolic_number): Initialize it to 1.
16070 (perform_symbolic_merge): Add n_ops from both operands into the new
16072 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
16073 without base_addr as useless if they need more than one operation.
16074 (bswap_replace): Handle !bswap case for NULL base_addr.
16076 2017-07-17 Tom de Vries <tom@codesourcery.com>
16079 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
16082 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
16084 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
16085 conditional builtin define __FIX_LEON3FT_B2BST.
16087 2017-07-17 Daniel Cederman <cederman@gaisler.com>
16089 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
16090 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
16093 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16095 PR rtl-optimization/81424
16096 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
16097 to remove potential trapping from operands if -fnon-call-exceptions.
16099 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16101 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
16102 profile_proability for scalling.
16103 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
16105 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16107 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
16109 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16111 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
16112 fixpoint arithmetics.
16114 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16116 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
16117 fixpoint arithmetics.
16119 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16121 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
16122 fixpoint arithmetics.
16124 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
16126 * profile-count.h (profile_probability::from_reg_br_prob_note,
16127 profile_probability::to_reg_br_prob_note): New functions.
16128 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
16129 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
16130 * predict.c (probability_reliable_p): Update.
16131 (edge_probability_reliable_p): Update.
16132 (br_prob_note_reliable_p): Update.
16133 (invert_br_probabilities): Update.
16134 (add_reg_br_prob_note): New function.
16135 (combine_predictions_for_insn): Update.
16136 * asan.c (asan_clear_shadow): Update.
16137 * cfgbuild.c (compute_outgoing_frequencies): Update.
16138 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
16139 (update_br_prob_note): Update.
16140 (rtl_verify_edges): Update.
16141 (purge_dead_edges): Update.
16142 (fixup_reorder_chain): Update.
16143 * emit-rtl.c (try_split): Update.
16144 * ifcvt.c (cond_exec_process_insns): Update.
16145 (cond_exec_process_if_block): Update.
16146 (dead_or_predicable): Update.
16147 * internal-fn.c (expand_addsub_overflow): Update.
16148 (expand_neg_overflow): Update.
16149 (expand_mul_overflow): Update.
16150 * loop-doloop.c (doloop_modify): Update.
16151 * loop-unroll.c (compare_and_jump_seq): Update.
16152 * optabs.c (emit_cmp_and_jump_insn_1): Update.
16153 * predict.h: Update.
16154 * reorg.c (mostly_true_jump): Update.
16156 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
16157 * config/alpha/alpha.c (emit_unlikely_jump): Update.
16158 * config/arc/arc.c: (emit_unlikely_jump): Update.
16159 * config/arm/arm.c: (emit_unlikely_jump): Update.
16160 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
16161 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
16162 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
16163 (ix86_print_operand): Update.
16164 (ix86_split_fp_branch): Update.
16165 (predict_jump): Update.
16166 * config/ia64/ia64.c (ia64_print_operand): Update.
16167 * config/mmix/mmix.c (mmix_print_operand): Update.
16168 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
16169 (rs6000_expand_split_stack_prologue): Update.
16170 * config/rs6000/rs6000.c: Update.
16171 * config/s390/s390.c (s390_expand_vec_strlen): Update.
16172 (s390_expand_vec_movstr): Update.
16173 (s390_expand_cs_tdsi): Update.
16174 (s390_expand_split_stack_prologue): Update.
16175 * config/sh/sh.c (sh_print_operand): Update.
16176 (expand_cbranchsi4): Update.
16177 (expand_cbranchdi4): Update.
16178 * config/sparc/sparc.c (output_v9branch): Update.
16179 * config/spu/spu.c (get_branch_target): Update.
16180 (ea_load_store_inline): Update.
16181 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
16182 * config/tilepro/tilepro.c: Update.
16184 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
16186 * gimplify.c (mostly_copy_tree_r): Revert latest change.
16187 (gimplify_save_expr): Likewise.
16189 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16191 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
16193 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16195 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
16197 * timevar.def (TV_IPA_FNSUMMARY): Define.
16199 2017-07-16 Daniel Cederman <cederman@gaisler.com>
16201 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
16202 to back store errata sensitive sequence from being generated.
16203 (sqrtdf2_fix): Likewise.
16205 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16207 * tree-ssa-threadupdate.c (compute_path_counts,
16208 update_joiner_offpath_counts): Use profile_probability.
16210 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
16213 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16215 * config/arm/arm-c.c (arm_cpu_builtins): Define
16216 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16218 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
16220 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
16221 array entries to represent __ieee128 versions of the
16222 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
16223 scalar_extract_sig, and scalar_insert_exp built-in functions.
16224 (altivec_resolve_overloaded_builtin): Add special case handling
16225 for the __builtin_scalar_insert_exp function, as represented by
16226 the P9V_BUILTIN_VEC_VSIEDP constant.
16227 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
16228 exponent support for __ieee128 argument.
16229 (VSESQP): Add scalar extract signature support for __ieee128
16231 (VSTDCNQP): Add scalar test negative support for __ieee128
16233 (VSIEQP): Add scalar insert exponent support for __int128 argument
16234 with __ieee128 result.
16235 (VSIEQPF): Add scalar insert exponent support for __ieee128
16236 argument with __ieee128 result.
16237 (VSTDCQP): Add scalar test data class support for __ieee128
16239 (VSTDCNQP): Add overload support for scalar test negative with
16240 __ieee128 argument.
16241 (VSTDCQP): Add overload support for scalar test data class
16242 __ieee128 argument.
16243 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
16244 UNSPEC_VSX_SXSIGDP.
16245 (UNSPEC_VSX_SIEXPQP): New constant.
16246 (xsxexpqp): New insn for VSX scalar extract exponent quad
16248 (xsxsigqp): New insn for VSX scalar extract significand quad
16250 (xsiexpqpf): New insn for VSX scalar insert exponent quad
16251 precision with floating point argument.
16252 (xststdcqp): New expand for VSX scalar test data class quad
16254 (xststdcnegqp): New expand for VSX scalar test negative quad
16256 (xststdcqp): New insn to match expansions for VSX scalar test data
16257 class quad precision and VSX scalar test negative quad precision.
16258 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
16259 special case operand checking to enforce that second operand of
16260 VSX scalar test data class with quad precision argument is a 7-bit
16262 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
16263 prototypes and descriptions of __ieee128 versions of
16264 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
16265 scalar_test_data_class, and scalar_test_neg built-in functions.
16267 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16269 PR tree-optimization/81162
16270 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
16271 replace a negate with an add.
16273 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
16275 * doc/invoke.texi (arm/-mcpu): Document +crypto.
16277 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16279 * config/arm/arm-c.c (arm_cpu_builtins): Define
16280 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
16282 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16284 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
16285 (armv8-r): Set ARM Cortex-R52 as default CPU.
16286 * config/arm/arm-tables.opt: Regenerate.
16287 * config/arm/arm-tune.md: Regenerate.
16288 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16290 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
16291 extension for -mcpu=cortex-r52.
16293 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
16295 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
16296 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
16297 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
16298 (fp-armv8): Define it as FP_ARMv8 only.
16299 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
16300 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
16302 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
16303 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
16304 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
16305 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
16306 than TARGET_FPU_ARMV8.
16307 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
16308 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
16309 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
16311 * config/arm/neon.md (neon_vrint): Likewise.
16312 (neon_vcvt): Likewise.
16313 (neon_<fmaxmin_op><mode>): Likewise.
16314 (<fmaxmin><mode>3): Likewise.
16315 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
16316 * config/arm/predicates.md (arm_cond_move_operator): Check against
16317 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
16319 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
16321 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
16322 to top of function.
16324 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16326 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
16327 loop in comment with memset.
16329 2017-07-14 Martin Liska <mliska@suse.cz>
16331 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
16332 * dwarf2out.c (is_java): Remove the function.
16333 (output_pubname): Remove usage of the function.
16334 (lower_bound_default): Remove usage of DW_LANG_Java.
16335 (gen_compile_unit_die): Likewise.
16336 * gcc.c: Remove compiler defaults for .java and .zip files.
16337 * gimple-expr.c (remove_suffix): Change as there's no longer
16338 extension than 4-letter one.
16339 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
16340 (gimplify_save_expr): Likewise.
16341 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
16342 as it's possible even for other languages than Java.
16343 * langhooks.h (struct lang_hooks): Remove Java from a comment.
16344 * lto-opts.c (lto_write_options): Remove reference to Java.
16345 * opts.c (strip_off_ending): Update file extension handling.
16346 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
16347 * tree-eh.c (lower_resx): Likewise.
16348 * tree.c (free_lang_data_in_type): Remove dead code.
16349 (find_decls_types_r): Likewise.
16350 (build_common_builtin_nodes): Remove Java from a comment.
16351 (verify_type): Remove dead code.
16352 * varasm.c (assemble_external): Remove Java from a comment.
16354 2017-07-14 Martin Liska <mliska@suse.cz>
16356 * opts.c (finish_options): Add quotes.
16357 (common_handle_option): Likewise.
16359 2017-07-14 Martin Liska <mliska@suse.cz>
16361 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
16362 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
16363 Remove N_SO_PASCAL.
16364 * dwarf2out.c (lower_bound_default): Do not handle
16366 (gen_compile_unit_die): Likewise.
16367 * gcc.c: Remove default extension binding for GNU Pascal.
16368 * stmt.c: Remove Pascal language from a comment.
16369 * xcoffout.c: Likewise.
16371 2017-07-13 David Malcolm <dmalcolm@redhat.com>
16374 * diagnostic-show-locus.c (fixit_cmp): New function.
16375 (layout::layout): Sort m_fixit_hints.
16376 (column_range::column_range): Assert that the values are valid.
16377 (struct char_span): New struct.
16378 (correction::overwrite): New method.
16379 (struct source_line): New struct.
16380 (line_corrections::add_hint): Add assertions. Reimplement memcpy
16381 calls in terms of classes source_line and char_span, and
16382 correction::overwrite.
16383 (selftest::test_overlapped_fixit_printing_2): New function.
16384 (selftest::diagnostic_show_locus_c_tests): Call it.
16386 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
16388 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
16389 early if there is no lhs.
16391 2017-07-13 Martin Liska <mliska@suse.cz>
16393 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
16394 (gen_reference_type_die): Likewise.
16395 * stor-layout.c: Remove Pascal-related comment.
16397 2017-07-13 Martin Liska <mliska@suse.cz>
16399 * opts.c (finish_options): Add quotes to error messages.
16400 (parse_sanitizer_options): Likewise.
16402 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
16404 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
16406 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
16408 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
16410 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16412 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
16414 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
16416 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16419 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
16420 provides the hardware capability bits, define the macro
16421 __BUILTIN_CPU_SUPPORTS__.
16422 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
16423 if GLIBC does not provide the hardware capability bits. Add a
16424 gcc_unreachable call if the built-in cpu function is neither
16425 __builtin_cpu_is nor __builtin_cpu_supports.
16426 (rs6000_get_function_versions_dispatcher): Change the warning
16427 that an old GLIBC is used which does not export the capability
16428 bits to be an error.
16429 * doc/extend.texi (target_clones attribute): Document the
16430 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
16431 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
16432 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
16433 the macros defined by GCC if the newer GLIBC is available.
16435 2017-07-12 Jeff Law <law@redhat.com>
16437 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
16438 remaining includes slightly.
16439 * config/riscv/riscv-builtins.c: Include profile-count.h.
16441 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16444 * config/avr/avr.c (avr_set_current_function): In diagnostic
16445 messages: Quote keywords and (parts of) identifiers.
16446 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
16449 2017-07-12 Carl Love <cel@us.ibm.com>
16451 * config/rs6000/rs6000-c.c: Add support for built-in functions
16452 vector bool char vec_revb (vector bool char);
16453 vector bool short vec_revb (vector short char);
16454 vector bool int vec_revb (vector bool int);
16455 vector bool long long vec_revb (vector bool long long);
16456 * doc/extend.texi: Update the built-in documentation file for the
16457 new built-in functions.
16459 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16461 * config/s390/s390.md: Remove movcc splitter.
16463 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
16465 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
16466 load/store on condition.
16468 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
16471 * config/avr/avr.c (avr_encode_section_info)
16472 [progmem && !TREE_READONLY]: Error if progmem object needs
16475 2017-07-11 Michael Collison <michael.collison@arm.com>
16477 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
16480 2017-07-11 Carl Love <cel@us.ibm.com>
16482 * config/rs6000/rs6000-c.c: Add support for builtins
16483 vector unsigned int vec_parity_lsbb (vector signed int);
16484 vector unsigned int vec_parity_lsbb (vector unsigned int);
16485 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
16486 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
16487 vector unsigned long long vec_parity_lsbb (vector signed long long);
16488 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
16489 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16490 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
16491 * doc/extend.texi: Update the built-in documentation file for the
16492 new built-in functions.
16494 2017-07-11 David Malcolm <dmalcolm@redhat.com>
16496 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
16497 (layout::m_primary_loc): New field.
16498 (layout::layout): Initialize new field. Move location filtering
16499 logic from here to...
16500 (layout::maybe_add_location_range): ...this new method. Add
16501 support for filtering to just the lines already specified by other
16503 (layout::will_show_line_p): New method.
16504 (gcc_rich_location::add_location_if_nearby): New method.
16505 (selftest::test_add_location_if_nearby): New test function.
16506 (selftest::diagnostic_show_locus_c_tests): Call it.
16507 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
16510 2017-07-11 Tom de Vries <tom@codesourcery.com>
16512 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
16513 (bb_first_real_insn): New function.
16514 (nvptx_single): Add extra initialization of broadcasted condition
16517 2017-07-11 Nathan Sidwell <nathan@acm.org>
16519 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
16521 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
16523 * doc/extend.texi (AVR Function Attributes): Remove weblink to
16524 Binutils doc as TEXI will mess them up.
16525 * doc/invoke.texi (AVR Options): Same here.
16527 2017-07-11 Daniel Cederman <cederman@gaisler.com>
16529 * config/sparc/sparc.opt (mfix-ut700): New option.
16530 (mfix-gr712rc): Likewise.
16531 (sparc_fix_b2bst): New variable.
16532 * doc/invoke.texi (SPARC options): Document them.
16533 (ARM options): Fix warnings.
16534 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
16535 instructions to prevent sequences that can trigger the store-store
16536 errata for certain LEON3FT processors.
16537 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
16538 (sparc_option_override): Set sparc_fix_b2bst appropriately.
16539 * config/sparc/sparc.md (fix_b2bst): New attribute.
16540 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
16542 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
16545 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
16547 (*rsqrtsf2_sse): Ditto.
16549 (div<mode>3): Macroize insn from divdf3 and divsf3
16550 using MODEF mode iterator.
16552 2017-07-10 Martin Sebor <msebor@redhat.com>
16554 PR tree-optimization/80397
16555 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
16556 instead of testing for equality to INTEGER_TYPE.
16558 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
16560 * config.gcc: Remove uclibc from arc target spec.
16562 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
16564 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
16566 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
16569 * lto-wrapper.c (remove_option): New function.
16570 (merge_and_complain): Merge PIC/PIE options more realistically.
16572 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16574 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
16578 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
16579 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
16580 * config.in: Regenerate.
16581 * configure: Regenerate.
16582 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
16583 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
16584 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
16585 (TARGET_GASISR_PROLOGUES): ...target mask.
16586 * common/config/avr/avr-common.c
16587 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
16588 Set -mgas-isr-prologues.
16589 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
16590 INSERT_PASS_BEFORE for it.
16591 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
16592 * config/avr/avr.c (avr_option_override)
16593 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
16594 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
16595 (avr_attribute_table) <no_gccisr>: Add new function attribute.
16596 (avr_set_current_function) <is_no_gccisr>: Init machine field.
16597 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
16599 (make_avr_pass_pre_proep): New function.
16600 (emit_push_sfr) <treg>: Add argument to function and use it
16601 instead of TMP_REG.
16602 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
16603 and set machine->gasisr.yes.
16604 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
16605 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
16606 __gcc_isr.n_pushed to .L__stack_usage.
16607 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
16608 (avr_asm_final_postscan_insn): ...this new static function.
16609 * config/avr/avr.h (machine_function)
16610 <is_no_gccisr, use_L__stack_usage>: New fields.
16611 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
16612 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
16613 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
16614 (gasisr, *gasisr): New expander and insn.
16615 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
16616 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
16617 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
16619 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
16621 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
16624 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
16626 Move jump-tables out of .text again.
16629 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
16630 (ASM_OUTPUT_ADDR_VEC): New function.
16631 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
16632 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
16633 INSN_ADDRESSes as asm comment.
16634 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
16635 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
16636 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
16637 * config/avr/avr.md (*tablejump): Adjust comment.
16638 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
16639 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
16641 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
16642 (avr_output_addr_vec): New proto.
16643 (avr_log_t) <insn_addresses>: New field.
16645 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
16648 * config/i386/i386.c (ix86_function_arg_advance): Set
16649 outgoing_args_on_stack to true if there are outgoing arguments
16651 (ix86_function_arg): Likewise.
16652 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
16653 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
16654 * config/i386/i386.h (machine_function): Add
16655 outgoing_args_on_stack.
16657 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
16659 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
16660 supporting pthreds.
16661 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
16663 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
16665 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
16666 (REAL_H): Remove $(MACHMODE_H).
16667 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
16669 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
16670 $(MACHMODE_H) and double-int.h.
16671 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
16673 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
16676 2017-07-07 Andrew Pinski <apinski@cavium.com>
16678 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
16679 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
16681 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
16683 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
16684 Add warning if GCC was not configured to link against a GLIBC that
16685 exports the hardware capability bits.
16686 (make_resolver_func): Make resolver function private and not a
16687 COMDAT function. Create the name with clone_function_name instead
16688 of make_unique_name.
16691 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
16692 correct operand in doing the split.
16694 2017-07-07 Carl Love <cel@us.ibm.com>
16696 * config/rs6000/rs6000-c: Add support for built-in function
16697 vector unsigned short vec_pack_to_short_fp32 (vector float,
16699 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
16700 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
16701 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
16702 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
16703 (convert_4f32_8i16): Add define_expand.
16704 * doc/extend.texi: Update the built-in documentation file for the
16705 new built-in function.
16707 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16709 * config/sparc/m8.md: New file.
16710 * config/sparc/sparc.md: Include m8.md.
16712 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16714 * config/sparc/sparc.opt: New option -mvis4b.
16715 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
16716 (sparc_option_override): Handle VIS4B.
16717 (enum sparc_builtins): Define
16718 SPARC_BUILTIN_DICTUNPACK{8,16,32},
16719 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
16720 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
16721 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
16722 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
16723 (check_constant_argument): New function.
16724 (sparc_vis_init_builtins): Define builtins
16725 __builtin_vis_dictunpack{8,16,32},
16726 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
16727 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
16728 __builtin_vis_fpcmpde{8,16,32}shl and
16729 __builtin_vis_fpcmpur{8,16,32}shl.
16730 (sparc_expand_builtin): Check that the constant operands to
16731 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
16732 constant and in range.
16733 * config/sparc/sparc-c.c (sparc_target_macros): Handle
16735 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
16736 (SPARC_IMM5_P): Likewise.
16737 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
16738 (enabled): Handle vis4b.
16739 (UNSPEC_DICTUNPACK): New unspec.
16740 (UNSPEC_FPCMPSHL): Likewise.
16741 (UNSPEC_FPUCMPSHL): Likewise.
16742 (UNSPEC_FPCMPDESHL): Likewise.
16743 (UNSPEC_FPCMPURSHL): Likewise.
16744 (cpu_feature): New CPU feature `vis4b'.
16745 (dictunpack{8,16,32}): New insns.
16746 (FPCSMODE): New mode iterator.
16747 (fpcscond): New code iterator.
16748 (fpcsucond): Likewise.
16749 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
16750 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
16751 (fpcmpde{8,16,32}{si,di}shl): Likewise.
16752 (fpcmpur{8,16,32}{si,di}shl): Likewise.
16753 * config/sparc/constraints.md: Define constraints `q' for unsigned
16754 2-bit integer constants and `t' for unsigned 5-bit integer
16756 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
16758 (imm5_operand_dictunpack16): Likewise.
16759 (imm5_operand_dictunpack32): Likewise.
16760 (imm2_operand): Likewise.
16761 * doc/invoke.texi (SPARC Options): Document -mvis4b.
16762 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
16763 ditunpack* and fpcmp*shl builtins.
16765 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16767 * config.gcc: Handle m8 in --with-{cpu,tune} options.
16768 * config.in: Add HAVE_AS_SPARC6 define.
16769 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
16771 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
16773 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
16774 (CPP_CPU_SPEC): Handle m8.
16775 (ASM_CPU_SPEC): Likewise.
16776 * config/sparc/sparc-opts.h (enum processor_type): Add
16778 * config/sparc/sparc.c (m8_costs): New struct.
16779 (sparc_option_override): Handle TARGET_CPU_m8.
16780 (sparc32_initialize_trampoline): Likewise.
16781 (sparc64_initialize_trampoline): Likewise.
16782 (sparc_issue_rate): Likewise.
16783 (sparc_register_move_cost): Likewise.
16784 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
16785 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
16786 (ASM_CPU64_DEFAULT_SPEC): Likewise.
16787 (CPP_CPU_SPEC): Handle M8.
16788 (ASM_CPU_SPEC): Likewise.
16789 (AS_M8_FLAG): Define.
16790 * config/sparc/sparc.md: Add m8 to the cpu attribute.
16791 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
16792 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
16794 * configure: Regenerate.
16795 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
16798 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16800 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
16802 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
16803 ("*movdi_insn_sp32"): Do not set v3pipe.
16804 ("*movsi_insn"): Likewise.
16805 ("*movdi_insn_sp64"): Likewise.
16806 ("*movsf_insn"): Likewise.
16807 ("*movdf_insn_sp32"): Likewise.
16808 ("*movdf_insn_sp64"): Likewise.
16809 ("*zero_extendsidi2_insn_sp64"): Likewise.
16810 ("*sign_extendsidi2_insn"): Likewise.
16811 ("*mov<VM32:mode>_insn"): Likewise.
16812 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16813 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16814 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16815 ("<vlop:code><VL:mode>3"): Likewise.
16816 ("*not_<vlop:code><VL:mode>3"): Likewise.
16817 ("*nand<VL:mode>_vis"): Likewise.
16818 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
16819 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
16820 ("one_cmpl<VL:mode>2"): Likewise.
16821 ("faligndata<VM64:mode>_vis"): Likewise.
16822 ("alignaddrsi_vis"): Likewise.
16823 ("alignaddrdi_vis"): Likweise.
16824 ("alignaddrlsi_vis"): Likewise.
16825 ("alignaddrldi_vis"): Likewise.
16826 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16827 ("bmaskdi_vis"): Likewise.
16828 ("bmasksi_vis"): Likewise.
16829 ("bshuffle<VM64:mode>_vis"): Likewise.
16830 ("cmask8<P:mode>_vis"): Likewise.
16831 ("cmask16<P:mode>_vis"): Likewise.
16832 ("cmask32<P:mode>_vis"): Likewise.
16833 ("pdistn<P:mode>_vis"): Likewise.
16834 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16836 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16838 * config/sparc/sparc.md ("subtype"): New insn attribute.
16839 ("*wrgsr_sp64"): Set insn subtype.
16840 ("*rdgsr_sp64"): Likewise.
16841 ("alignaddrsi_vis"): Likewise.
16842 ("alignaddrdi_vis"): Likewise.
16843 ("alignaddrlsi_vis"): Likewise.
16844 ("alignaddrldi_vis"): Likewise.
16845 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
16846 ("fexpand_vis"): Likewise.
16847 ("fpmerge_vis"): Likewise.
16848 ("faligndata<VM64:mode>_vis"): Likewise.
16849 ("bshuffle<VM64:mode>_vis"): Likewise.
16850 ("cmask8<P:mode>_vis"): Likewise.
16851 ("cmask16<P:mode>_vis"): Likewise.
16852 ("cmask32<P:mode>_vis"): Likewise.
16853 ("fchksm16_vis"): Likewise.
16854 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
16855 ("fmean16_vis"): Likewise.
16856 ("fp<plusminus_insn>64_vis"): Likewise.
16857 ("<plusminus_insn>v8qi3"): Likewise.
16858 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
16859 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
16860 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
16861 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
16862 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
16863 ("*movqi_insn"): Likewise.
16864 ("*movhi_insn"): Likewise.
16865 ("*movsi_insn"): Likewise.
16866 ("movsi_pic_gotdata_op"): Likewise.
16867 ("*movdi_insn_sp32"): Likewise.
16868 ("*movdi_insn_sp64"): Likewise.
16869 ("movdi_pic_gotdata_op"): Likewise.
16870 ("*movsf_insn"): Likewise.
16871 ("*movdf_insn_sp32"): Likewise.
16872 ("*movdf_insn_sp64"): Likewise.
16873 ("*zero_extendhisi2_insn"): Likewise.
16874 ("*zero_extendqihi2_insn"): Likewise.
16875 ("*zero_extendqisi2_insn"): Likewise.
16876 ("*zero_extendqidi2_insn"): Likewise.
16877 ("*zero_extendhidi2_insn"): Likewise.
16878 ("*zero_extendsidi2_insn_sp64"): Likewise.
16879 ("ldfsr"): Likewise.
16880 ("prefetch_64"): Likewise.
16881 ("prefetch_32"): Likewise.
16882 ("tie_ld32"): Likewise.
16883 ("tie_ld64"): Likewise.
16884 ("*tldo_ldub_sp32"): Likewise.
16885 ("*tldo_ldub1_sp32"): Likewise.
16886 ("*tldo_ldub2_sp32"): Likewise.
16887 ("*tldo_ldub_sp64"): Likewise.
16888 ("*tldo_ldub1_sp64"): Likewise.
16889 ("*tldo_ldub2_sp64"): Likewise.
16890 ("*tldo_ldub3_sp64"): Likewise.
16891 ("*tldo_lduh_sp32"): Likewise.
16892 ("*tldo_lduh1_sp32"): Likewise.
16893 ("*tldo_lduh_sp64"): Likewise.
16894 ("*tldo_lduh1_sp64"): Likewise.
16895 ("*tldo_lduh2_sp64"): Likewise.
16896 ("*tldo_lduw_sp32"): Likewise.
16897 ("*tldo_lduw_sp64"): Likewise.
16898 ("*tldo_lduw1_sp64"): Likewise.
16899 ("*tldo_ldx_sp64"): Likewise.
16900 ("*mov<VM32:mode>_insn"): Likewise.
16901 ("*mov<VM64:mode>_insn_sp64"): Likewise.
16902 ("*mov<VM64:mode>_insn_sp32"): Likewise.
16904 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16906 * config/sparc/sparc.md ("type"): New insn type viscmp.
16907 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
16909 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
16910 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
16911 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
16912 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
16914 ("n7_vis_logical_11cycle"): Likewise.
16915 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
16916 * config/sparc/niagara2.md ("niag3_vis": Likewise.
16917 * config/sparc/niagara.md ("niag_vis"): Likewise.
16918 * config/sparc/ultra3.md ("us3_fga"): Likewise.
16919 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
16921 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
16923 * config/sparc/sparc.md: New instruction type `bmask'.
16924 (bmaskdi_vis): Use the `bmask' type.
16925 (bmasksi_vis): Likewise.
16926 * config/sparc/ultra3.md (us3_array): Likewise.
16927 * config/sparc/niagara7.md (n7_array): Likewise.
16928 * config/sparc/niagara4.md (n4_array): Likewise.
16929 * config/sparc/niagara2.md (niag2_vis): Likewise.
16930 (niag3_vis): Likewise.
16931 * config/sparc/niagara.md (niag_vis): Likewise.
16933 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16935 * ipa-comdats.c: Remove optimize check from gate.
16936 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
16937 for functions not optimized.
16938 (ipa_fn_summary_read): Skip optimize check.
16939 (ipa_fn_summary_write): Likewise.
16940 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
16942 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
16944 (can_inline_edge_p): Check flag_pcc_struct_return for match.
16945 (check_callers): Give up on caller which is not optimized.
16946 (inline_small_functions): Likewise.
16947 (ipa_inline): Do not give up when not optimizing.
16948 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
16949 away unoptimizes cdtors.
16950 (whole_program_function_and_variable_visibility): Do
16951 ipa_discover_readonly_nonaddressable_vars in LTO mode.
16952 * ipa.c (process_references): Do not check optimize.
16953 (symbol_table::remove_unreachable_nodes): Update optimize check.
16954 (set_writeonly_bit): Update optimize check.
16955 (pass_ipa_cdtor_merge::gate): Do not check optimize.
16956 (pass_ipa_single_use::gate): Remove.
16958 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16960 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
16961 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
16962 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
16963 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
16964 permute_load, permute_store, adjust_extract, adjust_splat,
16965 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
16966 replace_swap_with_copy, dump_swap_insn_table,
16967 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
16968 recombine_lvx_pattern, recombine_stvx_pattern,
16969 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
16970 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
16971 to file rs6000-p8swap.c.
16972 * config/rs6000/rs6000-p8swap.c: New file.
16973 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
16974 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
16975 and rs6000*-*-* targets.
16977 2017-07-06 David Malcolm <dmalcolm@redhat.com>
16979 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
16981 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16983 * lto-wrapper.c (merge_and_complain): Do not merge
16984 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
16985 fsigned_zeros, ftrapping_math, fwrapv.
16986 (append_compiler_options): Do not track these options.
16987 (append_linker_options): Likewie
16989 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16991 * cgraphunit.c (cgraph_node::finalize_function): When
16992 !flag_toplevel_reorde set no_reorder flag.
16993 (varpool_node::finalize_decl): Likewise.
16994 (symbol_table::compile): Drop no toplevel reorder path.
16996 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
16998 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
16999 edges; zero probability is not better than uninitialized.
17001 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17003 * asan.h (asan_sanitize_allocas_p): Declare.
17004 * asan.c (asan_sanitize_allocas_p): New function.
17005 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
17006 (handle_builtin_alloca): Likewise.
17007 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
17008 if !asan_sanitize_allocas_p.
17009 * params.def (asan-instrument-allocas): Add new option.
17010 * params.h (ASAN_PROTECT_ALLOCAS): Define.
17011 * opts.c (common_handle_option): Disable allocas sanitization for
17014 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
17016 * asan.c: Include gimple-fold.h.
17017 (get_last_alloca_addr): New function.
17018 (handle_builtin_stackrestore): Likewise.
17019 (handle_builtin_alloca): Likewise.
17020 (asan_emit_allocas_unpoison): Likewise.
17021 (get_mem_refs_of_builtin_call): Add new parameter, remove const
17022 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
17023 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
17024 (instrument_builtin_call): Pass gimple iterator to
17025 get_mem_refs_of_builtin_call.
17026 (last_alloca_addr): New global.
17027 * asan.h (asan_emit_allocas_unpoison): Declare.
17028 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
17029 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
17030 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
17031 if function calls alloca.
17032 * gimple-fold.c (replace_call_with_value): Remove static keyword.
17033 * gimple-fold.h (replace_call_with_value): Declare.
17034 * internal-fn.c: Include asan.h.
17035 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
17036 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
17038 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17040 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
17041 (C_SELFTEST_FLAGS): New.
17042 (CPP_SELFTEST_FLAGS): New.
17043 (SELFTEST_DEPS): New, from deps of s-selftest.
17044 (C_SELFTEST_DEPS): New, from deps of s-selftest.
17045 (CPP_SELFTEST_DEPS): New.
17046 (selftest): Add dependency on s-selftest-c++.
17047 (s-selftest): Rename to...
17048 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
17049 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
17050 than SELFTEST_FLAGS.
17051 (selftest-gdb): Rename to...
17052 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
17054 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
17055 (selftest-valgrind): Rename to...
17056 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
17058 (selftest-valgrind): Reintroduce as an alias for
17059 selftest-c-valgrind.
17060 (s-selftest-c++): New.
17061 (selftest-c++-gdb): New.
17062 (selftest-c++-valgrind): New.
17064 2017-07-06 Olivier Hainque <hainque@adacore.com>
17066 * gcc.c (process_command): When deciding if undefined variables
17067 should be ignored when processing specs, accept "gcc -v" as well.
17069 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
17071 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
17072 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
17074 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
17076 * config/arm/arm-cpus.in (armv8-r): Add new entry.
17077 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
17078 * config/arm/arm-tables.opt: Regenerate.
17079 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
17081 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
17083 2017-07-06 Carl Love <cel@us.ibm.com>
17085 * ChangeLog: Clean up from mid air collision
17087 2017-07-06 Carl Love <cel@us.ibm.com>
17089 * config/rs6000/rs6000-c.c: Add support for built-in functions
17090 vector signed int vec_subc (vector signed int, vector signed int);
17091 vector signed __int128 vec_subc (vector signed __int128,
17092 vector signed __int128);
17093 vector unsigned __int128 vec_subc (vector unsigned __int128,
17094 vector unsigned __int128);
17095 vector signed int vec_sube (vector signed int, vector signed int,
17096 vector signed int);
17097 vector unsigned int vec_sube (vector unsigned int,
17098 vector unsigned int,
17099 vector unsigned int);
17100 vector signed __int128 vec_sube (vector signed __int128,
17101 vector signed __int128,
17102 vector signed__int128);
17103 vector unsigned __int128 vec_sube (vector unsigned __int128,
17104 vector unsigned __int128,
17105 vector unsigned __int128);
17106 vector signed int vec_subec (vector signed int, vector signed int,
17107 vector signed int);
17108 vector unsigned int vec_subec (vector unsigned int,
17109 vector unsigned int,
17110 vector unsigned int);
17111 vector signed __int128 vec_subec (vector signed __int128,
17112 vector signed __int128,
17113 vector signed__int128);
17114 vector unsigned __int128 vec_subec (vector unsigned __int128,
17115 vector unsigned __int128,
17116 vector unsigned __int128);
17117 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
17118 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
17119 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
17120 BU_ALTIVEC_OVERLOAD_X definitions.
17121 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
17122 * doc/extend.texi: Update the built-in documentation file for the new
17123 built-in functions.
17125 2017-07-06 David Malcolm <dmalcolm@redhat.com>
17128 * diagnostic-show-locus.c (layout::layout): Use start and finish
17129 spelling location for the start and finish of each range.
17130 * genmatch.c (linemap_client_expand_location_to_spelling_point):
17131 Add unused aspect param.
17132 * input.c (expand_location_1): Add "aspect" param, and use it
17133 to access the correct part of the location.
17134 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
17136 (expand_location_to_spelling_point): Likewise.
17137 (linemap_client_expand_location_to_spelling_point): Add "aspect"
17138 param, and pass it to expand_location_1.
17140 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
17142 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
17143 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
17144 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
17145 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
17146 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
17147 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
17148 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
17149 _mm_maskz_getmant_ss): New intrinsics.
17150 (__builtin_ia32_getexpss128_mask): Changed to ...
17151 __builtin_ia32_getexpss128_round ... this.
17152 (__builtin_ia32_getexpsd128_mask): Changed to ...
17153 __builtin_ia32_getexpsd128_round ... this.
17154 * config/i386/i386-builtin-types.def
17155 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
17156 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
17157 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
17158 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
17159 __builtin_ia32_getmantss_mask_round): New builtins.
17160 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
17161 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
17162 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
17163 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
17164 * config/i386/sse.md
17165 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
17166 avx512f_sgetexp<mode><mask_scalar_name>
17167 <round_saeonly_scalar_name> ... this.
17168 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
17169 %0, %1, %2<round_saeonly_op3>}): Changed to ...
17170 vgetexp<ssescalarmodesuffix>
17171 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17172 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
17173 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
17174 avx512f_vgetmant<mode><mask_scalar_name>
17175 <round_saeonly_scalar_name> ... this.
17176 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
17177 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
17178 vgetmant<ssescalarmodesuffix>
17179 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
17180 %0<mask_scalar_operand4>, %1, %2
17181 <round_saeonly_scalar_mask_op4>, %3} ... this.
17182 * config/i386/subst.md (mask_scalar_operand4,
17183 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
17184 round_saeonly_scalar_nimm_predicate): New subst attributes.
17186 2017-07-06 Julia Koval <julia.koval@intel.com>
17188 * config/i386/i386.c (ix86_erase_embedded_rounding):
17189 Remove code for old rounding pattern.
17191 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
17193 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
17195 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
17197 * doc/sourcebuild.texi (Test Directives, Variants of
17198 dg-require-support): Add documentation for dg-require-stack-check.
17200 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
17202 * config/i386/subst.md (mask_scalar, round_scalar,
17203 round_saeonly_scalar): New meta-templates.
17204 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
17205 round_scalar_mask_operand3, round_scalar_mask_op3,
17206 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
17207 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
17208 round_saeonly_scalar_constraint,
17209 round_saeonly_scalar_prefix): New subst attribute.
17210 * config/i386/sse.md
17211 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
17212 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
17213 <round_scalar_name> ... this.
17214 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
17215 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
17216 <round_scalar_name> ... this.
17217 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
17218 <sse>_vm<code><mode>3<mask_scalar_name>
17219 <round_saeonly_scalar_name> ... this.
17220 (v<plusminus_mnemonic><ssescalarmodesuffix>
17221 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17222 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17223 v<plusminus_mnemonic><ssescalarmodesuffix>
17224 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17225 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17226 (v<multdiv_mnemonic><ssescalarmodesuffix>
17227 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
17228 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
17229 v<multdiv_mnemonic><ssescalarmodesuffix>
17230 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17231 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
17232 (v<maxmin_float><ssescalarmodesuffix>
17233 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
17234 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
17235 v<maxmin_float><ssescalarmodesuffix>
17236 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
17237 %0<mask_scalar_operand3>, %1, %<iptr>2
17238 <round_saeonly_scalar_mask_op3>} ... this.
17240 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
17242 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
17243 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
17245 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
17246 Alan Hayward <alan.hayward@arm.com>
17247 David Sherwood <david.sherwood@arm.com>
17249 * combine.c (simplify_if_then_else): Remove "enum" before
17251 * compare-elim.c (can_eliminate_compare): Likewise.
17252 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
17254 (aarch64_lookup_simd_builtin_type): Likewise.
17255 (aarch64_simd_builtin_type): Likewise.
17256 (aarch64_init_simd_builtin_types): Likewise.
17257 (aarch64_simd_expand_args): Likewise.
17258 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
17260 (aarch64_reverse_mask): Likewise.
17261 (aarch64_simd_emit_reg_reg_move): Likewise.
17262 (aarch64_gen_adjusted_ldpstp): Likewise.
17263 (aarch64_ccmp_mode_to_code): Likewise.
17264 (aarch64_operands_ok_for_ldpstp): Likewise.
17265 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17266 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
17268 (aarch64_min_divisions_for_recip_mul): Likewise.
17269 (aarch64_reassociation_width): Likewise.
17270 (aarch64_get_condition_code_1): Likewise.
17271 (aarch64_simd_emit_reg_reg_move): Likewise.
17272 (aarch64_simd_attr_length_rglist): Likewise.
17273 (aarch64_reverse_mask): Likewise.
17274 (aarch64_operands_ok_for_ldpstp): Likewise.
17275 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
17276 (aarch64_gen_adjusted_ldpstp): Likewise.
17277 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
17279 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
17280 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
17281 (arm_lookup_simd_builtin_type): Likewise.
17282 (arm_simd_builtin_type): Likewise.
17283 (arm_init_simd_builtin_types): Likewise.
17284 (arm_expand_builtin_args): Likewise.
17285 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
17286 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
17287 (ft32_setup_incoming_varargs): Likewise.
17288 (ft32_function_arg): Likewise.
17289 (ft32_function_arg_advance): Likewise.
17290 (ft32_pass_by_reference): Likewise.
17291 (ft32_arg_partial_bytes): Likewise.
17292 (ft32_valid_pointer_mode): Likewise.
17293 (ft32_addr_space_pointer_mode): Likewise.
17294 (ft32_addr_space_legitimate_address_p): Likewise.
17295 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
17297 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
17298 (ix86_emit_outlined_ms2sysv_restore): Likewise.
17299 (iamcu_alignment): Likewise.
17300 (canonicalize_vector_int_perm): Likewise.
17301 (ix86_noce_conversion_profitable_p): Likewise.
17302 (ix86_mpx_bound_mode): Likewise.
17303 (ix86_operands_ok_for_move_multiple): Likewise.
17304 * config/microblaze/microblaze-protos.h
17305 (microblaze_expand_conditional_branch_reg): Likewise.
17306 * config/microblaze/microblaze.c
17307 (microblaze_expand_conditional_branch_reg): Likewise.
17308 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
17310 (rs6000_reassociation_width): Likewise.
17311 (rs6000_invalid_binary_op): Likewise.
17312 (fusion_p9_p): Likewise.
17313 (emit_fusion_p9_load): Likewise.
17314 (emit_fusion_p9_store): Likewise.
17315 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
17317 (riscv_hard_regno_mode_ok_p): Likewise.
17318 (riscv_address_insns): Likewise.
17319 (riscv_split_symbol): Likewise.
17320 (riscv_legitimize_move): Likewise.
17321 (riscv_function_value): Likewise.
17322 (riscv_hard_regno_nregs): Likewise.
17323 (riscv_expand_builtin): Likewise.
17324 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
17325 (riscv_build_integer): Likewise.
17326 (riscv_split_integer): Likewise.
17327 (riscv_legitimate_constant_p): Likewise.
17328 (riscv_cannot_force_const_mem): Likewise.
17329 (riscv_regno_mode_ok_for_base_p): Likewise.
17330 (riscv_valid_base_register_p): Likewise.
17331 (riscv_valid_offset_p): Likewise.
17332 (riscv_valid_lo_sum_p): Likewise.
17333 (riscv_classify_address): Likewise.
17334 (riscv_legitimate_address_p): Likewise.
17335 (riscv_address_insns): Likewise.
17336 (riscv_load_store_insns): Likewise.
17337 (riscv_force_binary): Likewise.
17338 (riscv_split_symbol): Likewise.
17339 (riscv_force_address): Likewise.
17340 (riscv_legitimize_address): Likewise.
17341 (riscv_move_integer): Likewise.
17342 (riscv_legitimize_const_move): Likewise.
17343 (riscv_legitimize_move): Likewise.
17344 (riscv_address_cost): Likewise.
17345 (riscv_subword): Likewise.
17346 (riscv_output_move): Likewise.
17347 (riscv_canonicalize_int_order_test): Likewise.
17348 (riscv_emit_int_order_test): Likewise.
17349 (riscv_function_arg_boundary): Likewise.
17350 (riscv_pass_mode_in_fpr_p): Likewise.
17351 (riscv_pass_fpr_single): Likewise.
17352 (riscv_pass_fpr_pair): Likewise.
17353 (riscv_get_arg_info): Likewise.
17354 (riscv_function_arg): Likewise.
17355 (riscv_function_arg_advance): Likewise.
17356 (riscv_arg_partial_bytes): Likewise.
17357 (riscv_function_value): Likewise.
17358 (riscv_pass_by_reference): Likewise.
17359 (riscv_setup_incoming_varargs): Likewise.
17360 (riscv_print_operand): Likewise.
17361 (riscv_elf_select_rtx_section): Likewise.
17362 (riscv_save_restore_reg): Likewise.
17363 (riscv_for_each_saved_reg): Likewise.
17364 (riscv_register_move_cost): Likewise.
17365 (riscv_hard_regno_mode_ok_p): Likewise.
17366 (riscv_hard_regno_nregs): Likewise.
17367 (riscv_class_max_nregs): Likewise.
17368 (riscv_memory_move_cost): Likewise.
17369 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
17370 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
17371 (rl78_addr_space_address_mode): Likewise.
17372 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
17374 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
17375 (rs6000_reassociation_width): Likewise.
17376 (rs6000_invalid_binary_op): Likewise.
17377 (fusion_p9_p): Likewise.
17378 (emit_fusion_p9_load): Likewise.
17379 (emit_fusion_p9_store): Likewise.
17380 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
17381 (ok_for_simple_move_operands): Likewise.
17382 (ok_for_simple_move_strict_operands): Likewise.
17383 (ok_for_simple_arith_logic_operands): Likewise.
17384 (visium_legitimize_reload_address): Likewise.
17385 (visium_select_cc_mode): Likewise.
17386 (output_cbranch): Likewise.
17387 (visium_split_double_move): Likewise.
17388 (visium_expand_copysign): Likewise.
17389 (visium_expand_int_cstore): Likewise.
17390 (visium_expand_fp_cstore): Likewise.
17391 * config/visium/visium.c (visium_pass_by_reference): Likewise.
17392 (visium_function_arg): Likewise.
17393 (visium_function_arg_advance): Likewise.
17394 (visium_libcall_value): Likewise.
17395 (visium_setup_incoming_varargs): Likewise.
17396 (visium_legitimate_constant_p): Likewise.
17397 (visium_legitimate_address_p): Likewise.
17398 (visium_legitimize_address): Likewise.
17399 (visium_secondary_reload): Likewise.
17400 (visium_register_move_cost): Likewise.
17401 (visium_memory_move_cost): Likewise.
17402 (prepare_move_operands): Likewise.
17403 (ok_for_simple_move_operands): Likewise.
17404 (ok_for_simple_move_strict_operands): Likewise.
17405 (ok_for_simple_arith_logic_operands): Likewise.
17406 (visium_function_value_1): Likewise.
17407 (rtx_ok_for_offset_p): Likewise.
17408 (visium_legitimize_reload_address): Likewise.
17409 (visium_split_double_move): Likewise.
17410 (visium_expand_copysign): Likewise.
17411 (visium_expand_int_cstore): Likewise.
17412 (visium_expand_fp_cstore): Likewise.
17413 (visium_split_cstore): Likewise.
17414 (visium_select_cc_mode): Likewise.
17415 (visium_split_cbranch): Likewise.
17416 (output_cbranch): Likewise.
17417 (visium_print_operand_address): Likewise.
17418 * expmed.c (flip_storage_order): Likewise.
17419 * expmed.h (emit_cstore): Likewise.
17420 (flip_storage_order): Likewise.
17421 * genrecog.c (validate_pattern): Likewise.
17422 * hsa-gen.c (gen_hsa_addr): Likewise.
17423 * internal-fn.c (expand_arith_overflow): Likewise.
17424 * ira-color.c (allocno_copy_cost_saving): Likewise.
17425 * lra-assigns.c (find_hard_regno_for_1): Likewise.
17426 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
17427 (process_invariant_for_inheritance): Likewise.
17428 * lra-eliminations.c (move_plus_up): Likewise.
17429 * omp-low.c (lower_oacc_reductions): Likewise.
17430 * simplify-rtx.c (simplify_subreg): Likewise.
17431 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
17432 (TARGET_CHKP_BOUND_MODE): Likewise..
17433 * targhooks.c (default_chkp_bound_mode): Likewise.
17434 (default_setup_incoming_vararg_bounds): Likewise.
17435 * targhooks.h (default_chkp_bound_mode): Likewise.
17436 (default_setup_incoming_vararg_bounds): Likewise.
17437 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
17438 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
17439 (have_whole_vector_shift): Likewise.
17440 * tree-vect-stmts.c (vectorizable_load): Likewise.
17441 * doc/tm.texi: Regenerate.
17443 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17445 Graceful degrade if Binutils PR21472 is not available.
17448 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
17449 .rodata in flash test fails.
17450 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
17451 * confgure: Regenerate.
17452 * config.in: Regenerate.
17453 * config/avr/avr.c (avr_asm_named_section)
17454 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
17455 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
17456 (avr_asm_init_sections): Same.
17458 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17460 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
17461 (fma<VH:mode>4_intrinsic): Likewise.
17462 (*fmsub<VCVTF:mode>4): Likewise.
17463 (*fmsub<VH:mode>4_intrinsic): Likewise.
17465 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
17468 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
17469 Don't depend on "optimize > 0".
17470 (out_movhi_r_mr, out_movqi_mr_r): Same.
17471 (out_movhi_mr_r, out_movqi_r_mr): Same.
17472 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
17473 io_address_operand on "optimize > 0".
17475 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17477 * tree-loop-distribution.c: Add general explanantion on the pass.
17478 (generate_loops_for_partition): Mark distributed loop.
17479 (pg_add_dependence_edges): New parameter. Handle alias data
17480 dependence specially and record it in the parameter if asked.
17481 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
17482 (init_partition_graph_vertices, add_partition_graph_edge): New.
17483 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
17484 (free_partition_graph_vdata, build_partition_graph): New.
17485 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
17486 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
17487 (data_ref_segment_size, latch_dominated_by_data_ref): New.
17488 (compute_alias_check_pairs, version_loop_by_alias_check): New.
17489 (version_for_distribution_p, finalize_partitions): New.
17490 (distribute_loop): Handle alias data dependence specially. Factor
17491 out loop fusion code as functions and call these functions.
17493 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17495 * tree-loop-distribution.c (classify_partition): New parameter and
17496 better handle reduction statement.
17497 (rdg_build_partitions): Revise comment.
17498 (distribute_loop): Compute statements in all partitions and pass it
17499 to classify_partition.
17501 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17503 * tree-loop-distribution.c (enum partition_type): New.
17504 (struct partition): New field type.
17505 (partition_merge_into): Add parameter. Update partition type.
17506 (data_dep_in_cycle_p, update_type_for_merge): New functions.
17507 (build_rdg_partition_for_vertex): Compute partition type.
17508 (rdg_build_partitions): Dump partition type.
17509 (distribute_loop): Update calls to partition_merge_into.
17511 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17513 * tree-loop-distribution.c (struct ddr_hasher): New.
17514 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
17516 (classify_partition): Call get_data_dependence.
17517 (pg_add_dependence_edges): Ditto.
17518 (distribute_loop): Release data dependence hash table.
17520 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17522 * tree-loop-distribution.c (ref_base_address): Delete.
17523 (similar_memory_accesses): Rename ...
17524 (share_memory_accesses): ... to this. Check if partitions access
17525 the same memory reference.
17526 (distribute_loop): Call share_memory_accesses.
17528 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17530 * tree-loop-distribution.c (struct partition): New field recording
17531 its data reference.
17532 (partition_alloc, partition_free): Init and release data refs.
17533 (partition_merge_into): Merge data refs.
17534 (build_rdg_partition_for_vertex): Collect data refs for partition.
17535 (pg_add_dependence_edges): Change parameters from vector to bitmap.
17537 (distribute_loop): Remve data refs from vertice data of partition
17540 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17542 * tree-loop-distribution.c (params.h): Include header file.
17543 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
17544 (datarefs_vec): New global var.
17545 (create_rdg_vertices): Use datarefs_vec directly.
17546 (free_rdg): Don't free data references.
17547 (build_rdg): Update use. Don't free data references.
17548 (distribute_loop): Compute global variable for data references.
17549 Bail out if there are too many data references.
17551 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17553 * tree-loop-distribution.c (loop_nest): New global var.
17554 (build_rdg): Use loop directly, rather than loop nest.
17555 (pg_add_dependence_edges): Remove loop nest parameter. Use global
17557 (distribute_loop): Compute global variable loop nest. Update use.
17559 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17561 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
17562 (partition_merge_into): New parameter. Dump reason for fusion.
17563 (distribute_loop): Update use of partition_merge_into.
17565 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17567 * tree-loop-distribution.c (bb_top_order_index): New.
17568 (bb_top_order_index_size, bb_top_order_cmp): New.
17569 (stmts_from_loop): Use topological order.
17570 (pass_loop_distribution::execute): Compute and release topological
17571 order for basic blocks.
17573 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17575 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
17578 2017-07-05 Bin Cheng <bin.cheng@arm.com>
17580 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
17581 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
17582 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
17583 * internal-fn.def (LOOP_DIST_ALIAS): New.
17584 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
17585 (fold_loop_internal_call): ... this.
17586 (vect_loop_dist_alias_call): New function.
17587 (set_uid_loop_bbs): Call fold_loop_internal_call.
17588 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
17591 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17594 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
17595 Require dead FLAGS_REG at the beginning of a peephole.
17597 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
17600 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
17601 arguments in the call to __builtin_ia32_sbb_u32.
17602 (_subborrow_u64): Swap _X and _Y arguments in the call to
17603 __builtin_ia32_sbb_u64.
17605 2017-07-04 Jakub Jelinek <jakub@redhat.com>
17608 * tree-vrp.c (compare_assert_loc): Turn into a function template
17609 with stable template parameter. Only test if a->e is NULL,
17610 !a->e == !b->e has been verified already. Use e == NULL or
17611 e != NULL instead of e or ! e tests. If stable is true, don't use
17612 iterative_hash_expr, on the other side allow a or b or both NULL
17613 and sort the NULLs last.
17614 (process_assert_insertions): Sort using compare_assert_loc<false>
17615 instead of compare_assert_loc, later sort using
17616 compare_assert_loc<true> before calling process_assert_insertions_for
17617 in a loop. Use break instead of continue once seen NULL pointer.
17619 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
17621 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
17622 Cortex-R7 and Cortex-R8 processors.
17624 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17626 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
17627 uninitialized while src is not.
17629 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
17631 * common/config/arm/arm-common.c: Adjust include path for
17633 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
17634 (arm-cpu.h): Create in build directory. Adjust dependency rules.
17635 (arm-cpu-data.h): Likewise.
17636 (arm-cpu-cdata.h): Likewise.
17637 * config/arm/arm-cpu.h: Delete.
17638 * config/arm/arm-cpu-cdata.h: Delete.
17639 * config/arm/arm-cpu-data.h: Delete.
17641 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
17643 * config/arm/arm-cpus.in (cortex-a55): New.
17644 (cortex-a75): Likewise.
17645 (cortex-a75.cortex-a55): Likewise.
17646 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
17648 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
17649 * config/arm/arm-cpu-cdata.h: Regenerate.
17650 * config/arm/arm-cpu-data.h: Regenerate.
17651 * config/arm/arm-cpu.h: Regenerate.
17652 * config/arm/arm-tables.opt: Regenerate.
17653 * config/arm/arm-tune.md: Regenerate.
17655 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17657 * haifa-sched.c (sched_create_recovery_edges): Update profile.
17659 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
17661 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
17664 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
17666 PR tree-optimization/81292
17667 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
17668 full_string_p, also call adjust_related_strinfos if the adjustment
17669 is simple, otherwise invalidate related strinfos.
17671 2017-07-04 Martin Liska <mliska@suse.cz>
17674 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
17675 newly created variable as DECL_IGNORED_P.
17677 2017-07-04 Martin Liska <mliska@suse.cz>
17680 * ipa-inline.c (inline_small_functions):
17681 Use xstrdup_for_dump.
17683 2017-07-04 Tom de Vries <tom@codesourcery.com>
17685 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
17687 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17690 * config/darwin.c (darwin_function_switched_text_sections):
17693 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
17695 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
17697 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17699 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
17701 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17703 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
17704 min_profitable_iters, and th as inclusive lower bounds.
17705 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
17706 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
17707 for min_profitable_iters and min_profitable_estimate.
17708 (vect_transform_loop): Treat th as an inclusive lower bound.
17709 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
17711 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
17714 * config/darwin.c (darwin_function_switched_text_sections):
17715 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
17716 in two pieces, and suppress the use of buf.
17718 2017-07-03 Nathan Sidwell <nathan@acm.org>
17720 * hash-table.h (hash_table_mod1): Fix indentation.
17722 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17724 PR middle-end/81290
17725 * predict.c (force_edge_cold): Be more careful about propagation
17727 * profile-count.h (profile_probability::guessed,
17728 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
17730 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
17732 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
17734 * doc/invoke.texi (rcpc architecture extension): Document it.
17736 2017-07-03 Richard Biener <rguenther@suse.de>
17738 PR tree-optimization/60510
17739 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
17740 the scalar reduction PHI and use it.
17741 (vectorizable_reduction): Properly guard the single_defuse_cycle
17742 path for non-SLP reduction chains where we cannot use it.
17743 Rework reduc_def/index and vector type deduction. Rework
17744 vector operand gathering during reduction op code-gen.
17745 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
17746 chains dissolve the chain and leave it to non-SLP reduction
17749 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17751 * tree-data-ref.h (dr_alignment): Declare.
17752 * tree-data-ref.c (dr_alignment): New function.
17753 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
17754 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
17756 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
17758 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17760 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
17761 and base_misalignment fields.
17762 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
17763 * tree-data-ref.c: Include builtins.h.
17764 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
17765 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
17766 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
17767 * tree-vect-data-refs.c: Include tree-cfg.h.
17768 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
17769 fields instead of calculating an alignment here.
17770 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
17771 innermost_loop_behavior fields.
17773 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17775 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
17777 (DR_STEP_ALIGNMENT): New macro.
17778 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
17779 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
17780 (create_data_ref): Print it.
17781 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
17782 to tell whether the step preserves vector (mis)alignment.
17783 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17784 Move the check for an integer step and generalise to all INTEGER_CST.
17785 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
17786 Print the outer step alignment.
17788 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17790 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
17791 with offset_alignment.
17792 (DR_ALIGNED_TO): Delete.
17793 (DR_OFFSET_ALIGNMENT): New macro.
17794 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
17795 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
17796 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
17797 (create_data_ref): Likewise.
17798 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
17799 (vect_analyze_data_refs): Likewise.
17800 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
17801 creating dummy innermost behavior.
17803 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17805 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
17806 with a "innermost_loop_behavior *" and refeence tree.
17807 * tree-data-ref.c (dr_analyze_innermost): Likewise.
17808 (create_data_ref): Update call accordingly.
17809 * tree-predcom.c (find_looparound_phi): Likewise.
17811 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17813 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
17814 fields with dr_wrt_vec_loop.
17815 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
17816 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
17817 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
17818 (vect_dr_behavior): New function.
17819 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17820 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
17821 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
17822 track whether the step preserves the misalignment.
17823 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
17824 Use vect_dr_behavior.
17825 (vect_setup_realignment): Update call accordingly.
17826 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
17827 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
17828 call to vect_create_addr_base_for_vector_ref.
17829 (vect_create_cond_for_align_checks): Likewise.
17830 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
17831 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
17832 (vect_recog_mask_conversion_pattern): Likewise.
17833 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
17834 (new_stmt_vec_info): Remove redundant zeroing.
17836 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
17838 * common/config/arm/arm-common.c (arm_be8_option): New function.
17839 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
17840 (ISA_ARMv6): Add isa_bit_be8.
17841 * config/arm/arm.h (arm_be8_option): Add prototype.
17842 (BE8_SPEC_FUNCTION): New define.
17843 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
17844 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
17845 (mlittle-endian): Similarly.
17846 (mbe8, mbe32): New options.
17847 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
17848 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
17850 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17852 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
17854 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17856 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
17857 (cleanup_tree_cfg_bb): Use it.
17858 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
17860 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
17862 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17865 * loop-doloop.c (add_test): Update profile.
17867 2017-07-03 Martin Liska <mliska@suse.cz>
17870 * sanopt.c (rewrite_usage_of_param): New function.
17871 (sanitize_rewrite_addressable_params): Likewise.
17872 (pass_sanopt::execute): Call rewrite_usage_of_param.
17874 2017-07-03 Richard Biener <rguenther@suse.de>
17876 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
17877 back to using VIEW_CONVERT_EXPR.
17879 2017-07-03 Martin Liska <mliska@suse.cz>
17882 * doc/extend.texi: Document when a resolver function is
17883 generated for target_clones.
17885 2017-07-03 Martin Liska <mliska@suse.cz>
17887 * asan.c (asan_emit_stack_protection): Unpoison just red zones
17888 and shadow memory of auto variables which are subject of
17889 use-after-scope sanitization.
17890 (asan_expand_mark_ifn): Add do set only when is_poison.
17892 2016-07-03 Richard Biener <rguenther@suse.de>
17894 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
17896 (vect_force_simple_reduction): Record reduction def -> phi mapping.
17897 (vectorizable_reduction): Perform reduction PHI creation when
17898 visiting a reduction PHI and adjust and simplify code generation
17899 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
17900 (vect_transform_loop): Visit reduction PHIs.
17901 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
17902 defs into the SLP tree.
17903 (vect_build_slp_tree): Reduction defs terminate the recursion.
17904 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
17906 (vect_get_vec_defs_for_stmt_copy): Export.
17907 (vect_get_vec_defs): Likewise.
17908 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
17910 (vect_get_vec_defs_for_stmt_copy): Declare.
17911 (vect_get_vec_defs): Likewise.
17913 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
17915 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
17916 parameter with a "loop" parameter and use it instead of the
17917 loop containing DR_STMT. Don't check simple_iv when doing
17918 BB analysis. Describe the two analysis modes in the comment.
17920 2017-07-03 Tom de Vries <tom@codesourcery.com>
17922 PR tree-optimization/69468
17923 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
17924 (find_same_succ_bb): Handle ignore_edge_flags.
17926 2017-07-03 Tom de Vries <tom@codesourcery.com>
17928 PR tree-optimization/81192
17929 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
17931 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
17933 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
17935 2017-07-03 Tom de Vries <tom@codesourcery.com>
17937 PR tree-optimization/81192
17938 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
17939 BB_SAME_SUCC (bb) == NULL.
17941 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17943 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
17946 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17948 * dumpfile.c: Include profile-count.h
17949 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
17951 (insert_cond_bb): Update profile.
17952 * tree-cfg.h (insert_cond_bb): Update prototype.
17953 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
17954 * tree-dump.c: Do not include tree-cfg.
17956 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17958 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
17960 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17962 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
17965 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
17967 * tree-complex.c (expand_complex_div_wide): update profile.
17969 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
17970 Alan Hayward <alan.hayward@arm.com>
17971 David Sherwood <david.sherwood@arm.com>
17973 * Makefile.in (MACHMODE_H): Remove insn-modes.h
17974 (CORETYPES_H): New define.
17975 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
17976 (insn-modes-inline.h, s-modes-inline-h): New rules.
17977 (generated_files): Add insn-modes-inline.h.
17978 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
17979 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
17980 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
17981 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
17982 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
17983 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
17984 (build/gencodes.o, build/genconditions.o): Likewise.
17985 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
17986 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
17987 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
17988 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
17989 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
17990 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
17991 * coretypes.h: Include everything up to real.h for generators.
17992 Include insn-modes.h first. Include wide-int-print.h after
17993 wide-int.h. Include insn-modes-inline.h and then machmode.h.
17994 * machmode.h: Don't include insn-modes.h here.
17995 * function-tests.c: Remove includes of signop.h, machmode.h,
17996 double-int.h and wide-int.h.
17998 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
18000 * optc-save-gen.awk: Likewise.
18001 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
18002 * godump.c: Remove include of wide-int-print.h.
18003 * pretty-print.h: Likewise.
18004 * wide-int-print.cc: Likewise.
18005 * wide-int.cc: Likewise.
18006 * hash-map-tests.c: Remove include of signop.h.
18007 * hash-set-tests.c: Likewise.
18008 * rtl-tests.c: Likewise.
18009 * mkconfig.sh: Remove include of machmode.h.
18010 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
18012 (emit_insn_modes_inline_h): ...this new function. Emit the code
18013 into an insn-modes-inline.h header file, adding appropriate
18014 include guards and end comments.
18015 (emit_insn_modes_c_header): Remove include of machmode.h.
18016 (emit_min_insn_modes_c_header): Include coretypes.h rather than
18018 (main): Handle -i flag and call emit_insn_modes_inline_h when
18021 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18023 * tree-ssa-strlen.c (strinfo): Rename the length field to
18024 nonzero_chars. Add a full_string_p field.
18025 (compare_nonzero_chars, zero_length_string_p): New functions.
18026 (get_addr_stridx): Add an offset_out parameter.
18027 Use compare_nonzero_chars.
18028 (get_stridx): Update accordingly. Use compare_nonzero_chars.
18029 (new_strinfo): Update after above changes to strinfo.
18030 (set_endptr_and_length): Set full_string_p.
18031 (get_string_length): Update after above changes to strinfo.
18032 (unshare_strinfo): Update call to new_strinfo.
18033 (maybe_invalidate): Likewise.
18034 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
18035 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
18036 as a uhwi instead of an shwi. Update after above changes to
18037 strinfo and new_strinfo.
18038 (zero_length_string): Assert that chainsi contains full strings.
18039 Use zero_length_string_p. Update call to new_strinfo.
18040 (adjust_related_strinfos): Update after above changes to strinfo.
18041 Copy full_string_p from origsi.
18042 (adjust_last_stmt): Use zero_length_string_p.
18043 (handle_builtin_strlen): Update after above changes to strinfo and
18044 new_strinfo. Install the lhs as the string length if the previous
18045 entry didn't describe a full string.
18046 (handle_builtin_strchr): Update after above changes to strinfo
18048 (handle_builtin_strcpy): Likewise.
18049 (handle_builtin_strcat): Likewise.
18050 (handle_builtin_malloc): Likewise.
18051 (handle_pointer_plus): Likewise.
18052 (handle_builtin_memcpy): Likewise. Track nonzero characters
18053 that aren't necessarily followed by a nul terminator.
18054 (handle_char_store): Likewise.
18056 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18058 PR tree-optimization/80769
18059 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
18060 for malloc and calloc. Document the new invariant that all related
18061 strinfos have delayed lengths or none do.
18062 (verify_related_strinfos): Move earlier in file.
18063 (set_endptr_and_length): New function, split out from...
18064 (get_string_length): ...here. Also set the lengths of related
18066 (zero_length_string): Assert that chainsi has known (rather than
18068 (adjust_related_strinfos): Likewise.
18070 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
18072 PR tree-optimization/81136
18073 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
18074 assert that two references with the same misalignment have the same
18075 compile-time misalignment if those compile-time misalignments
18078 2017-07-01 Andi Kleen <ak@linux.intel.com>
18080 * print-tree.c (print_node): Print all attributes.
18082 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18084 * cfg.c (scale_bbs_frequencies): New function.
18085 * cfg.h (scale_bbs_frequencies): Declare it.
18086 * cfgloopanal.c (single_likely_exit): Cleanup.
18087 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
18089 (scale_loop_profile): Likewise.
18090 (loop_version): Likewise.
18091 (create_empty_loop_on_edge): Update.
18092 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
18093 scale_loop_frequencies, scale_loop_profile, loopify,
18094 loop_version): Update prototypes.
18095 * modulo-sched.c (sms_schedule): Update.
18096 * predict.c (unlikely_executed_edge_p): Also check probability.
18097 (probably_never_executed_edge_p): Fix typo.
18098 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18099 * tree-parloops.c (gen_parallel_loop): Update.
18100 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
18101 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18102 * tree-ssa-loop-split.c (split_loop): Update.
18103 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18104 * tree-vect-loop-manip.c (vect_do_peeling): Update.
18105 (vect_loop_versioning): Update.
18106 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18108 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18110 * trans-mem.c (split_bb_make_tm_edge): Update profile.
18112 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18114 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
18115 to keep profile consistent.
18117 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18119 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
18120 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
18121 * profile-count.h (max_safe_multiplier): Make unsigned.
18122 (profile_count::guessed_zero): New.
18124 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18126 * bb-reorder.c (fix_up_crossing_landing_pad,
18127 fix_crossing_conditional_branches): Use make_single_succ_edge
18128 to keep profile consistent.
18130 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
18132 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
18135 2017-07-01 Jakub Jelinek <jakub@redhat.com>
18138 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
18139 the right scopes, make sure cond_jump isn't preserved between multiple
18140 iterations. Search for fallthru edge whenever there are 3+ edges and
18141 use find_fallthru_edge for it.
18143 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18145 Patch by Alexander Monakov <amonakov@ispras.ru>
18146 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
18147 probabilities consistently.
18149 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18151 * pa.c (pa_expand_compare_and_swap_loop): Update call of
18152 emit_cmp_and_jump_insns.
18154 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18157 * tree-inline.c (expand_call_inline): Combine profile statuses.
18159 2017-06-30 Andrew Pinski <apinski@cavium.com>
18161 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
18162 fold_stmt returned true.
18164 2017-06-30 Nathan Sidwell <nathan@acm.org>
18166 * ggc.h (empty_string): Delete.
18167 * cfgexpand.c (expand_asm_stmt): Use plain "".
18168 * optabs.c (expand_asm_memory_barrier): Likewise.
18169 * stringpool.c (empty_string): Delete.
18170 (digit_vector, digit_string): Delete.
18171 (ggc_alloc_string): Use plain "", don't optimize single digit
18172 strings. Use ggc_alloc_atomic.
18174 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
18176 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
18177 comparison set and one other set, use the cost of the non-comparison
18180 2017-06-30 Nathan Sidwell <nathan@acm.org>
18182 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
18185 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
18187 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
18188 loops. Remove now unneeded calls to gimple_switch_set_label() that
18189 just set removed labels to NULL_TREE.
18191 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
18193 * tree-ssanames.c (set_range_info_raw): Abstract from ...
18194 (set_range_info): ...here. Only call set_range_info_raw if domain
18196 (set_nonzero_bits): Call set_range_info_raw.
18197 * tree-ssanames.h (set_range_info_raw): New.
18199 2017-06-30 Jakub Jelinek <jakub@redhat.com>
18202 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
18203 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
18204 of nonimmediate_operand and <store_mask_constraint> instead of m
18205 for the input operand. For V8FI iterator, always split if input
18206 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
18207 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
18208 <store_mask_predicate> instead of register_operand and
18209 <store_mask_constraint> instead of v for the input operand. Make
18210 sure both operands aren't MEMs for if not <mask_applied>.
18212 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
18214 * lto-wrapper.c (copy_file) Close both file descriptors before
18217 2017-06-30 Martin Liska <mliska@suse.cz>
18220 * multiple_target.c (create_dispatcher_calls): Make ifunc
18221 also for function that don't have calls or are not referenced.
18223 2017-06-30 Richard Biener <rguenther@suse.de>
18225 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
18226 analyze the first scalar stmt. Move vector type computation
18227 for the BB case here from ...
18228 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
18229 live operation processing in the SLP case properly.
18231 2017-06-30 Richard Biener <rguenther@suse.de>
18233 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
18235 2017-06-30 Martin Liska <mliska@suse.cz>
18238 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
18239 before BUILT_IN_UNWIND_RESUME when ASAN is used.
18241 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
18243 * doc/invoke.texi (AArch64): Add missing options and remove redundant
18246 2017-06-30 Richard Biener <rguenther@suse.de>
18248 PR tree-optimization/81249
18249 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
18250 condition reduction result to original scalar type.
18252 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18254 * profile-count.h (enum profile_quality): Fix typos and whitespace
18257 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18259 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
18260 type for branch probabilities.
18262 2017-06-29 Julian Brown <julian@codesourcery.com>
18263 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18265 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
18266 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
18267 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
18268 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
18270 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18272 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
18273 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
18274 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
18275 CC usage from generic code to here.
18276 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
18277 CC usage into the target macros.
18279 2017-06-29 Maya Rashish <coypu@sdf.org>
18281 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
18284 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18286 * arm/arm-builtins.c: Include profile-count.h
18287 * except.c (sjlj_emit_function_enter): Use
18288 profile_probability::unlikely.
18290 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
18292 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
18293 and tocrel_offset be pointer args rather than implicitly using
18295 (legitimate_constant_pool_address_p, rs6000_emit_move,
18296 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
18297 tocrel_offset and use in toc_relative_expr_p call.
18298 (print_operand, print_operand_address): Use static tocrel_base_oac
18299 and tocrel_offset_oac.
18300 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
18303 2017-06-29 Maya Rashish <coypu@sdf.org>
18305 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
18307 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
18309 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
18310 objects, take into account only the alignment of 'op0' and 'mode1' if
18313 2017-06-29 Steve Ellcey <sellcey@cavium.com>
18315 * ccmp.c (ccmp_tree_comparison_p): New function.
18316 (ccmp_candidate_p): Update to use above function.
18317 (get_compare_parts): New function.
18318 (expand_ccmp_next): Update to use new functions.
18319 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
18321 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
18322 take mode as argument.
18323 * ccmp.h (expand_ccmp_expr): Add mode as argument.
18324 * expr.c (expand_expr_real_1): Pass mode as argument.
18326 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
18328 * combine.c (combine_instructions): Print insns to dump_file, together
18331 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
18333 * asan.c (asan_emit_stack_protection): Update.
18334 (create_cond_insert_point): Update.
18335 * auto-profile.c (afdo_propagate_circuit): Update.
18336 * basic-block.h (struct edge_def): Turn probability to
18337 profile_probability.
18338 (EDGE_FREQUENCY): Update.
18339 * bb-reorder.c (find_traces_1_round): Update.
18340 (better_edge_p): Update.
18341 (sanitize_hot_paths): Update.
18342 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
18343 (make_single_succ_edge): Update.
18344 (check_bb_profile): Update.
18345 (dump_edge_info): Update.
18346 (update_bb_profile_for_threading): Update.
18347 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
18348 probabilitycount to 0.
18349 * cfgbuild.c (compute_outgoing_frequencies): Update.
18350 * cfgcleanup.c (try_forward_edges): Update.
18351 (outgoing_edges_match): Update.
18352 (try_crossjump_to_edge): Update.
18353 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
18354 (expand_gimple_tailcall): Update.
18355 (construct_init_block): Use make_single_succ_edge.
18356 (construct_exit_block): Use make_single_succ_edge.
18357 * cfghooks.c (verify_flow_info): Update.
18358 (redirect_edge_succ_nodup): Update.
18359 (split_edge): Update.
18360 (account_profile_record): Update.
18361 * cfgloopanal.c (single_likely_exit): Update.
18362 * cfgloopmanip.c (scale_loop_profile): Update.
18363 (set_zero_probability): Remove.
18364 (duplicate_loop_to_header_edge): Update.
18365 * cfgloopmanip.h (loop_version): Update prototype.
18366 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
18367 (force_nonfallthru_and_redirect): Update.
18368 (update_br_prob_note): Update.
18369 (rtl_verify_edges): Update.
18370 (purge_dead_edges): Update.
18371 (rtl_lv_add_condition_to_bb): Update.
18372 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
18373 * cgraphunit.c (init_lowered_empty_function): Update.
18374 (cgraph_node::expand_thunk): Update.
18375 * cilk-common.c: Include profile-count.h
18376 * dojump.c (inv): Remove.
18377 (jumpifnot): Update.
18378 (jumpifnot_1): Update.
18379 (do_jump_1): Update.
18381 (do_jump_by_parts_greater_rtx): Update.
18382 (do_compare_rtx_and_jump): Update.
18383 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
18384 do_jump_1. do_compare_rtx_and_jump): Update prototype.
18385 * dwarf2cfi.c: Include profile-count.h
18386 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
18387 (sjlj_emit_dispatch_table): Likewise.
18388 * explow.c: Include profile-count.h
18389 * expmed.c (emit_store_flag_force): Update.
18390 (do_cmp_and_jump): Update.
18391 * expr.c (compare_by_pieces_d::generate): Update.
18392 (compare_by_pieces_d::finish_mode): Update.
18393 (emit_block_move_via_loop): Update.
18394 (store_expr_with_bounds): Update.
18395 (store_constructor): Update.
18396 (expand_expr_real_2): Update.
18397 (expand_expr_real_1): Update.
18398 * expr.h (try_casesi, try_tablejump): Update prototypes.
18399 * gimple-pretty-print.c (dump_probability): Update.
18400 (dump_profile): New.
18401 (dump_gimple_label): Update.
18402 (dump_gimple_bb_header): Update.
18403 * graph.c (draw_cfg_node_succ_edges): Update.
18404 * hsa-gen.c (convert_switch_statements): Update.
18405 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
18406 (find_if_case_1): Update.
18407 (find_if_case_2): Update.
18408 * internal-fn.c (expand_arith_overflow_result_store): Update.
18409 (expand_addsub_overflow): Update.
18410 (expand_neg_overflow): Update.
18411 (expand_mul_overflow): Update.
18412 (expand_vector_ubsan_overflow): Update.
18413 * ipa-cp.c (good_cloning_opportunity_p): Update.
18414 * ipa-split.c (split_function): Use make_single_succ_edge.
18415 * ipa-utils.c (ipa_merge_profiles): Update.
18416 * loop-doloop.c (add_test): Update.
18417 (doloop_modify): Update.
18418 * loop-unroll.c (compare_and_jump_seq): Update.
18419 (unroll_loop_runtime_iterations): Update.
18420 * lra-constraints.c (lra_inheritance): Update.
18421 * lto-streamer-in.c (input_cfg): Update.
18422 * lto-streamer-out.c (output_cfg): Update.
18423 * mcf.c (adjust_cfg_counts): Update.
18424 * modulo-sched.c (sms_schedule): Update.
18425 * omp-expand.c (expand_omp_for_init_counts): Update.
18426 (extract_omp_for_update_vars): Update.
18427 (expand_omp_ordered_sink): Update.
18428 (expand_omp_for_ordered_loops): Update.
18429 (expand_omp_for_generic): Update.
18430 (expand_omp_for_static_nochunk): Update.
18431 (expand_omp_for_static_chunk): Update.
18432 (expand_cilk_for): Update.
18433 (expand_omp_simd): Update.
18434 (expand_omp_taskloop_for_outer): Update.
18435 (expand_omp_taskloop_for_inner): Update.
18436 * omp-simd-clone.c (simd_clone_adjust): Update.
18437 * optabs.c (expand_doubleword_shift): Update.
18438 (expand_abs): Update.
18439 (emit_cmp_and_jump_insn_1): Update.
18440 (expand_compare_and_swap_loop): Update.
18441 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
18442 * predict.c (predictable_edge_p): Update.
18443 (edge_probability_reliable_p): Update.
18444 (set_even_probabilities): Update.
18445 (combine_predictions_for_insn): Update.
18446 (combine_predictions_for_bb): Update.
18447 (propagate_freq): Update.
18448 (estimate_bb_frequencies): Update.
18449 (force_edge_cold): Update.
18450 * profile-count.c (profile_count::dump): Add missing space into dump.
18451 (profile_count::debug): Add newline.
18452 (profile_count::differs_from_p): Explicitly convert to unsigned.
18453 (profile_count::stream_in): Update.
18454 (profile_probability::dump): New member function.
18455 (profile_probability::debug): New member function.
18456 (profile_probability::differs_from_p): New member function.
18457 (profile_probability::differs_lot_from_p): New member function.
18458 (profile_probability::stream_in): New member function.
18459 (profile_probability::stream_out): New member function.
18460 * profile-count.h (profile_count_quality): Rename to ...
18461 (profile_quality): ... this one.
18462 (profile_probability): New.
18463 (profile_count): Update.
18464 * profile.c (compute_branch_probabilities): Update.
18465 * recog.c (peep2_attempt): Update.
18466 * sched-ebb.c (schedule_ebbs): Update.
18467 * sched-rgn.c (find_single_block_region): Update.
18468 (compute_dom_prob_ps): Update.
18469 (schedule_region): Update.
18470 * sel-sched-ir.c (compute_succs_info): Update.
18471 * stmt.c (struct case_node): Update.
18472 (do_jump_if_equal): Update.
18473 (get_outgoing_edge_probs): Update.
18474 (conditional_probability): Update.
18475 (emit_case_dispatch_table): Update.
18476 (expand_case): Update.
18477 (expand_sjlj_dispatch_table): Update.
18478 (emit_case_nodes): Update.
18479 * targhooks.c: Update.
18480 * tracer.c (better_p): Update.
18481 (find_best_successor): Update.
18482 * trans-mem.c (expand_transaction): Update.
18483 * tree-call-cdce.c: Update.
18484 * tree-cfg.c (gimple_split_edge): Upate.
18485 (move_sese_region_to_fn): Upate.
18486 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
18487 * tree-eh.c (lower_resx): Upate.
18488 (cleanup_empty_eh_move_lp): Upate.
18489 * tree-if-conv.c (version_loop_for_if_conversion): Update.
18490 * tree-inline.c (copy_edges_for_bb): Update.
18491 (copy_cfg_body): Update.
18492 * tree-parloops.c (gen_parallel_loop): Update.
18493 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
18494 (gimple_gen_time_profiler): Update.
18495 * tree-ssa-dce.c (remove_dead_stmt): Update.
18496 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
18497 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
18498 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
18499 (unloop_loops): Update.
18500 (try_peel_loop): Update.
18501 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
18502 * tree-ssa-loop-split.c (connect_loops): Update.
18503 (split_loop): Update.
18504 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
18505 (hoist_guard): Update.
18506 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
18507 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
18508 (value_replacement): Update.
18509 * tree-ssa-reassoc.c (branch_fixup): Update.
18510 * tree-ssa-tail-merge.c (replace_block_by): Update.
18511 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
18512 (create_edge_and_update_destination_phis): Update.
18513 (compute_path_counts): Update.
18514 (recompute_probabilities): Update.
18515 (update_joiner_offpath_counts): Update.
18516 (freqs_to_counts_path): Update.
18517 (duplicate_thread_path): Update.
18518 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
18519 (struct switch_conv_info): Update.
18520 (gen_inbound_check): Update.
18521 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
18522 (vect_do_peeling): Update.
18523 (vect_loop_versioning): Update.
18524 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
18525 (optimize_mask_stores): Update.
18526 * ubsan.c (ubsan_expand_null_ifn): Update.
18527 * value-prof.c (gimple_divmod_fixed_value): Update.
18528 (gimple_divmod_fixed_value_transform): Update.
18529 (gimple_mod_pow2): Update.
18530 (gimple_mod_pow2_value_transform): Update.
18531 (gimple_mod_subtract): Update.
18532 (gimple_mod_subtract_transform): Update.
18533 (gimple_ic): Update.
18534 (gimple_stringop_fixed_value): Update.
18535 (gimple_stringops_transform): Update.
18536 * value-prof.h: Update.
18538 2017-06-29 Carl Love <cel@us.ibm.com>
18540 * config/rs6000/rs6000-c.c: Add support for built-in functions
18541 vector signed int vec_signed (vector float);
18542 vector signed long long vec_signed (vector double);
18543 vector signed int vec_signed2 (vector double, vector double);
18544 vector signed int vec_signede (vector double);
18545 vector signed int vec_signedo (vector double);
18546 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
18547 instruction generator.
18548 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18549 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
18550 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
18552 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
18553 vunsignede_v2df): Add define_expands.
18554 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
18555 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
18556 VEC_UNSIGNEDO): Add definitions.
18557 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
18558 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
18559 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
18560 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
18561 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
18562 * config/rs6000/altivec.h (vec_signed, vec_signed2,
18563 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
18564 vec_unsignede, vec_unsignedo): Add builtin defines.
18565 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
18567 * doc/extend.texi: Update the built-in documentation file for the
18568 new built-in functions.
18570 2017-06-29 Richard Biener <rguenther@suse.de>
18572 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
18573 reduction chains to LOOP_VINFO_REDUCTIONS.
18574 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
18575 SLP reductions after processing reduction chains.
18577 2017-06-29 Nathan Sidwell <nathan@acm.org>
18579 * builtins.c (fold_builtin_FUNCTION): Use
18580 lang_hooks.decl_printable_name.
18582 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
18584 PR middle-end/81194
18585 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
18586 with only one label.
18587 * stmt.c (expand_case): Assert NCASES is greater than one.
18589 2017-06-29 Richard Biener <rguenther@suse.de>
18591 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
18593 (group_case_labels): Likewise.
18594 (find_taken_edge): Push sanity checking on val to workers...
18595 (find_taken_edge_cond_expr): ... here
18596 (find_taken_edge_switch_expr): ... and here, handle cases
18597 with just a default label.
18598 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
18599 (group_case_labels): Likewise.
18600 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
18601 group_case_labels does anything cleanup the CFG again.
18603 2017-06-29 Bin Cheng <bin.cheng@arm.com>
18605 PR tree-optimization/81196
18606 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
18607 exit condition comparing two IVs.
18609 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
18611 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
18612 profile to the dummy entry at the end of the list of architectures.
18613 * config/arm/arm-cpu-cdata.h: Regenerated.
18615 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18616 Michael Collison <michael.collison@arm.com>
18619 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
18621 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
18622 (*aarch64_reg_<mode>3_minus_mask): New pattern.
18623 (*aarch64_<optab>_reg_di3_mask2): New pattern.
18624 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
18625 of shift when the shift amount is masked with constant equal to
18626 the size of the mode.
18627 * config/aarch64/predicates.md (subreg_lowpart_operator): New
18630 2017-06-29 Martin Liska <mliska@suse.cz>
18632 * config/i386/i386.opt: Change range from [1,5] to [0,5].
18634 2017-06-29 Yury Gribov <tetra2005@gmail.com>
18637 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
18640 (edge_growth_cache_entry::edge_growth_cache_entry): New
18642 (reset_edge_growth_cache): Update to use constructor.
18644 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18646 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
18647 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18648 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
18650 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
18652 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
18653 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
18655 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
18657 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
18658 (*-linux-uclibc*): Add t-uclibc tmake_file.
18659 * config/t-musl: New.
18660 * config/t-uclibc: New.
18662 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
18664 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
18666 (gen_comm_data): Emit architectural setting of arch_prof.
18667 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
18669 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
18670 (armv8-m.base, armv8-m.main): Likewise.
18671 * arm-protos.h (arm_build_target): Add profile field.
18672 (arch_option): Likewise.
18673 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
18675 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
18676 arm_active_target.profile.
18678 2017-06-28 Richard Biener <rguenther@suse.de>
18680 PR middle-end/81227
18681 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
18682 TYPE_OVERFLOW_WRAPS.
18683 * match.pd (negate_expr_p): Likewise.
18684 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
18685 fold_build2, not fold_binary.
18687 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18689 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
18690 Convert memory address to Pmode.
18691 (aarch64_print_operand): Assert MEM operands are always Pmode.
18693 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
18696 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
18697 Remove redundant if.
18698 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
18699 * config/arm/aarch-common-protos.h
18700 (aarch_forward_to_shift_is_not_shifted_re): Remove.
18701 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
18703 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
18706 * multiple_target.c (create_dispatcher_calls): Set the default
18707 clone to be static, not public.
18709 2017-06-28 Richard Biener <rguenther@suse.de>
18711 * tree-vect-loop.c (vectorizable_reduction): Move special
18712 cond reduction IV var creation ...
18713 (vect_create_epilog_for_reduction): ... here. Remove induction_index
18714 parameter. Use STMT_VINFO_VECTYPE.
18715 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
18718 2017-06-28 Martin Liska <mliska@suse.cz>
18721 * ipa-visibility.c (non_local_p): Handle visibility.
18723 2017-06-28 Martin Liska <mliska@suse.cz>
18726 * common.opt: Add IntegerRange to various options.
18727 * opt-functions.awk (integer_range_info): New function.
18728 * optc-gen.awk: Add integer_range_info to cl_options struct.
18729 * opts-common.c (decode_cmdline_option): Handle
18730 CL_ERR_INT_RANGE_ARG.
18731 (cmdline_handle_error): Likewise.
18732 * opts.c (print_filtered_help): Show valid interval in
18733 when --help is provided.
18734 * opts.h (struct cl_option): Add range_min and range_max fields.
18735 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
18737 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
18739 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
18740 (x * C EQ/NE y * C): New transformation.
18742 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
18744 * genmultilib (combination_space): Accept '+' in option names.
18746 2017-06-28 Martin Liska <mliska@suse.cz>
18749 * asan.c (instrument_derefs): Bail out inner references
18750 that are hard register variables.
18752 2017-06-28 Jakub Jelinek <jakub@redhat.com>
18755 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
18756 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
18758 2017-06-28 Richard Biener <rguenther@suse.de>
18760 * tree-vectorizer.h (vect_get_vec_defs): Remove.
18761 (vect_get_slp_defs): Adjust.
18762 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
18764 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
18766 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
18767 get_initial_defs_for_reduction instead of vect_get_vec_defs.
18768 (vectorizable_reduction): Adjust.
18769 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
18771 (vect_get_slp_defs): Likewise.
18772 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
18773 (vectorizable_bswap): Adjust.
18774 (vectorizable_call): Likewise.
18775 (vectorizable_conversion): Likewise.
18776 (vectorizable_assignment): Likewise.
18777 (vectorizable_shift): Likewise.
18778 (vectorizable_operation): Likewise.
18779 (vectorizable_store): Likewise.
18780 (vectorizable_condition): Likewise.
18781 (vectorizable_comparison): Likewise.
18783 2017-06-28 Michael Collison <michael.collison@arm.com>
18786 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
18788 (arm_gen_movmemqi): Removed unused variable 'i'.
18789 Convert 'for' loop into 'while' loop.
18790 (arm_expand_prologue): Remove last unnecessary set of insn.
18791 (thumb_pop): Remove unused variable 'pushed_words'.
18792 (thumb_exit): Remove last unnecessary set of regs_to_pop.
18794 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
18796 * config/s390/predicates.md: Use s390_rel_address_ok_p.
18797 * config/s390/s390-protos.h: Add prototype of
18798 s390_rel_address_ok_p.
18799 * config/s390/s390.c (s390_got_symbol): New function.
18800 (s390_rel_address_ok_p): New function.
18801 (legitimize_pic_address): Use s390_rel_address_ok_p.
18802 (s390_load_got): Use s390_got_symbol.
18803 (s390_option_override): Issue error if
18804 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
18805 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
18807 * config/s390/s390.opt: New option mpic-data-is-text-relative.
18809 2017-06-27 Andrew Pinski <apinski@cavium.com>
18811 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
18812 (X * copysign (1.0, X)): New pattern.
18813 (X * copysign (1.0, -X)): New pattern.
18814 (copysign (-1.0, CST)): New pattern.
18816 2017-06-27 Joseph Myers <joseph@codesourcery.com>
18818 * genmultilib (combination_space): Remove variable.
18819 Validate reuse rules against regular expression for any sequence
18820 of multilib options in any order.
18822 2017-06-27 Michael Collison <michael.collison@arm.com>
18824 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
18825 call aarch64_split_simd_combine.
18826 * (aarch64_combine_internal<mode>): Delete pattern.
18827 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
18828 Allow register and subreg operands.
18830 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18832 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
18833 specific need, just fallback on defaults.
18834 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
18836 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18837 Olivier Hainque <hainque@adacore.com>
18839 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
18841 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
18842 targets. Pick a default if no particular attempt applied.
18843 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
18846 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18848 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
18849 (x86_64-wrs-vxworks7): Likewise.
18851 2017-06-27 Marek Polacek <polacek@redhat.com>
18854 * ubsan.c (instrument_null): Check get_base_address's result for null.
18856 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18858 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
18860 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
18862 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
18863 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
18864 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
18865 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
18866 New function types.
18867 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
18868 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
18869 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
18870 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
18871 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
18872 BUILT_IN_FEUPDATEENV): New builtins.
18873 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
18874 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
18875 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
18876 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
18878 (builtin_structptr_types): Adjust size.
18879 * tree.c (builtin_structptr_types): Add four entries.
18881 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18882 Olivier Hainque <hainque@adacore.com>
18884 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
18885 (TLS_SYM): New local macro, forcing reference to __tls__ on
18886 link command lines for VxWorks 7 RTPs, triggering initialization
18888 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
18889 OS features TLS support, true for RTPs on VxWorks 7.
18890 * config/vxworks.c (vxworks_override_options): Setup emutls
18893 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18895 * predict.c (test_prediction_value_range): Use -1U instead of -1
18896 to avoid narrowing conversion warning.
18897 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
18898 to avoid narrowing conversion warning.
18899 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
18901 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
18903 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
18905 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
18906 64bit configurations.
18907 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
18908 (SIZE_TYPE): Likewise.
18909 * config/vxworks.c (vxworks_emutls_var_fields): Use
18910 long_unsigned_type_node instead of unsigned_type_node as the offset
18911 field type, which is "pointer" mode in emutls.c.
18913 2017-06-27 Jakub Jelinek <jakub@redhat.com>
18916 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
18918 PR middle-end/81207
18919 * gimple-fold.c (replace_call_with_call_and_fold): Handle
18920 gimple_vuse copying separately from gimple_vdef copying.
18922 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
18924 * value-prof.c (free_hist): Remove call to memset and the enclosing if
18927 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18928 Olivier Hainque <hainque@adacore.com>
18930 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
18931 for all vxworks7 targets.
18932 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
18933 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
18934 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
18935 variations for VX6/VX7 and 32/64bits later on in ...
18936 (VXWORKS_LIB_SPEC): Leverage new macros.
18937 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
18938 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
18940 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
18942 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
18943 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
18945 2017-06-26 Carl Love <cel@us.ibm.com>
18947 * config/rs6000/rs6000-c.c: Add support for built-in functions
18948 vector bool char vec_reve (vector bool char);
18949 vector signed char vec_reve (vector signed char);
18950 vector unsigned char vec_reve (vector unsigned char);
18951 vector bool int vec_reve (vector bool int);
18952 vector signed int vec_reve (vector signed int);
18953 vector unsigned int vec_reve (vector unsigned int);
18954 vector bool long long vec_reve (vector bool long long);
18955 vector signed long long vec_reve (vector signed long long);
18956 vector unsigned long long vec_reve (vector unsigned long long);
18957 vector bool short vec_reve (vector bool short);
18958 vector signed short vec_reve (vector signed short);
18959 vector double vec_reve (vector double);
18960 vector float vec_reve (vector float);
18961 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
18962 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
18963 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
18964 (altivec_vreve): New pattern.
18965 * config/rs6000/altivec.h (vec_reve): New define.
18966 * doc/extend.texi (vec_rev): Update the built-in documentation file
18967 for the new built-in functions.
18969 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
18971 PR tree-optimization/71815
18972 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
18974 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
18976 (slsr_process_phi): Likewise.
18977 (replace_uncond_cands_and_profitable_phis): Don't replace a
18978 multiply candidate with a stride of 1 (copy or cast).
18979 (phi_incr_cost): Call uses_consumed_by_stmt rather than
18981 (lowest_cost_path): Likewise.
18982 (total_savings): Likewise.
18984 2017-06-26 Richard Biener <rguenther@suse.de>
18987 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
18988 Use def_builtin_pure for all gather builtins.
18990 2017-06-26 Richard Biener <rguenther@suse.de>
18992 PR tree-optimization/81203
18993 * tree-tailcall.c (find_tail_calls): Do not move stmts into
18994 non-dominating BBs.
18996 2017-06-26 Marek Polacek <polacek@redhat.com>
18999 * doc/invoke.texi: Document -Wmultistatement-macros.
19001 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
19003 * doc/sourcebuild.texi (ARM-specific attributes): Document new
19004 arm_neon_ok_no_float_abi effective target.
19006 2017-06-26 Richard Biener <rguenther@suse.de>
19008 PR tree-optimization/80928
19009 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
19010 (copy_bbs): Set BB_DUPLICATED flag early.
19011 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
19013 (execute_on_shrinking_pred): Likewise.
19014 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
19015 BB_DUPLICATED blocks.
19016 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
19017 iterate over all PHIs considering removal of *gsi.
19019 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
19021 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
19024 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19026 * config/rs6000/rs6000-string.c: (expand_block_clear,
19027 do_load_for_compare, select_block_compare_mode,
19028 compute_current_alignment, expand_block_compare,
19029 expand_strncmp_align_check, expand_strn_compare,
19030 expand_block_move, rs6000_output_load_multiple)
19031 Move functions related to string/block move/compare
19032 to a separate file.
19033 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
19034 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
19035 for this function which is now used in two files.
19036 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
19037 * config.gcc: Add rs6000-string.o to extra_objs for
19038 targets powerpc*-*-* and rs6000*-*-*.
19040 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
19043 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
19044 32-bit, since indexed is not valid for DImode.
19045 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
19046 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
19047 (define_peephole2 for Altivec d-form load): Add 32-bit support.
19048 (define_peephole2 for Altivec d-form store): Likewise.
19051 * multiple_target.c (create_dispatcher_calls): Only create the
19052 dispatcher call if the function is the default clone of a
19053 versioned function.
19055 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
19057 PR middle-end/80902
19058 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
19059 a call, force the call to not be a tail call.
19061 2017-06-23 Jeff Law <law@redhat.com>
19063 * doc/contrib.texi: Add entry for Steven Pemberton's work on
19066 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
19068 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
19069 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
19070 handling for early expansion of vector shifts (sl,sr,sra,rl).
19071 (builtin_function_type): Add vector shift right instructions
19072 to the unsigned argument list.
19074 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
19076 rtl-optimizatoin/79286
19077 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
19078 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
19079 trap. PIC register plus a const unspec without offset can never trap.
19081 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
19083 * tree.h (builtin_structptr_type): New type.
19084 (builtin_structptr_types): Declare new array.
19085 * tree.c (builtin_structptr_types): New array.
19086 (free_lang_data, build_common_tree_nodes): Use it.
19088 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
19091 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
19094 2017-06-22 Matt Turner <mattst88@gmail.com>
19096 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
19097 Lake models to skylake case. Assume skylake for unknown
19098 models with clflushopt.
19100 2017-06-22 Jeff Law <law@redhat.com>
19102 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
19103 frame sizes that do not satisfy aarch64_uimm12_shift.
19105 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
19107 * profile-count.h (apply_probability,
19108 apply_scale, probability_in): Fix checks for zero.
19110 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19112 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
19113 * doc/cppdiropts.texi (-I @var{dir}): Document it.
19115 2016-06-22 Richard Biener <rguenther@suse.de>
19117 * tree-vect-loop.c (vect_model_reduction_cost): Handle
19118 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
19119 REDUC_MAX_EXPR support.
19120 (vectorizable_reduction): Likewise.
19121 (vect_create_epilog_for_reduction): Likewise.
19123 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
19125 * match.pd (A / (1 << B) -> A >> B): New.
19126 * generic-match-head.c: Include optabs-tree.h.
19127 * gimple-match-head.c: Likewise.
19128 * optabs-tree.h (target_supports_op_p): New.
19129 * optabs-tree.c (target_supports_op_p): New.
19131 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19133 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
19134 $gcc_cv_ld --help output.
19135 (gcc_cv_ld_demangle): Likewise.
19136 (gcc_cv_ld_eh_frame_hdr): Likewise.
19137 (gcc_cv_ld_pie): Likewise.
19138 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
19139 (gcc_cv_ld_buildid): Likewise.
19140 (gcc_cv_ld_sysroot): Likewise.
19141 (ld_bndplt_support): Likewise.
19142 (ld_pushpopstate_support): Likewise.
19143 * configure: Regenerate.
19144 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
19146 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19149 * config/i386/sse.md (round<mode>2): Renumber match_dup and
19150 operands indexes to avoid gap between operands and match_dups.
19152 2017-06-21 Andrew Pinski <apinski@cavium.com>
19154 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
19155 Increment Arith_shift and Arith_shift_reg by 1.
19156 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
19158 * config/aarch64/aarch64.c (thunderx_tunings): Enable
19159 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
19160 (aarch64_strip_extend): Add new argument and test for it.
19161 (aarch64_cheap_mult_shift_p): New function.
19162 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
19163 add a cost if it is true.
19164 Update calls to aarch64_strip_extend.
19165 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
19167 2017-06-21 Andrew Pinski <apinski@cavium.com>
19169 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
19171 (thunderxt88): Likewise.
19172 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
19173 (thunderx_prefetch_tune): New variable.
19174 (thunderx2t99_prefetch_tune): Update for the correct values.
19175 (thunderxt88_tunings): New variable.
19176 (thunderx_tunings): Use thunderx_prefetch_tune instead of
19177 generic_prefetch_tune.
19178 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
19180 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19182 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
19183 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
19184 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
19185 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
19186 (aarch64_atomic_cas<mode>, GPI): Likewise.
19188 2017-06-21 Martin Liska <mliska@suse.cz>
19190 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
19191 statements on cold and hot labels.
19192 * predict.c (tree_estimate_probability_bb): Remove the
19193 prediction from this place.
19195 2017-06-21 Martin Liska <mliska@suse.cz>
19197 PR tree-optimization/79489
19198 * gimplify.c (maybe_add_early_return_predict_stmt): New
19200 (gimplify_return_expr): Call the function.
19201 * predict.c (tree_estimate_probability_bb): Remove handling
19203 * predict.def: Update comment about early return predictor.
19204 * gimple-predict.h (is_gimple_predict): New function.
19205 * predict.def: Change default value of early return to 66.
19206 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
19208 * passes.def: Put pass_strip_predict_hints to the beginning of
19211 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
19213 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
19214 FUNCTION_DECL declarations.
19215 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
19217 (dwaf2out_decl): Likewise.
19218 * godump.c (go_early_global_decl): Skip call to the real debug hook
19219 for FUNCTION_DECL declarations.
19220 * passes.c (rest_of_decl_compilation): Skip call to the
19221 early_global_decl debug hook for FUNCTION_DECL declarations, unless
19222 -fdump-go-spec is passed.
19224 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19226 * config/i386/i386.c (struct builtin_isa): New field pure_p.
19227 Reorder for compactness.
19228 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
19229 (def_builtin_pure, def_builtin_pure2): New functions.
19230 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
19232 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
19234 * match.pd (nop_convert): New predicate.
19235 ((A +- CST1) +- CST2): Allow some NOP conversions.
19237 2017-06-21 Jakub Jelinek <jakub@redhat.com>
19240 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
19241 with ctors/dtors if GOVD_SHARED is set.
19243 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19245 * config/aarch64/aarch64.md (movti_aarch64):
19246 Emit mov rather than orr.
19247 (movtf_aarch64): Likewise.
19248 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
19249 Emit mov rather than orr.
19251 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19253 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
19254 Swap alternatives, make integer dup more expensive.
19256 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
19258 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
19259 Return true for non-tls symbols.
19261 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
19263 * config/aarch64/aarch64-cores.def (cortex-a55): New.
19264 (cortex-a75): Likewise.
19265 (cortex-a75.cortex-a55): Likewise.
19266 * config/aarch64/aarch64-tune.md: Regenerate.
19267 * doc/invoke.texi (-mtune): Document new values for -mtune.
19269 2017-06-21 Tom de Vries <tom@codesourcery.com>
19271 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
19272 stack_size feature.
19273 (Effective-Target Keywords, Other attributes): Suggest using
19274 dg-add-options stack_size feature to get stack limit in stack_size
19275 effective target documentation.
19277 2017-06-21 Julian Brown <julian@codesourcery.com>
19278 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
19280 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
19281 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
19282 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
19284 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
19285 attribute type list for neon_multiply.
19286 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
19287 attribute type list for neon_multiply.
19288 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
19289 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
19290 attribute type list for neon_multiply.
19291 * config/arm/types.md (crypto_pmull): Add.
19292 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
19293 attribute type list.
19295 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
19297 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
19300 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19302 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
19303 to make sure not to dereference a NULL cost_classes_ptr pointer.
19305 2017-06-20 Carl Love <cel@us.ibm.com>
19307 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19308 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
19309 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
19310 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
19311 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
19312 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
19313 VMULOSW): New enum "unspec" values.
19314 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
19315 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
19316 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
19317 altivec_vmulosw): New patterns.
19318 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
19319 VMULOSW): Add definitions.
19321 2017-06-20 Julia Koval <julia.koval@intel.com>
19323 * config/i386/i386.c: Fix rounding expand for new pattern.
19324 * config/i386/subst.md: Fix pattern (parallel -> unspec).
19326 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19328 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
19329 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
19331 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19333 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
19336 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
19338 * config/aarch64/aarch64-cores.def: Rearrange to sort by
19339 architecture, then by implementer ID.
19340 * config/aarch64/aarch64-tune.md: Regenerate.
19342 2017-06-20 Richard Biener <rguenther@suse.de>
19344 PR middle-end/81097
19345 * fold-const.c (split_tree): Fold to type before negating.
19347 2017-06-20 David Malcolm <dmalcolm@redhat.com>
19349 * diagnostic-show-locus.c
19350 (selftest::test_fixit_deletion_affecting_newline): New function.
19351 (selftest::diagnostic_show_locus_c_tests): Call it.
19353 2017-06-20 Andreas Schwab <schwab@suse.de>
19356 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
19359 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
19361 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
19362 __ARM_FEATURE_COPROC according to support.
19364 2017-06-20 Jakub Jelinek <jakub@redhat.com>
19366 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19367 Rewritten to avoid overflow for > 32-bit pointers.
19370 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
19371 by removing enum keyword.
19372 (ubsan_type_descriptor): Likewise. Formatting fix.
19375 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
19376 splitter): Require TARGET_SSE2 in the condition.
19378 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
19381 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
19382 for doing vector set of SFmode on ISA 3.0.
19383 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
19384 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
19386 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
19387 SFmode value into a V4SF variable that was extracted from another
19388 V4SF variable without converting the element to double precision
19389 and back to single precision vector format.
19390 (vsx_insert_extract_v4sf_p9_2): Likewise.
19392 2017-06-19 Jakub Jelinek <jakub@redhat.com>
19394 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
19395 in UWHI to avoid undefined overflow.
19398 * ubsan.h (enum ubsan_encode_value_phase): New.
19399 (ubsan_encode_value): Change second argument to
19400 enum ubsan_encode_value_phase with default value of
19401 UBSAN_ENCODE_VALUE_GENERIC.
19402 * ubsan.c (ubsan_encode_value): Change second argument to
19403 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
19404 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
19405 create_tmp_var_raw instead of create_tmp_var and use a
19407 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
19408 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
19409 ubsan_encode_value callers.
19412 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
19413 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
19414 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
19416 2017-06-19 Richard Biener <rguenther@suse.de>
19418 PR middle-end/81118
19419 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
19420 estimates if we changed anything.
19422 2017-06-19 Richard Biener <rguenther@suse.de>
19424 PR tree-optimization/80887
19425 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
19426 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
19427 simplified lookups, then reset mprts_hook.
19428 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
19430 (try_to_simplify): Likewise.
19432 2017-06-19 Martin Liska <mliska@suse.cz>
19435 * gimplify.c (gimplify_switch_expr):
19436 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
19438 2017-06-19 Martin Liska <mliska@suse.cz>
19440 * doc/install.texi: Document that PGO runs in 4 stages.
19442 2017-06-19 Martin Liska <mliska@suse.cz>
19445 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
19446 to dispatcher function name.
19447 * multiple_target.c (replace_function_decl): New function.
19448 (create_dispatcher_calls): Redirect both edges and references.
19450 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
19452 * profile-count.c (profile_count::dump): Dump quality.
19453 (profile_count::differs_from_p): Update for unsigned val.
19454 * profile-count.h (profile_count_quality): New enum.
19455 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
19457 2017-06-19 Richard Biener <rguenther@suse.de>
19459 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
19460 struct function as arg.
19461 (estimate_numbers_of_iterations): Export overload with loop arg.
19462 (free_numbers_of_iterations_estimates_loop): Use an overload of
19463 free_numbers_of_iterations_estimates instead.
19464 * tree-cfg.c (remove_bb): Adjust.
19465 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
19466 * tree-parloops.c (gen_parallel_loop): Likewise.
19467 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
19469 (tree_unroll_loops_completely): Likewise.
19470 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
19471 Use an overload instead and export.
19472 (estimated_loop_iterations): Adjust.
19473 (max_loop_iterations): Likewise.
19474 (likely_max_loop_iterations): Likewise.
19475 (estimate_numbers_of_iterations): Take struct function as arg
19477 (loop_exits_before_overflow): Adjust.
19478 (free_numbers_of_iterations_estimates_loop): Use an overload.
19479 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
19480 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
19482 2017-06-19 Richard Biener <rguenther@suse.de>
19485 * ipa-prop.c (find_constructor_constant_at_offset): Handle
19486 RANGE_EXPR conservatively.
19488 2017-06-16 Carl Love <cel@us.ibm.com>
19490 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
19491 definitions for vec_float, vec_float2, vec_floato,
19492 vec_floate built-ins.
19493 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
19494 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
19496 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
19497 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
19498 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
19499 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
19500 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
19501 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
19502 vec_floato): Add builtin defines.
19503 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
19504 Update the built-in documentation file for the new built-in
19507 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19509 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
19510 (mthumb): Mark as the negative of -marm.
19512 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19514 * doc/invoke.texi (ARM Options, -mcpu): Document supported
19516 (ARM Options, -mtune): Document that this accepts the same
19517 extension options as -mcpu.
19518 (ARM Options, -mfpu): Document addition of -mfpu=auto.
19520 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19522 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
19523 permitted extensions.
19525 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19527 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
19528 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
19529 (armv8-m.main): Add option +nodsp.
19530 * config/arm/arm-cpu-cdata.h: Regenerated.
19532 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19534 * config/arm/t-fuchsia: New file.
19535 * config.gcc (arm*-*-fuchsia*): Use it.
19537 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19539 * config/arm/t-symbian: Rewrite for new option infrastructure.
19541 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19543 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
19544 (MULTILIB_REQUIRED): Likewise.
19546 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19548 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
19549 (MULTILIB_RESUE): Likewise.
19550 (MULTILIB_MATCHES): Likewise.
19551 (MULTLIB_REQUIRED): Likewise.
19553 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19555 * config/arm/t-rtems: Rewrite for new option framework.
19557 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19559 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
19560 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
19561 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
19562 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
19563 * config/arm/t-multilib: ... here.
19564 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
19565 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
19566 armv7-a and armv8*-a when A-profile libraries have not been built.
19567 * config/arm/t-rmprofile: Rewrite.
19569 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19571 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
19572 with a backslash. Remove the backslash after substituting unescaped
19574 * doc/fragments.texi (MULTILIB_REUSE): Document it.
19576 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19578 * config.gcc: (arm*-*-*): When building a-profile libraries, force
19579 the driver to pass through the default setting of -mfloat-abi.
19580 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
19582 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
19583 (all_feat_combs): New rule.
19584 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
19586 * config/arm/t-aprofile: Rewrite.
19588 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19590 * config/arm/arm.h (FPUTYPE_AUTO): Define.
19591 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
19592 fpu is not specified by the user/command-line.
19593 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
19594 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
19595 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
19596 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
19597 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
19598 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
19600 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19602 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
19603 * config/arm/t-arm-elf: Rewritten.
19605 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19607 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
19608 have some floating-point instructions.
19609 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
19610 (TARGET_MAYBE_HARD_FLOAT): New macro.
19611 * config/arm/arm-builtins.c (arm_init_builtins): Use
19612 TARGET_MAYBE_HARD_FLOAT.
19613 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
19615 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19617 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
19618 (configargs.h): Include it.
19619 (arm_print_hint_for_fpu_option): New function.
19620 (arm_parse_fpu_option): New function.
19621 (candidate_extension): New class.
19622 (arm_canon_for_multilib): New function.
19623 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
19624 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
19625 (ARCH_CANONICAL_SPECS): New macro.
19626 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
19628 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19630 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
19631 are set after handling multilib fragments. Set target_cpu_default2
19634 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19636 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
19638 (arm*-*-*): Set target_cpu_default2 to a quoted string.
19639 * config/arm/parsecpu.awk (check_cpu): Validate any extension
19641 (check_arch): Likewise.
19642 * config/arm/arm.c (arm_configure_build_target): Handle
19643 TARGET_CPU_DEFAULT being a string constant. Scan any feature
19644 options in the default.
19646 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19648 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
19649 when an option is an alias of another.
19650 * config/arm/parsecpu.awk (optalias): New parser token.
19651 (gen_comm_data): Mark non-alias options as such. Emit entries
19652 for extension aliases.
19653 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
19654 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
19655 (armv6kz, armv6zk, armv6t2): Likewise.
19656 (armv7): Make vfpv3-d16 an alias.
19657 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
19659 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
19660 Sort in canonical order.
19661 (armv8-a): Sort in canonical order.
19662 (armv8.1-a, armv8.2-a): Likewise.
19663 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
19665 (cortex-a9): Sort in canonical order.
19666 * config/arm/arm.c (selftests.h): Include it.
19667 (arm_test_cpu_arch_data): New function.
19668 (arm_run_self_tests): New function.
19669 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
19670 (targetm): Move declaration to the end of the file.
19671 * arm-cpu-cdata.h: Regenerated.
19673 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19675 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
19676 call to target_mode_check describing the type of option passed.
19677 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
19678 (arm_target_thumb_only): Use arm_parse_arch_option_name or
19679 arm_parse_cpu_option_name to match parameters against list of
19681 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
19682 arm_arch_core_flags data structure.
19683 * config/arm/arm-cpu_cdata.h: Regenerated.
19685 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19687 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
19689 (arm_print_hint_for_cpu_option): Likewise.
19690 (arm_print_hint_for_arch_option): Likewise.
19691 (arm_parse_cpu_option_name): Likewise.
19692 (arm_parse_arch_option_name): Likewise.
19693 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
19694 of entries in the all_fpus list.
19695 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
19696 (arm_parse_cpu_option_name): Declare.
19697 (arm_parse_arch_option_name): Declare.
19698 (arm_parse_option_features): Declare.
19699 (arm_intialize_isa): Declare.
19700 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
19702 (gen_comm_data): ... here. Make definitions non-static.
19703 * config/arm/arm-cpu-data.h: Regenerated.
19704 * config/arm/arm-cpu-cdata.h: Regenerated.
19706 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19708 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
19709 (cpu_arch_extension): New structure.
19710 (cpu_arch_option, arch_option, cpu_option): New structures.
19711 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
19712 architecture types.
19713 (gen_data): Generate new format data tables.
19714 * config/arm/arm.c (cpu_tune): New structure.
19715 (cpu_option, processors): Delete.
19716 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
19717 (arm_print_hint_for_cpu_option): ... this and ...
19718 (arm_print_hint_for_arch_option): ... this.
19719 (arm_parse_arch_cpu_name): Delete. Replace with ...
19720 (arm_parse_cpu_option_name): ... this and ...
19721 (arm_parse_arch_option_name): ... this.
19722 (arm_unrecognized_feature): Change type of target parameter to
19724 (arm_parse_arch_cpu_features): Delete. Replace with ...
19725 (arm_parse_option_features): ... this.
19726 (arm_configure_build_target): Rework to use new configuration data
19728 (arm_print_tune_info): Rework for new configuration data tables.
19729 * config/arm/arm-cpu-data.h: Regenerated.
19730 * config/arm/arm-cpu.h: Regenerated.
19732 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19734 * Makefile.in (OBJS): Move sbitmap.o from here ...
19735 (OBJS-libcommon): ... to here.
19737 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19739 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
19740 (ISA_ALL_CRYPTO): New macro.
19741 (ISA_ALL_SIMD): New macro
19742 (ISA_ALL_FP): New macro.
19743 * config/arm/arm.c (fpu_bitlist): Update initializer.
19744 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
19746 (arm9e): Add fpu. Add option for nofp
19747 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
19748 (arm926ej-s, arm1026ej-s): Likewise.
19749 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
19750 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
19751 neon-fp16, neon-vfpv4, nofp and nosimd.
19752 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
19753 (cortex-a8): Add fpu. Add option for nofp.
19754 (cortex-a9): Add fpu. Add options for nosimd and nofp.
19755 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
19756 (cortex-r4f): Add fpu.
19757 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
19758 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
19760 (cortex-r8): Likewise.
19761 (cortex-m4): Add fpu. Add option for nofp.
19762 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
19763 (cortex-a17.cortex-a7): Likewise.
19764 (cortex-a32): Add fpu. Add options for crypto and nofp.
19765 (cortex-a35, cortex-a53): Likewise.
19766 (cortex-a57): Add fpu. Add option for crypto.
19767 (cortex-a72, cortex-a73): Likewise.
19768 (exynos-m1): Likewise.
19769 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
19770 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
19771 (cortex-m33): Add fpu. Add option for nofp.
19772 * config/arm/arm-cpu-cdata.h: Regenerated
19773 * config/arm/arm-cpu-data.h: Regenerated.
19775 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19777 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
19778 (armv5te, armv5tej): Likewise.
19779 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
19780 (armv7): Add options fp and vfpv3-d16.
19781 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
19782 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
19784 (armv7ve): Likewise.
19785 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
19786 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
19787 (armv8-a): Add nocrypto option.
19788 (armv8.1-a, armv8.2-a): Likewise.
19789 (armv8-m.main): add options fp, fp.dp and nofp.
19791 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19793 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
19795 (armv8-a+crc): Delete.
19796 (armv8.1-a): Add options simd, crypto and nofp.
19797 (armv8.2-a): Add options fp16, simd, crypto and nofp.
19798 (armv8.2-a+fp16): Delete.
19799 (armv8-m.main): Add option dsp.
19800 (armv8-m.main+dsp): Delete.
19801 (cortex-a8): Add fpu. Add nofp option.
19802 (cortex-a9): Add fpu. Add nofp and nosimd options.
19803 * config/arm/parsecpu.awk (gen_data): Generate option tables and
19804 link to main cpu and architecture data structures.
19805 (gen_comm_data): Only put isa attributes from the main architecture
19807 (option): New statement for architecture and CPU entries.
19808 * arm.c (struct cpu_option): New structure.
19809 (struct processors): Add entry for options.
19810 (arm_unrecognized_feature): New function.
19811 (arm_parse_arch_cpu_name): Ignore any characters after the first
19813 (arm_parse_arch_cpu_feature): New function.
19814 (arm_configure_build_target): Separate out any CPU and architecture
19815 features and parse separately. Don't error out if -mfpu=auto is
19816 used with only an architecture string.
19817 (arm_print_asm_arch_directives): New function.
19818 (arm_file_start): Call it.
19819 * config/arm/arm-cpu-cdata.h: Regenerated.
19820 * config/arm/arm-cpu-data.h: Likewise.
19821 * config/arm/arm-tables.opt: Likewise.
19823 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19825 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
19826 assembler when it is not -mfpu=auto.
19828 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19830 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
19831 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
19832 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
19833 (ASM_CPU_SPEC): Rewrite.
19834 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
19835 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
19836 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
19837 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
19838 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
19839 copied string is NUL-terminated. Also strip any characters prefixed
19841 (arm_rewrite_selected_arch): New function.
19842 (arm_rewrite_march): New function.
19844 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
19846 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
19847 (x_arm_cpu_string, x_arm_tune_string): Likewise.
19848 (march, mcpu, mtune): Convert to string-based options.
19849 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
19850 (arm_parse_arch_cpu_name): New function.
19851 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
19852 identify selected architecture or CPU.
19853 (arm_option_save): New function.
19854 (TARGET_OPTION_SAVE): Redefine.
19855 (arm_option_restore): Restore string options.
19856 (arm_option_print): Print string options.
19858 2017-06-16 Martin Sebor <msebor@redhat.com>
19860 PR tree-optimization/80933
19861 PR tree-optimization/80934
19862 * builtins.c (fold_builtin_3): Do not handle bcmp here.
19863 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
19864 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
19865 (gimple_fold_builtin): Call them.
19867 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19869 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
19870 as unlikely; update profile.
19872 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19874 * predict.c (force_edge_cold): Handle declaring edges impossible
19877 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19879 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
19881 (try_unroll_loop_completely): Fix reporting.
19883 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19885 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
19887 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
19890 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
19891 if given a non-constant argument for an intrinsic which requires a
19894 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19896 * profile.c (compare_freqs): New function.
19897 (branch_prob): Sort edge list.
19898 (find_spanning_tree): Assume that the list is priority sorted.
19900 2017-06-16 Richard Biener <rguenther@suse.de>
19902 PR tree-optimization/81090
19903 * passes.def (pass_record_bounds): Remove.
19904 * tree-pass.h (make_pass_record_bounds): Likewise.
19905 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
19906 make_pass_record_bounds): Likewise.
19907 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
19908 not free niter estimates at the beginning but at the end.
19909 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
19911 2017-06-16 Richard Biener <rguenther@suse.de>
19913 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
19914 initializer to workaround ICE in host GCC 4.8.
19916 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
19918 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
19920 (clone_inlined_nodes): Update.
19922 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19924 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
19925 prefetch settings, and enable prefetching by default at -O3.
19927 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19929 * config/aarch64/aarch64.c (aarch64_override_options_internal):
19930 Set flag_prefetch_loop_arrays according to tuning data.
19932 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19934 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
19935 New tune structure.
19936 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
19937 [Unrelated to main purpose of the patch] Place the pointer field last
19938 to enable type checking errors when tune structure are wrongly merged.
19939 * config/aarch64/aarch64.c (generic_prefetch_tune,)
19940 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
19941 (thunderx2t99_prefetch_tune): New tune constants.
19942 (tune_params *_tunings): Update all tunings (no functional change).
19943 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
19944 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
19945 from tunings structures.
19947 2017-06-16 Jakub Jelinek <jakub@redhat.com>
19950 * ubsan.c (instrument_null): Add T argument, use it instead
19951 of computing it based on IS_LHS.
19952 (instrument_object_size): Likewise.
19953 (pass_ubsan::execute): Adjust instrument_null and
19954 instrument_object_size callers to pass gimple_get_lhs or
19955 gimple_assign_rhs1 result to it. Use instrument_null instead of
19956 calling get_base_address and instrument_mem_ref. Handle
19957 aggregate call arguments for object-size sanitization.
19959 2017-06-16 Yury Gribov <tetra2005@gmail.com>
19961 PR tree-optimization/81089
19962 * tree-vrp.c (is_masked_range_test): Validate operands of
19965 2017-06-15 Martin Sebor <msebor@redhat.com>
19968 * dumpfile.c (dump_register): Avoid calling memset to initialize
19969 a class with a default ctor.
19970 * gcc.c (struct compiler): Remove const qualification.
19971 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
19972 * hash-table.h: Ditto.
19973 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
19975 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
19976 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
19978 * params.h (struct param_info): Make struct members non-const.
19979 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
19980 with default initialization.
19981 * vec.h (vec_copy_construct, vec_default_construct): New helper
19983 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
19984 with vec_copy_construct.
19985 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
19986 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
19987 * doc/invoke.texi (-Wclass-memaccess): Document.
19989 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
19991 * emit-rtl.h (is_leaf): Update comment about local
19992 register allocator.
19994 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19997 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
19998 for a variable to have a section before checking if the section has a
20000 Set section to.persistent if persistent attribute is set.
20001 Warn if .persistent attribute is used on an automatic variable.
20003 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
20005 PR rtl-optimization/80474
20006 * reorg.c (update_block): Do not ignore instructions in a delay slot.
20008 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
20010 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
20013 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
20015 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
20016 (casesi): Emit bounds checking as RTL.
20017 (casesi_internal_mips16_<mode>): Remove bounds checking.
20019 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
20021 * config/xtensa/xtensa.c (xtensa_option_override): Append
20022 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
20023 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
20024 xtensa_doloop_hooks): Define unconditionally.
20025 (xtensa_reorg_loops): Only call reorg_loops in the presence of
20027 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
20028 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
20029 for it in xtensa_option_override.
20030 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
20031 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
20033 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
20035 * doc/cppopts.texi: Document '-' special value to -MF.
20037 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
20039 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
20040 (cortex_a53_fconst): Likewise.
20041 (cortex_a53_fpmul): Likewise.
20042 (cortex_a53_f_load_64): Likewise.
20043 (cortex_a53_f_load_many): Likewise.
20044 (cortex_a53_advsimd_alu): Likewise.
20045 (cortex_a53_advsimd_alu_q): Likewise.
20046 (cortex_a53_advsimd_mul): Likewise.
20047 (cortex_a53_advsimd_mul_q): Likewise.
20048 (fpmac bypass): Add new bypass for fpmac-fpmac case.
20049 Add missing fmul, r2f_cvt and fconst cases.
20051 2017-06-14 Richard Biener <rguenther@suse.de>
20053 PR middle-end/81088
20054 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
20056 (fold_binary_loc): When associating do not treat pre-existing
20057 TREE_OVERFLOW on literal constants as a reason to allow
20058 TREE_OVERFLOW on associated literal constants.
20060 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20062 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
20063 (MASK_FEATURES): New macro.
20064 * config/sparc/sparc.c (sparc_option_override): Remove the special
20065 handling of -mfpu and generalize it to all MASK_FEATURES switches.
20067 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20069 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
20070 a division of 0 if non-call exceptions are enabled.
20072 2017-06-14 Andrew Pinski <apinski@cavium.com>
20073 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
20076 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
20077 Improve vector initialization code gen for only variable case.
20079 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
20081 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
20083 2017-06-14 Richard Biener <rguenther@suse.de>
20085 PR tree-optimization/81083
20086 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
20089 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20091 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
20092 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
20093 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
20094 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
20095 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
20096 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
20098 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20100 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
20101 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
20103 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20105 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
20107 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20109 * config/rs6000/t-rtems: Don't handle SPE.
20111 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20113 * config/rs6000/t-linux: Don't handle SPE.
20115 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20117 * config/rs6000/eabispe.h: Delete file.
20119 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20121 * config/rs6000/t-spe: Delete file.
20123 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20125 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
20126 (rs6000_legitimate_offset_address_p): Return false for anything in
20127 V2SImode or V2SFmode.
20129 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
20131 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
20132 except V2SF and V2SI. Rearrange the vector modes, and add comments.
20133 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
20135 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
20136 (rs6000_legitimate_offset_address_p): Ditto.
20137 (rs6000_emit_move): Ditto.
20138 (rs6000_init_builtins): Remove V4HI_type_node.
20140 2017-06-13 Martin Liska <mliska@suse.cz>
20143 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
20144 (gate_asan): Likewise.
20145 * asan.h (asan_no_sanitize_address_p): Remove the function.
20146 (sanitize_flags_p): New function.
20147 * builtins.def: Fix coding style.
20148 * common.opt: Use renamed enum value.
20149 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
20150 * doc/extend.texi: Document no_sanitize attribute.
20151 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
20152 to SANITIZE_UNDEFINED_NONDEFAULT.
20153 * gcc.c (sanitize_spec_function): Use the renamed enum value.
20154 * gimple-fold.c (optimize_atomic_compare_exchange_p):
20155 Use sanitize_flags_p.
20156 * gimplify.c (gimplify_function_tree): Likewise.
20157 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
20158 * opts.c (parse_no_sanitize_attribute): New function.
20159 (common_handle_option): Use renamed enum value.
20160 * opts.h (parse_no_sanitize_attribute): Declare.
20161 * tree.c (sanitize_flags_p): New function.
20162 * tree.h: Declared here.
20163 * tsan.c: Use sanitize_flags_p.
20164 * ubsan.c (ubsan_expand_null_ifn): Likewise.
20165 (instrument_mem_ref): Likewise.
20166 (instrument_bool_enum_load): Likewise.
20167 (do_ubsan_in_current_function): Remove the function.
20168 (pass_ubsan::execute): Use sanitize_flags_p.
20169 * ubsan.h: Remove do_ubsan_in_current_function
20170 * tree-cfg.c (print_no_sanitize_attr_value): New function.
20171 (dump_function_to_file): Use it here.
20173 2017-06-13 Martin Jambor <mjambor@suse.cz>
20175 PR tree-optimization/80803
20176 PR tree-optimization/81063
20177 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
20178 (propagate_subaccesses_across_link): Enqueue subtree whenever
20179 necessary instead of relying on the caller.
20181 2017-06-13 Martin Jambor <mjambor@suse.cz>
20183 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
20184 that have a first_link.
20185 (sort_and_splice_var_accesses): Do not check first_link before
20187 (subtree_mark_written_and_enqueue): Likewise.
20188 (propagate_all_subaccesses): Likewise and do not stop at first
20189 parent with a first_link.
20191 2017-06-13 Martin Jambor <mjambor@suse.cz>
20193 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
20196 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20198 * match.pd: New pattern.
20200 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20202 * tree-vrp.c (is_masked_range_test): New function.
20203 (register_edge_assert_for): Determine ranges for
20206 2017-06-13 Yury Gribov <tetra2005@gmail.com>
20208 PR tree-optimization/67328
20209 * fold-const.c (maskable_range_p): New function.
20210 (build_range_check): Generate bittests if possible.
20212 2017-06-13 Martin Liska <mliska@suse.cz>
20214 * gimple-pretty-print.c (dump_probability): Add new argument.
20215 (dump_edge_probability): Dump both probability and count.
20216 (dump_gimple_label): Likewise.
20217 (dump_gimple_bb_header): Likewise.
20219 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
20222 * config/avr/avr-devices.c: Fix indentation.
20223 * config/avr/gen-avr-mmcu-specs.c: Dito.
20225 2017-06-13 Richard Biener <rguenther@suse.de>
20227 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
20228 instead get vector type from stmt_info.
20229 (vectorizable_reduction): Adjust. Remove dead code.
20231 2017-06-13 Richard Biener <rguenther@suse.de>
20233 PR middle-end/81065
20234 * fold-const.c (extract_muldiv_1): Remove bogus distribution
20235 case of C * (x * C2 + C3).
20236 (fold_addr_of_array_ref_difference): Properly fold index difference.
20238 2017-06-12 David S. Miller <davem@davemloft.net>
20241 * config/sparc/sparc.md (return expander): Emit frame blockage if
20242 function uses alloca.
20244 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
20246 * combine.c (make_field_assignment): Check len rather than the mode
20247 precision when calling force_to_mode.
20249 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
20251 Support multilibs and devices that see flash in RAM address range.
20254 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
20255 (avr_mcu_t) <flash_pm_offset>: New field.
20256 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
20257 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
20258 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
20259 (AVR_TINY_PM_OFFSET): Remove macro.
20260 * config/avr/avr.opt (-mshort-calls): New option.
20261 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
20262 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
20263 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
20264 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
20265 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
20266 instead of avr_arch->have_jmp_call.
20267 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
20268 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
20269 avr_arch->flash_pm_offset to define.
20270 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
20271 new field flash_pm_offset. Add entry for avrxmega3.
20272 (avr_texinfo): Add entry for avrxmega3.
20273 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
20274 attiny212, attiny214,
20275 attiny412, attiny414, attiny416, attiny417,
20276 attiny814, attiny816, attiny817,
20277 attiny1614, attiny1616, attiny1617,
20278 attiny3214, attiny3216, attiny3217.
20279 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
20280 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
20281 (avr_print_operand_address) [AVR_TINY]: Same.
20282 (avr_asm_init_sections) <readonly_data_section>: Only patch
20283 callback if avr_arch->flash_pm_offset = 0.
20284 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
20285 for rodata if avr_arch->flash_pm_offset != 0.
20286 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
20287 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
20288 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
20289 (m_options): Append opt_rcall.
20290 (m_dirnames): Append dir_rcall.
20291 * config/avr/t-multilib: Regenerate.
20293 * configure.ac [target=avr]: Check whether avrxmega3 default
20294 linker description file works as needed.
20295 * configure: Regenerate.
20296 * doc/avr-mmcu.texi: Regenerate.
20297 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
20298 <__AVR_ARCH__>: Document avrxmega3 and 103.
20299 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
20300 <__AVR_SHORT_CALLS__>: Document it.
20301 <__AVR_PM_BASE_ADDRESS__>: Document it.
20302 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
20303 (AVR Variable Attributes) <progmem>: Document this is
20304 not needed for avrxmega3.
20305 (AVR Named Address Spaces) <__flash>: Dito.
20307 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
20309 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
20311 2017-06-12 Doug Rupp <rupp@adacore.com>
20313 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
20314 Append vxworks-stdint.h to the tm_file list.
20315 * config/vxworks-stdint.h: New file.
20317 2017-06-12 Martin Liska <mliska@suse.cz>
20319 PR tree-optimization/81041
20320 * tree-profile.c (gimple_gen_ic_func_profiler):
20321 Create an extra BB in profile-generate
20322 (gimple_gen_time_profiler): Likewise.
20324 2017-06-12 Jakub Jelinek <jakub@redhat.com>
20326 PR tree-optimization/81003
20327 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
20328 (update_range_test): Use it instead of force_gimple_operand_gsi.
20330 2017-06-12 Richard Biener <rguenther@suse.de>
20332 PR tree-optimization/81053
20333 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
20334 with backedge value not defined in loop. Simplify def stmt
20337 2017-06-11 Tom de Vries <tom@codesourcery.com>
20340 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
20342 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
20343 nvptx_cannot_force_const_mem.
20345 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20347 * opts.c (finish_options): Move test for flag_split_stack after
20348 it has been initialized.
20350 2017-06-11 Jason Merrill <jason@redhat.com>
20352 * tree.h (id_equal): New.
20353 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
20354 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
20355 instead of strcmp of IDENTIFIER_POINTER.
20357 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20359 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
20360 (mark_all_inlined_calls_cdtor): Fix formating.
20361 (inline_transform): Rescale profile before inlining.
20363 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20365 * cgraph.h (cgraph_edge::clone): Update prototype.
20366 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
20367 (cgraph_node::create_clone): Update.
20368 (cgraph_node::create_version_clone): Update.
20369 * tree-inline.c (copy_bb): Update.
20370 (expand_call_inline): Update.
20372 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20374 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
20375 factored out from ...
20376 (rs6000_emit_prologue): ... here.
20378 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
20380 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
20381 factored out from ...
20382 (rs6000_emit_prologue): ... here.
20384 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20386 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
20388 (handle_missing_profiles): Fix computation of tp_first_run.
20389 (counts_to_freqs): Do not touch freqs when count is 0.
20391 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
20393 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
20396 2017-06-10 Tom de Vries <tom@codesourcery.com>
20398 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
20399 attributes): Document signal effective target.
20401 2017-06-10 Tom de Vries <tom@codesourcery.com>
20403 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
20404 Document effective target stack_size.
20406 2017-06-09 David Malcolm <dmalcolm@redhat.com>
20408 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
20409 to the edit_context if they can be auto-applied.
20411 2017-06-9 Ian Lance Taylor <iant@golang.org>
20413 * opts.c (finish_options): If -fsplit-stack, disable implicit
20414 -forder-blocks-and-partition.
20415 * doc/invoke.texi (Optimize Options): Document that when using
20416 -fsplit-stack -forder-blocks-and-partition is not implicitly
20419 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20421 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
20422 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
20423 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
20424 * builtins.def (abort, trap, unreachable): Declare cold.
20425 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
20426 * tree-core.h (ECF_COLD): New.
20427 * tree.c (set_call_expr_flags): Handle ECF_COLD.
20428 (build_common_builtin_nodes): Mark unreachable and abort as cold.
20430 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20432 * predict.c (unlikely_executed_stmt_p): Cleanup.
20434 2017-06-09 Richard Biener <rguenther@suse.de>
20436 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
20437 model if the ref is always written to.
20439 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20441 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
20443 2017-06-09 Tamar Christina <tamar.christina@arm.com>
20445 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
20448 2017-06-09 Tom de Vries <tom@codesourcery.com>
20451 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
20452 "target cannot support label values" when encountering LABEL_REF.
20454 2017-06-09 Martin Liska <mliska@suse.cz>
20456 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
20457 (gimple_gen_ic_func_profiler): Emit direct comparison
20458 of __gcov_indirect_call_callee with NULL.
20459 (gimple_gen_time_profiler): Change probability from
20460 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
20462 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20464 * profile.c (edge_gcov_counts): Turn to pointer.
20465 (compute_branch_probabilities, compute_branch_probabilities): Update.
20466 (branch_prob): Do not clear edge_gcov_count.
20467 * profile.h (edge_gcov_counts): Turn to pointer.
20468 (edge_gcov_count): Update.
20470 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
20472 * gimple.h (gimple_check_failed): Mark cold.
20474 2017-06-09 Richard Biener <rguenther@suse.de>
20476 PR tree-optimization/66623
20477 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
20478 refactor check_reduction into two parts, properly computing
20479 whether we have to check reduction validity for outer loop
20482 2017-06-09 Richard Biener <rguenther@suse.de>
20484 PR tree-optimization/79483
20485 * graphite-scop-detection.c (order): New global.
20486 (get_order): Compute bb to order mapping that satisfies code
20487 generation constraints.
20488 (cmp_pbbs): New helper.
20489 (build_scops): Start domwalk at entry block, sort generated
20492 2017-06-09 Richard Biener <rguenther@suse.de>
20494 PR middle-end/81007
20495 * ipa-polymorphic-call.c
20496 (ipa_polymorphic_call_context::restrict_to_inner_class):
20497 Skip FIELD_DECLs with error_mark_node type.
20498 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
20501 2017-06-09 Martin Liska <mliska@suse.cz>
20503 * predict.c (struct branch_predictor): New struct.
20504 (test_prediction_value_range): New test.
20505 (predict_c_tests): New function.
20506 * selftest-run-tests.c (selftest::run_tests): Run the function.
20507 * selftest.h: Declare new tests.
20509 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
20512 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
20513 gen_add3_insn did not fail.
20514 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
20515 r0, construct that number in a temporary reg and add that reg to r0.
20516 If asked to put the result in r0 as well, fail.
20518 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
20520 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20521 for early expansion of vec_eqv.
20523 2017-06-08 Jakub Jelinek <jakub@redhat.com>
20525 PR middle-end/81005
20526 * ubsan.c (instrument_null): Avoid pointless code temporary.
20527 (pass_ubsan::execute): Instrument aggregate arguments of calls.
20529 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
20533 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
20536 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
20539 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20541 * predict.c (unlikely_executed_edge_p): Move ahead.
20542 (probably_never_executed_edge_p): Use it.
20544 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20546 PR middle-end/79988
20547 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
20548 gimple_call_builtin_p call.
20550 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20552 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
20553 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
20554 rtl_check_failed_type2, rtl_check_failed_code1,
20555 rtl_check_failed_code2, rtl_check_failed_code_mode,
20556 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
20557 rtvec_check_failed_bounds, rtl_check_failed_flag,
20558 _fatal_insn_not_found, _fatal_insn): Likewise.
20559 * tree.h (tree_contains_struct_check_failed,
20560 tree_check_failed, tree_not_check_failed,
20561 tree_class_check_failed, tree_range_check_failed,
20562 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
20563 tree_vec_elt_check_failed, phi_node_elt_check_failed,
20564 tree_operand_check_failed, omp_clause_check_failed,
20565 omp_clause_operand_check_failed, omp_clause_range_check_failed):
20568 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20570 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
20571 flag_branch_probabilities.
20572 * ipa-inline.c (edge_badness): Likewise.
20573 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
20574 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
20575 * predict.c (maybe_hot_frequency_p): Likewise.
20576 (probably_never_executed): Likewise.
20577 * sched-ebb.c (schedule_ebbs): Likewise.
20578 * sched-rgn.c (find_single_block_region): Likewise.
20579 * tracer.c (tail_duplicate): Likewise.
20581 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20583 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
20584 longer requires x_flag_profile_use.
20586 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20588 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
20589 instead of flag_reorder_blocks_and_partition.
20590 * dbxout.c (dbxout_function_end): Likewise.
20591 * dwarf2out.c (gen_subprogram_die): Likewise.
20592 * haifa-sched.c (sched_create_recovery_edges): Likewise.
20593 * hw-doloop.c (reorg_loops): Likewise.
20594 * varasm.c (assemble_start_function,
20595 assemble_end_function): Likewise.
20596 (decide_function_section): Do not check for
20597 flag_reorder_blocks_and_partition.
20599 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
20601 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
20603 (chkp_get_hard_register_fake_addr_expr): Ditto.
20604 (chkp_build_addr_expr): Add check for hard reg case.
20605 (chkp_parse_array_and_component_ref): Ditto.
20606 (chkp_find_bounds_1): Ditto.
20607 (chkp_process_stmt): Don't generate bounds store for
20610 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20612 * predict.c (maybe_hot_bb_p): Do not check profile status.
20613 (maybe_hot_edge_p): Likewise.
20614 (probably_never_executed): Check for zero counts even if profile
20616 (unlikely_executed_edge_p): New function.
20617 (unlikely_executed_stmt_p): New function.
20618 (unlikely_executed_bb_p): New function.
20619 (set_even_probabilities): Use unlikely predicates.
20620 (combine_predictions_for_bb): Likewise.
20621 (predict_paths_for_bb): Likewise.
20622 (predict_paths_leading_to_edge): Likewise.
20623 (determine_unlikely_bbs): New function.
20624 (estimate_bb_frequencies): Use it.
20625 (compute_function_frequency): Use zero counts even if profile is
20627 * profile-count.h: Fix typo.
20629 2017-08-08 Julia Koval <julia.koval@intel.com>
20631 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
20632 _mm512_mask_cvtsepi16_storeu_epi8,
20633 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
20634 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
20635 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
20636 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
20637 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
20638 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
20639 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
20640 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
20641 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
20642 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
20643 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
20644 __builtin_ia32_pmovuswb256mem_mask,
20645 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
20646 __builtin_ia32_pmovwb512mem_mask): New builtins.
20648 2017-08-08 Julia Koval <julia.koval@intel.com>
20650 PR target/73350,80862
20651 * config/i386/subst.md (round): Fix round pattern.
20652 * config/i386/i386.c (ix86_erase_embedded_rounding):
20653 Fix erasing rounding for the fixed pattern.
20655 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
20657 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
20659 2017-06-08 Martin Liska <mliska@suse.cz>
20661 PR gcov-profile/80911
20662 * gcov.c (block_info::block_info): New constructor.
20664 2017-06-07 Carl Love <cel@us.ibm.com>
20666 * config/rs6000/rs6000-c: The return type of the following
20667 built-in functions was implemented as int not long long. Fix sign
20668 of return value for the unsigned version of vec_mulo and vec_mule.
20669 vector unsigned long long vec_bperm (vector unsigned long long,
20670 vector unsigned char)
20671 vector signed long long vec_mule (vector signed int,
20673 vector unsigned long long vec_mule (vector unsigned int,
20674 vector unsigned int)
20675 vector signed long long vec_mulo (vector signed int,
20677 vector unsigned long long vec_mulo (vector unsigned int,
20678 vector unsigned int)
20679 * doc/extend.texi: Fix the documentation for the built-in
20682 2017-06-07 Carl Love <cel@us.ibm.com>
20685 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
20688 2017-06-07 Carl Love <cel@us.ibm.com>
20690 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
20691 support, Generate doublehv for signed int/float for BE case only.
20693 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
20695 * doc/invoke.texi (mcx16): Rewrite.
20697 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20699 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
20700 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
20701 *mov<mode>_softfloat, and an anonymous splitter): Use
20702 nonimmediate_operand instead of rs6000_nonimmediate_operand.
20704 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20706 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
20708 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20709 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
20710 (rs6000_debug_reg_global): Adjust.
20711 (rs6000_init_hard_regno_mode_ok): Adjust.
20712 (rs6000_dbx_register_number): Adjust.
20713 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
20714 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20715 Remove SPE_ACC and SPEFSCR.
20716 (REG_ALLOC_ORDER): Ditto.
20717 (FRAME_POINTER_REGNUM): Change to 111.
20718 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
20719 (REG_CLASS_NAMES): Ditto.
20720 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
20721 (REGISTER_NAMES): Ditto.
20722 (ADDITIONAL_REG_NAMES): Ditto.
20723 (rs6000_reg_names): Ditto.
20724 * config/rs6000/rs6000.md: Renumber some register number
20727 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20729 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
20731 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
20732 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
20734 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
20735 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
20736 Delete the SPE high registers.
20737 (REG_ALLOC_ORDER): Ditto.
20738 (enum reg_class): Remove SPE_HIGH_REGS.
20739 (REG_CLASS_NAMES): Ditto.
20740 (REG_CLASS_CONTENTS): Delete the SPE high registers.
20741 (REGISTER_NAMES): Ditto.
20742 (rs6000_reg_names): Ditto.
20743 * doc/tm.texi.in: Remove SPE as example.
20744 * doc/tm.texi: Regenerate.
20746 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20748 * config/rs6000/8540.md (ppc8540_brinc): Delete.
20749 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
20750 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
20751 * config/rs6000/rs6000.md (type): Remove "brinc".
20753 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20755 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
20756 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
20757 * config/rs6000/linuxspe.h: Delete file.
20758 * config/rs6000/rs6000.md: Don't include spe.md.
20759 * config/rs6000/spe.h: Delete file.
20760 * config/rs6000/spe.md: Delete file.
20761 * config/rs6000/t-rs6000: Remove spe.md.
20763 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20765 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
20766 (reg_or_none500mem_operand): Delete.
20767 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
20768 instead of reg_or_none500mem_operand.
20770 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20772 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20773 handling of SPE flags.
20774 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
20776 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20778 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
20780 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
20781 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
20782 paired_divv2sf3): Similar.
20783 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
20784 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
20785 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
20787 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
20788 Rename the paired_* instruction patterns.
20789 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
20791 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
20792 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
20793 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
20794 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
20795 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
20796 PAIRED_VECTOR_MODE.
20797 (struct machine_function): Delete field spe_insn_chain_scanned_p.
20798 (spe_func_has_64bit_regs_p): Delete.
20799 (spe_expand_predicate_builtin): Delete.
20800 (spe_expand_evsel_builtin): Delete.
20801 (TARGET_DWARF_REGISTER_SPAN): Do not define.
20802 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
20803 (invalid_e500_subreg): Delete.
20804 (rs6000_legitimize_address): Always force_reg op2 as well, for
20805 paired single memory accesses.
20806 (rs6000_member_type_forces_blk): Delete.
20807 (rs6000_spe_function_arg): Delete.
20808 (rs6000_expand_unop_builtin): Delete SPE handling.
20809 (rs6000_expand_binop_builtin): Ditto.
20810 (spe_expand_stv_builtin): Delete.
20811 (bdesc_2arg_spe): Delete.
20812 (spe_expand_builtin): Delete.
20813 (spe_expand_predicate_builtin): Delete.
20814 (spe_expand_evsel_builtin): Delete.
20815 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
20816 (spe_init_builtins): Delete.
20817 (spe_func_has_64bit_regs_p): Delete.
20818 (savres_routine_name): Delete "info" parameter. Adjust callers.
20819 (rs6000_emit_stack_reset): Ditto.
20820 (rs6000_dwarf_register_span): Delete.
20821 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
20822 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
20823 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
20825 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
20827 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
20828 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
20829 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
20830 mulv2sf3, divv2sf3): Delete expanders.
20832 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20834 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
20836 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20838 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
20839 * config/rs6000/rs6000.c: Ditto.
20841 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20843 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
20844 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
20845 comparison_operator.
20847 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20849 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
20850 * config/rs6000/rs6000.opt: Ditto.
20851 * config/rs6000/t-rtems: Ditto.
20853 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20855 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
20856 TARGET_E500_SINGLE by 0, simplify.
20857 * config/rs6000/rs6000.c: Ditto.
20858 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
20859 (spe_build_register_parallel): Delete.
20860 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
20861 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
20862 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
20863 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
20864 (E500_CONVERT): Delete.
20865 * config/rs6000/spe.md: Remove many patterns and all define_constants.
20867 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
20869 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
20870 * config/rs6000/dfp.md: Ditto.
20871 (negdd2, *negdd2_fpr): Merge.
20872 (absdd2, *absdd2_fpr): Merge.
20873 (negtd2, *negtd2_fpr): Merge.
20874 (abstd2, *abstd2_fpr): Merge.
20875 * config/rs6000/e500.h: Delete file.
20876 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
20877 TARGET_FPRS by 1 and simplify.
20878 * config/rs6000/rs6000-c.c: Ditto.
20879 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
20880 TARGET_DF_SPE by 0.
20881 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
20883 * config/rs6000/rs6000.md: Ditto.
20884 (floatdidf2, *floatdidf2_fpr): Merge.
20885 (move_from_CR_gt_bit): Delete.
20886 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
20887 (E500_CR_IOR_COMPARE): Delete.
20888 (All patterns that require !TARGET_FPRS): Delete.
20889 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
20891 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20893 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
20895 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20897 * graphds.c (add_edge): Intitialize edge's attached data.
20898 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
20899 pointer parameter. Call pointed function on each edge during
20900 graph traversing. Skip traversing the edge when the function
20902 (graphds_dfs, graphds_scc): Ditto.
20903 (for_each_edge): New parameter. Pass the new parameter to callback
20905 * graphds.h (skip_edge_callback): New function pointer type.
20906 (graphds_dfs, graphds_scc): New function pointer parameter.
20907 (graphds_edge_callback, for_each_edge): New parameter.
20909 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20911 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
20912 out code checking if runtime alias check is possible to below ...
20913 Call the new function.
20914 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
20915 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
20917 2017-06-07 Marek Polacek <polacek@redhat.com>
20920 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
20921 TYPE_OVERFLOW_WRAPS check.
20923 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20925 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
20926 if versioning is required.
20927 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
20928 peeling with the check for versioning.
20930 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20932 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
20933 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
20934 Set true to new parameter if new ssa variable is defined.
20935 (vect_gen_vector_loop_niters): Refactor. Set range information
20936 for the new vector loop bound variable.
20937 (vect_do_peeling): Ditto.
20939 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20941 * tree-affine.c (ssa.h): Include header file.
20942 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
20943 has wrapping overflow behavior.
20945 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20947 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
20949 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20951 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
20952 (tree_to_aff_combination): ... here.
20954 2017-06-07 Bin Cheng <bin.cheng@arm.com>
20956 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
20957 reg_pressure model function.
20958 (ivopts_global_cost_for_size): Delete.
20959 (determine_set_costs, iv_ca_recount_cost): Call new model function
20960 ivopts_estimate_reg_pressure.
20962 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20964 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
20965 expensive than udiv. Remove floating point cases from mod.
20967 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20969 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
20970 Increase idiv cost.
20972 2017-06-07 Tamar Christina <tamar.christina@arm.com>
20974 * config/aarch64/aarch64.md
20975 (copysignsf3): Fix mask generation.
20977 2017-06-07 Jakub Jelinek <jakub@redhat.com>
20979 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
20981 (class dump_manager): Add register_dumps method.
20982 * dumpfile.c: Include langhooks.h.
20983 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
20984 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
20985 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
20986 (dump_manager::dump_register): Start with 512 entries instead of 32.
20987 (dump_manager::register_dumps): New method.
20988 * toplev.c (general_init): Instead of invoking register_dumps
20989 langhook, invoke register_dumps method on the dump manager.
20990 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
20993 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
20995 * doc/md.texi: Clarify the restrictions on a define_insn condition.
20996 Say that # requires an associated define_split to exist, and that
20997 the define_split must be suitable for use after register allocation.
20999 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21001 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
21002 (compute_outgoing_frequencies): Also initialize zero counts.
21003 (find_many_sub_basic_blocks): Do not produce uninitialized profile
21004 around loops; preserve more of profile when nothing changes.
21006 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
21008 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
21010 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
21011 * config/arm/arm-cpu-cdata.h: Regenerate.
21012 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
21013 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
21014 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
21015 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
21016 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
21018 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
21020 * config/arm/t-rmprofile: Likewise.
21021 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
21023 2017-06-06 David S. Miller <davem@davemloft.net>
21026 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
21027 blockage if function uses alloca.
21029 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21031 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
21032 New "uid" fields to hold pretty-print IDs of group and ref.
21033 Memory references are now identified as <group_id>:<ref_id>
21034 instead of using [random] addresses.
21035 (dump_mem_details): Simplify, no functional change.
21036 (dump_mem_ref): Simplify and make output more concise.
21037 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
21038 (find_or_create_group): Initialize group uid.
21039 (record_ref): Initialize ref uid. Improve debug output.
21040 (prune_group_by_reuse, should_issue_prefetch_p,)
21041 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
21042 (mark_nontemporal_store, determine_loop_nest_reuse):
21043 Improve debug output.
21045 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
21047 * dbgcnt.def (prefetch): New debug counter.
21048 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
21049 (schedule_prefetches): Stop issueing prefetches if debug counter
21052 2017-06-06 Tom de Vries <tom@codesourcery.com>
21054 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
21055 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
21057 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21059 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
21060 Use aarch64_reg_or_zero predicate for operand 4.
21061 (aarch64_compare_and_swap<mode> define_insn_and_split):
21062 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
21063 (aarch64_store_exclusive<mode>): Likewise for operand 2.
21065 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
21067 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
21068 (arm_compute_save_core_reg_mask): This.
21069 (thumb1_compute_save_reg_mask): Rename into ...
21070 (thumb1_compute_save_core_reg_mask): This.
21071 (arm_compute_save_reg0_reg12_mask): Adapt comment.
21072 (arm_compute_frame_layout): Likewise.
21074 2017-06-06 Richard Biener <rguenther@suse.de>
21076 PR tree-optimization/80974
21077 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
21078 keep or clear leaders SSA info.
21080 2017-06-06 Tom de Vries <tom@codesourcery.com>
21082 * config/nvptx/nvptx.c (split_mode_p): New function.
21083 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
21085 2017-06-06 Tom de Vries <tom@codesourcery.com>
21087 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
21089 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21092 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
21095 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21097 * shrink-wrap.c (handle_simple_exit): Update profile.
21098 (try_shrink_wrapping): Upate profile.
21100 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21102 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
21103 (tree_guess_outgoing_edge_probabilities): New.
21104 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
21105 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
21107 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21109 * ipa-split.c (split_function): Initialize return bb profile.
21111 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
21113 * profile.c (compute_branch_probabilities): Also initialize
21114 EXIT_BLOCK profile.
21116 2017-06-06 Richard Biener <rguenther@suse.de>
21118 PR tree-optimization/80928
21119 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
21120 (vect_analyze_loop_operations): Properly guard analysis for
21122 (vect_transform_loop): Likewise.
21123 (vect_analyze_loop_2): Also reset SLP type on PHIs.
21124 (vect_model_induction_cost): Do not cost for pure SLP.
21125 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
21126 of induction in inner loop vectorization.
21127 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
21128 (vect_get_and_check_slp_defs): Handle vect_induction_def.
21129 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
21131 (vect_analyze_slp_cost_1): Cost induction.
21132 (vect_detect_hybrid_slp_stmts): Handle PHIs.
21133 (vect_get_slp_vect_defs): Likewise.
21134 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
21135 (vect_transform_stmt): Handle SLP reductions.
21136 * tree-vectorizer.h (vectorizable_induction): Adjust.
21138 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21140 * config/rs6000/rs6000.c (make_resolver_func): Update
21141 init_lowered_empty_function call.
21143 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
21145 * doc/invoke.texi: Document the -fprofile-abs-path option.
21146 * common.opt (fprofile-abs-path): New option.
21147 * gcov-io.h (gcov_write_filename): Declare.
21148 * gcov-io.c (gcov_write_filename): New function.
21149 * coverage.c (coverage_begin_function): Use gcov_write_filename.
21150 * profile.c (output_location): Likewise.
21152 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21154 * shring-wrap.c: Revert accidental commit.
21156 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
21158 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
21160 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21162 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
21164 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
21165 profile in callgraph edge.
21166 * profile-count.h (apply_probability): If THIS is 0, then result is 0
21167 (apply_scale): Likewise.
21168 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
21169 Also scale profile when inlining function with zero profile.
21170 (initialize_cfun): Update exit block profile even when it is zero.
21171 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
21172 when profile is read.
21174 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
21176 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
21177 (CLONE_*): New constants to define the processors we can generate
21178 code for with the target_clone attribute.
21179 (rs6000_clone_map): New array to identify which clone processors
21180 the current program is running on.
21181 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
21182 target_clone attribute.
21183 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
21184 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
21185 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
21186 (cpu_expand_builtin): Add support for target_clone attribute.
21187 (rs6000_valid_attribute_p): Allow "default" attribute.
21188 (get_decl_name): New debug function to simplify printing the
21189 current function name in debugging statements.
21190 (rs6000_clone_priority): New functions to support the target_clone
21191 attribute, and be able to generate code to switch between ISA 2.05
21192 through ISA 3.0 (power6 through power9).
21193 (rs6000_compare_version_priority): Likewise.
21194 (rs6000_get_function_versions_dispatcher): Likewise.
21195 (make_resolver_func): Likewise.
21196 (add_condition_to_bb): Likewise.
21197 (dispatch_function_versions): Likewise.
21198 (rs6000_generate_version_dispatcher_body): Likewise.
21199 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
21200 (fusion_gpr_load_p): Fix a spacing issue.
21201 * doc/extend.texi (Common Function Attributes): Document that the
21202 PowerPC supports the target_clone attribute.
21204 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
21206 * config/arm/arm.h: explain F symbol found in description of ARM
21207 register allocation in its legend.
21209 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21211 * config/mips/frame-header-opt.c: Include profile-count.h.
21212 * config/riscv/riscv.c: Include profile-count.h
21214 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
21216 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
21218 (sm_set_flag_if_changed): Add bbs field.
21219 (execute_sm_if_changed_flag_set): Pass BBS.
21220 (execute_sm): Update.
21222 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21224 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
21227 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21229 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
21230 (peephole2): New peephole2 to emit the above.
21231 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
21233 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21235 * config/aarch64/aarch64.c (define_peephole2 above
21236 *sub_<shift>_<mode>): New peephole.
21238 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21240 * config/i386/i386.c (make_resolver_func): Update.
21241 * Makefile.in: Add profile-count.h and profile-count.o
21242 * auto-profile.c (afdo_indirect_call): Update to new API.
21243 (afdo_set_bb_count): Update.
21244 (afdo_propagate_edge): Update.
21245 (afdo_propagate_circuit): Update.
21246 (afdo_calculate_branch_prob): Update.
21247 (afdo_annotate_cfg): Update.
21248 * basic-block.h: Include profile-count.h
21249 (struct edge_def): Turn count to profile_count.
21250 (struct basic_block_def): Likewie.
21251 (REG_BR_PROB_BASE): Move to profile-count.h
21252 (RDIV): Move to profile-count.h
21253 * bb-reorder.c (max_entry_count): Turn to profile_count.
21254 (find_traces): Update.
21255 (rotate_loop):Update.
21256 (connect_traces):Update.
21257 (sanitize_hot_paths):Update.
21258 * bt-load.c (migrate_btr_defs): Update.
21259 * cfg.c (RDIV): Remove.
21260 (init_flow): Use alloc_block.
21261 (alloc_block): Uninitialize count.
21262 (unchecked_make_edge): Uninitialize count.
21263 (check_bb_profile): Update.
21264 (dump_edge_info): Update.
21265 (dump_bb_info): Update.
21266 (update_bb_profile_for_threading): Update.
21267 (scale_bbs_frequencies_int): Update.
21268 (scale_bbs_frequencies_gcov_type): Update.
21269 (scale_bbs_frequencies_profile_count): New.
21270 * cfg.h (update_bb_profile_for_threading): Update.
21271 (scale_bbs_frequencies_profile_count): Declare.
21272 * cfgbuild.c (compute_outgoing_frequencies): Update.
21273 (find_many_sub_basic_blocks): Update.
21274 * cfgcleanup.c (try_forward_edges): Update.
21275 (try_crossjump_to_edge): Update.
21276 * cfgexpand.c (expand_gimple_tailcall): Update.
21277 (construct_exit_block): Update.
21278 * cfghooks.c (verify_flow_info): Update.
21279 (dump_bb_for_graph): Update.
21280 (split_edge): Update.
21281 (make_forwarder_block): Update.
21282 (duplicate_block): Update.
21283 (account_profile_record): Update.
21284 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
21285 (get_estimated_loop_iterations): Update.
21286 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
21287 (single_likely_exit): Update.
21288 * cfgloopmanip.c (scale_loop_profile): Update.
21290 (set_zero_probability): Update.
21291 (lv_adjust_loop_entry_edge): Update.
21292 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
21293 (purge_dead_edges): Update.
21294 (rtl_account_profile_record): Update.
21295 * cgraph.c (cgraph_node::create): Uninitialize count.
21296 (symbol_table::create_edge): Uninitialize count.
21297 (cgraph_update_edges_for_call_stmt_node): Update.
21298 (cgraph_edge::dump_edge_flags): Update.
21299 (cgraph_node::dump): Update.
21300 (cgraph_edge::maybe_hot_p): Update.
21301 * cgraph.h: Include profile-count.h
21302 (create_clone), create_edge, create_indirect_edge): Update.
21303 (cgraph_node): Turn count to profile_count.
21304 (cgraph_edge0: Likewise.
21305 (make_speculative, clone): Update.
21306 (create_edge): Update.
21307 (init_lowered_empty_function): Update.
21308 * cgraphclones.c (cgraph_edge::clone): Update.
21309 (duplicate_thunk_for_node): Update.
21310 (cgraph_node::create_clone): Update.
21311 * cgraphunit.c (cgraph_node::analyze): Update.
21312 (cgraph_node::expand_thunk): Update.
21313 * final.c (dump_basic_block_info): Update.
21314 * gimple-streamer-in.c (input_bb): Update.
21315 * gimple-streamer-out.c (output_bb): Update.
21316 * graphite.c (print_global_statistics): Update.
21317 (print_graphite_scop_statistics): Update.
21318 * hsa-brig.c: Include basic-block.h.
21319 * hsa-dump.c: Include basic-block.h.
21320 * hsa-gen.c (T sum_slice): Update.
21321 (convert_switch_statements):Update.
21322 * hsa-regalloc.c: Include basic-block.h.
21323 * ipa-chkp.c (chkp_produce_thunks): Update.
21324 * ipa-cp.c (struct caller_statistics): Update.
21325 (init_caller_stats): Update.
21326 (gather_caller_stats): Update.
21327 (ipcp_cloning_candidate_p): Update.
21328 (good_cloning_opportunity_p): Update.
21329 (get_info_about_necessary_edges): Update.
21330 (dump_profile_updates): Update.
21331 (update_profiling_info): Update.
21332 (update_specialized_profile): Update.
21333 (perhaps_add_new_callers): Update.
21334 (decide_about_value): Update.
21335 (ipa_cp_c_finalize): Update.
21336 * ipa-devirt.c (struct odr_type_warn_count): Update.
21337 (struct decl_warn_count): Update.
21338 (struct final_warning_record): Update.
21339 (possible_polymorphic_call_targets): Update.
21340 (ipa_devirt): Update.
21341 * ipa-fnsummary.c (redirect_to_unreachable): Update.
21342 * ipa-icf.c (sem_function::merge): Update.
21343 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
21344 * ipa-inline.c (compute_uninlined_call_time): Update.
21345 (compute_inlined_call_time): Update.
21346 (want_inline_small_function_p): Update.
21347 (want_inline_self_recursive_call_p): Update.
21348 (edge_badness): Update.
21349 (lookup_recursive_calls): Update.
21350 (recursive_inlining): Update.
21351 (inline_small_functions): Update.
21352 (dump_overall_stats): Update.
21353 (dump_inline_stats): Update.
21354 * ipa-profile.c (ipa_profile_generate_summary): Update.
21355 (ipa_propagate_frequency): Update.
21356 (ipa_profile): Update.
21357 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21358 * ipa-utils.c (ipa_merge_profiles): Update.
21359 * loop-doloop.c (doloop_modify): Update.
21360 * loop-unroll.c (report_unroll): Update.
21361 (unroll_loop_runtime_iterations): Update.
21362 * lto-cgraph.c (lto_output_edge): Update.
21363 (lto_output_node): Update.
21364 (input_node): Update.
21365 (input_edge): Update.
21366 (merge_profile_summaries): Update.
21367 * lto-streamer-in.c (input_cfg): Update.
21368 * lto-streamer-out.c (output_cfg): Update.
21369 * mcf.c (create_fixup_graph): Update.
21370 (adjust_cfg_counts): Update.
21371 (sum_edge_counts): Update.
21372 * modulo-sched.c (sms_schedule): Update.
21373 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
21374 * predict.c (maybe_hot_count_p): Update.
21375 (probably_never_executed): Update.
21376 (dump_prediction): Update.
21377 (combine_predictions_for_bb): Update.
21378 (propagate_freq): Update.
21379 (handle_missing_profiles): Update.
21380 (counts_to_freqs): Update.
21381 (rebuild_frequencies): Update.
21382 (force_edge_cold): Update.
21383 * predict.h: Include profile-count.h
21384 (maybe_hot_count_p, counts_to_freqs): UPdate.
21385 * print-rtl-function.c: Do not include cfg.h
21386 * print-rtl.c: Include basic-block.h
21387 * profile-count.c: New file.
21388 * profile-count.h: New file.
21389 * profile.c (is_edge_inconsistent): Update.
21390 (correct_negative_edge_counts): Update.
21391 (is_inconsistent): Update.
21392 (set_bb_counts): Update.
21393 (read_profile_edge_counts): Update.
21394 (compute_frequency_overlap): Update.
21395 (compute_branch_probabilities): Update; Initialize and deinitialize
21397 (branch_prob): Update.
21398 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
21399 (edge_gcov_count): New.
21400 (bb_gcov_count): New.
21401 * shrink-wrap.c (try_shrink_wrapping): Update.
21402 * tracer.c (better_p): Update.
21403 * trans-mem.c (expand_transaction): Update.
21404 (ipa_tm_insert_irr_call): Update.
21405 (ipa_tm_insert_gettmclone_call): Update.
21406 * tree-call-cdce.c: Update.
21407 * tree-cfg.c (gimple_duplicate_sese_region): Update.
21408 (gimple_duplicate_sese_tail): Update.
21409 (gimple_account_profile_record): Update.
21410 (execute_fixup_cfg): Update.
21411 * tree-inline.c (copy_bb): Update.
21412 (copy_edges_for_bb): Update.
21413 (initialize_cfun): Update.
21414 (freqs_to_counts): Update.
21415 (copy_cfg_body): Update.
21416 (expand_call_inline): Update.
21417 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
21418 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
21419 (try_unroll_loop_completely): Update.
21420 (try_peel_loop): Update.
21421 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
21422 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
21423 * tree-ssa-loop-split.c (connect_loops): Update.
21424 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
21425 * tree-ssa-reassoc.c (branch_fixup): Update.
21426 * tree-ssa-tail-merge.c (replace_block_by): Update.
21427 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
21428 (compute_path_counts): Update.
21429 (update_profile): Update.
21430 (recompute_probabilities): Update.
21431 (update_joiner_offpath_counts): Update.
21432 (estimated_freqs_path): Update.
21433 (freqs_to_counts_path): Update.
21434 (clear_counts_path): Update.
21435 (ssa_fix_duplicate_block_edges): Update.
21436 (duplicate_thread_path): Update.
21437 * tree-switch-conversion.c (case_bit_test_cmp): Update.
21438 (struct switch_conv_info): Update.
21439 * tree-tailcall.c (decrease_profile): Update.
21440 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
21441 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
21442 * value-prof.c (check_counter): Update.
21443 (gimple_divmod_fixed_value): Update.
21444 (gimple_mod_pow2): Update.
21445 (gimple_mod_subtract): Update.
21446 (gimple_ic_transform): Update.
21447 (gimple_stringop_fixed_value): Update.
21448 * value-prof.h (gimple_ic): Update.
21450 2017-06-02 Carl Love <cel@us.ibm.com>
21452 * config/rs6000/rs6000-c: Add support for built-in functions
21453 vector double vec_doublee (vector signed int);
21454 vector double vec_doublee (vector unsigned int);
21455 vector double vec_doublee (vector float);
21456 vector double vec_doubleh (vector signed int);
21457 vector double vec_doubleh (vector unsigned int);
21458 vector double vec_doubleh (vector float);
21459 vector double vec_doublel (vector signed int);
21460 vector double vec_doublel (vector unsigned int);
21461 vector double vec_doublel (vector float);
21462 vector double vec_doubleo (vector signed int);
21463 vector double vec_doubleo (vector unsigned int);
21464 vector double vec_doubleo (vector float);.
21465 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
21466 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
21468 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
21469 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
21470 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
21472 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
21473 vec_doublel, vec_doubleh.
21474 * doc/extend.texi: Update the built-in documentation file for the
21475 new built-in functions.
21477 2017-06-02 David Malcolm <dmalcolm@redhat.com>
21480 * ipa-inline-analysis.c (free_growth_caches): Set
21481 edge_removal_hook_holder to NULL after removing it.
21483 2017-06-02 Sudakshina Das <sudi.das@arm.com>
21485 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
21486 comparision with zero.
21488 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21489 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21490 for early expansion of vec_min and vec_max builtins.
21491 (builtin_function_type): Add min/max unsigned variants to those
21492 identified as having unsigned arguments.
21494 2017-06-02 Olivier Hainque <hainque@adacore.com>
21496 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
21498 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21500 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
21501 Use VALL_F16 iterator rather than VALL.
21503 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21505 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
21506 Emit CBNZ inside loop when doing a strong exchange and comparing
21507 against zero. Generate the CC flags after the loop.
21509 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
21511 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
21512 (dl_section_ref): New.
21513 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
21514 On AIX, append an expression to subtract the size of the
21515 section length to dl_section_ref.
21517 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
21519 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
21520 for early expansion of vector absolute builtins.
21522 2017-06-02 Richard Biener <rguenther@suse.de>
21524 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
21525 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
21527 2017-06-02 Richard Biener <rguenther@suse.de>
21529 PR tree-optimization/80948
21530 * tree-tailcall.c (find_tail_calls): Track stmts to move in
21531 stmt order as well.
21533 2017-06-02 Richard Biener <rguenther@suse.de>
21535 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
21537 * tree-vect-stmts.c (process_use): Do not mark backedge defs
21538 for inductions as relevant.
21540 2017-06-02 Richard Biener <rguenther@suse.de>
21542 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
21543 (vectorizable_induction): ... this. Remove dead code.
21545 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
21547 * builtins. (expand_builtin_alloca): Remove second parameter and
21548 infer its value from the first parameter instead.
21549 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
21551 2017-06-02 Jakub Jelinek <jakub@redhat.com>
21553 PR rtl-optimization/80903
21554 * loop-doloop.c (add_test): Unshare sequence.
21556 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
21558 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
21560 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
21562 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
21564 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
21565 xlogue_layout::get_instance, logue_layout::xlogue_layout,
21566 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
21567 (xlogue_layout::get_stub_rtx): Make static.
21568 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
21569 (xlogue_layout::compute_stub_managed_regs): Rename to...
21570 (xlogue_layout::count_stub_managed_regs): ...this.
21571 (xlogue_layout::is_stub_managed_reg): New function.
21572 (xlogue_layout::m_stub_names): Rename to...
21573 (xlogue_layout::s_stub_names): ...this, make static.
21574 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
21575 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
21576 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
21577 xlogue_layout::s_stub_names): Instantiate statics.
21578 (stub_managed_regs): Remove.
21579 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
21580 (disable_call_ms2sysv_xlogues): Rename to...
21581 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
21582 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
21584 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
21585 change after reload_completed.
21586 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
21588 (ix86_expand_prologue): Likewise.
21589 (ix86_expand_epilogue): Likewise.
21590 (ix86_expand_split_stack_prologue): Likewise.
21591 (ix86_compute_frame_layout): Remove frame parameter ...
21592 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
21593 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
21595 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
21596 (ix86_frame): Move from here ...
21597 * config/i386/i386.h (ix86_frame): ... to here.
21598 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
21599 complete ix86_frame data structure instead. Remove some_ld_name.
21601 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21603 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
21604 symbols that hold a DECL_VALUE_EXPR.
21606 2017-06-01 Martin Jambor <mjambor@suse.cz>
21608 PR tree-optimization/80898
21609 * tree-sra.c (process_subtree_disqualification): Removed.
21610 (disqualify_candidate): Do not acll
21611 process_subtree_disqualification.
21612 (subtree_mark_written_and_enqueue): New function.
21613 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
21614 RHS has been disqualified and re-queue LHS if necessary. Apart
21615 from that, ignore disqualified RHS.
21617 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21619 * config/s390/s390.c (s390_emit_epilogue): Disable early return
21620 address fetch for z10 or later.
21622 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21624 * config/arc/arc.md (tst_movb): Add guard when splitting.
21626 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21628 * config/arc/arc.c (arc_can_eliminate): Test against
21629 arc_frame_pointer_needed.
21631 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21633 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
21634 to prevent store reordering.
21635 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
21636 (type): Add block type.
21637 (stack_tie): Define special instruction to be used in
21640 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21642 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
21643 constraint. It is not valid for the pattern.
21644 (noncommutative_binary_comparison): Likewise.
21646 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21648 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
21651 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21653 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
21654 be used by the reg-alloc.
21656 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21658 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
21659 reg-alloc when having mul64 or mul32x16 instructions.
21660 (mulsidi3): Likewise.
21661 (umulsidi3): Likewise.
21662 (mulsi32x16): New pattern.
21663 (mulsi64): Likewise.
21664 (mulsidi64): Likewise.
21665 (umulsidi64): Likewise.
21666 (MUL32x16_REG): Define.
21667 (mul64_600): Use MUL32x16_REG.
21668 (mac64_600): Likewise.
21669 (umul64_600): Likewise.
21670 (umac64_600): Likewise.
21672 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
21674 * config/arc/arc.md (mulsi3_700): Make it commutative.
21676 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
21678 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
21680 (*sign_extendsidi2_insn): Likewise for movstosw.
21682 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
21684 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
21685 the type of the input discriminant value. Convert the
21686 discriminant value of signedness vary.
21688 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
21690 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
21691 Add to -Wall section.
21693 2017-06-01 Richard Biener <rguenther@suse.de>
21695 PR middle-end/66313
21696 * fold-const.c (fold_plusminus_mult_expr): If the factored
21697 factor may be zero use a wrapping type for the inner operation.
21698 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
21699 and handle moved defs.
21700 (process_assignment): Properly guard the unary op case. Return a
21701 tri-state indicating that moving the stmt before the call may allow
21702 to continue. Pass through to_move.
21703 (find_tail_calls): Handle moving unrelated defs before
21706 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
21709 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
21710 splitter result in the canonical way.
21712 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
21714 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
21715 also for 32bit target. Update insn attributes.
21716 (zero-extendsidi2 splitter): Allow all registers for operand 1.
21718 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
21720 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
21721 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
21722 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
21723 (_mm_maskz_min_ss): New intrinsics.
21725 2017-05-31 Martin Liska <mliska@suse.cz>
21727 * tree-vect-loop.c (vect_create_epilog_for_reduction):
21728 Change comment style to one we normally use.
21729 (vectorizable_reduction): Likewise.
21730 (vectorizable_induction): Likewise.
21731 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
21732 (vectorizable_call): Likewise.
21733 (vectorizable_simd_clone_call): Likewise.
21734 (vectorizable_conversion): Likewise.
21735 (vectorizable_assignment): Likewise.
21736 (vectorizable_shift): Likewise.
21737 (vectorizable_operation): Likewise.
21738 (vectorizable_store): Likewise.
21739 (vectorizable_load): Likewise.
21740 * tree-vectorizer.h: Likewise.
21742 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
21744 * passes.c (emergency_dump_function): New.
21745 * tree-pass.h (emergency_dump_function): Declare.
21746 * plugin.c (plugins_internal_error_function): Remove.
21747 * plugin.h (plugins_internal_error_function): Remove declaration.
21748 * toplev.c (internal_error_function): New static function. Use it...
21749 (general_init): ...here.
21751 2017-05-31 Graham Markall <graham.markall@embecosm.com>
21753 * config/arc/arc.c (arc_print_operand): Handle constant operands.
21754 (arc_rtx_costs): Add costs for new patterns.
21755 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
21756 * config/arc/predicates.md: Add _1_2_3_operand predicate.
21758 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21760 * tree-ssa-strlen.c (get_next_strinfo): New function.
21761 (get_stridx_plus_constant): Use it.
21762 (zero_length_string): Likewise.
21763 (adjust_related_strinfos): Likewise.
21764 (adjust_last_stmt): Likewise.
21766 2017-05-31 Richard Biener <rguenther@suse.de>
21769 * config/i386/i386.c (ix86_expand_builtin): Remove assert
21770 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
21772 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
21774 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
21775 loop_vinfo argument and use of dependence distance vectors.
21776 Check instead whether the two references differ only in their
21777 initial value and assume that they have the same alignment if the
21778 difference is a multiple of the vector alignment.
21779 (vect_analyze_data_refs_alignment): Update call accordingly.
21781 2017-05-31 Martin Liska <mliska@suse.cz>
21784 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
21786 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21788 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
21789 (create_intersect_range_checks): Move from ...
21790 * tree-data-ref.c (create_intersect_range_checks_index)
21791 (create_intersect_range_checks): ... to here.
21792 (create_runtime_alias_checks): New function factored from ...
21793 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
21794 here. Call above function.
21795 * tree-data-ref.h (create_runtime_alias_checks): New function.
21797 2017-05-31 Bin Cheng <bin.cheng@arm.com>
21799 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
21800 segment length for dr_b and compute it in wide_int.
21802 2017-05-31 Richard Biener <rguenther@suse.de>
21804 PR tree-optimization/80906
21805 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
21806 and pass through iv_map.
21807 (copy_bb_and_scalar_dependences): Adjust.
21808 (translate_pending_phi_nodes): Likewise.
21809 (copy_loop_close_phi_args): Handle code-generating IVs instead
21812 2017-05-30 David Malcolm <dmalcolm@redhat.com>
21814 * diagnostic-color.c (color_dict): Add "type-diff".
21815 (parse_gcc_colors): Update comment.
21816 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
21817 -fdiagnostics-show-template-tree and -fno-elide-type.
21818 (GCC_COLORS): Add type-diff to example.
21820 (-fdiagnostics-show-template-tree): New.
21821 (-fno-elide-type): New.
21822 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
21823 the pp_format_decoder callback. Call any m_format_postprocessor's
21825 (pretty_printer::pretty_printer): Initialize
21826 m_format_postprocessor.
21827 (pretty_printer::~pretty_printer): Delete any
21828 m_format_postprocessor.
21829 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
21830 (class format_postprocessor): New class.
21831 (struct pretty_printer::format_decoder): Document the new parameters.
21832 (struct pretty_printer::m_format_postprocessor): New field.
21833 * tree-diagnostic.c (default_tree_printer): Update for new
21834 bool and const char ** params.
21835 * tree-diagnostic.h (default_tree_printer): Likewise.
21837 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
21839 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
21840 (lwa_operand): Delete rs6000_gen_cell_microcode test.
21841 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
21842 rs6000_gen_cell_microcode code.
21843 (rs6000_final_prescan_insn): Delete.
21844 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
21845 "warn-cell-microcode" entries.
21846 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
21847 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
21848 throughout. Change cc_reg_not_micro_cr0_operand to
21849 cc_reg_not_cr0_operand throughout.
21850 (*extendhi<mode>2_noload): Delete.
21851 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
21852 (mwarn-cell-microcode): Delete.
21853 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
21854 -mgen-cell-microcode and -mwarn-cell-microcode.
21856 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
21859 * config/i386/constraints.md (Yd): New constraint.
21861 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
21862 and (?Yd, r) alternatives. Update insn attributes.
21863 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
21864 and (?*Yd, r) alternatives. Update insn attributes.
21865 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
21867 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
21869 * gimplify.c (gimplify_modify_expr): Don't create a
21870 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
21873 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
21875 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
21877 2017-05-30 Richard Biener <rguenther@suse.de>
21879 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
21880 and reduc_def fields.
21881 (STMT_VINFO_REDUC_TYPE): New define.
21882 (STMT_VINFO_REDUC_DEF): Likewise.
21883 (vect_force_simple_reduction): Adjust prototype.
21884 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
21885 (vect_is_simple_reduction): Remove check_reduction argument.
21886 (vect_force_simple_reduction): Adjust and set
21887 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21888 (vectorizable_reduction): Do not re-do reduction analysis
21889 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
21890 * tree-parloops.c (gather_scalar_reductions): Adjust.
21892 2017-05-30 Richard Biener <rguenther@suse.de>
21894 PR middle-end/80901
21895 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
21898 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
21900 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
21901 Introduce unknown_misalignment parameter and remove vf.
21902 (vect_peeling_hash_get_lowest_cost):
21903 Pass unknown_misalignment parameter.
21904 (vect_enhance_data_refs_alignment):
21905 Fix unsupportable data ref treatment.
21907 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21909 * tree-vect-data-refs.c (vect_get_data_access_cost):
21910 Workaround for SLP handling.
21911 (vect_enhance_data_refs_alignment):
21912 Compute costs for doing no peeling at all, compare to the best
21913 peeling costs so far and avoid peeling if cheaper.
21915 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21917 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
21918 Return peeling info and set costs to zero for unlimited cost
21920 (vect_enhance_data_refs_alignment): Also inspect all datarefs
21921 with unknown misalignment. Compute and costs for unknown
21922 misalignment, compare them to the costs for known misalignment
21923 and choose the cheapest for peeling.
21925 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21927 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
21928 (vect_get_peeling_costs_all_drs): Create function.
21929 (vect_peeling_hash_get_lowest_cost):
21930 Use vect_get_peeling_costs_all_drs.
21931 (vect_peeling_supportable): Create function.
21932 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
21934 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
21936 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
21937 DR_HAS_NEGATIVE_STEP.
21938 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
21939 (vect_enhance_data_refs_alignment): Use.
21940 (vect_duplicate_ssa_name_ptr_info): Use.
21941 * tree-vectorizer.h (dr_misalignment): Use.
21942 (known_alignment_for_access_p): Use.
21944 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
21947 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
21949 (has_section_name): New function.
21951 2017-05-30 Martin Liska <mliska@suse.cz>
21954 * auto-profile.c (get_function_decl_from_block): Fix
21957 2017-05-30 Richard Biener <rguenther@suse.de>
21959 PR middle-end/80876
21960 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
21962 2017-05-30 Martin Liska <mliska@suse.cz>
21964 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
21965 * dumpfile.h (struct dump_file_info): Remove ctors.
21967 2017-05-30 Martin Liska <mliska@suse.cz>
21969 * predict.def: Fix GNU coding style.
21971 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21973 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
21974 Mark 'to' argument with ATTRIBUTE_UNUSED.
21976 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
21978 * config/xtensa/xtensa.c (xtensa_emit_call): Use
21979 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
21980 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
21983 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
21985 * doc/install.texi (Options specification): Restore entry of
21986 --enable-sjlj-exceptions.
21988 2017-05-27 Michael Eager <eager@eagercon.com>
21991 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
21993 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
21995 * config/microblaze/microblaze.h
21996 (FIXED_REGISTERS): Update in macro.
21997 (CALL_USED_REGISTERS): Update in macro.
21999 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
22001 * doc/install.texi: Add links to macOS binary distributions.
22003 2017-05-27 Jakub Jelinek <jakub@redhat.com>
22007 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22009 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22011 2017-05-26 Martin Liska <mliska@suse.cz>
22013 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
22015 2017-05-26 Martin Liska <mliska@suse.cz>
22017 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
22018 always leading ';; '.
22019 (dump_bb_info): Likewise.
22020 (brief_dump_cfg): Likewise.
22021 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
22022 * dumpfile.c: Remove usage of TDF_VERBOSE.
22023 * dumpfile.h (enum dump_kind): Likewise.
22024 (dump_gimple_bb_header): Do not use TDF_COMMENT.
22025 * print-tree.c (debug_verbose): Remove.
22026 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
22027 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
22028 * tree-diagnostic.c (default_tree_printer): Replace
22029 TDF_DIAGNOSTIC with TDF_SLIM.
22031 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22033 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
22034 in parameter loop, rather than loop_vinfo.
22035 (create_intersect_range_checks): Ditto.
22036 (vect_create_cond_for_alias_checks): Update call to above functions.
22038 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22040 PR tree-optimization/80815
22041 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
22042 for merging runtime alias checks. Handle negative DR_STEPs.
22044 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22046 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
22048 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
22049 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
22050 out code pruning runtime alias checks.
22051 * tree-data-ref.c (prune_runtime_alias_test_list): New function
22052 factored out from above.
22053 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22055 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
22057 (prune_runtime_alias_test_list): New decalaration.
22059 2017-05-26 Bin Cheng <bin.cheng@arm.com>
22061 * tree-vect-data-refs.c (compare_tree): Rename and move ...
22062 * tree-data-ref.c (data_ref_compare_tree): ... to here.
22063 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
22064 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
22065 (operator==, comp_dr_with_seg_len_pair): Ditto.
22066 (vect_prune_runtime_alias_test_list): Ditto.
22068 2017-05-26 Martin Liska <mliska@suse.cz>
22071 * params.def: Bound partial-inlining-entry-probability param.
22073 2017-05-26 Marek Polacek <polacek@redhat.com>
22076 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
22079 2017-05-26 Richard Biener <rguenther@suse.de>
22081 PR tree-optimization/80842
22082 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
22085 2017-05-26 Richard Biener <rguenther@suse.de>
22087 PR tree-optimization/80844
22088 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
22090 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
22092 * doc/md.texi (Machine Constraints): Update x86 family
22093 machine constraints section to match 'config/i386/constraints.md'.
22095 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
22097 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
22099 2017-05-25 Nathan Sidwell <nathan@acm.org>
22101 * doc/invoke.texi (--enable-languages): Update documentation.
22103 2017-05-25 Martin Liska <mliska@suse.cz>
22105 * dumpfile.c: Add TDF_FOLDING.
22106 * dumpfile.h (enum dump_kind): Likewise.
22107 * genmatch.c (dt_simplify::gen_1): Use it.
22109 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22111 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
22113 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22115 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
22116 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
22118 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
22120 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
22121 * match.pd (X == C): Rewrite it here.
22122 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
22123 with_certain_nonzero_bits2): New predicates.
22124 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
22126 2017-05-24 Nathan Sidwell <nathan@acm.org>
22128 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
22131 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
22134 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22136 * config/powerpcspe: New port. Files are copied from the rs6000
22137 port, with "rs6000" in filenames replaced by "powerpcspe".
22139 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
22141 PR rtl-optimization/80754
22142 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
22144 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
22146 * config/sparc/sparc.md (length): Return the correct value for -mflat
22147 sibcalls to match output_sibcall.
22149 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
22153 * config/rs6000/rs6000.c (struct machine_function): Add new field
22155 (rs6000_get_separate_components): Init that field, use it.
22156 (rs6000_components_for_bb): Use the field.
22158 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
22160 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
22162 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
22164 PR middle-end/80823
22165 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
22167 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
22170 * config/s390/s390.c (s390_check_qrst_address): Check incoming
22171 address against address_operand predicate.
22172 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
22174 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
22176 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
22177 parameters passed indirectly.
22179 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
22181 * config/i386/i386.md (*movdi_internal): Remove SSE4
22182 alternative 18 (?r, *v). Update insn attributes.
22183 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
22184 Update insn attributes.
22185 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
22186 Update insn attributes.
22187 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
22188 alternative 1 (r, v). Remove isa attribute.
22189 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
22190 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
22191 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
22193 2017-05-23 Tom de Vries <tom@codesourcery.com>
22195 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
22198 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22200 * cgraphunit.c (symbol_table::process_new_functions): Update.
22201 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
22202 (inline_generate_summary): Rename to ...
22203 (ipa_fn_summary_generate): ... this one.
22204 (inline_read_summary): Rename to ...
22205 (ipa_fn_summary_read): ... this one.
22206 (inline_write_summary): Rename to ...
22207 (ipa_fn_summary_write): ... this one.
22208 (inline_free_summary): Rename to ...
22209 (ipa_free_fn_summary): ... this one.
22210 (pass_data_local_fn_summary, pass_local_fn_summary,
22211 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
22212 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
22213 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
22214 make_pass_ipa_fn_summary): New.
22215 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
22216 inline_write_summary, inline_free_summary): Remove.
22217 (ipa_free_fn_summary) : New.
22218 * ipa-inline.c (ipa_inline): Update.
22219 (pass_ipa_inline): Do not generate summaries.
22220 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
22222 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
22223 and add pass_ipa_fn_summary.
22224 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
22226 (make_pass_inline_parameters): Remove.
22228 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
22230 * omp-low.c (struct omp_context): Remove "default_kind" member.
22233 * omp-offload.c (execute_oacc_device_lower): Remove the
22234 parallelism dimensions function attributes for unparallelized
22235 OpenACC kernels constructs.
22237 2017-05-23 Martin Liska <mliska@suse.cz>
22239 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
22241 (cgraph_edge::make_speculative): Likewise.
22242 (cgraph_edge::resolve_speculation): Likewise.
22243 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
22244 (cgraph_node::dump): Likewise.
22245 * cgraph.h: Likewise.
22246 * cgraphunit.c (analyze_functions): Likewise.
22247 (symbol_table::compile): Likewise.
22248 * ipa-cp.c (print_all_lattices): Likewise.
22249 (determine_versionability): Likewise.
22250 (initialize_node_lattices): Likewise.
22251 (ipcp_verify_propagated_values): Likewise.
22252 (estimate_local_effects): Likewise.
22253 (update_profiling_info): Likewise.
22254 (create_specialized_node): Likewise.
22255 (perhaps_add_new_callers): Likewise.
22256 (decide_about_value): Likewise.
22257 (decide_whether_version_node): Likewise.
22258 (identify_dead_nodes): Likewise.
22259 (ipcp_store_bits_results): Likewise.
22260 * ipa-devirt.c (dump_targets): Likewise.
22261 (ipa_devirt): Likewise.
22262 * ipa-icf.c (sem_item::dump): Likewise.
22263 (sem_function::equals): Likewise.
22264 (sem_variable::equals): Likewise.
22265 (sem_item_optimizer::read_section): Likewise.
22266 (sem_item_optimizer::execute): Likewise.
22267 (congruence_class::dump): Likewise.
22268 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
22269 (dump_inline_summary): Likewise.
22270 (estimate_node_size_and_time): Likewise.
22271 (inline_analyze_function): Likewise.
22272 * ipa-inline-transform.c (inline_call): Likewise.
22273 * ipa-inline.c (report_inline_failed_reason): Likewise.
22274 (want_early_inline_function_p): Likewise.
22275 (edge_badness): Likewise.
22276 (update_edge_key): Likewise.
22277 (inline_small_functions): Likewise.
22278 * ipa-profile.c (ipa_profile): Likewise.
22279 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
22280 (ipa_make_edge_direct_to_target): Likewise.
22281 (remove_described_reference): Likewise.
22282 (ipa_impossible_devirt_target): Likewise.
22283 (propagate_controlled_uses): Likewise.
22284 (ipa_print_node_params): Likewise.
22285 (ipcp_transform_function): Likewise.
22286 * ipa-pure-const.c (pure_const_read_summary): Likewise.
22287 (propagate_pure_const): Likewise.
22288 * ipa-reference.c (generate_summary): Likewise.
22289 (read_write_all_from_decl): Likewise.
22290 (propagate): Likewise.
22291 (ipa_reference_read_optimization_summary): Likewise.
22292 * ipa-utils.c (ipa_merge_profiles): Likewise.
22293 * ipa.c (walk_polymorphic_call_targets): Likewise.
22294 (symbol_table::remove_unreachable_nodes): Likewise.
22295 (ipa_single_use): Likewise.
22296 * passes.c (execute_todo): Likewise.
22297 * predict.c (drop_profile): Likewise.
22298 * symtab.c (symtab_node::get_dump_name): New function.
22299 (symtab_node::dump_name): Likewise.
22300 (symtab_node::dump_asm_name): Likewise.
22301 (symtab_node::dump_references): Likewise.
22302 (symtab_node::dump_referring): Likewise.
22303 (symtab_node::dump_base): Likewise.
22304 (symtab_node::debug_symtab): Likewise.
22305 * tree-sra.c (convert_callers_for_node): Likewise.
22306 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
22307 * value-prof.c (init_node_map): Likewise.
22309 2017-05-23 Martin Liska <mliska@suse.cz>
22311 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
22312 and symtab_node::debug_symtab to symbol_table::debug.
22313 * cgraphunit.c (analyze_functions): Use the renamed function.
22314 (symbol_table::compile): Likewise.
22315 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
22316 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
22317 * passes.c (execute_todo): Likewise.
22318 * symtab.c (symbol_table::dump): New function.
22319 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
22321 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22323 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
22324 that nonconst implies exec.
22326 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
22328 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22329 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22330 (inline_edge_summary_vec): Turn into ...
22331 (ipa_call_summaries): ... this one.
22332 (redirect_to_unreachable, edge_set_predicate,
22333 evaluate_properties_for_edge, inline_summary_alloc,
22334 reset_ipa_call_summary, reset_inline_summary,
22335 inline_summary_t::duplicate): Update.
22336 (inline_edge_duplication_hook): Turn to ...
22337 (ipa_call_summary_t::duplicate): ... this one.
22338 (inline_edge_removal_hook): Turn to ...
22339 (ipa_call_summary_t::remove): ... this one.
22340 (dump_inline_edge_summary): Turn to ...
22341 (dump_ipa_call_summary): ... this one.
22342 (estimate_function_body_sizes): Update.
22343 (inline_update_callee_summaries): Update.
22344 (remap_edge_change_prob): Update.
22345 (remap_edge_summaries): Update.
22346 (inline_merge_summary): Update.
22347 (do_estimate_edge_time): Update.
22348 (inline_generate_summary): Update.
22349 (inline_read_section): Update.
22350 (inline_read_summary): Update.
22351 (inline_free_summary): Update.
22352 * ipa-inline.c (can_inline_edge_p): Update.
22353 (compute_inlined_call_time): Update.
22354 (want_inline_small_function_p): Update.
22355 (edge_badness): Update.
22356 (early_inliner): Update.
22357 * ipa-inline.h (inline_edge_summary): Turn to ...
22358 (ipa_call_summary): ... this one.
22359 (ipa_call_summary_t): New class.
22360 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22361 (ipa_call_summaries): New.
22362 (inline_edge_summary): Remove.
22363 (estimate_edge_growth): Update.
22364 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22365 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22366 * ipa-split.c (execute_split_functions): Update.
22367 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22369 2017-05-23 Tom de Vries <tom@codesourcery.com>
22371 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22372 attributes): Document rdrand effective target.
22374 2017-05-23 Tom de Vries <tom@codesourcery.com>
22376 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
22377 attributes): Sort alphabetically.
22379 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
22381 * config/avr/genmultilib.awk: Use gsub instead of gensub.
22383 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
22386 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
22387 V2DF/V2DI splat into two separate patterns, one that handles
22388 registers, and the other that only handles memory. Drop support
22389 for splatting from a GPR on ISA 2.07 and then splitting the
22390 splat into direct move and splat.
22391 (vsx_splat_<mode>_reg): Likewise.
22392 (vsx_splat_<mode>_mem): Likewise.
22394 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
22396 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
22398 2017-05-22 Jakub Jelinek <jakub@redhat.com>
22400 PR middle-end/80809
22401 * omp-low.c (finish_taskreg_remap): New function.
22402 (finish_taskreg_scan): If unit size of ctx->record_type
22403 is non-constant, unshare the size expression and replace
22404 decls in it with possible outer var refs.
22406 PR middle-end/80809
22407 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
22408 GOVD_SHARED rather than GOVD_PRIVATE with it.
22409 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
22410 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
22412 PR middle-end/80853
22413 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
22414 as last argument to build_outer_var_ref for pointer bases of array
22415 section reductions.
22417 2017-05-19 Martin Sebor <msebor@redhat.com>
22419 * print-tree.c (print_node): Print DECL_READ_P flag.
22421 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22423 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
22424 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
22425 * cgraph.c: Likewise.
22426 * cgraphunit.c: Likewise.
22427 * gengtype.c: Likewise.
22428 * ipa-cp.c: Likewise.
22429 * ipa-devirt.c: Likewise.
22430 * ipa-icf.c: Likewise.
22431 * ipa-predicate.c: Likewise.
22432 * ipa-profile.c: Likewise.
22433 * ipa-prop.c: Likewise.
22434 * ipa-split.c: Likewise.
22436 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
22437 edge_predicate_pool, dump_inline_hints,
22438 inline_summary::account_size_time, redirect_to_unreachable,
22439 edge_set_predicate, set_hint_predicate,
22440 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
22441 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
22442 inline_summary_t::remove, remap_hint_predicate_after_duplication,
22443 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
22444 ipa_call_summary_t::remove, initialize_growth_caches,
22445 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
22446 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
22447 mark_modified, unmodified_parm_1, unmodified_parm,
22448 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
22449 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
22450 compute_bb_predicates, will_be_nonconstant_expr_predicate,
22451 will_be_nonconstant_predicate, record_modified_bb_info,
22452 get_minimal_bb, record_modified, param_change_prob,
22453 phi_result_unknown_predicate, predicate_for_phi_result,
22454 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
22455 estimate_function_body_sizes, compute_inline_parameters,
22456 compute_inline_parameters_for_curren, pass_data_inline_parameters,
22457 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
22458 inline_update_callee_summaries, remap_edge_change_prob,
22459 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
22460 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
22461 inline_analyze_function, inline_summary_t::insert,
22462 inline_generate_summary, read_ipa_call_summary, inline_read_section,
22463 inline_read_summary, write_ipa_call_summary, inline_write_summary,
22464 inline_free_summary): Move to ipa-fnsummary.h
22465 (predicate_t): Remove.
22466 * ipa-fnsummary.c: New file.
22467 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
22468 (enum inline_hints_vals, inline_hints, agg_position_info,
22469 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
22470 inline_summaries, ipa_call_summary, ipa_call_summary_t,
22471 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
22472 dump_inline_summary, dump_inline_hints, inline_generate_summary,
22473 inline_read_summary, inline_write_summary, inline_free_summary,
22474 inline_analyze_function, initialize_inline_failed,
22475 inline_merge_summary, inline_update_overall_summary,
22476 compute_inline_parameters): Move to ipa-fnsummary.h
22477 * ipa-fnsummary.h: New file.
22478 * ipa-inline-transform.h: Include ipa-inline.h.
22479 * ipa-inline.c: LIkewise.
22481 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22483 * ipa-inline.c (edge_badness): Use inlined_time instead of
22484 inline_summaries->get.
22486 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22488 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
22490 2017-05-22 Nathan Sidwell <nathan@acm.org>
22492 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
22493 (fdump-lang): Document 'raw' option.
22494 * dumpfile.h (TDI_tu): Delete.
22495 * dumpfile.c (dump_files): Remove translation-unit.
22496 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
22498 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
22500 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
22501 command option from $(AWK) call.
22502 * config/avr/genmultilib.awk: Simplify and rewrite so that it
22503 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
22504 [FORMAT]: Remove handling of variable.
22505 * config/avr/t-multilib: Regenerate.
22507 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22509 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
22511 (dump_inline_summary): Do not print self_time.
22512 (estimate_function_body_sizes): Do not set self_time.
22513 (compute_inline_parameters): Likewise.
22514 (inline_read_section, inline_write_summary): Do not stream self_time.
22515 * ipa-inline.h (inline_summary): Drop self_time.
22517 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
22519 * ipa-inline-analysis.c (account_size_time): Rename to ...
22520 (inline_summary::account_size_time): ... this one.
22521 (reset_ipa_call_summary): Turn to ...
22522 (ipa_call_summary::reset): ... this one.
22523 (reset_inline_summary): Turn to ...
22524 (inline_summary::reset): ... this one.
22525 (inline_summary_t::remove): Update.
22526 (inline_summary_t::duplicate): Update.
22527 (ipa_call_summary_t::remove): Update.
22528 (dump_inline_summary): Update.
22529 (estimate_function_body_sizes): Update.
22530 (compute_inline_parameters): Update.
22531 (estimate_node_size_and_time): Update.
22532 (inline_merge_summary): Update.
22533 (inline_update_overall_summary): Update.
22534 (inline_read_section): Update.
22535 (inline_write_summary): Update.
22536 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
22537 add account_size_time and reset member functions.
22538 (ipa_call_summary): Add reset function.
22539 * ipa-predicate.h (predicate::operator &): Constify.
22541 2017-05-22 Richard Biener <rguenther@suse.de>
22543 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
22545 2017-05-19 Jason Merrill <jason@redhat.com>
22547 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
22549 2017-05-19 Marek Polacek <polacek@redhat.com>
22552 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
22553 TYPE_OVERFLOW_WRAPS checks.
22555 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
22557 * tree-core.h (enum omp_clause_default_kind): Add
22558 "OMP_CLAUSE_DEFAULT_PRESENT".
22559 * tree-pretty-print.c (dump_omp_clause): Handle it.
22560 * gimplify.c (enum gimplify_omp_var_data): Add
22561 "GOVD_MAP_FORCE_PRESENT".
22562 (gimplify_adjust_omp_clauses_1): Map it to
22563 "GOMP_MAP_FORCE_PRESENT".
22564 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
22566 * gimplify.c (oacc_default_clause): Clarify.
22568 2017-05-19 Nathan Sidwell <nathan@acm.org>
22570 LANG_HOOK_REGISTER_DUMPS
22571 * toplev.c (general_init): Call register dump lang hook.
22572 * doc/invoke.texi: Document -fdump-lang option family.
22573 * dumpfile.c (dump_files): Remove class dump here.
22574 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
22575 * dumpfile.h (tree_dump_index): Remove TDI_class.
22576 * langhooks-def.h (lhd_register_dumps): Declare.
22577 (LANG_HOOKS_REGISTER_DUMPS): Define.
22578 (LANG_HOOKS_INITIALIZER): Add it.
22579 * langhooks.c (lhd_register_dumps): Define.
22580 * langhooks.h (struct lang_hooks): Add register_dumps.
22582 2017-05-19 Nathan Sidwell <nathan@acm.org>
22584 * context.h (context::set_passes): New.
22585 * context.c (context::context): Do not create pass manager.
22586 * toplev.c (general_init): Create pass manager here.
22588 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
22590 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
22591 use this splitter if two add or or instructions would also work for
22592 the constant we want to generate.
22594 2017-05-19 Richard Biener <rguenther@suse.de>
22597 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
22598 predicate evaluation.
22600 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22602 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
22604 * ipa-inline.c (want_inline_small_function_p): Do not cast to
22607 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
22609 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
22610 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
22611 (inline_edge_summary_vec): Turn into ...
22612 (ipa_call_summaries): ... this one.
22613 (redirect_to_unreachable, edge_set_predicate,
22614 evaluate_properties_for_edge, inline_summary_alloc,
22615 reset_ipa_call_summary, reset_inline_summary,
22616 inline_summary_t::duplicate): Update.
22617 (inline_edge_duplication_hook): Turn to ...
22618 (ipa_call_summary_t::duplicate): ... this one.
22619 (inline_edge_removal_hook): Turn to ...
22620 (ipa_call_summary_t::remove): ... this one.
22621 (dump_inline_edge_summary): Turn to ...
22622 (dump_ipa_call_summary): ... this one.
22623 (estimate_function_body_sizes): Update.
22624 (inline_update_callee_summaries): Update.
22625 (remap_edge_change_prob): Update.
22626 (remap_edge_summaries): Update.
22627 (inline_merge_summary): Update.
22628 (do_estimate_edge_time): Update.
22629 (inline_generate_summary): Update.
22630 (inline_read_section): Update.
22631 (inline_read_summary): Update.
22632 (inline_free_summary): Update.
22633 * ipa-inline.c (can_inline_edge_p): Update.
22634 (compute_inlined_call_time): Update.
22635 (want_inline_small_function_p): Update.
22636 (edge_badness): Update.
22637 (early_inliner): Update.
22638 * ipa-inline.h (inline_edge_summary): Turn to ...
22639 (ipa_call_summary): ... this one.
22640 (ipa_call_summary_t): New class.
22641 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
22642 (ipa_call_summaries): New.
22643 (inline_edge_summary): Remove.
22644 (estimate_edge_growth): Update.
22645 * ipa-profile.c (ipa_propagate_frequency_1): Update.
22646 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
22647 * ipa-split.c (execute_split_functions): Update.
22648 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
22650 2017-05-19 Richard Biener <rguenther@suse.de>
22652 PR middle-end/80764
22653 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
22655 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
22657 * config/rs6000/rs6000.c (struct machine_function): Add field
22658 fpr_is_wrapped_separately.
22659 (rs6000_get_separate_components): Use 64 components. Handle the
22660 new FPR components.
22661 (rs6000_components_for_bb): Handle the FPR components.
22662 (rs6000_emit_prologue_components): Handle the FPR components.
22663 (rs6000_emit_epilogue_components): Handle the FPR components.
22664 (rs6000_set_handled_components): Handle the FPR components.
22665 (rs6000_emit_prologue): Don't output prologue code for those FPRs
22666 that are already separately shrink-wrapped.
22667 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
22668 that are already separately shrink-wrapped.
22670 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
22673 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
22676 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
22677 (define_peephole2 for Altivec d-form load): Add peepholes to catch
22678 cases where the register allocator uses a move and an offsettable
22679 memory operation to/from a FPR register on ISA 2.06/2.07.
22680 (define_peephole2 for Altivec d-form store): Likewise.
22682 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
22685 * config/i386/mmx.md (*mov<mode>_internal): Enable
22686 alternatives 11, 12, 13 and 14 also for 32bit targets.
22687 Remove alternatives 15, 16, 17 and 18.
22688 * config/i386/sse.md (vec_concatv2di): Change
22689 alternative (!x, *y) to (x, ?!*Yn).
22691 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
22693 * dumpfile.h (enum dump_kind): Remove stray comma.
22695 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22697 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
22698 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
22699 predicate::num_conditions
22700 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
22701 (CHANGED): turn into predicate::changed.
22702 (agg_position_info): Move to ipa-predicate.h
22703 (add_condition, predicate::add_clause, predicate::operator &=,
22704 predicate::or_with, predicate::evaluate, predicate::probability,
22705 dump_condition, dump_clause, predicate::dump,
22706 predicate::remap_after_duplication, predicate::remap_after_inlining,
22707 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
22708 (evaluate_conditions_for_known_args): Update.
22709 (set_cond_stmt_execution_predicate): Update.
22710 * ipa-inline.h: Include ipa-predicate.h
22711 (condition, inline_param_summary, conditions, agg_position_info,
22712 predicate): Move to ipa-predicate.h
22713 * ipa-predicate.c: New file.
22714 * ipa-predicate.h: New file.
22716 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
22718 * final.c (leaf_function_p): Check we are not in a sequence.
22720 2017-05-18 Martin Liska <mliska@suse.cz>
22722 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
22723 * dumpfile.c (dump_register): Use new enum dump_kind.
22724 (get_dump_file_name): Likewise.
22725 (dump_enable_all): Likewise.
22726 (dump_switch_p_1): Likewise.
22727 (enable_rtl_dump_file): Remove usage of TDF_RTL.
22728 * dumpfile.h (enum dump_kind): New enum type.
22729 (struct dump_file_info): Create constructor and
22730 format fields and comments.
22731 * passes.c (pass_manager::register_one_dump_file):
22733 * statistics.c (statistics_early_init): Likewise.
22734 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
22735 TDF_TREE with TDF_SLIM.
22736 (gather_memory_references_ref): Likewise.
22738 2017-05-18 Martin Liska <mliska@suse.cz>
22740 * vec.h (struct vnull): Use it.
22742 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
22744 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
22745 (true_predicate, false_predicate, true_predicate_p,
22746 false_predicate_p): Remove.
22747 (single_cond_predicate, not_inlined_predicate): Turn to member function
22749 (add_condition): Update.
22750 (add_clause): Turn to...
22751 (predicate::add_clause): ... this one; update; allow passing NULL
22753 (and_predicates): Turn to ...
22754 (predicate::operator &=): ... this one.
22755 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
22756 (or_predicates): Turn to ...
22757 (predicate::or_with): ... this one.
22758 (evaluate_predicate): Turn to ...
22759 (predicate::evaluate): ... this one.
22760 (predicate_probability): Turn to ...
22761 (predicate::probability): ... this one.
22762 (dump_condition): Update.
22763 (dump_predicate): Turn to ...
22764 (predicate::dump): ... this one.
22765 (account_size_time): Update.
22766 (edge_set_predicate): Update.
22767 (set_hint_predicate): UPdate.
22768 (evaluate_conditions_for_known_args): Update.
22769 (evaluate_properties_for_edge): Update.
22770 (remap_predicate_after_duplication): Turn to...
22771 (predicate::remap_after_duplication): ... this one.
22772 (remap_hint_predicate_after_duplication): Update.
22773 (inline_summary_t::duplicate): UPdate.
22774 (dump_inline_edge_summary): Update.
22775 (dump_inline_summary): Update.
22776 (set_cond_stmt_execution_predicate): Update.
22777 (set_switch_stmt_execution_predicate): Update.
22778 (compute_bb_predicates): Update.
22779 (will_be_nonconstant_expr_predicate): Update.
22780 (will_be_nonconstant_predicate): Update.
22781 (phi_result_unknown_predicate): Update.
22782 (predicate_for_phi_result): Update.
22783 (array_index_predicate): Update.
22784 (estimate_function_body_sizes): Update.
22785 (estimate_node_size_and_time): Update.
22786 (estimate_ipcp_clone_size_and_time): Update.
22787 (remap_predicate): Rename to ...
22788 (predicate::remap_after_inlining): ... this one.
22789 (remap_hint_predicate): Update.
22790 (inline_merge_summary): Update.
22791 (inline_update_overall_summary): Update.
22792 (estimate_size_after_inlining): Update.
22793 (read_predicate): Rename to ...
22794 (predicate::stream_in): ... this one.
22795 (read_inline_edge_summary): Update.
22796 (write_predicate): Rename to ...
22797 (predicate::stream_out): ... this one.
22798 (write_inline_edge_summary): Update.
22799 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
22800 (clause_t): Turn to uint32_t
22801 (predicate): Turn to class; implement constructor and operators
22803 (size_time_entry): Update.
22804 (inline_summary): Update.
22805 (inline_edge_summary): Update.
22807 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
22809 * fold-const.c (fold_binary_loc): Move transformation...
22810 * match.pd (C - X CMP X): ... here.
22812 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
22814 * config/sparc/sparc.c (sparc_option_override): Set function
22815 alignment for -mcpu=niagara7 to 64 to match the I$ line.
22816 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
22818 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
22820 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
22822 2017-05-18 Marek Polacek <polacek@redhat.com>
22825 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
22826 (pass_ubsan::execute): Call gimple_assign_single_p instead of
22827 gimple_assign_load_p.
22829 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
22831 PR middle-end/80692
22832 * real.c (do_compare): Give decimal_do_compare preference over
22833 comparing just the signs.
22835 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
22837 * doc/md.texi (Canonicalization of Instructions): Describe the
22838 canonical form of instructions that inherently set a condition
22841 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
22843 PR middle-end/80775
22844 * tree-cfg.c: Move deletion of unreachable case statements to after
22845 the merging of consecutive case labels.
22847 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
22849 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
22850 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
22851 restoring of callee-saved registers.
22853 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
22855 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
22856 * config/visium/visium.c (single_set_and_flags): Likewise.
22857 * config/visium/visium.md (Substitutions): Likewise.
22859 2017-05-17 Martin Liska <mliska@suse.cz>
22861 * cfg.c: Introduce dump_flags_t type and
22862 use it instead of int type.
22864 * cfghooks.c: Likewise.
22865 * cfghooks.h (struct cfg_hooks): Likewise.
22866 * cfgrtl.c: Likewise.
22867 * cfgrtl.h: Likewise.
22868 * cgraph.c (cgraph_node::get_body): Likewise.
22869 * coretypes.h: Likewise.
22870 * domwalk.c: Likewise.
22871 * domwalk.h: Likewise.
22872 * dumpfile.c (struct dump_option_value_info): Likewise.
22873 (dump_enable_all): Likewise.
22874 (dump_switch_p_1): Likewise.
22875 (opt_info_switch_p): Likewise.
22876 * dumpfile.h (enum tree_dump_index): Likewise.
22877 (struct dump_file_info): Likewise.
22878 * genemit.c: Likewise.
22879 * generic-match-head.c: Likewise.
22880 * gengtype.c (open_base_files): Likewise.
22881 * gimple-pretty-print.c: Likewise.
22882 * gimple-pretty-print.h: Likewise.
22883 * graph.c (print_graph_cfg): Likewise.
22884 * graphite-scop-detection.c (dot_all_sese): Likewise.
22885 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
22886 * loop-unroll.c (report_unroll): Likewise.
22887 * passes.c (pass_manager::register_one_dump_file): Likewise.
22888 * print-tree.c: Likewise.
22889 * statistics.c: Likewise.
22890 * tree-cfg.c: Likewise.
22891 * tree-cfg.h: Likewise.
22892 * tree-dfa.c: Likewise.
22893 * tree-dfa.h: Likewise.
22894 * tree-dump.c (dump_function): Likewise.
22895 * tree-dump.h (struct dump_info): Likewise.
22896 * tree-pretty-print.c: Likewise.
22897 * tree-pretty-print.h: Likewise.
22898 * tree-ssa-live.c: Likewise.
22899 * tree-ssa-live.h: Likewise.
22900 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
22901 * tree-vect-loop.c: Likewise.
22902 * tree-vect-slp.c: Likewise.
22904 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
22905 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22907 PR tree-optimization/80457
22908 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
22909 of all arguments to a statement as scalar_to_vec operations.
22910 (vectorizable_call): Adjust call to vect_model_simple_cost for
22912 (vectorizable_conversion): Likewise.
22913 (vectorizable_assignment): Likewise.
22914 (vectorizable_shift): Likewise.
22915 (vectorizable_operation): Likewise.
22916 (vectorizable_comparison): Likewise.
22917 (vect_is_simple_cond): Record the def types for operands.
22918 (vectorizable_condition): Likewise, call vect_model_simple_cost.
22919 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
22920 for statement argument count.
22922 2017-05-16 Carl Love <cel@us.ibm.com>
22924 * config/rs6000/rs6000-c: Add support for built-in functions
22925 vector unsigned long long vec_bperm (vector unsigned long long,
22926 vector unsigned char)
22927 vector signed long long vec_mule (vector signed int,
22929 vector unsigned long long vec_mule (vector unsigned int,
22930 vector unsigned int)
22931 vector signed long long vec_mulo (vector signed int,
22933 vector unsigned long long vec_mulo (vector unsigned int,
22934 vector unsigned int)
22935 vector signed char vec_sldw (vector signed char,
22936 vector signed char,
22938 vector unsigned char vec_sldw (vector unsigned char,
22939 vector unsigned char,
22941 vector signed short vec_sldw (vector signed short,
22942 vector signed short,
22944 vector unsigned short vec_sldw (vector unsigned short,
22945 vector unsigned short,
22947 vector signed int vec_sldw (vector signed int,
22950 vector unsigned int vec_sldw (vector unsigned int,
22951 vector unsigned int,
22953 vector signed long long vec_sldw (vector signed long long,
22954 vector signed long long,
22956 vector unsigned long long vec_sldw (vector unsigned long long,
22957 vector unsigned long long,
22959 * config/rs6000/rs6000-c: Add support for built-in functions
22960 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
22961 * config/rs6000/altivec.h: Add defintion for vec_sldw.
22962 * doc/extend.texi: Update the built-in documentation for the
22963 new built-in functions.
22965 2017-05-16 Marek Polacek <polacek@redhat.com>
22969 * tree.c (save_expr): Don't fold the expression.
22971 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
22973 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
22974 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
22975 and (?*y,m). Update insn attributes.
22977 2017-05-16 Martin Liska <mliska@suse.cz>
22979 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
22980 flags argument of print_gimple_stmt, print_gimple_expr,
22981 print_generic_stmt and print_generic_expr.
22982 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
22983 * coretypes.h: Likewise.
22984 * except.c (dump_eh_tree): Likewise.
22985 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
22986 * gimple-pretty-print.h: Likewise.
22987 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
22988 (backprop::push_to_worklist): Likewise.
22989 (backprop::pop_from_worklist): Likewise.
22990 (backprop::process_use): Likewise.
22991 (backprop::intersect_uses): Likewise.
22992 (note_replacement): Likewise.
22993 * gimple-ssa-store-merging.c
22994 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
22995 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
22996 (pass_store_merging::execute): Likewise.
22997 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
22998 (ssa_base_cand_dump_callback): Likewise.
22999 (dump_incr_vec): Likewise.
23000 (replace_refs): Likewise.
23001 (replace_mult_candidate): Likewise.
23002 (create_add_on_incoming_edge): Likewise.
23003 (create_phi_basis): Likewise.
23004 (insert_initializers): Likewise.
23005 (all_phi_incrs_profitable): Likewise.
23006 (introduce_cast_before_cand): Likewise.
23007 (replace_one_candidate): Likewise.
23008 * gimplify.c (gimplify_expr): Likewise.
23009 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
23010 (set_rename): Likewise.
23011 (rename_uses): Likewise.
23012 (copy_loop_phi_nodes): Likewise.
23013 (add_close_phis_to_merge_points): Likewise.
23014 (copy_loop_close_phi_args): Likewise.
23015 (copy_cond_phi_args): Likewise.
23016 (graphite_copy_stmts_from_block): Likewise.
23017 (translate_pending_phi_nodes): Likewise.
23018 * graphite-poly.c (print_pdr): Likewise.
23019 (dump_gbb_cases): Likewise.
23020 (dump_gbb_conditions): Likewise.
23021 (print_scop_params): Likewise.
23022 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
23023 (build_cross_bb_scalars_use): Likewise.
23024 (gather_bbs::before_dom_children): Likewise.
23025 * hsa-dump.c (dump_hsa_immed): Likewise.
23026 * ipa-cp.c (print_ipcp_constant_value): Likewise.
23027 (get_replacement_map): Likewise.
23028 * ipa-inline-analysis.c (dump_condition): Likewise.
23029 (estimate_function_body_sizes): Likewise.
23030 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
23031 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
23032 * ipa-prop.c (ipa_dump_param): Likewise.
23033 (ipa_print_node_jump_functions_for_edge): Likewise.
23034 (ipa_modify_call_arguments): Likewise.
23035 (ipa_modify_expr): Likewise.
23036 (ipa_dump_param_adjustments): Likewise.
23037 (ipa_dump_agg_replacement_values): Likewise.
23038 (ipcp_modif_dom_walker::before_dom_children): Likewise.
23039 * ipa-pure-const.c (check_stmt): Likewise.
23040 (pass_nothrow::execute): Likewise.
23041 * ipa-split.c (execute_split_functions): Likewise.
23042 * omp-offload.c (dump_oacc_loop_part): Likewise.
23043 (dump_oacc_loop): Likewise.
23044 * trans-mem.c (tm_log_emit): Likewise.
23045 (tm_memopt_accumulate_memops): Likewise.
23046 (dump_tm_memopt_set): Likewise.
23047 (dump_tm_memopt_transform): Likewise.
23048 * tree-cfg.c (gimple_verify_flow_info): Likewise.
23049 (print_loop): Likewise.
23050 * tree-chkp-opt.c (chkp_print_addr): Likewise.
23051 (chkp_gather_checks_info): Likewise.
23052 (chkp_get_check_result): Likewise.
23053 (chkp_remove_check_if_pass): Likewise.
23054 (chkp_use_outer_bounds_if_possible): Likewise.
23055 (chkp_reduce_bounds_lifetime): Likewise.
23056 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
23057 (chkp_mark_completed_bounds): Likewise.
23058 (chkp_register_incomplete_bounds): Likewise.
23059 (chkp_mark_invalid_bounds): Likewise.
23060 (chkp_maybe_copy_and_register_bounds): Likewise.
23061 (chkp_build_returned_bound): Likewise.
23062 (chkp_get_bound_for_parm): Likewise.
23063 (chkp_build_bndldx): Likewise.
23064 (chkp_get_bounds_by_definition): Likewise.
23065 (chkp_generate_extern_var_bounds): Likewise.
23066 (chkp_get_bounds_for_decl_addr): Likewise.
23067 * tree-chrec.c (chrec_apply): Likewise.
23068 * tree-data-ref.c (dump_data_reference): Likewise.
23069 (dump_subscript): Likewise.
23070 (dump_data_dependence_relation): Likewise.
23071 (analyze_overlapping_iterations): Likewise.
23072 * tree-inline.c (expand_call_inline): Likewise.
23073 (tree_function_versioning): Likewise.
23074 * tree-into-ssa.c (dump_defs_stack): Likewise.
23075 (dump_currdefs): Likewise.
23076 (dump_names_replaced_by): Likewise.
23077 (dump_update_ssa): Likewise.
23078 (update_ssa): Likewise.
23079 * tree-object-size.c (pass_object_sizes::execute): Likewise.
23080 * tree-parloops.c (build_new_reduction): Likewise.
23081 (try_create_reduction_list): Likewise.
23082 (ref_conflicts_with_region): Likewise.
23083 (oacc_entry_exit_ok_1): Likewise.
23084 (oacc_entry_exit_single_gang): Likewise.
23085 * tree-pretty-print.h: Likewise.
23086 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
23087 (get_scalar_evolution): Likewise.
23088 (add_to_evolution): Likewise.
23089 (get_loop_exit_condition): Likewise.
23090 (analyze_evolution_in_loop): Likewise.
23091 (analyze_initial_condition): Likewise.
23092 (analyze_scalar_evolution): Likewise.
23093 (instantiate_scev): Likewise.
23094 (number_of_latch_executions): Likewise.
23095 (gather_chrec_stats): Likewise.
23096 (final_value_replacement_loop): Likewise.
23097 (scev_const_prop): Likewise.
23098 * tree-sra.c (dump_access): Likewise.
23099 (disqualify_candidate): Likewise.
23100 (create_access): Likewise.
23101 (reject): Likewise.
23102 (maybe_add_sra_candidate): Likewise.
23103 (create_access_replacement): Likewise.
23104 (analyze_access_subtree): Likewise.
23105 (analyze_all_variable_accesses): Likewise.
23106 (sra_modify_assign): Likewise.
23107 (initialize_constant_pool_replacements): Likewise.
23108 (find_param_candidates): Likewise.
23109 (decide_one_param_reduction): Likewise.
23110 (replace_removed_params_ssa_names): Likewise.
23111 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
23112 * tree-ssa-copy.c (dump_copy_of): Likewise.
23113 (copy_prop_visit_cond_stmt): Likewise.
23114 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
23115 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
23116 (record_equivalences_from_stmt): Likewise.
23117 * tree-ssa-dse.c (compute_trims): Likewise.
23118 (delete_dead_call): Likewise.
23119 (delete_dead_assignment): Likewise.
23120 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
23121 (forward_propagate_into_cond): Likewise.
23122 (pass_forwprop::execute): Likewise.
23123 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
23124 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
23126 (move_computations_worker): Likewise.
23127 (execute_sm): Likewise.
23128 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
23129 (remove_exits_and_undefined_stmts): Likewise.
23130 (remove_redundant_iv_tests): Likewise.
23131 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
23132 (adjust_iv_update_pos): Likewise.
23133 * tree-ssa-math-opts.c (bswap_replace): Likewise.
23134 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
23135 (value_replacement): Likewise.
23136 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
23137 * tree-ssa-pre.c (print_pre_expr): Likewise.
23138 (get_representative_for): Likewise.
23139 (create_expression_by_pieces): Likewise.
23140 (insert_into_preds_of_block): Likewise.
23141 (eliminate_insert): Likewise.
23142 (eliminate_dom_walker::before_dom_children): Likewise.
23143 (eliminate): Likewise.
23144 (remove_dead_inserted_code): Likewise.
23145 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
23146 * tree-ssa-reassoc.c (get_rank): Likewise.
23147 (eliminate_duplicate_pair): Likewise.
23148 (eliminate_plus_minus_pair): Likewise.
23149 (eliminate_not_pairs): Likewise.
23150 (undistribute_ops_list): Likewise.
23151 (eliminate_redundant_comparison): Likewise.
23152 (update_range_test): Likewise.
23153 (optimize_range_tests_var_bound): Likewise.
23154 (optimize_vec_cond_expr): Likewise.
23155 (rewrite_expr_tree): Likewise.
23156 (rewrite_expr_tree_parallel): Likewise.
23157 (linearize_expr): Likewise.
23158 (break_up_subtract): Likewise.
23159 (linearize_expr_tree): Likewise.
23160 (attempt_builtin_powi): Likewise.
23161 (attempt_builtin_copysign): Likewise.
23162 (transform_stmt_to_copy): Likewise.
23163 (transform_stmt_to_multiply): Likewise.
23164 (dump_ops_vector): Likewise.
23165 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
23166 (print_scc): Likewise.
23167 (set_ssa_val_to): Likewise.
23168 (visit_reference_op_store): Likewise.
23169 (visit_use): Likewise.
23170 (sccvn_dom_walker::before_dom_children): Likewise.
23171 (run_scc_vn): Likewise.
23172 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
23174 (expr_hash_elt::print): Likewise.
23175 (const_and_copies::pop_to_marker): Likewise.
23176 (const_and_copies::record_const_or_copy_raw): Likewise.
23177 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
23178 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
23179 (dump_predicates): Likewise.
23180 (find_uninit_use): Likewise.
23181 (warn_uninitialized_phi): Likewise.
23182 (pass_late_warn_uninitialized::execute): Likewise.
23183 * tree-ssa.c (verify_vssa): Likewise.
23184 (verify_ssa): Likewise.
23185 (maybe_optimize_var): Likewise.
23186 * tree-vrp.c (dump_value_range): Likewise.
23187 (dump_all_value_ranges): Likewise.
23188 (dump_asserts_for): Likewise.
23189 (register_edge_assert_for_2): Likewise.
23190 (vrp_visit_cond_stmt): Likewise.
23191 (vrp_visit_switch_stmt): Likewise.
23192 (vrp_visit_stmt): Likewise.
23193 (vrp_visit_phi_node): Likewise.
23194 (simplify_cond_using_ranges_1): Likewise.
23195 (fold_predicate_in): Likewise.
23196 (evrp_dom_walker::before_dom_children): Likewise.
23197 (evrp_dom_walker::push_value_range): Likewise.
23198 (evrp_dom_walker::pop_value_range): Likewise.
23199 (execute_early_vrp): Likewise.
23201 2017-05-16 Richard Biener <rguenther@suse.de>
23203 * dwarf2out.c (loc_list_from_tree_1): Do not create
23204 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
23206 2017-05-16 Richard Biener <rguenther@suse.de>
23208 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
23210 (note_variable_value_in_expr): If we resolved the decl ref
23211 do not push to the stack.
23213 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
23215 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
23216 operations in fast-math mode.
23217 (vaddq_f16): Likewise.
23218 (vmul_f16): Likewise.
23219 (vmulq_f16): Likewise.
23220 (vsub_f16): Likewise.
23221 (vsubq_f16): Likewise.
23222 * config/arm/neon.md (add<mode>3): New.
23224 (fma:<VH:mode>3): New. Also remove outdated comment.
23227 2017-05-16 Martin Liska <mliska@suse.cz>
23231 * ipa-devirt.c (warn_types_mismatch): Fix typo.
23232 (odr_types_equivalent_p): Likewise.
23234 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
23236 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
23238 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
23241 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
23242 non-interunit SSE move alternatives with '?'.
23243 (zero-extendsidi peephole2): New peephole to skip intermediate
23244 general register in SSE zero-extend sequence.
23246 2017-05-15 Jeff Law <law@redhat.com>
23248 * reorg.c (relax_delay_slots): Create a new variable to hold
23249 the temporary target rather than clobbering TARGET_LABEL.
23251 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
23252 missing argument to extract_bit_field call.
23253 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
23255 2017-05-15 Martin Liska <mliska@suse.cz>
23258 * gcc.c (process_command): Do not allow empty argument of -o option.
23260 2017-05-15 Renlin Li <renlin.li@arm.com>
23262 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
23263 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
23264 * config/aarch64/constraints.md (Usf): Add long call check.
23265 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
23266 (call_value): Likewise.
23267 (sibcall): Likewise.
23268 (sibcall_value): Likewise.
23270 (call_value_insn): New.
23271 (sibcall_insn): Update rtx pattern.
23272 (sibcall_value_insn): Likewise.
23273 (call_internal): Remove.
23274 (call_value_internal): Likewise.
23275 (sibcall_internal): Likewise.
23276 (sibcall_value_internal): Likewise.
23277 (call_reg): Likewise.
23278 (call_symbol): Likewise.
23279 (call_value_reg): Likewise.
23280 (call_value_symbol): Likewise.
23282 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
23285 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
23287 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
23289 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
23290 compatible with CCGOCmode and with CCZmode.
23292 2017-05-14 Martin Sebor <msebor@redhat.com>
23294 PR middle-end/77671
23295 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
23296 (gimple_fold_builtin_snprintf): Same.
23297 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
23298 (gimple_fold_builtin_snprintf): Same.
23299 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
23300 of character types.
23301 (is_call_safe): New function.
23302 (try_substitute_return_value): Call it.
23303 (try_simplify_call): New function.
23304 (pass_sprintf_length::handle_gimple_call): Call it.
23306 2017-05-14 Martin Sebor <msebor@redhat.com>
23308 PR middle-end/80669
23309 * builtins.c (expand_builtin_stpncpy): Simplify.
23311 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
23313 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
23314 * config/i386/i386.h
23315 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
23316 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
23317 (struct machine_function): Add new members call_ms2sysv,
23318 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
23319 (struct machine_frame_state): New fields sp_realigned and
23320 sp_realigned_offset.
23321 * config/i386/i386.c
23322 (enum xlogue_stub): New enum.
23323 (enum xlogue_stub_sets): New enum.
23324 (class xlogue_layout): New class.
23325 (struct ix86_frame): New fields stack_realign_allocate_offset,
23326 stack_realign_offset and outlined_save_offset. Modify comments to
23327 detail stack layout when using out-of-line stubs.
23328 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
23329 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
23330 -mcall-ms2sysv-xlogues.
23331 (stub_managed_regs): New static variable.
23332 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
23333 registers managed by out-of-line stub.
23334 (disable_call_ms2sysv_xlogues): New function.
23335 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
23336 m->call_ms2sysv when appropriate and compute frame layout for
23338 (sp_valid_at, fp_valid_at): New inline functions.
23339 (choose_basereg): New function.
23340 (choose_baseaddr): Add align parameter, use choose_basereg and modify
23342 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
23343 Use align parameter of choose_baseaddr to generated aligned SSE movs
23345 (pro_epilogue_adjust_stack): Modify to track
23346 machine_frame_state::sp_realigned.
23347 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
23348 (ix86_nsaved_sseregs): Likewise.
23349 (ix86_emit_save_regs): Likewise.
23350 (ix86_emit_save_regs_using_mov): Likewise.
23351 (ix86_emit_save_sse_regs_using_mov): Likewise.
23352 (get_scratch_register_on_entry): Likewise.
23353 (gen_frame_set): New function.
23354 (gen_frame_load): Likewise.
23355 (gen_frame_store): Likewise.
23356 (emit_outlined_ms2sysv_save): Likewise.
23357 (emit_outlined_ms2sysv_restore): Likewise.
23358 (ix86_expand_prologue): Modify stack re-alignment code and call
23359 emit_outlined_ms2sysv_save when appropriate.
23360 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
23361 parameter rtx_insn *insn, which allows the function to be used to only
23362 generate the notes.
23363 (ix86_expand_epilogue): Modify validity checks of frame and stack
23364 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
23365 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
23366 * config/i386/predicates.md
23367 (save_multiple): New predicate.
23368 (restore_multiple): Likewise.
23369 * config/i386/sse.md
23370 (save_multiple<mode>): New pattern.
23371 (save_multiple_realign<mode>): Likewise.
23372 (restore_multiple<mode>): Likewise.
23373 (restore_multiple_and_return<mode>): Likewise.
23374 (restore_multiple_leave_return<mode>): Likewise.
23375 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
23377 2017-05-14 Julia Koval <julia.koval@intel.com>
23379 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
23380 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
23381 (__builtin_ia32_xsetbv): New builtins.
23382 * config/i386/i386.c (ix86_expand_special_args_builtin):
23384 (ix86_expand_builtin): Special expand for new intrinsics.
23385 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
23386 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
23387 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
23389 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23391 * cfganal.c (inverted_post_order_compute): Change argument type
23393 * cfganal.h (inverted_post_order_compute): Adjust prototype.
23394 * df-core.c (rest_of_handle_df_initialize): Adjust.
23395 (rest_of_handle_df_finish): Likewise.
23396 (df_analyze_1): Likewise.
23397 (df_analyze): Likewise.
23398 (loop_inverted_post_order_compute): Change argument to be a vec *.
23399 (df_analyze_loop): Adjust.
23400 (df_get_n_blocks): Likewise.
23401 (df_get_postorder): Likewise.
23402 * df.h (struct df_d): Change field to be a vec.
23403 * lcm.c (compute_laterin): Adjust.
23404 (compute_available): Likewise.
23405 * lra-lives.c (lra_create_live_ranges_1): Likewise.
23406 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
23407 * tree-ssa-pre.c (compute_antic): Likewise.
23409 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23411 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
23412 (depth_first_search::depth_first_search): Change structure init
23413 function to this constructor.
23414 (depth_first_search::add_bb): Rename function to this member.
23415 (depth_first_search::execute): Likewise.
23416 (flow_dfs_compute_reverse_finish): Adjust.
23418 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23420 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
23421 (longest_simple_path): Likewise.
23422 * shrink-wrap.c (spread_components): Likewise.
23423 (disqualify_problematic_components): Likewise.
23424 (emit_common_heads_for_components): Likewise.
23425 (emit_common_tails_for_components): Likewise.
23426 (insert_prologue_epilogue_for_components): Likewise.
23428 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23430 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
23433 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23435 * df-core.c (df_set_blocks): Start using auto_bitmap.
23436 (df_compact_blocks): Likewise.
23437 * df-problems.c (df_rd_confluence_n): Likewise.
23438 * df-scan.c (df_insn_rescan_all): Likewise.
23439 (df_process_deferred_rescans): Likewise.
23440 (df_update_entry_block_defs): Likewise.
23441 (df_update_exit_block_uses): Likewise.
23442 (df_entry_block_bitmap_verify): Likewise.
23443 (df_exit_block_bitmap_verify): Likewise.
23444 (df_scan_verify): Likewise.
23445 * lra-constraints.c (lra_constraints): Likewise.
23446 (undo_optional_reloads): Likewise.
23447 (lra_undo_inheritance): Likewise.
23448 * lra-remat.c (calculate_gen_cands): Likewise.
23449 (do_remat): Likewise.
23450 * lra-spills.c (assign_spill_hard_regs): Likewise.
23451 (spill_pseudos): Likewise.
23452 * tree-ssa-pre.c (bitmap_set_and): Likewise.
23453 (bitmap_set_subtract_values): Likewise.
23455 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23457 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
23458 management with auto_bitmap.
23459 (fix_inter_tick): Likewise.
23460 (fix_recovery_deps): Likewise.
23461 * ira.c (add_store_equivs): Likewise.
23462 (find_moveable_pseudos): Likewise.
23463 (split_live_ranges_for_shrink_wrap): Likewise.
23464 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
23465 (rtx_reuse_manager::seen_def_p): Likewise.
23466 (rtx_reuse_manager::set_seen_def): Likewise.
23467 * print-rtl.h (class rtx_reuse_manager): Likewise.
23469 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23471 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
23473 (migrate_btr_def): Likewise.
23474 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
23475 * df-core.c (loop_post_order_compute): Likewise.
23476 (loop_inverted_post_order_compute): Likewise.
23477 * hsa-common.h: Likewise.
23478 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
23479 * init-regs.c (initialize_uninitialized_regs): Likewise.
23480 * ipa-inline.c (resolve_noninline_speculation): Likewise.
23481 (inline_small_functions): Likewise.
23482 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
23483 * ira.c (combine_and_move_insns): Likewise.
23484 (build_insn_chain): Likewise.
23485 * loop-invariant.c (find_invariants): Likewise.
23486 * lower-subreg.c (propagate_pseudo_copies): Likewise.
23487 * predict.c (tree_predict_by_opcode): Likewise.
23488 (predict_paths_leading_to): Likewise.
23489 (predict_paths_leading_to_edge): Likewise.
23490 (estimate_loops_at_level): Likewise.
23491 (estimate_loops): Likewise.
23492 * shrink-wrap.c (try_shrink_wrapping): Likewise.
23493 (spread_components): Likewise.
23494 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
23495 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
23496 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
23497 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
23498 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
23499 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
23500 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
23501 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
23502 (mark_threaded_blocks): Likewise.
23503 (thread_through_all_blocks): Likewise.
23504 * tree-ssa.c (verify_ssa): Likewise.
23505 (execute_update_addresses_taken): Likewise.
23506 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
23508 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23510 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
23512 (post_order_compute): Likewise.
23513 (inverted_post_order_compute): Likewise.
23514 (pre_and_rev_post_order_compute_fn): Likewise.
23516 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23518 * genrecog.c (int_set::int_set): Explicitly construct our
23519 auto_vec base class.
23520 * vec.h (auto_vec::auto_vec): New constructor.
23522 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23524 * bitmap.h (class auto_bitmap): New constructor taking
23525 bitmap_obstack * argument.
23527 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
23529 * bitmap.h (class auto_bitmap): Change type of m_bits to
23530 bitmap_head, and adjust ctor / dtor and member operators.
23532 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
23534 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
23535 when returned register mode doesn't match original mode.
23537 2017-05-12 Jeff Law <law@redhat.com>
23538 Jakub Jelinek <jakub@redhat.com>
23540 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
23541 we look for cc setter after the compare-elim changes.
23542 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
23543 within the vector to match what compare-elim now expects.
23544 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
23545 (xorsi3_flags, one_cmplsi2_flags): Likewise.
23547 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
23548 after the compare-elim changes.
23549 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
23550 the vector to match what compare-elim now expects.
23551 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
23552 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
23553 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
23554 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
23555 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
23557 * config/visium/visium.c (single_set_and_flags): Fix where
23558 we look for cc setter after the compare-elim changes.
23559 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
23560 with the vector to match what compare-elim now expects.
23561 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
23562 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
23563 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
23564 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
23565 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
23566 (neg<mode>2_insn_set_overflow): Likewise.
23568 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
23570 PR middle-end/79794
23571 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
23572 maybe_expand_insn call, set ops[0].target. If still set after call,
23573 set alt_rtl. Add extra arg to recursive calls.
23574 (extract_bit_field): Add alt_rtl argument. Pass to
23576 * expmed.h (extract_bit_field): Fix prototype.
23577 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
23578 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
23579 to extract_bit_field_calls.
23580 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
23581 Pass alt_rtl to extract_bit_field calls.
23582 * calls.c (store_unaligned_arguments_into_psuedos)
23583 load_register_parameters): Pass extra NULL to extract_bit_field calls.
23584 * optabs.c (maybe_legitimize_operand): Clear op->target when call
23586 * optabs.h (struct expand_operand): Add target bitfield.
23588 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23590 * compare-elim.c (try_eliminate_compare): Canonicalize
23591 operation with embedded compare to
23592 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
23593 (set (reg) (operation)].
23595 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
23597 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
23600 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
23601 cost of adding a carry flag for ADC instruction.
23602 [case MINUS]: Ignore the cost of subtracting a carry flag
23603 for SBB instruction.
23605 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
23607 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
23609 * config/rs6000/bmiintrin.h: New file.
23610 * config/rs6000/bmi2intrin.h: New file.
23611 * config/rs6000/x86intrin.h: New file.
23613 2017-05-12 Jeff Law <law@redhat.com>
23615 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
23618 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
23620 PR middle-end/80707
23621 * tree-cfg.c: Remove cfg edges of unreachable case statements.
23623 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23625 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23626 early expansion of vector divide builtins.
23627 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
23628 builtins identified as having unsigned arguments.
23630 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23632 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
23633 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
23634 expansion of vector logical operations (and, andc, or, xor,
23637 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23639 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
23640 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
23642 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
23644 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
23645 early GIMPLE expansion of vector multiplies.
23647 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23649 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
23650 TARGET_HAVE_MOVT conditional.
23651 (movt splitter): Likewise.
23653 2017-05-12 Richard Biener <rguenther@suse.de>
23655 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
23656 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23657 Fold all stmts not inplace.
23659 2017-05-12 Richard Biener <rguenther@suse.de>
23661 PR tree-optimization/80713
23662 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
23663 inserted_exprs bit for not removed stmts.
23665 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
23667 PR middle-end/69921
23668 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
23669 parallelized" attribute for parallelized OpenACC kernels.
23670 * omp-offload.c (execute_oacc_device_lower): Use it.
23672 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
23673 Set "oacc kernels" attribute.
23674 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
23675 parameter. Adjust all users.
23676 (oacc_fn_attrib_kernels_p): Remove function.
23677 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
23678 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
23679 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
23680 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
23681 assert "oacc kernels" attribute is set.
23683 2017-05-11 Carl Love <cel@us.ibm.com>
23685 * config/rs6000/rs6000-c: Add support for built-in functions
23686 vector unsigned char vec_popcnt (vector signed char)
23687 vector unsigned char vec_popcnt (vector unsigned char)
23688 vector unsigned short vec_popcnt (vector signed short)
23689 vector unsigned short vec_popcnt (vector unsigned short)
23690 vector unsigned int vec_popcnt (vector signed int)
23691 vector unsigned int vec_popcnt (vector unsigned int)
23692 vector unsigned long long vec_popcnt (vector signed long long)
23693 vector unsigned long long vec_popcnt (vector unsigned long long)
23694 vector signed long long vec_slo (vector signed long long,
23695 vector signed char)
23696 vector signed long long vec_slo (vector signed long long,
23697 vector unsigned char)
23698 vector unsigned long long vec_slo (vector unsigned long long,
23699 vector signed char)
23700 vector unsigned long long vec_slo (vector unsigned long long,
23701 vector unsigned char)
23702 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
23703 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
23704 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
23705 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
23706 * doc/extend.texi: Update the built-in documentation file for the
23707 new built-in functions.
23709 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
23711 * attribs.h (sorted_attr_string): Move machine independent
23712 functions for target clone support from the i386 port to common
23713 code. Rename ix86_function_versions to common_function_versions.
23714 Rename make_name to make_unique_name.
23715 (common_function_versions): Likewise.
23716 (make_unique_name): Likewise.
23717 (make_dispatcher_decl): Likewise.
23718 (is_function_default_version): Likewise.
23719 * attribs.c (attr_strcmp): Likewise.
23720 (sorted_attr_string): Likewise.
23721 (common_function_versions): Likewise.
23722 (make_unique_name): Likewise.
23723 (make_dispatcher_decl): Likewise.
23724 (is_function_default_version): Likewise.
23725 * config/i386/i386.c (attr_strcmp): Likewise.
23726 (sorted_attr_string): Likewise.
23727 (ix86_function_versions): Likewise.
23728 (make_name): Likewise.
23729 (make_dispatcher_decl): Likewise.
23730 (is_function_default_version): Likewise.
23731 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
23733 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23736 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
23737 Account for direct move costs for vec_construct of integer
23740 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
23743 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
23744 (UNSPEC_STX_ATOMIC): Ditto.
23745 (loaddi_via_sse): New insn.
23746 (storedi_via_sse): Ditto.
23747 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
23748 Update corresponding peephole2 patterns.
23749 (atomic_storedi_fpu): Ditto.
23751 2017-05-11 Julia Koval <julia.koval@intel.com>
23753 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
23754 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
23756 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
23757 (__builtin_ia32_rsqrt14ss_mask): New builtins.
23758 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
23760 2017-05-11 Nathan Sidwell <nathan@acm.org>
23762 * graphite-poly.c: Include dumpfile.h.
23764 * dumpfle.h (dump_function): Declare here ...
23765 * tree-dump.h (dump_function): ... not here.
23766 * dumpfile.c: #include tree-cfg.h.
23767 (dump_function): Move here from ...
23768 * tree-dump.c (dump_function): ... here.
23769 * gimplify.c: #include splay-tree.h, not tree-dump.h.
23770 * graphite-poly.c: Don't include tree-dump.h.
23771 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
23772 * print-tree.c: Likewise.
23773 * stor-layout.c: Likewise.
23774 * tree-nested.c: Likewise.
23776 * dumpfile.c (dump_start): Use TDF_FLAGS.
23777 (dump_enable_all): Fix TDF_KIND check thinko.
23779 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
23781 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
23782 array entries to represent two legal parameterizations of the
23783 overloaded __builtin_cmpb function, as represented by the
23784 P6_OV_BUILTIN_CMPB constant.
23785 (altivec_resolve_overloaded_builtin): Add special case handling
23786 for the __builtin_cmpb function, as represented by the
23787 P6_OV_BUILTIN_CMPB constant.
23788 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
23789 (BU_P6_64BIT_2): New macro.
23790 (BU_P6_OVERLOAD_2): New macro
23791 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
23792 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
23793 (CMPB): Add overload support to represent both 32-bit and 64-bit
23794 compare-bytes function.
23795 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
23796 support for TARGET_CMPB.
23797 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
23798 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
23799 documentation of the __builtin_cmpb overloaded built-in function.
23801 2017-05-11 Richard Biener <rguenther@suse.de>
23803 PR tree-optimization/80705
23804 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
23805 bases are not vectorizable.
23807 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23809 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
23810 when counting register pressure.
23812 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23814 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
23815 (iv_ca_more_deps): Renamed to ...
23816 (iv_ca_compare_deps): ... this.
23817 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
23819 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23821 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
23823 (determine_group_iv_costs): ... here.
23824 (find_inv_vars_cb): Record inv var if it's not recorded before.
23826 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23828 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
23829 (get_shiftadd_cost): Ditto.
23831 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23833 * tree-ssa-address.c: Include header file.
23834 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
23836 (add_to_parts): Refactor.
23837 (addr_to_parts): New parameter. Update use of move_hint_to_base.
23838 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
23841 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23843 PR tree-optimization/53090
23844 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
23846 (extract_cond_operands): Detect condition with IV on both sides
23847 and return COMP_IV_EXPR_2.
23848 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
23849 (rewrite_use_compare): Simplify by removing call to function
23850 extract_cond_operands.
23852 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23854 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
23855 (extract_cond_operands): Detect condition comparing against non-
23856 invariant bound and return appropriate enum value.
23857 (find_interesting_uses_cond): Update use of extract_cond_operands.
23858 Handle its return value accordingly.
23859 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
23861 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23863 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
23864 nonlinear iv_use computation in loop invariant sensitive way.
23866 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23868 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
23869 (find_iv_candidates): Call relate_compare_use_with_all_cands.
23871 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23873 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
23874 (dump_cand): Support iv_cand.inv_exprs.
23875 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
23877 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
23880 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23882 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
23884 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
23885 as local function. Include necessary header files.
23886 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
23888 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23890 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
23892 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23894 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
23895 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
23896 RSHIFT_EXPR and BIT_NOT_EXPR.
23898 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23900 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
23901 (adjust_setup_cost): New parameter supporting round up adjustment.
23902 (struct address_cost_data): Delete.
23903 (force_expr_to_var_cost): Don't bound cost with spill_cost.
23904 (split_address_cost, ptr_difference_cost): Delete.
23905 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
23906 (struct ainc_cost_data): New struct.
23907 (get_address_cost_ainc): New function.
23908 (get_address_cost, get_computation_cost): Reimplement.
23909 (determine_group_iv_cost_address): Record inv_expr for all uses of
23911 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
23912 (iv_ca_has_deps): Reimplemented to ...
23913 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
23915 (iv_ca_extend): Call iv_ca_more_deps.
23917 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23919 * tree-ssa-address.c (struct mem_address): Move to header file.
23920 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
23921 * tree-ssa-address.h (struct mem_address): Move from C file.
23922 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
23924 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23926 * tree-affine.h (aff_combination_type): New interface.
23927 (aff_combination_zero_p): Remove static.
23928 (aff_combination_const_p): New interface.
23929 (aff_combination_singleton_var_p): New interfaces.
23931 2017-05-11 Richard Biener <rguenther@suse.de>
23933 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
23934 Skip unreachable blocks and destinations.
23935 (eliminate): Move stmt removal and fixup ...
23936 (fini_eliminate): ... here. Skip inserted exprs.
23937 (pass_pre::execute): Move fini_pre after fini_eliminate.
23938 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
23939 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
23940 PRE to get rid of dead code that has invalid SSA form and
23941 split critical edges again.
23943 2017-05-11 Bin Cheng <bin.cheng@arm.com>
23945 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23947 2017-05-11 Richard Biener <rguenther@suse.de>
23949 * passes.c (execute_function_todo): Verify loops if they are
23950 said to be up-to-date.
23951 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
23952 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
23954 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
23957 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
23958 handle calling assemble_external ourself.
23961 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
23962 modes with zero size. Enhance comment.
23964 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23966 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
23967 built-ins for vec_xl and vec_xst with short and char pointer
23970 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23972 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
23973 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
23974 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
23975 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
23976 (_mm_maskz_min_round_ss): New intrinsics.
23977 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
23978 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
23979 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
23980 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
23981 (__builtin_ia32_minss_mask_round): New builtins.
23982 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23983 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23984 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
23986 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
23987 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
23989 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
23992 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
23994 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
23995 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
23996 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
23997 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
23998 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
23999 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
24000 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
24001 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24002 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
24003 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
24004 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
24005 (__builtin_ia32_mulss_mask_round): New builtins.
24006 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
24007 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
24008 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
24010 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
24011 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
24013 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
24016 2017-05-10 Julia Koval <julia.koval@intel.com>
24018 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
24019 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
24020 (_mm256_setr_m128i): New intrinsics.
24022 2017-05-10 Julia Koval <julia.koval@intel.com>
24024 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
24025 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
24026 (_mm_maskz_rcp14_ss): New intrinsics.
24027 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
24028 (__builtin_ia32_rcp14ss_mask): New builtins.
24029 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
24031 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
24033 PR tree-optimization/51513
24034 * tree-cfg.c (gimple_seq_unreachable_p): New function.
24035 (assert_unreachable_fallthru_edge_p): Use it.
24036 (group_case_labels_stmt): Likewise.
24037 * tree-cfg.h: Prototype it.
24038 * stmt.c: Include cfghooks.h and tree-cfg.h.
24039 (emit_case_dispatch_table) <gap_label>: New local variable.
24040 Use it to fill dispatch table gaps.
24041 Test for default_label before updating probabilities.
24042 (expand_case) <default_label>: Remove unneeded initialization.
24043 Test for unreachable default case statement and remove its edge.
24044 Set default_label accordingly.
24045 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
24047 2017-05-10 Carl Love <cel@us.ibm.com>
24049 * config/rs6000/rs6000-c: Add support for built-in functions
24050 vector signed char vec_neg (vector signed char)
24051 vector signed short int vec_neg (vector short int)
24052 vector signed int vec_neg (vector signed int)
24053 vector signed long long vec_neg (vector signed long long)
24054 vector float vec_neg (vector float)
24055 vector double vec_neg (vector double)
24056 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
24058 * config/rs6000/altivec.h: Add define for vec_neg
24059 * doc/extend.texi: Update the built-in documentation for the
24060 new built-in functions.
24062 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24064 PR tree-optimization/77644
24065 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
24067 2017-05-10 Nathan Sidwell <nathan@acm.org>
24069 * dumpfile.h (TDI_lang_all): New.
24070 (TDF_KIND): New. Renumber others
24071 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
24073 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
24075 (get_dump_file_name): Adjust suffix generation.
24076 (dump_enable_all): Use TDF_KIND.
24077 * doc/invoke.texi (-fdump-lang-all): Document.
24079 * dumpfile.h: Tabify.
24081 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
24084 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
24085 Move member access before delete.
24087 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
24089 * tree-inline.c (expand_call_inline): Split block at stmt
24092 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
24095 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
24096 are now unused after splitting mov{sf,sd}_hardfloat.
24097 (f32_lr2): Likewise.
24098 (f32_lm): Likewise.
24099 (f32_lm2): Likewise.
24100 (f32_li): Likewise.
24101 (f32_li2): Likewise.
24102 (f32_lv): Likewise.
24103 (f32_sr): Likewise.
24104 (f32_sr2): Likewise.
24105 (f32_sm): Likewise.
24106 (f32_sm2): Likewise.
24107 (f32_si): Likewise.
24108 (f32_si2): Likewise.
24109 (f32_sv): Likewise.
24110 (f32_dm): Likewise.
24111 (f32_vsx): Likewise.
24112 (f32_av): Likewise.
24113 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
24114 For movsf, order stores so the VSX stores occur before the GPR
24115 store which encourages the register allocator to use a traditional
24116 FPR instead of a GPR. For movsd, order the stores so that the GPR
24117 store comes before the VSX stores to allow the power6 to work.
24118 This is due to the power6 not having a 32-bit integer store
24119 instruction from a FPR.
24120 (movsf_hardfloat): Likewise.
24121 (movsd_hardfloat): Likewise.
24123 2017-05-09 Martin Sebor <msebor@redhat.com>
24125 PR translation/80280
24126 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
24129 PR translation/80280
24130 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
24131 data member added in r247778.
24132 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
24134 2017-05-09 Nathan Sidwell <nathan@acm.org>
24136 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
24138 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
24141 2017-05-09 Marek Polacek <polacek@redhat.com>
24143 * doc/invoke.texi: Fix typo.
24145 2017-05-09 Richard Biener <rguenther@suse.de>
24147 * tree-vrp.c (vrp_val_is_max): Adjust comment.
24148 (vrp_val_is_min): Likewise.
24149 (set_value_range_to_value): Likewise.
24150 (set_value_range_to_nonnegative): Likewise.
24151 (gimple_assign_nonzero_p): Likewise.
24152 (gimple_stmt_nonzero_p): Likewise.
24153 (vrp_int_const_binop): Likewise. Remove unreachable case.
24154 (adjust_range_with_scev): Adjust comments.
24155 (compare_range_with_value): Likewise.
24156 (extract_range_from_phi_node): Likewise.
24157 (test_for_singularity): Likewise.
24159 2017-05-09 Richard Biener <rguenther@suse.de>
24161 * tree-vrp.c (get_single_symbol): Add assert that we don't
24162 get overflowed constants as invariant part.
24163 (compare_values_warnv): Add comment before the TREE_NO_WARNING
24164 checks. Use wi::cmp instead of recursing for integer constants.
24165 (compare_values): Just ignore whether we assumed undefined
24166 overflow instead of failing the compare.
24167 (extract_range_for_var_from_comparison_expr): Add comment before the
24168 TREE_NO_WARNING sets.
24169 (test_for_singularity): Likewise.
24170 (extract_range_from_comparison): Do not disable optimization
24171 when we assumed undefined overflow.
24172 (extract_range_basic): Remove init of unused var.
24174 2017-05-09 Richard Biener <rguenther@suse.de>
24176 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
24177 (extract_range_from_multiplicative_op_1): Adjust.
24178 (extract_range_from_binary_expr_1): Use int_const_binop.
24180 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
24183 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
24184 rs6000_store_data_bypass_p in seven define_bypass directives and
24185 in several comments.
24186 * config/rs6000/rs6000-protos.h: Add prototype for
24187 rs6000_store_data_bypass_p function.
24188 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
24189 function implements slightly different (rs6000-specific) semantics
24190 than store_data_bypass_p, returning false rather than aborting
24191 with assertion error when arguments do not satisfy the
24192 requirements of store data bypass.
24193 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
24194 rs6000_store_data_bypass_p.
24196 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
24198 * config/xtensa/xtensa-protos.h
24199 (xtensa_initial_elimination_offset): New declaration.
24200 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
24201 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
24202 macro definition, add case for FRAME_POINTER_REGNUM when
24203 FRAME_GROWS_DOWNWARD.
24204 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
24205 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
24206 xtensa_initial_elimination_offset.
24208 2017-05-08 Nathan Sidwell <nathan@acm.org>
24210 * doc/invoke.texi: Alphabetize -fdump options.
24212 2017-05-08 Martin Sebor <msebor@redhat.com>
24214 PR translation/80280
24215 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
24217 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
24219 * target.def (compute_frame_layout): New optional target hook.
24220 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
24221 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
24222 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
24224 * reload1.c (verify_initial_elim_offsets): Likewise.
24225 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
24226 (use_simple_return_p): Call arm_compute_frame_layout if needed.
24227 (arm_get_frame_offsets): Split up into this ...
24228 (arm_compute_frame_layout): ... and this function.
24230 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
24232 * config/aarch64/constraints.md (Usa): New constraint.
24233 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
24235 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
24237 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
24238 with_multilib_list after it has been checked.
24240 2017-05-08 Richard Biener <rguenther@suse.de>
24242 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
24243 (bitmap_set_subtract_values): Likewise.
24245 2017-05-08 Richard Biener <rguenther@suse.de>
24247 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
24248 (gimple_assign_nonzero): ... this and remove strict_overflow_p
24250 (gimple_stmt_nonzero_warnv_p): Rename to ...
24251 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
24253 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
24254 (extract_range_basic): Adjust, do not disable propagation on
24255 strict overflow sensitive simplification.
24256 (vrp_visit_cond_stmt): Likewise.
24258 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
24260 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
24261 body size unconditionally.
24263 2017-05-07 Jeff Law <law@redhat.com>
24266 2017-05-06 Jeff Law <law@redhat.com>
24267 PR tree-optimization/78496
24268 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24271 PR tree-optimization/78496
24272 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24273 (simplify_stmt_using_ranges): Call it.
24274 (vrp_dom_walker::before_dom_children): Extract equivalences
24275 from an ASSERT_EXPR with an equality comparison against a
24278 2017-05-06 Jeff Law <law@redhat.com>
24280 PR tree-optimization/78496
24281 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
24284 PR tree-optimization/78496
24285 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
24286 (simplify_stmt_using_ranges): Call it.
24287 (vrp_dom_walker::before_dom_children): Extract equivalences
24288 from an ASSERT_EXPR with an equality comparison against a
24291 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24293 * lra-constraints.c (lra_copy_reg_equiv): New function.
24294 (split_reg): Use it to copy equivalence information from the
24295 original register to the spill register.
24297 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
24299 PR rtl-optimization/75964
24300 * simplify-rtx.c (simplify_const_relational_operation): Remove
24301 invalid handling of comparisons of integer ABS.
24303 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
24305 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
24306 initialize to zero.
24307 (init_regs): Remove declaration.
24308 (function_arg_advance_32): Initialize error_p as boolean variable.
24310 2017-05-05 Nathan Sidwell <nathan@acm.org>
24312 * store-motion.c (remove_reachable_equiv_notes): Reformat long
24313 lines. Use for (;;).
24315 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24317 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
24318 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
24319 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
24320 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
24321 VF=2 that require versioning.
24323 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24325 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
24328 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24330 * diagnostic.h (diagnostic_override_option_index): Convert from
24331 macro to inline function.
24333 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24335 * diagnostic.c (last_module_changed_p): New function.
24336 (set_last_module): New function.
24337 (diagnostic_report_current_module): Convert macro usage to
24338 the above functions.
24339 * diagnostic.h (diagnostic_context::last_module): Strengthen
24340 from const line_map * to const line_map_ordinary *.
24341 (diagnostic_last_module_changed): Delete macro.
24342 (diagnostic_set_last_module): Delete macro.
24344 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24346 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
24347 with diagnostic_report_diagnostic.
24348 (diagnostic_n_impl_richloc): Likewise.
24349 * diagnostic.h (report_diagnostic): Delete macro.
24350 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
24351 with diagnostic_report_diagnostic.
24352 * substring-locations.c (format_warning_va): Likewise.
24354 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24356 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
24357 save/restor of format_spec. Move option-printing code to...
24358 (print_option_information): ...this new function, and
24359 reimplement by simply printing to the pretty_printer,
24360 rather than appending to the format string.
24362 2017-05-05 David Malcolm <dmalcolm@redhat.com>
24364 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
24365 handling logic into...
24366 (update_effective_level_from_pragmas): ...this new function.
24368 2017-05-04 Andrew Waterman <andrew@sifive.com>
24370 * config/riscv/riscv.opt (mstrict-align): New option.
24371 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
24372 (SLOW_UNALIGNED_ACCESS): Define.
24373 (riscv_slow_unaligned_access): Declare.
24374 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
24376 (riscv_slow_unaligned_access): New variable.
24377 (rocket_tune_info): Set slow_unaligned_access to true.
24378 (optimize_size_tune_info): Set slow_unaligned_access to false.
24379 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
24380 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
24381 (riscv_option_override): Set riscv_slow_unaligned_access.
24382 * doc/invoke.texi: Add -mstrict-align to RISC-V.
24384 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
24386 * config/riscv/riscv.md: Unify indentation.
24388 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
24393 * config/rs6000/rs6000.md (u code attribute): Add FIX and
24395 (extendsi<mode>2): Add support for doing sign extension via
24396 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
24397 don't have ISA 3.0 instructions.
24398 (extendsi<mode>2 splitter): Likewise.
24399 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
24400 generate the normal insns since SImode can now go in vector
24401 registers. Disallow the special UNSPECs needed for previous
24402 machines to hide SImode being used. Add new insns
24403 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
24404 (fix_trunc<mode>si2_stfiwx): Likewise.
24405 (fix_trunc<mode>si2_internal): Likewise.
24406 (fixuns_trunc<mode>si2): Likewise.
24407 (fixuns_trunc<mode>si2_stfiwx): Likewise.
24408 (fctiw<u>z_<mode>_smallint): Likewise.
24409 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
24410 of floating point to 32-bit integer from doing a direct move to
24411 the GPR registers to do a store.
24412 (fctiwz_<mode>): Break long line.
24414 2017-05-05 Bin Cheng <bin.cheng@arm.com>
24416 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
24417 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
24418 (addr_list, addr_offset_valid_p): New.
24419 (split_address_groups): Check offset validity with above function.
24420 (gt-tree-ssa-loop-ivopts.h): Include header file.
24422 2017-05-05 Nathan Sidwell <nathan@acm.org>
24424 * config.gcc (arm*-*-*): Add missing 'fi'.
24426 2017-05-05 Steve Ellcey <sellcey@cavium.com>
24428 * doc/invoke.texi (-fopt-info): Explicitly say order of options
24429 included in -fopt-info does not matter.
24430 * doc/optinfo.texi (-fopt-info): Fix description of default
24431 behavour. Explicitly say order of options included in -fopt-info
24434 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
24436 * config.gcc: Allow combinations of aprofile and rmprofile values for
24437 --with-multilib-list.
24438 * config/arm/t-multilib: New file.
24439 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
24440 variables. Remove setting of ISA and floating-point ABI in
24441 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
24442 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
24443 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
24444 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
24446 * config/arm/t-rmprofile: Likewise except for the matches changes.
24447 * doc/install.texi (--with-multilib-list): Document the combination of
24448 aprofile and rmprofile values and warn about pitfalls in doing that.
24450 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24452 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
24453 (movdi_aarch64): Likewise.
24455 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24457 PR tree-optimization/80632
24458 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
24460 (build_arrays): Initialize it for virtual phis.
24461 (fix_phi_nodes): Use it for virtual phis.
24463 PR tree-optimization/80558
24464 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
24465 [x, y] op z into [x op, y op z] for op & or | if conditions
24468 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
24469 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24472 * config/arm/arm.md (use_literal_pool): Remove.
24473 (64-bit immediate split): No longer takes cost into consideration
24474 if arm_disable_literal_pool is enabled.
24475 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
24476 used when arm_disable_literal_pool is enabled.
24477 (arm_max_const_double_inline_cost): Remove use of
24478 arm_disable_literal_pool.
24479 (push_minipool_fix): Add assert.
24480 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
24481 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
24482 (no_literal_pool_sf_immediate): New.
24484 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24486 PR tree-optimization/80613
24487 * tree-ssa-dce.c (propagate_necessity): Remove cases for
24488 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
24490 2017-05-05 Richard Biener <rguenther@suse.de>
24492 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
24494 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
24496 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
24497 of this flag from insn conditions due to removal from r247495.
24499 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
24501 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
24503 (arm_early_store_addr_dep_ptr): Likewise.
24504 * config/arm/aarch-common-protos.h
24505 (arm_early_load_addr_dep_ptr): Add prototype.
24506 (arm_early_store_addr_dep_ptr): Likewise.
24507 * config/arm/cortex-a53.md: Add new bypasses.
24509 2017-05-05 Jakub Jelinek <jakub@redhat.com>
24511 * tree.c (next_type_uid): Change type to unsigned.
24512 (type_hash_canon): Decrement back next_type_uid if
24513 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
24514 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
24517 2017-05-04 Martin Sebor <msebor@redhat.com>
24519 * builtins.c: Fix a trivial typo in a comment.
24521 PR middle-end/79234
24522 * builtins.c (check_sizes): Adjust to handle reading past the end.
24523 Avoid printing excessive upper bound of ranges. Use %E to print
24524 tree nodes instead of converting them to %wu.
24525 (expand_builtin_memchr): New function.
24526 (compute_dest_size): Rename...
24527 (compute_objsize): ...to this.
24528 (expand_builtin_memcpy): Adjust.
24529 (expand_builtin_mempcpy): Adjust.
24530 (expand_builtin_strcat): Adjust.
24531 (expand_builtin_strcpy): Adjust.
24532 (check_strncat_sizes): Adjust.
24533 (expand_builtin_strncat): Adjust.
24534 (expand_builtin_strncpy): Adjust and simplify.
24535 (expand_builtin_memset): Adjust.
24536 (expand_builtin_bzero): Adjust.
24537 (expand_builtin_memcmp): Adjust.
24538 (expand_builtin): Handle memcmp.
24539 (maybe_emit_chk_warning): Check strncat just once.
24541 2017-05-04 Martin Sebor <msebor@redhat.com>
24543 PR preprocessor/79214
24544 PR middle-end/79222
24545 PR middle-end/79223
24546 * builtins.c (check_sizes): Add inlining context and issue
24547 warnings even when -Wno-system-headers is set.
24548 (check_strncat_sizes): Same.
24549 (expand_builtin_strncat): Same.
24550 (expand_builtin_memmove): New function.
24551 (expand_builtin_stpncpy): Same.
24552 (expand_builtin): Handle memmove and stpncpy.
24554 2017-05-04 Bin Cheng <bin.cheng@arm.com>
24556 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
24557 which is not used any more.
24559 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24561 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
24563 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24565 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
24566 (cortexa53_tunings): Likewise.
24567 (cortexa57_tunings): Likewise.
24568 (cortexa72_tunings): Likewise.
24569 (cortexa73_tunings): Likewise.
24571 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24573 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
24574 Set loop alignment to 8.
24576 2017-05-04 Martin Sebor <msebor@redhat.com>
24578 PR translation/80280
24579 * builtins.c (expand_builtin_object_size): Add missing quoting to
24580 %D and like directives.
24581 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
24582 (hsa_type_for_tree_type): Same.
24583 (verify_function_arguments): Same.
24584 * symtab.c (symbol_table::change_decl_assembler_name): Same.
24585 * varasm.c (get_section): Same.
24588 2017-05-04 Martin Sebor <msebor@redhat.com>
24590 PR translation/80280
24591 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
24593 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
24595 * config/aarch64/aarch64.c (generic_addrcost_table):
24596 Change HI/TI mode setting.
24598 2017-05-04 Martin Jambor <mjambor@suse.cz>
24600 PR tree-optimization/80622
24601 * tree-sra.c (comes_initialized_p): New function.
24602 (build_accesses_from_assign): Only set write lazily when
24603 comes_initialized_p is false.
24604 (analyze_access_subtree): Use comes_initialized_p.
24605 (propagate_subaccesses_across_link): Assert !comes_initialized_p
24606 instead of testing for PARM_DECL.
24608 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24610 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
24611 constraint on operand 0 to allow more general addressing modes.
24612 Adjust output template.
24613 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
24615 * config/aarch64/aarch64-protos.h
24616 (aarch64_address_valid_for_prefetch_p): Declare prototype.
24617 * config/aarch64/constraints.md (Dp): New address constraint.
24618 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
24621 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
24623 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
24624 update use of estimate_ipcp_clone_size_and_time.
24625 (estimate_local_effects): Update use of
24626 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
24627 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
24628 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
24629 Return nonspecialized time.
24631 2017-05-04 Richard Biener <rguenther@suse.de>
24633 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
24634 for the last VUSE which def dominates the PHI. Directly call
24636 (get_continuation_for_phi_1): Remove.
24638 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
24640 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
24641 to explain the use of truncating division. Cap the number of
24642 iterations to the maximum given by nb_iterations_upper_bound,
24645 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24647 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
24648 * configure: Regenerate.
24649 * config.in: Regenerate.
24650 * config/i386/driver-mingw32.c: new file.
24651 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
24652 * config.host: Link driver-mingw32.o on MinGW host.
24653 * doc/install.texi: Document new --enable-mingw-wildcard configure
24656 2017-05-04 Marek Polacek <polacek@redhat.com>
24658 PR tree-optimization/80612
24659 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
24661 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24662 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
24664 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
24665 (movt splitter): Likewise.
24666 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
24667 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
24668 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
24669 block for Thumb-1 with MOVT.
24670 (thumb2_legitimate_address_p): Move code block ...
24671 (can_avoid_literal_pool_for_label_p): ... into this new function.
24672 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
24674 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
24675 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
24676 "M-profile targets with the MOVT instruction".
24678 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
24680 * config/arm/arm-builtins.c (arm_init_builtins): Rename
24681 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
24682 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
24684 2017-05-04 Martin Liska <mliska@suse.cz>
24686 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
24687 variable cond_code.
24689 2017-05-04 Richard Biener <rguenther@suse.de>
24691 * tree.c (array_at_struct_end_p): Handle arrays at struct
24692 end with flexarrays more conservatively. Refactor and treat
24693 arrays of arrays or aggregates more strict. Fix
24694 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
24695 * tree.c (array_at_struct_end_p): Adjust prototype.
24696 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
24697 * gimple-fold.c (get_range_strlen): Likewise.
24698 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
24700 2017-05-04 Richard Biener <rguenther@suse.de>
24702 PR tree-optimization/31130
24703 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
24705 (supports_overflow_infinity): Likewise.
24706 (is_negative_overflow_infinity): Likewise.
24707 (is_positive_overflow_infinity): Likewise.
24708 (is_overflow_infinity): Likewise.
24709 (stmt_overflow_infinity): Likewise.
24710 (overflow_infinity_range_p): Likewise.
24711 (usable_range_p): Remove as always returning true.
24712 (make_overflow_infinity): Remove.
24713 (negative_overflow_infinity): Likewise.
24714 (positive_overflow_infinity): Likewise.
24715 (avoid_overflow_infinity): Likewise.
24716 (set_value_range): Adjust accordingly.
24717 (set_value_range_to_nonnegative): Likewise, remove now unused
24718 overflow_infinity arg.
24719 (vrp_operand_equal_p): Adjust.
24720 (update_value_range): Likewise.
24721 (range_int_cst_singleton_p): Likewise.
24722 (operand_less_p): Likewise.
24723 (compare_values_warnv): Likewise.
24724 (extract_range_for_var_from_comparison_expr): Likewise.
24725 (vrp_int_const_binop): Likewise.
24726 (zero_nonzero_bits_from_vr): Likewise.
24727 (extract_range_from_multiplicative_op_1): Likewise.
24728 (extract_range_from_binary_expr_1): Likewise.
24729 (extract_range_from_unary_expr): Likewise.
24730 (extract_range_from_comparison): Likewise.
24731 (extract_range_basic): Likewise.
24732 (adjust_range_with_scev): Likewise.
24733 (compare_ranges): Likewise.
24734 (compare_range_with_value): Likewise.
24735 (dump_value_range): Likewise.
24736 (test_for_singularity): Likewise, remove strict_overflow_p parameter
24738 (simplify_cond_using_ranges): Adjust.
24740 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
24742 * brig-builtins.def: Added a builtin for class_f64.
24743 * builtin-types.def: Added a builtin type needed by class_f64.
24745 2017-05-03 Jason Merrill <jason@redhat.com>
24747 * timevar.def: Add TV_CONSTEXPR.
24749 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24751 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
24753 2017-05-03 Martin Jambor <mjambor@suse.cz>
24755 * ipa-prop.c (ipa_update_after_lto_read): Removed.
24756 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
24757 * ipa-cp.c (ipcp_propagate_stage): Do not call
24758 ipa_update_after_lto_read.
24759 * ipa-inline.c (ipa_inline): Likewise.
24761 2017-05-03 Martin Jambor <mjambor@suse.cz>
24763 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
24764 tag. Added a default constructor and a destructor.
24765 (ipa_edge_args_sum_t): New class;
24766 (ipa_edge_args_sum): Declare.
24767 (ipa_edge_args_vector): Remove declaration.
24768 (IPA_EDGE_REF): Use ipa_edge_args_sum.
24769 (ipa_free_edge_args_substructures): Remove declaration.
24770 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
24771 (ipa_edge_args_info_available_for_edge_p): Likewise.
24772 * ipa-prop.c (ipa_edge_args_vector): Removed.
24773 (edge_removal_hook_holder): Likewise.
24774 (edge_duplication_hook_holder): Likewise.
24775 (ipa_edge_args_sum): New variable.
24776 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
24777 ipa_edge_args_vector.
24778 (ipa_free_edge_args_substructures): Likewise.
24779 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
24780 ipa_edge_args_vector.
24781 (ipa_edge_removal_hook): Turned into method
24782 ipa_edge_args_sum_t::remove.
24783 (ipa_edge_duplication_hook): Turned into method
24784 ipa_edge_args_sum_t::duplicate.
24785 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
24786 registering edge hooks.
24787 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
24788 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
24789 ipa_edge_args_sum instead of ipa_edge_args_vector.
24790 * ipa-profile.c (ipa_profile): Likewise.
24792 2017-05-03 Martin Jambor <mjambor@suse.cz>
24794 * symbol-summary.h (function_summary): New method exists.
24795 (function_summary::symtab_removal): Deallocate through release.
24796 (call_summary): New class.
24797 (gt_ggc_mx): New overload.
24798 (gt_pch_nx): Likewise.
24799 (gt_pch_nx): Likewise.
24801 2017-05-03 Jeff Law <law@redhat.com>
24803 PR tree-optimization/78496
24804 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
24805 from simplify_cond_using_ranges. Split off code to walk
24806 backwards through casts into ...
24807 (simplify_cond_using_ranges_2): New function.
24808 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
24809 (execute_vrp): After identifying jump threads, call
24810 simplify_cond_using_ranges_2.
24812 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
24815 * ipa-inline.h (inline_summary): Add ctor.
24816 (create_ggc): Do not use ggc_cleared_alloc.
24818 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
24819 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24821 * gcc.c (handle_braces): Support escaping in switch matching
24823 * doc/invoke.texi (Spec Files): Document it.
24824 Remove superfluous @code markup in items.
24826 2017-05-03 David Malcolm <dmalcolm@redhat.com>
24828 * diagnostic-show-locus.c (struct column_range): New struct.
24829 (get_affected_columns): New function.
24830 (get_printed_columns): New function.
24831 (struct correction): New struct.
24832 (correction::ensure_capacity): New function.
24833 (correction::ensure_terminated): New function.
24834 (struct line_corrections): New struct.
24835 (line_corrections::~line_corrections): New dtor.
24836 (line_corrections::add_hint): New function.
24837 (layout::print_trailing_fixits): Reimplement in terms of the new
24839 (selftest::test_overlapped_fixit_printing): New function.
24840 (selftest::diagnostic_show_locus_c_tests): Call it.
24842 2017-05-03 Nathan Sidwell <nathan@acm.org>
24844 Canonicalize canonical type hashing
24845 * tree.h (type_hash_canon_hash): Declare.
24846 * tree.c (type_hash_list, attribute_hash_list): Move into
24847 type_hash_canon_hash.
24848 (build_type_attribute_qual_variant): Break out hash code calc into
24849 type_hash_canon_hash.
24850 (type_hash_canon_hash): New. Generic type hash computation.
24851 (build_range_type_1, build_array_type_1, build_function_type,
24852 build_method_type_directly, build_offset_type, build_complex_type,
24853 make_vector_type): Call it.
24855 2017-05-03 Richard Biener <rguenther@suse.de>
24857 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
24858 When all DRs have unknown misaligned do not always peel
24859 when there is a store but apply the same costing model as if
24860 there were only loads.
24862 2017-05-03 Richard Biener <rguenther@suse.de>
24865 PR tree-optimization/80492
24866 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24867 compare_base_decls returning dont-know properly.
24869 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
24871 * config/arm/iterators.md (CCSI): New mode iterator.
24872 (arch): New mode attribute.
24873 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
24874 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
24875 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
24876 code iterator for success result mode.
24877 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
24878 the corresponding new insn generators.
24880 2017-05-03 Bin Cheng <bin.cheng@arm.com>
24883 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24884 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
24886 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
24888 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
24889 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
24890 (DDR_A): Wrap DDR argument in brackets.
24891 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
24892 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
24893 (DDR_REVERSED_P): Likewise.
24895 2017-05-03 Jakub Jelinek <jakub@redhat.com>
24897 PR tree-optimization/79472
24898 * tree-switch-conversion.c (struct switch_conv_info): Add
24899 contiguous_range and default_case_nonstandard fields.
24900 (collect_switch_conv_info): Compute contiguous_range and
24901 default_case_nonstandard fields, don't clear final_bb if
24902 contiguous_range and only the default case doesn't have the required
24904 (check_all_empty_except_final): Set default_case_nonstandard instead
24905 of failing if contiguous_range and the default case doesn't have empty
24907 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
24908 and only the default case doesn't have the required constants. Skip
24910 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
24911 if default_case_nonstandard.
24912 (build_constructors): Build constant 1 just once. Assert that default
24913 values aren't inserted in between cases if contiguous_range. Skip
24915 (build_arrays): Skip virtual phis.
24916 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
24917 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
24918 Handle virtual phis.
24919 (gen_inbound_check): Handle default_case_nonstandard case.
24920 (process_switch): Adjust check_final_bb caller. Call
24921 gather_default_values with the first non-default case instead of
24922 default case if default_case_nonstandard.
24924 2017-05-02 Nathan Sidwell <nathan@acm.org>
24926 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
24927 check. Fix formatting.
24929 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
24931 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
24932 errors when comparing specialized and unspecialized times.
24934 2017-05-02 David Malcolm <dmalcolm@redhat.com>
24936 * diagnostic-show-locus.c
24937 (layout::should_print_annotation_line_p): Make private.
24938 (layout::print_annotation_line): Make private.
24939 (layout::annotation_line_showed_range_p): Make private.
24940 (layout::show_ruler): Make private.
24941 (layout::print_source_line): Make private. Pass in line and
24942 line_width, rather than calling location_get_source_line. Drop
24944 (layout::print_leading_fixits): New method.
24945 (layout::print_any_fixits): Rename to...
24946 (layout::print_trailing_fixits): ...this, and make private.
24947 Don't print newline fixits.
24948 (diagnostic_show_locus): Move logic for printing one row into...
24949 (layout::print_line): ...this new function. Move the
24950 location_get_source_line call and error-handling from
24951 print_source_line to here. Call print_leading_fixits, and rename
24952 print_any_fixits to print_trailing_fixits.
24953 (selftest::test_fixit_insert_containing_newline): Update now that
24954 newlines are partially supported.
24955 (selftest::test_fixit_insert_containing_newline_2): New test.
24956 (selftest::test_fixit_replace_containing_newline): Update comments.
24957 (selftest::diagnostic_show_locus_c_tests): Call the new test.
24958 * edit-context.c (class added_line): New class.
24959 (class edited_line): Describe newline handling in comment.
24960 (edited_line::actually_edited_p): New method.
24961 (edited_line::print_content): Delete redundant decl.
24962 (edited_line::m_predecessors): New field.
24963 (edited_file::print_content): Call edited_line::print_content.
24964 (edited_file::print_diff): Update to support newlines.
24965 (edited_file::print_diff_hunk): Likewise.
24966 (edited_file::print_run_of_changed_lines): New function.
24967 (edited_file::print_diff_line): Convert to...
24968 (print_diff_line): ...this.
24969 (edited_file::get_effective_line_count): New function.
24970 (edited_line::edited_line): Initialize new field m_predecessors.
24971 (edited_line::~edited_line): Clean up m_predecessors.
24972 (edited_line::apply_fixit): Handle newlines.
24973 (edited_line::get_effective_line_count): New function.
24974 (edited_line::print_content): New function.
24975 (edited_line::print_diff_lines): New function.
24976 (selftest::test_applying_fixits_insert_containing_newline): New
24978 (selftest::test_applying_fixits_replace_containing_newline): New
24980 (selftest::insert_line): New function.
24981 (selftest::test_applying_fixits_multiple_lines): Add example of
24983 (selftest::edit_context_c_tests): Call the new tests.
24985 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24987 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
24988 parameter cand. Update dump information.
24989 (get_computation_cost): Update uses.
24991 2017-05-02 Bin Cheng <bin.cheng@arm.com>
24993 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
24994 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
24995 (get_computation_at, rewrite_use_address): Update use of
24996 get_computation_aff.
24998 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25000 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
25001 (get_computation): Delete.
25002 (get_computation_cost): Implement like get_computation_cost_at.
25003 Use get_computation_at.
25004 (get_computation_cost_at): Delete.
25005 (rewrite_use_nonlinear_expr): Use get_computation_at.
25006 (rewrite_use_compare, remove_unused_ivs): Ditto.
25008 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25010 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
25012 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25014 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
25015 (ivopts_global_cost_for_size): Rename parameter and update uses.
25016 (iv_ca_recount_cost): Update uses.
25017 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
25018 candidates seperately in n_invs and n_cands.
25019 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
25021 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25023 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
25024 (find_inv_vars_cb): New.
25025 (find_depends): Renamed to ...
25026 (find_inv_vars): ... this.
25027 (add_candidate_1, force_var_cost): Call find_inv_vars.
25028 (split_address_cost, determine_group_iv_cost_cond): Ditto.
25030 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25032 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
25033 inv_vars. Add inv_exprs.
25034 (struct iv_cand): Rename depends_on to inv_vars.
25035 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
25036 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
25037 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
25038 (dump_cand): Dump inv_vars.
25039 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
25040 (record_invariant, find_depends, add_candidate_1): Ditto.
25041 (set_group_iv_cost, force_var_cost): Ditto.
25042 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
25043 (get_computation_cost_at, get_computation_cost): Ditto.
25044 (determine_group_iv_cost_generic): Ditto.
25045 (determine_group_iv_cost_address): Ditto.
25046 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
25047 (determine_group_iv_costs): Ditto.
25048 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
25049 (iv_ca_set_remove_invariants): Renamed to ...
25050 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
25051 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
25052 (iv_ca_set_add_invariants): Renamed to ...
25053 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
25054 (iv_ca_set_cp): Use iv_ca_set_add_invs.
25055 (iv_ca_has_deps): Support inv_vars and inv_exprs.
25056 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
25057 (create_new_ivs): Remove useless dump.
25059 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25061 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
25063 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
25064 (iv_ca_set_no_cp, create_new_iv): Ditto.
25066 2017-05-02 Bin Cheng <bin.cheng@arm.com>
25068 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
25070 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25072 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
25073 function tree_check2.
25075 2017-05-02 Martin Liska <mliska@suse.cz>
25077 * doc/gcov.texi: Add missing preposition.
25078 * gcov.c (function_info::function_info): Properly fill up
25079 all member variables.
25081 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25083 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
25085 2017-05-02 Tamar Christina <tamar.christina@arm.com>
25087 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
25089 2017-05-02 Martin Liska <mliska@suse.cz>
25092 * lto-streamer-in.c (lto_read_tree_1): Remove
25093 LTO_STREAMER_DEBUG.
25094 * lto-streamer.c (struct tree_hash_entry): Likewise.
25095 (struct tree_entry_hasher): Likewise.
25096 (tree_entry_hasher::hash): Likewise.
25097 (tree_entry_hasher::equal): Likewise.
25098 (lto_streamer_init): Likewise.
25099 (lto_orig_address_map): Likewise.
25100 (lto_orig_address_get): Likewise.
25101 (lto_orig_address_remove): Likewise.
25102 * lto-streamer.h: Likewise.
25103 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
25104 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
25106 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
25108 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
25109 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
25110 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
25111 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
25112 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
25113 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
25114 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
25115 (mm_maskz_sub_ss): New intrinsics.
25116 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25117 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
25118 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
25119 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
25120 (__builtin_ia32_subss_mask_round): New builtins.
25121 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
25122 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
25123 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
25125 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
25126 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
25128 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
25131 2017-05-02 Martin Jambor <mjambor@suse.cz>
25133 PR tree-optimization/78687
25134 * tree-sra.c (access): New field parent.
25135 (process_subtree_disqualification): New function.
25136 (disqualify_candidate): Call it.
25137 (build_accesses_from_assign): Reset write flag if creating an
25139 (build_access_subtree): Fill in parent field and also prpagate
25140 down grp_write flag.
25141 (create_artificial_child_access): New parameter set_grp_write, set
25142 grp_write to its value.
25143 (propagate_subaccesses_across_link): Also propagate grp_write flag
25145 (propagate_all_subaccesses): Push the closest parent back to work
25146 queue if add_access_to_work_queue returned true.
25148 2017-05-02 Richard Biener <rguenther@suse.de>
25150 * common.opt (fstrict-overflow): Alias negative to fwrapv.
25151 * doc/invoke.texi (fstrict-overflow): Remove all traces of
25152 -fstrict-overflow documentation.
25153 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
25154 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
25155 flag_strict_overflow.
25156 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
25157 * lto-opts.c (lto_write_options): Do not stream it.
25158 * lto-wrapper.c (merge_and_complain): Do not handle it.
25159 * opts.c (default_options_table): Do not set -fstrict-overflow.
25160 (finish_options): Likewise do not clear it when sanitizing.
25161 * simplify-rtx.c (simplify_const_relational_operation): Do not
25162 test flag_strict_overflow.
25164 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25166 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
25167 using enabled attribute.
25168 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
25169 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
25170 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
25171 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
25172 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
25173 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
25174 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
25175 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
25176 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
25177 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
25179 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
25181 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
25183 2017-05-02 Richard Biener <rguenther@suse.de>
25185 PR tree-optimization/80591
25187 2017-04-10 Richard Biener <rguenther@suse.de>
25189 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25192 2017-05-02 Richard Biener <rguenther@suse.de>
25194 PR tree-optimization/80549
25195 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
25196 (cleanup_tree_cfg_noloop): Create forwarders to known loop
25197 headers if they do not have a preheader.
25199 2017-05-02 Martin Liska <mliska@suse.cz>
25202 * common.opt: Fix typo.
25203 * doc/invoke.texi: Likewise.
25205 2017-05-01 Jan Beulich <jbeulich@suse.com>
25207 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
25208 swapping, add (x,x,m,x,n) alternative.
25210 2017-05-01 Nathan Sidwell <nathan@acm.org>
25212 * calls.c (combine_pending_stack_adjustment_and_call): Remove
25213 unnecessary unadjusted_alignment check.
25215 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
25218 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
25220 * gimplify.c (gimplify_cilk_detach): New function.
25221 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
25222 * tree-core.h: Document EXPR_CILK_SPAWN.
25223 * tree.h (EXPR_CILK_SPAWN): Define.
25225 2017-05-01 David Malcolm <dmalcolm@redhat.com>
25227 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
25228 to use new fixit_hint representation, using the "replace" logic.
25229 (get_line_span_for_fixit_hint): Likewise.
25230 (layout::print_any_fixits): Likewise.
25231 (selftest::test_one_liner_many_fixits): Rename to...
25232 (selftest::test_one_liner_many_fixits_1): ...this, and update
25233 comment and expected output to reflect that the multiple fix-it
25234 hints are now consolidated into one insertion.
25235 (selftest::test_one_liner_many_fixits_2): New test.
25236 (selftest::test_diagnostic_show_locus_one_liner): Update for
25238 (selftest::test_fixit_consolidation): Update for fix-it API
25240 * diagnostic.c (print_parseable_fixits): Likewise.
25241 * edit-context.c (edited_line::m_line_events): Convert from
25242 auto_vec <line_event *> to auto_vec <line_event>.
25243 (class line_event): Convert from abstract base class to a concrete
25244 class, taking over the role of replace_event.
25245 (class insert_event): Delete.
25246 (class replace_event): Rename to class line_event. Convert to
25248 (edit_context::add_fixits): Reimplement.
25249 (edit_context::apply_insert): Delete.
25250 (edit_context::apply_replace): Rename to...
25251 (edit_context::apply_fixit): ...this. Convert to half-open range.
25252 (edited_file::apply_insert): Delete.
25253 (edited_file::apply_replace): Rename to...
25254 (edited_file::apply_fixit): ...this.
25255 (edited_line::~edited_line): Drop deletion of events.
25256 (edited_line::apply_insert): Delete.
25257 (edited_line::apply_replace): Rename to...
25258 (edited_line::apply_fixit): ...this. Convert to half-open range.
25259 Update for change to type of m_line_events.
25260 * edit-context.h (edit_context::apply_insert): Delete.
25261 (edit_context::apply_replace): Rename to...
25262 (edit_context::apply_fixit): ...this.
25264 2017-05-01 Martin Sebor <msebor@redhat.com>
25266 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
25269 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
25272 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
25273 __get_cpuid_max returns 0.
25274 (__get_cpuid_count): Ditto.
25276 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
25278 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
25279 replacement expression is another instance of one of its arguments.
25281 2017-05-01 Jakub Jelinek <jakub@redhat.com>
25284 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
25285 check for stack push/pop autoinc.
25286 * config/i386/i386.c (ix86_agi_dependent): Return false
25287 if the only reason why modified_in_p returned true is that
25288 addr is SP based and set_insn is a push or pop.
25290 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25292 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
25295 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
25298 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
25299 (account_size_time): Use two predicates - exec_pred and
25301 (evaluate_conditions_for_known_args): Compute both clause and
25303 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
25304 (inline_summary_t::duplicate): Update.
25305 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
25307 (compute_inline_parameters): Likewise.
25308 (estimate_edge_size_and_time): Update caluclation of time.
25309 (estimate_node_size_and_time): Compute both time and nonspecialized
25311 (estimate_ipcp_clone_size_and_time): Update.
25312 (inline_merge_summary): Update.
25313 (do_estimate_edge_time): Update.
25314 (do_estimate_edge_size): Update.
25315 (do_estimate_edge_hints): Update.
25316 (inline_read_section, inline_write_summary): Stream both new predicates.
25317 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
25319 (compute_inlined_call_time): Cleanup.
25320 (big_speedup_p): Update.
25321 (edge_badness): Update.
25322 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
25323 (size_time_entry): Replace predicate by exec_predicate and
25324 nonconst_predicate.
25325 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
25326 (estimate_edge_time): Return also nonspec_time.
25327 (reset_edge_growth_cache): Update.
25329 2017-04-29 Jakub Jelinek <jakub@redhat.com>
25331 PR rtl-optimization/80491
25332 * ifcvt.c (noce_process_if_block): When looking for x setter
25333 with missing else_bb, don't check only the insn right before
25334 cond_earliest, but look for the last insn that x is modified in
25335 within the same bb.
25337 PR rtl-optimization/80491
25338 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
25340 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
25342 PR tree-optimization/80487
25343 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
25345 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
25347 PR tree-optimization/79697
25348 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
25349 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
25350 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
25352 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
25353 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
25355 2017-04-28 Martin Sebor <msebor@redhat.com>
25357 PR tree-optimization/80523
25358 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
25359 (init_target_to_host_charmap, target_to_host, target_strtol10): New
25361 (maybe_warn, format_directive, parse_directive): Use new functions.
25362 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
25364 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
25366 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
25368 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
25370 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
25371 target_header_dir): Set correctly.
25372 * configure: Regenerated.
25373 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
25374 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
25375 instead of SYSTEM_HEADER_DIR.
25377 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
25379 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
25380 (estimate_local_effects): Likewise.
25381 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
25382 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
25383 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
25384 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
25385 do_estimate_edge_time, estimate_edge_time): Likewise.
25386 * ipa-inline-analysis.c (estimate_node_size_and_time,
25387 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
25388 (estimate_time_after_inlining): Remove.
25390 2017-04-28 Martin Liska <mliska@suse.cz>
25392 * doc/gcov.texi: Enhance documentation of gcov.
25394 2017-04-28 Martin Liska <mliska@suse.cz>
25396 * doc/gcov.texi: Sort options in alphabetic order.
25397 * doc/gcov-dump.texi: Likewise.
25398 * doc/gcov-tool.texi: Likewise.
25399 * gcov.c (print_usage): Likewise.
25400 * gcov-dump.c (print_usage): Likewise.
25401 * gcov-tool.c (print_merge_usage_message): Likewise.
25402 (print_rewrite_usage_message): Likewise.
25403 (print_overlap_usage_message): Likewise.
25405 2017-04-28 Martin Liska <mliska@suse.cz>
25407 PR gcov-profile/53915
25408 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
25410 2017-04-28 Martin Liska <mliska@suse.cz>
25412 PR gcov-profile/79891
25413 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
25414 is marked by compiler as living on a line.
25415 (get_cycles_count): Remove usage of the union.
25416 (output_intermediate_file): Likewise.
25417 (find_source): Fix GNU coding style.
25418 (accumulate_line_counts): Remove old non-all block mode.
25419 (output_lines): Remove usage of the union.
25420 * profile.c (output_location): Include all BBs, even if
25421 belonging to a same line (and file) as a previous BB.
25423 2017-04-28 Martin Liska <mliska@suse.cz>
25425 * gcov.c (process_args): Handle new argument 'w'.
25426 (read_graph_file): Assign ID to BBs.
25427 (output_branch_count): Display BB # if verbose flag is set.
25428 (output_lines): Likewise for arcs.
25429 (print_usage): Add '--verbose' option help.
25430 * doc/gcov.texi: Document --verbose (-w) option.
25432 2017-04-28 Martin Liska <mliska@suse.cz>
25434 * gcov.c (struct block_location_info): New struct.
25435 (process_file): Fill up the new structure.
25436 (read_graph_file): Replace usage of encoding by the newly added
25438 (add_line_counts): Likewise.
25439 (accumulate_line_counts): Remove usage of the union.
25440 (function_info::function_info): New function.
25441 (function_info::~function_info): Likewise.
25442 (process_file): Call delete instead of release_function.
25443 (release_function): Release the function.
25444 (release_structures): Call delete instead of release_function.
25445 (solve_flow_graph): Replace usage of num_blocks.
25446 (find_exception_blocks): Likewise.
25447 (output_lines): Fix GNU coding style.
25449 2017-04-28 Martin Liska <mliska@suse.cz>
25452 * coverage.c (coverage_remove_note_file): New function.
25453 * coverage.h: Declare the function.
25454 * toplev.c (finalize): Clean if an error has been seen.
25456 2017-04-28 Martin Liska <mliska@suse.cz>
25458 PR gcov-profile/80031
25459 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
25460 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
25461 * gcov.c (read_graph_file): Read just number of blocks.
25462 * profile.c (branch_prob): Do not stream 0 flags per a basic
25465 2017-04-28 Martin Liska <mliska@suse.cz>
25467 * gcov-dump.c (tag_*): Add new argument to declarations.
25468 (dump_gcov_file): Likewise.
25469 (tag_blocks): Add and use new argument depth.
25470 (tag_arcs): Likewise.
25471 (tag_lines): Likewise.
25472 (tag_counters): Likewise.
25473 (tag_summary): Likewise.
25474 (dump_working_sets): Use depth to do a proper indentation.
25476 2017-04-28 Jakub Jelinek <jakub@redhat.com>
25479 * cgraph.h (symtab_node::debug_symtab): No longer inline.
25480 * symtab.c (symtab_node::debug_symtab): Move definition here.
25482 2017-04-28 Richard Biener <rguenther@suse.de>
25484 * lto-streamer.h (LTO_major_version): Bump to 7.
25486 2017-04-28 Richard Biener <rguenther@suse.de>
25488 * tree-vrp.c (assert_info): New struct.
25489 (add_assert_info): New helper.
25490 (register_edge_assert_for_2): Refactor to add asserts to a vector
25492 (register_edge_assert_for_1): Likewise.
25493 (register_edge_assert_for): Likewise.
25494 (finish_register_edge_assert_for): New helper actually registering
25495 asserts where live on edge.
25496 (find_conditional_asserts): Adjust.
25497 (find_switch_asserts): Likewise.
25498 (evrp_dom_walker::try_find_new_range): Generalize.
25499 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
25501 2017-04-27 Marek Polacek <polacek@redhat.com>
25504 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
25505 arg10 and arg11 to itype.
25507 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
25509 * doc/extend.texi (Object Size Checking): Improve grammar.
25511 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
25514 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
25515 that the logic for permitting reciprocal estimates matches that
25518 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25521 * tree.c (type_cache_hasher::equal): Only compare
25522 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
25523 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
25524 non-aggregate element types.
25525 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
25526 about the flag on ARRAY_TYPEs in the comment, formatting fix.
25528 2017-04-27 Richard Biener <rguenther@suse.de>
25530 PR middle-end/80533
25531 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
25532 stripping ARRAY_REFs from MEM_EXPR make sure we're not
25533 keeping a reference to a trailing array.
25535 2017-04-27 Richard Biener <rguenther@suse.de>
25537 PR middle-end/80539
25538 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
25539 being in loop-closed SSA form conservatively.
25540 (chrec_fold_multiply_poly_poly): Likewise.
25542 2017-04-27 Tamar Christina <tamar.christina@arm.com>
25544 PR middle-end/79665
25545 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
25546 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
25548 2017-04-27 Jakub Jelinek <jakub@redhat.com>
25551 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
25552 (aarch64_function_arg_alignment): Return unsigned int again, but still
25553 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
25554 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25555 Don't emit -Wpsabi note.
25556 (aarch64_function_arg_boundary): Likewise.
25557 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25560 2017-04-26 Nathan Sidwell <nathan@acm.org>
25562 * tree.h (crc32_unsigned_n): Declare.
25563 (crc32_unsigned, crc32_unsigned): Make inline.
25564 * tree.c (crc32_unsigned_bits): Replace with ...
25565 (crc32_unsigned_n): ... this.
25566 (crc32_unsigned, crc32_byte): Remove.
25567 (crc32_string): Remove unnecessary braces.
25569 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25571 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
25572 * ipa-inline-analysis.c (MAX_TIME): Remove.
25573 (account_size_time): Use sreal for time.
25574 (dump_inline_summary): Update.
25575 (estimate_function_body_sizes): Update.
25576 (estimate_edge_size_and_time): Update.
25577 (estimate_calls_size_and_time): Update.
25578 (estimate_node_size_and_time): Update.
25579 (inline_merge_summary): Update.
25580 (inline_update_overall_summary): Update.
25581 (estimate_time_after_inlining): Update.
25582 (inline_read_section): Update.
25583 (inline_write_summary): Update.
25584 * ipa-inline.c (compute_uninlined_call_time): Update.
25585 (compute_inlined_call_time): Update.
25586 (recursive_inlining): Update.
25587 (inline_small_functions): Update.
25588 (dump_overall_stats): Update.
25589 * ipa-inline.h: Include sreal.h.
25590 (size_time_entry): Turn time to sreal.
25591 (inline_summary): Turn self_time nad time to sreal.
25593 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
25595 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
25597 (sreal::stream_out, sreal::stream_in): New.
25598 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
25600 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25602 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
25605 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
25608 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
25610 (dimode_scalar_chain::compute_convert_gain): Ditto.
25611 (dimode_scalar_chain::make_vector_copies): Ditto.
25612 (dimode_scalar_chain::convert_reg): Ditto.
25613 (dimode_scalar_chain::convert_insn): Ditto.
25614 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
25615 (VI248_AVX512BW_1): New mode iterator.
25616 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
25617 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
25620 2017-04-25 Martin Sebor <msebor@redhat.com>
25622 PR tree-optimization/80497
25623 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
25624 constants are representable in HOST_WIDE_INT.
25625 (parse_directive): Ditto.
25627 2017-04-25 Martin Sebor <msebor@redhat.com>
25630 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
25631 (new_zero_array): Adjust signature.
25632 (dom_info::dom_init): Used unsigned rather that size_t.
25633 (dom_info::dom_info): Same.
25635 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25636 Jakub Jelinek <jakub@redhat.com>
25639 * config/arm/arm.c: Include gimple.h.
25640 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25641 returns negative, increment ncrn only if it returned positive.
25642 (arm_needs_doubleword_align): Return int instead of bool,
25643 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
25644 members, but if there is any such non-FIELD_DECL
25645 > PARM_BOUNDARY aligned decl, return -1 instead of false.
25646 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
25647 returns negative, increment nregs only if it returned positive.
25648 (arm_setup_incoming_varargs): Likewise.
25649 (arm_function_arg_boundary): Emit -Wpsabi note if
25650 arm_needs_doubleword_align returns negative, return
25651 DOUBLEWORD_ALIGNMENT only if it returned positive.
25653 2017-04-25 Marek Polacek <polacek@redhat.com>
25656 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25657 first argument to type.
25659 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
25662 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
25663 type checks to test for compatibility instead of equality.
25665 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
25666 Jakub Jelinek <jakub@redhat.com>
25669 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
25671 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
25672 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
25673 the alignment computation, but return their maximum in warn_alignment.
25674 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
25675 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
25677 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
25678 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
25681 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25683 * config/arc/simdext.md (dmpyh): Fix typo.
25685 2017-04-25 Richard Biener <rguenther@suse.de>
25687 PR tree-optimization/80492
25688 * alias.c (compare_base_decls): Handle registers with asm
25689 specification conservatively.
25690 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
25691 compare_base_decls returning dont-know properly.
25693 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25695 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
25696 (legitimate_offset_address_p): New function.
25697 (arc_legitimate_address_p): Use above function.
25699 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25701 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
25703 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25705 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
25706 ACCH registers whenever they are available.
25708 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25710 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
25711 double regs fix when not used.
25713 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25715 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
25717 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
25718 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
25720 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25722 * config/arc/arc.c (arc_output_addsi): Check for h-register class
25723 when emitting short ADD instructions.
25725 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
25727 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
25729 (cmpsi_cc_c_insn): Likewise.
25730 (cbranchsi4_scratch): Compute proper instruction length using
25731 compact_hreg_operand.
25732 * config/arc/predicates.md (compact_hreg_operand): New predicate.
25734 2017-04-25 Richard Biener <rguenther@suse.de>
25736 PR middle-end/80509
25737 * passes.c (pass_manager::pass_manager): Initialize
25738 m_name_to_pass_map.
25740 2017-04-25 Richard Biener <rguenther@suse.de>
25742 PR tree-optimization/79201
25743 * tree-ssa-sink.c (statement_sink_location): Handle calls.
25745 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25748 * config/s390/vector.md: Split MEM->GPR vector moves for
25749 non-s_operand addresses.
25751 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25754 * config/s390/predicates.md (reload_const_wide_int_operand): New
25756 * config/s390/s390.md ("movti"): Remove d/P alternative.
25757 ("movti_bigconst"): New pattern definition.
25759 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25762 * s390-protos.h (s390_expand_cs_hqi): Removed.
25763 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
25764 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
25765 modes as well as CCZ1mode and CCZmode.
25766 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
25767 signature of s390_emit_compare_and_swap.
25768 (s390_expand_cs_hqi): Likewise, make static.
25769 (s390_expand_cs_tdsi): Generate an explicit compare before trying
25770 compare-and-swap, in some cases.
25771 (s390_expand_cs): Wrapper function.
25772 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
25774 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
25775 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
25776 patterns for small and large integers. Forbid symref memory operands.
25777 Move expander to s390.c. Require cc register.
25778 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
25779 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
25780 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
25781 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
25782 symref memory operands. Remove CC mode and call s390_match_ccmode
25784 ("atomic_exchange<mode>"): Allow and implement all integer modes.
25786 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25788 * config/s390/s390.md (define_peephole2): New peephole to help
25789 combining the load-and-test pattern with volatile memory.
25791 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
25793 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
25794 with CCZmode for TARGET_Z196.
25796 2017-04-25 Jakub Jelinek <jakub@redhat.com>
25798 PR rtl-optimization/80501
25799 * combine.c (make_compound_operation_int): Set subreg_code to SET
25800 even for AND with mask of the sign bit of mode.
25802 PR rtl-optimization/80500
25803 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
25804 sum's initial value.
25806 2017-04-25 Julian Brown <julian@codesourcery.com>
25807 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25809 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
25811 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
25813 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
25815 2017-04-25 Julian Brown <julian@codesourcery.com>
25816 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25818 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
25819 (thunderx2t99_sha): New Reservation.
25821 2017-04-25 Julian Brown <julian@codesourcery.com>
25822 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
25824 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
25825 type for 1-element load.
25827 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
25829 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
25831 2017-04-24 Martin Jambor <mjambor@suse.cz>
25833 PR tree-optimization/80293
25834 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
25835 char arrays not totally scalarizable if it is false.
25836 (analyze_all_variable_accesses): Pass correct value in the new
25837 parameter. Add a statistics counter.
25839 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
25841 PR middle-end/79931
25842 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
25844 2017-04-24 Richard Biener <rguenther@suse.de>
25846 PR tree-optimization/80494
25847 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
25848 out for complex types.
25850 2017-04-24 Richard Biener <rguenther@suse.de>
25852 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
25853 * tree-ssa-sccvn.c (print_scc): Print SCC size.
25854 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
25855 (DFS): Adjust and never fail.
25856 (sccvn_dom_walker::fail): Remove.
25857 (sccvn_dom_walker::before_dom_children): Adjust.
25858 (run_scc_vn): Likewise and never fail.
25859 * tree-ssa-pre.c (pass_pre::execute): Adjust.
25860 (pass_fre::execute): Likewise.
25862 2017-04-24 Richard Biener <rguenther@suse.de>
25864 PR tree-optimization/79725
25865 * tree-ssa-sink.c (statement_sink_location): Return whether
25866 failure reason was zero uses. Move that check later.
25867 (sink_code_in_bb): Deal with zero uses by removing the stmt
25870 2017-04-24 Richard Biener <rguenther@suse.de>
25873 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
25874 pointer-based references.
25876 2017-04-24 Richard Biener <rguenther@suse.de>
25879 * pass_manager.h (pass_manager::operator new): Remove.
25880 (pass_manager::operator delete): Likewise.
25881 * passes.c (pass_manager::operator new): Remove.
25882 (pass_manager::operator delete): Likewise.
25883 (pass_manager::pass_manager): Zero individual pass members.
25885 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
25888 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
25889 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
25890 Check "XEXP (src, 1)" operand here.
25891 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
25892 Check "XEXP (src, 1)" operand here.
25893 (dimode_scalar_chain::make_vector_copies): Detect count register
25894 of a shift instruction. Zero extend count register from QImode
25895 to DImode to satisfy vector shift pattern count operand predicate.
25896 Substitute vector shift count operand with a DImode copy.
25897 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
25900 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
25902 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
25903 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25904 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
25905 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
25906 (UNSPEC_NOREX_MEM): Remove definition.
25908 2017-04-21 Richard Biener <rguenther@suse.de>
25910 PR tree-optimization/79547
25911 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
25912 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
25913 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
25914 without any constraints.
25916 2017-04-21 Richard Biener <rguenther@suse.de>
25918 PR tree-optimization/78847
25919 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
25921 2017-04-21 Richard Biener <rguenther@suse.de>
25923 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
25924 (build_distinct_type_copy): Likewise.
25925 (build_variant_type_copy): Likewise.
25926 * tree.c (build_qualified_type): Pass down mem-stat info.
25927 (build_distinct_type_copy): Likewise.
25928 (build_variant_type_copy): Likewise.
25930 2017-04-21 Richard Biener <rguenther@suse.de>
25932 PR tree-optimization/80237
25933 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
25935 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
25936 for a simplified result.
25938 2016-04-21 Richard Biener <rguenther@suse.de>
25940 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
25941 sth as strict as a simple_iv but a chrec without symbols and an
25942 operand defined in the loop we are peeling (and not some subloop).
25943 (propagate_constants_for_unrolling): Propagate all constants.
25945 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25948 * config/i386/i386.c (print_reg): Remove assert for disalowed
25949 regno values, call output_operand_lossage instead.
25951 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25954 * config/i386/constraints.md (Yc): New register constraint.
25955 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25956 Use Yc constraint for alternative 2 of operand 0. Remove
25957 preferred_for_speed attribute.
25959 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
25961 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
25962 lastprivate clauses in SIMT case.
25964 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
25966 * doc/invoke.texi (-Wextra-semi): Document new warning option.
25968 2017-04-20 Richard Biener <rguenther@suse.de>
25970 PR tree-optimization/57796
25971 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
25972 as N scalar stores.
25973 (vect_model_load_cost): Cost gathers as N scalar loads.
25975 2017-04-20 Richard Biener <rguenther@suse.de>
25977 * ggc-page.c (ggc_allocated_p): Rename to ...
25978 (safe_lookup_page_table_entry): ... this and return the lookup
25980 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
25982 2017-04-20 Richard Biener <rguenther@suse.de>
25984 PR tree-optimization/80453
25985 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
25986 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
25987 from the conditions.
25988 (vn_phi_eq): Pass them down.
25989 (vn_phi_lookup): Record them.
25990 (vn_phi_insert): Likewise.
25992 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
25994 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
25995 uninitialized variable warning to avoid buffer overrun.
25997 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
26000 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
26001 is suppressed for '{ 0 }' in C.
26003 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26005 * BASE-VER: Set to 8.0.0.
26007 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
26009 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
26010 priority .init_array and .fini_array section with SECTION_NOTYPE
26013 2017-04-20 Jakub Jelinek <jakub@redhat.com>
26015 PR middle-end/80423
26016 * tree.h (build_array_type): Add typeless_storage default argument.
26017 * tree.c (type_cache_hasher::equal): Also compare
26018 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
26019 (build_array_type): Add typeless_storage argument, set
26020 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
26022 (build_nonshared_array_type): Adjust build_array_type_1 caller.
26023 (build_array_type): Likewise. Add typeless_storage argument.
26025 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26026 Jakub Jelinek <jakub@redhat.com>
26028 PR tree-optimization/80426
26029 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
26030 operation on symbolic operands, also compute the overflow for the
26031 invariant part when the operation degenerates into a negation.
26033 2017-04-19 Jakub Jelinek <jakub@redhat.com>
26036 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
26037 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
26040 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
26042 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
26045 * config/avr/avr.c (tree.h): Include it.
26046 (cgraph.h): Include it.
26047 (avr_encode_section_info): Don't warn for uninitialized progmem
26048 variable if it's just an alias.
26050 2017-04-19 Richard Biener <rguenther@suse.de>
26053 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
26054 when needed by AutoPGO.
26056 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
26059 * doc/lto.texi: Remove an extra 'that'.
26061 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
26063 PR rtl-optimization/80429
26064 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
26065 are only used in debug insns.
26067 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
26068 Vladimir Makarov <vmakarov@redhat.com>
26070 * config/sparc/predicates.md (input_operand): Add comment. Return
26071 true for any memory operand when LRA is in progress.
26072 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
26074 2017-04-18 Jeff Law <law@redhat.com>
26077 * mips.md ({return,simple_return}_internal): Do not overwrite
26080 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26082 PR tree-optimization/80443
26083 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
26084 instead of adding 1, subtract -1 and similarly instead of subtracting
26087 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
26089 PR rtl-optimization/80357
26090 * haifa-sched.c (tmp_bitmap): New variable.
26091 (model_recompute): Handle duplicate use records.
26092 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
26093 (free_global_sched_pressure_data): Free it.
26095 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
26098 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
26099 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
26100 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
26101 instead of SYSTEM_HEADER_DIR.
26103 2017-04-18 Jeff Law <law@redhat.com>
26105 PR middle-end/80422
26106 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
26107 predecessors after walking up the insn chain.
26109 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26112 * dwarf2out.c (modified_type_die): Try harder not to emit internal
26113 sizetype type into debug info.
26115 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
26118 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
26119 unneeded test for TARGET_UPPER_REGS_SF.
26120 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
26122 2017-04-18 Jakub Jelinek <jakub@redhat.com>
26125 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
26126 instead of gsi_after_labels.
26128 2017-04-18 Jeff Law <law@redhat.com>
26130 * regcprop.c (maybe_mode_change): Avoid creating copies of the
26134 2017-04-13 Jeff Law <law@redhat.com>
26135 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26136 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26138 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
26141 * config/avr/avr.c (intl.h): Include it.
26142 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
26144 2017-04-18 Martin Liska <mliska@suse.cz>
26146 PR gcov-profile/78783
26147 * gcov-tool.c (gcov_output_files): Validate that destination
26148 file is either removed by the tool or by a user.
26150 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26151 Guy Benyei <guybe@mellanox.com>
26153 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
26154 block, and do not negate it, the stored id is already negative.
26156 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
26158 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
26160 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26163 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
26164 masks of options that should be turned off if the VSX vector
26165 options are turned off.
26166 (OTHER_P8_VECTOR_MASKS): Likewise.
26167 (OTHER_VSX_VECTOR_MASKS): Likewise.
26168 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
26169 rs6000_disable_incompatible_switches to validate no type switches
26171 (rs6000_incompatible_switch): New function to disallow turning on
26172 other vector options if -mno-vsx, -mno-power8-vector, or
26173 -mno-power9-vector are specified.
26175 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26177 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
26179 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26181 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
26182 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
26183 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
26184 (ARG_POINTER_CFA_OFFSET): Likewise.
26186 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
26188 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
26189 conditions to take advantage of various optimizations.
26191 2017-04-13 Jeff Law <law@redhat.com>
26193 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
26194 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
26195 (zero_extendsidi2_dext): Likewise.
26197 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26200 * fold-const.c (fold_ternary_loc): Revert
26201 use op0 instead of fold_convert_loc (loc, type, arg0) part of
26204 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
26206 PR rtl-optimization/80343
26207 * lra-remat.c (update_scratch_ops): Assign original hard reg to
26208 new scratch pseudo.
26210 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
26213 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
26214 to ubsan_encode_value.
26216 2017-04-13 Jeff Law <law@redhat.com>
26218 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
26219 appearing in DEBUG_INSNs.
26221 2017-04-13 Martin Liska <mliska@suse.cz>
26223 PR gcov-profile/80413
26224 * gcov-io.c (gcov_write_string): Copy to buffer just when
26225 allocated size is greater than zero.
26227 2017-04-13 Jakub Jelinek <jakub@redhat.com>
26230 * dwarf2out.c (decls_for_scope): Ignore declarations of
26231 current_function_decl in BLOCK_NONLOCALIZED_VARS.
26233 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
26236 * ipa-visibility.c (non_local_p): Fix typos.
26237 (localize_node): When localizing symbol in same comdat group,
26238 dissolve the group only when we know external symbols are going
26240 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
26242 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26244 PR tree-optimization/79390
26245 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
26246 order does not result in usable sequence, retry with reversed operand
26252 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
26253 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
26254 op0 instead of fold_convert_loc (loc, type, arg0).
26256 2017-04-12 Jeff Law <law@redhat.com>
26258 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
26259 has a delay slot in the generated code.
26261 * config/cris/cris.md (cris_preferred_reload_class): Return
26262 GENNONACR_REGS rather than GENERAL_REGS.
26264 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26267 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
26268 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
26269 signedness of the result type.
26271 2017-04-12 Richard Biener <rguenther@suse.de>
26272 Jeff Law <law@redhat.com>
26274 PR tree-optimization/80359
26275 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
26276 trim stores to TARGET_MEM_REFs.
26278 2017-04-12 Richard Biener <rguenther@suse.de>
26280 PR tree-optimization/79390
26281 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
26282 threading case even more.
26284 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
26287 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
26288 for quad_address_p for TImode, instead of just not indexed_address.
26290 2017-04-12 Richard Biener <rguenther@suse.de>
26291 Bernd Edlinger <bernd.edlinger@hotmail.de>
26293 PR middle-end/79671
26294 * alias.c (component_uses_parent_alias_set_from): Handle
26295 TYPE_TYPELESS_STORAGE.
26296 (get_alias_set): Likewise.
26297 * tree-core.h (tree_type_common): Add typeless_storage flag.
26298 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
26299 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
26300 for types containing members with TYPE_TYPELESS_STORAGE.
26301 (place_field): Likewise.
26302 (layout_type): Likewise for ARRAY_TYPE.
26303 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
26304 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
26305 TYPE_TYPELESS_STORAGE.
26306 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
26308 2017-04-12 Jakub Jelinek <jakub@redhat.com>
26311 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
26312 first argument to type.
26314 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26318 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
26319 CONST0_RTX (mode) rather than const0_rtx where appropriate.
26320 (rs6000_expand_binop_builtin): Likewise.
26321 (rs6000_expand_ternop_builtin): Likewise; also add missing
26322 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
26323 vshasigma built-ins.
26324 * doc/extend.texi: Document that vec_xxpermdi's third argument
26325 must be a constant.
26327 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
26329 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
26330 Use shift_const cost parameter when calculating gain of STV shifts.
26332 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
26334 PR rtl-optimization/70478
26335 * lra-constraints.c (process_alt_operands): Check memory for
26336 disfavoring memory insn operand.
26338 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26340 PR middle-end/80100
26341 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
26342 left shift in unsigned HOST_WIDE_INT type.
26344 PR rtl-optimization/80385
26345 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
26346 (not (neg X)) into (plus X -1) for complex or non-integral modes.
26349 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
26350 if they have any depend clauses.
26352 2017-04-11 Martin Liska <mliska@suse.cz>
26355 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
26356 * ipa-split.c (split_function): Create a local comdat symbol
26357 if caller is in a comdat group.
26359 2017-04-11 Martin Liska <mliska@suse.cz>
26362 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
26365 2017-04-11 Martin Sebor <msebor@redhat.com>
26367 PR middle-end/80364
26368 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
26369 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
26371 (directive::set_width, directive::set_precision, format_character):
26373 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
26376 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
26379 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
26380 conflict, set target->arch_name instead of target->cpu_name.
26382 2017-04-11 Richard Biener <rguenther@suse.de>
26384 PR tree-optimization/80374
26385 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
26386 build_zero_cst, remove fold_convertible_p check again.
26388 2017-04-11 Martin Liska <mliska@suse.cz>
26391 * ubsan.c (instrument_object_size): Do not instrument register
26394 2017-04-11 Jakub Jelinek <jakub@redhat.com>
26397 * config/i386/i386-builtin-types.def
26398 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
26399 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
26400 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
26401 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
26402 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
26403 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
26404 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
26405 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
26406 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
26407 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
26408 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
26409 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
26410 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
26411 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
26412 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
26413 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
26414 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
26415 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
26416 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
26417 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
26418 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
26419 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
26420 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
26421 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
26422 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
26423 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
26424 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
26425 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
26426 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
26427 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
26428 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
26429 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
26430 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
26431 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
26432 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
26433 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
26434 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
26435 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
26436 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
26437 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
26438 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
26439 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
26440 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
26441 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
26442 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
26443 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
26445 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
26446 flag to second_arg_count, handle 4 argument function type _COUNT
26447 aliases, handle second_arg_count on second argument rather than last.
26449 2017-04-10 Jeff Law <law@redhat.com>
26451 PR tree-optimization/80374
26452 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
26453 record anything if we can not convert integer_zero_node to the
26456 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
26459 * config/rs6000/rs6000.c (rs6000_option_override_internal):
26460 Enhance special handling given to the TARGET_P9_MINMAX option in
26461 relation to certain other options.
26463 2017-04-10 Bin Cheng <bin.cheng@arm.com>
26465 PR tree-optimization/80153
26466 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
26467 remove POINTER_PLUS_EXPR's base part directly, rather than through
26470 2017-04-10 Richard Biener <rguenther@suse.de>
26471 Bin Cheng <bin.cheng@arm.com>
26473 PR tree-optimization/80153
26474 * tree-affine.c (aff_combination_to_tree): Get base pointer from
26475 the first element of pointer type aff_tree. Build result expr in
26477 (add_elt_to_tree): Convert to type unconditionally. Remove other
26478 fold_convert calls.
26479 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
26480 (rewrite_use_nonlinear_expr): Check invariant using iv information.
26482 2017-04-10 Richard Biener <rguenther@suse.de>
26484 * tree-ssa-structalias.c (find_func_aliases): Properly handle
26487 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
26489 PR rtl-optimization/70478
26490 * lra-constraints.c (curr_small_class_check): New.
26491 (update_and_check_small_class_inputs): New.
26492 (process_alt_operands): Update curr_small_class_check. Disfavor
26493 alternative insn memory operands. Check available regs for small
26496 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26499 * config/mips/mips.opt (-mvirt): Update description.
26500 * doc/invoke.texi (-mvirt): Likewise.
26502 2017-04-10 Richard Biener <rguenther@suse.de>
26504 PR middle-end/80362
26505 * fold-const.c (fold_binary_loc): Look at unstripped ops when
26506 looking for NEGATE_EXPR in -A / -B to A / B folding.
26508 2017-04-10 Martin Liska <mliska@suse.cz>
26510 PR gcov-profile/80224
26511 * gcov.c (print_usage): Fix usage string.
26512 (get_gcov_intermediate_filename): Remove.
26513 (output_gcov_file): Use both for normal and intermediate format.
26514 (generate_results): Do not initialize special file for
26515 intermediate format.
26517 2017-04-10 Richard Biener <rguenther@suse.de>
26519 PR tree-optimization/80304
26520 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
26523 2017-04-10 Nathan Sidwell <nathan@acm.org>
26526 * config/rs6000/rs6000.c (rs6000_vector_type): New.
26527 (rs6000_init_builtins): Use it.
26529 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26531 * config/arm/arm.md (<mrc>): Add mode to SET source.
26532 (<mrrc>): Likewise.
26534 2017-04-10 Richard Biener <rguenther@suse.de>
26536 PR middle-end/80344
26537 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
26539 2017-04-10 Jakub Jelinek <jakub@redhat.com>
26542 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
26543 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
26544 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
26545 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
26546 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
26547 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
26548 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
26549 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
26550 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
26551 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
26552 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
26553 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
26554 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
26555 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
26556 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
26557 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
26558 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
26559 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
26560 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
26561 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
26562 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
26563 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
26564 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
26566 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
26568 PR rtl-optimization/70478
26569 * lra-constraints.c: Reverse the last patch.
26571 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
26573 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
26574 Add comment for WCHAR_T.
26576 2017-04-08 Martin Liska <mliska@suse.cz>
26579 2017-04-07 Martin Liska <mliska@suse.cz>
26582 * ipa-split.c (split_function): Add function part to a same comdat
26585 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26588 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
26590 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
26592 * rs6000/rs6000.c (vec_load_pendulum): Rename...
26593 (vec_pairing): ...to this.
26594 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
26595 (rs6000_sched_init): Adjust for name change.
26596 (struct rs6000_sched_context): Likewise.
26597 (rs6000_init_sched_context): Likewise.
26598 (rs6000_set_sched_context): Likewise.
26600 2017-04-07 Jakub Jelinek <jakub@redhat.com>
26606 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
26608 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
26609 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
26610 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
26612 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
26614 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
26616 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26618 PR rtl-optimization/70703
26619 * ira-color.c (update_conflict_hard_regno_costs): Use
26620 int64_t instead of HOST_WIDE_INT.
26622 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
26624 PR rtl-optimization/70478
26625 * lra-constraints.c (process_alt_operands): Disfavor alternative
26626 insn memory operands.
26628 2017-04-07 Jeff Law <law@redhat.com>
26630 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
26631 CALL and NOTE_INSN_CALL_ARG_LOCATION.
26633 2017-04-07 Martin Liska <mliska@suse.cz>
26636 * config/aarch64/aarch64.c (aarch64_process_target_attr):
26637 Show error message instead of an ICE.
26639 2017-04-07 Martin Liska <mliska@suse.cz>
26642 * ipa-split.c (split_function): Add function part to a same comdat
26645 2017-04-07 Richard Biener <rguenther@suse.de>
26647 PR middle-end/80341
26648 * tree.c (get_unwidened): Also handle ! for_type case for
26650 * convert.c (do_narrow): Split out from ...
26651 (convert_to_integer_1): ... here. Do not pass final truncation
26652 type to get_unwidened for TRUNC_DIV_EXPR.
26654 2017-04-07 Richard Biener <rguenther@suse.de>
26656 * tree-affine.c (wide_int_ext_for_comb): Take type rather
26658 (aff_combination_const): Adjust.
26659 (aff_combination_scale): Likewise.
26660 (aff_combination_add_elt): Likewise.
26661 (aff_combination_add_cst): Likewise.
26662 (aff_combination_convert): Likewise.
26663 (add_elt_to_tree): Likewise. Remove unused argument.
26664 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
26666 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
26668 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
26670 * config/arm/arm.c (arm_default_short_enums): Use
26671 ARM_DEFAULT_SHORT_ENUMS.
26672 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
26674 2017-04-06 Jakub Jelinek <jakub@redhat.com>
26677 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
26678 members with redundant out-of-class redeclaration.
26680 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26683 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
26684 * config/i386/i386.md (*zero_extendsidi2):
26685 Add (?*x,*x) and (?*v,*v) alternatives.
26687 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
26690 * config/i386/i386.c (ix86_expand_builtin)
26691 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
26692 mode from insn data. Convert operands to insn operand mode.
26693 Copy operands that don't satisfy insn predicate to a register.
26695 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
26697 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
26700 2017-04-06 Richard Biener <rguenther@suse.de>
26702 PR tree-optimization/80334
26703 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
26704 preserve alignment of accesses.
26706 2017-04-06 Richard Biener <rguenther@suse.de>
26708 PR tree-optimization/80262
26709 * tree-sra.c (build_ref_for_offset): Preserve address-space
26711 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
26712 Drop useless address-space information on MEM_REF offsets.
26714 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
26716 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
26718 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26720 PR rtl-optimization/70703
26721 * ira-color.c (update_conflict_hard_regno_costs): Use
26722 HOST_WIDE_INT instead of long.
26724 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
26727 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
26728 not defined for x86_64 target. Add -mmmx target option when __SSE2__
26730 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
26731 for x86_64 target. Handle -m3dnowa option.
26733 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
26735 PR rtl-optimization/70703
26736 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
26737 (update_conflict_hard_regno_costs): Use long instead of unsigned
26738 arithmetic for cost calculation.
26740 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26741 Bernd Edlinger <bernd.edlinger@hotmail.de>
26744 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
26747 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
26750 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
26751 ptr_mode with Pmode throughout.
26752 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
26753 into probe_stack_range and use DImode.
26755 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
26758 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
26759 call_eh_return is true.
26761 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26763 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
26764 Initialize last_match_fntype_index.
26766 2017-04-05 Jakub Jelinek <jakub@redhat.com>
26769 * tree-nvr.c: Include internal-fn.h.
26770 (pass_return_slot::execute): Ignore internal calls without
26773 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26774 Richard Biener <rguenther@suse.de>
26777 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
26778 captures used multiple times, except for the last use.
26779 * generic-match-head.c: Include gimplify.h.
26781 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26783 PR tree-optimization/79390
26784 * target.h (struct noce_if_info): Declare.
26785 * targhooks.h (default_noce_conversion_profitable_p): Declare.
26786 * target.def (noce_conversion_profitable_p): New target hook.
26787 * ifcvt.h (struct noce_if_info): New type, moved from ...
26788 * ifcvt.c (struct noce_if_info): ... here.
26789 (noce_conversion_profitable_p): Renamed to ...
26790 (default_noce_conversion_profitable_p): ... this. No longer
26792 (noce_try_store_flag_constants, noce_try_addcc,
26793 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
26794 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
26795 instead of noce_conversion_profitable_p.
26796 * config/i386/i386.c: Include ifcvt.h.
26797 (ix86_option_override_internal): Don't override
26798 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
26799 (ix86_noce_conversion_profitable_p): New function.
26800 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
26801 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
26802 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
26803 * doc/tm.texi: Regenerated.
26805 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26807 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
26810 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
26813 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
26814 instructions for small multiply cores.
26816 2017-04-04 Jeff Law <law@redhat.com>
26818 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
26820 (mips_expand_vec_perm_const): Initialize elements in orig_perm
26821 that are not set by the loop over the elements.
26823 2017-04-04 Jakub Jelinek <jakub@redhat.com>
26826 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
26827 int mode, convert_modes it to mode as unsigned, otherwise use
26828 lowpart_subreg to mode rather than SImode.
26829 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
26830 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
26831 Use DImode instead of SImode for the shift count operand.
26832 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
26835 2017-04-04 Richard Biener <rguenther@suse.de>
26837 PR middle-end/80281
26838 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
26839 arithmetic done for the negate or the plus. Simplify.
26840 (A - (-B) -> A + B): Likewise.
26841 * fold-const.c (split_tree): Make sure to not negate pointers.
26843 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
26845 PR rtl-optimization/60818
26846 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
26847 a compare of comparisons with the thing compared if this results
26848 in a different machine mode.
26850 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
26852 * alias.c (base_alias_check): Fix typo in comment.
26853 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
26854 * cgraphunit.c (symbol_table::compile): Likewise.
26855 * collect2.c (maybe_run_lto_and_relink): Likewise.
26856 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
26857 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
26858 * config/avr/avr.c (avr_map_op_t): Likewise.
26859 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
26860 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
26861 * config/epiphany/epiphany.md (movcc): Likewise.
26862 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
26863 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
26865 * config/mips/mips.c (mips_save_restore_reg): Likewise.
26866 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
26867 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
26868 * config/sh/sh.c (sh_rtx_costs): Likewise.
26869 * fold-const.c (fold_truth_andor): Likewise.
26870 * genautomata.c (collapse_flag): Likewise.
26871 * gengtype.h (struct type::u::s): Likewise.
26872 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
26873 * input.c (FORMAT_AMOUNT): Likewise.
26874 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
26875 (known_aggs_to_agg_replacement_list): Likewise.
26876 * ipa-inline-analysis.c: Likewise.
26877 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
26878 * ipa-polymorphic-call.c
26879 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
26880 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
26881 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
26883 * modulo-sched.c (apply_reg_moves): Likewise.
26884 * omp-expand.c (build_omp_regions_1): Likewise.
26885 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
26886 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
26887 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
26888 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
26889 * value-prof.c: Likewise.
26890 * var-tracking.c (val_reset): Likewise.
26892 2017-04-03 Richard Biener <rguenther@suse.de>
26894 PR tree-optimization/80275
26895 * fold-const.c (split_address_to_core_and_offset): Handle
26898 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
26900 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
26901 descriptors is at least equal to that of functions.
26903 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26905 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
26907 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
26910 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
26911 (mov<IMOD4:mode>): New expander.
26912 (*mov<IMOD4:mode>_internal): New insn and split pattern.
26914 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
26916 PR rtl-optimization/79405
26917 * fwprop.c (propagations_left): New variable.
26918 (forward_propagate_into): Decrement it.
26919 (fwprop_init): Initialize it.
26920 (fw_prop): If the variable has reached zero, stop propagating.
26921 (fwprop_addr): Ditto.
26923 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26926 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
26927 a FUNCTION_DECL, pass it as decl instead of origin to
26930 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
26932 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
26935 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
26938 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
26939 TARGET_VSX_SMALL_INTEGER.
26941 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26943 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
26944 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
26946 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
26948 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
26949 extraction from odd-numbered MSA register.
26951 2017-03-31 Jakub Jelinek <jakub@redhat.com>
26953 PR middle-end/80173
26954 * expmed.c (store_bit_field_1): Don't attempt to create
26955 a word subreg out of hard registers wider than word if they
26956 have HARD_REGNO_NREGS of 1 for their mode.
26958 PR middle-end/80163
26959 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
26960 conversions to integer types wider than word and pointer.
26963 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
26964 (rtx_equal_for_cselib_p): Pass 0 to it.
26965 * cselib.c (cselib_hasher::equal): Likewise.
26966 (rtx_equal_for_cselib_1): Add depth argument. If depth
26967 is 128, don't look up VALUE locs and punt. Increment
26968 depth in recursive calls when walking VALUE locs.
26970 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
26972 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
26973 (make_gcov_file_name): Use the canonical path name for generating
26975 (read_line): Fix handling of files with ascii null bytes.
26977 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
26979 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
26980 to initialise a vector register instead
26981 of using a const_int.
26983 2017-03-30 Jakub Jelinek <jakub@redhat.com>
26985 PR translation/80189
26986 * gimplify.c (omp_default_clause): Use %qs instead of %s in
26987 diagnostic messages.
26989 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
26992 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
26993 (dfp_diex_<mode>): Update mode of operand 1.
26994 * doc/extend.texi (dxex, dxexq): Document change to return type.
26995 (diex, diexq): Document change to argument type.
26997 2017-03-30 Martin Jambor <mjambor@suse.cz>
27000 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
27001 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
27002 it reflects the signature changes performed at the callee side.
27003 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
27004 to cgraph_build_function_type_skip_args.
27005 (build_function_decl_skip_args): Adjust call to the above function.
27007 2017-03-30 Jakub Jelinek <jakub@redhat.com>
27010 * config/i386/sse.md
27011 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
27012 register as dest whenever it is a MEM not rtx_equal_p to the
27013 corresponding dup operand, and when forcing into reg move the
27014 reg into the memory afterwards.
27015 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
27016 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
27017 for the force_reg mode.
27018 (avx512vl_vextractf128<mode>): Use register as dest either
27019 always when a MEM, or when it is a MEM not rtx_equal_p to the
27020 corresponding dup operand, or even not when it is a CONST_VECTOR
27021 depending on the mode and lo vs. hi.
27022 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
27024 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
27025 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
27026 Likewise. Require that operands[2] is even.
27027 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
27028 Remove extraneous parens. Require that operands[2] is a multiple
27030 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
27031 operands[0] is a MEM if <mask_applied>, the predicates/constraints
27032 disallow memory then.
27034 2017-03-30 Richard Biener <rguenther@suse.de>
27036 PR tree-optimization/77498
27037 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
27038 to non-constants over backedges.
27040 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
27042 PR rtl-optimization/80233
27043 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
27044 as last_combined_insn. Do not test for BARRIER_P separately.
27046 2017-03-29 Andreas Schwab <schwab@suse.de>
27049 * calls.c (prepare_call_address): Convert funexp to Pmode before
27050 copying to temp reg.
27052 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27054 PR tree-optimization/80158
27055 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
27056 Handle possible future case of more than one alternate
27058 (replace_rhs_if_not_dup): Likewise.
27059 (replace_one_candidate): Likewise.
27061 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
27063 PR rtl-optimization/80193
27064 * ira.c (ira): Do not check allocation for LRA.
27066 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
27068 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
27069 (nvptx_output_simt_exit): Declare.
27070 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
27071 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
27072 (init_softstack_frame): Move initialization of crtl->is_leaf to...
27073 (nvptx_declare_function_name): ...here. Emit declaration of local
27074 memory space buffer for omp_simt_enter insn.
27075 (nvptx_output_unisimt_switch): New.
27076 (nvptx_output_softstack_switch): New.
27077 (nvptx_output_simt_enter): New.
27078 (nvptx_output_simt_exit): New.
27079 * config/nvptx/nvptx.h (struct machine_function): New fields
27080 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
27081 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
27082 (UNSPECV_SIMT_EXIT): Ditto.
27083 (omp_simt_enter_insn): New insn.
27084 (omp_simt_enter): New expansion.
27085 (omp_simt_exit): New insn.
27086 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
27088 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
27089 (expand_GOMP_SIMT_ENTER_ALLOC): New.
27090 (expand_GOMP_SIMT_EXIT): New.
27091 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
27092 (GOMP_SIMT_ENTER_ALLOC): Ditto.
27093 (GOMP_SIMT_EXIT): Ditto.
27094 * target-insns.def (omp_simt_enter): New insn.
27095 (omp_simt_exit): Ditto.
27096 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
27098 (lower_rec_simd_input_clauses): Implement SIMT privatization.
27099 (lower_rec_input_clauses): Likewise.
27100 (lower_lastprivate_clauses): Handle SIMT privatization.
27102 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
27103 (ompdevlow_adjust_simt_enter): New.
27104 (find_simtpriv_var_op): New.
27105 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
27106 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
27108 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
27109 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
27110 (copy_decl_for_dup_finish): Ditto.
27112 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
27114 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
27117 * config/i386/i386.c (ix86_option_override_internal): Always
27118 allow -mpreferred-stack-boundary=3 for 64-bit targets.
27120 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27122 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
27124 2017-03-28 Bin Cheng <bin.cheng@arm.com>
27126 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
27127 mark new edge's irreducible flag accordign to it.
27128 (vect_do_peeling): Check loop preheader edge's irreducible flag
27129 and pass it to function slpeel_add_loop_guard.
27131 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
27133 PR tree-optimization/80218
27134 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
27135 Update block frequencies and counts.
27137 2017-03-28 Richard Biener <rguenther@suse.de>
27139 PR tree-optimization/78644
27140 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
27141 of a simplification result we may not use it at all.
27143 2017-03-28 Richard Biener <rguenther@suse.de>
27146 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
27147 without arguments, generate default definition of a SSA name.
27149 2017-03-28 Richard Biener <rguenther@suse.de>
27151 PR middle-end/80222
27152 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
27153 TYPE_REF_CAN_ALIAS_ALL references.
27154 * fold-const.c (fold_indirect_ref_1): Likewise.
27156 2017-03-28 Martin Liska <mliska@suse.cz>
27159 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
27160 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
27162 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27163 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
27165 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
27166 (EXTRA_SPECS): Define.
27167 (SUBTARGET_EXTRA_SPECS): Likewise.
27168 (SUBTARGET_CPP_SPEC): Likewise.
27169 * config/arc/elf.h (EXTRA_SPECS): Renamed to
27170 SUBTARGET_EXTRA_SPECS.
27171 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
27173 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
27175 * config/arc/simdext.md (vst64_insn): Update pattern.
27176 (vld32wh_insn): Likewise.
27177 (vld32wl_insn): Likewise.
27178 (vld64_insn): Likewise.
27179 (vld32_insn): Likewise.
27181 2017-03-28 Marek Polacek <polacek@redhat.com>
27184 * fold-const.c (fold_comparison): Use protected_set_expr_location
27185 instead of SET_EXPR_LOCATION.
27187 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
27189 * tree.c (add_expr): Avoid name lookup warning.
27191 2017-03-27 Jeff Law <law@redhat.com>
27193 PR tree-optimization/80216
27194 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
27195 function name. Limit recursion depth.
27196 (record_temporary_equivalences): Corresponding changes.
27198 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
27200 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
27203 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27206 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
27208 * cfgcleanup.c (reg_note_cfa_p): New array.
27209 (insns_have_identical_cfa_notes): New function.
27210 (old_insns_match_p): Don't cross-jump in between /f
27211 and non-/f instructions. If both i1 and i2 are frame related,
27212 verify all CFA notes, their order and content.
27214 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
27217 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
27218 HImode and SImode with zero extend to DImode to one insn.
27219 (bswap<mode>2_extenddi): Likewise.
27220 (bswapsi2_extenddi): Likewise.
27221 (bswaphi2_extendsi): Likewise.
27222 (bswaphi2): Combine bswap HImode and SImode into one insn.
27223 Separate memory insns from swapping register.
27224 (bswapsi2): Likewise.
27225 (bswap<mode>2): Likewise.
27226 (bswaphi2_internal): Delete, no longer used.
27227 (bswapsi2_internal): Likewise.
27228 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
27229 store, and gpr<-gpr swap insns.
27230 (bswap<mode>2_store): Likewise.
27231 (bswaphi2_reg): Register only splitter, combine with the splitter.
27232 (bswaphi2 splitter): Likewise.
27233 (bswapsi2_reg): Likewise.
27234 (bswapsi2 splitter): Likewise.
27235 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
27236 the insns into load, store, and register/register insns.
27237 (bswapdi2_ldbrx): Likewise.
27238 (bswapdi2_load): Likewise.
27239 (bswapdi2_store): Likewise.
27240 (bswapdi2_reg): Likewise.
27242 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
27244 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
27245 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
27247 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
27250 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
27252 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27253 special handling for target option conflicts between dform
27254 options (-mpower9-dform, -mpower9-dform-vector,
27255 -mpower9-dform-scalar) and -mno-direct-move.
27257 2017-03-27 Richard Biener <rguenther@suse.de>
27259 PR tree-optimization/80181
27260 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
27262 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27264 * config/arc/predicates.md (move_double_src_operand): Replace the
27265 call to move_double_src_operand with a call to address_operand.
27267 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27269 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
27270 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
27271 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
27273 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27275 * config/arc/predicates.md (long_immediate_loadstore_operand):
27276 Consider scaled addresses cases.
27278 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
27280 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
27281 restored when in interrupt.
27282 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
27283 doesn't have delay slot.
27285 2017-03-27 Richard Biener <rguenther@suse.de>
27288 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
27289 inlined thunk clones.
27291 2017-03-27 Jakub Jelinek <jakub@redhat.com>
27294 * asan.c (instrument_derefs): Copy over last operand from
27295 original COMPONENT_REF to the new COMPONENT_REF with
27296 DECL_BIT_FIELD_REPRESENTATIVE.
27297 * ubsan.c (instrument_object_size): Likewise.
27299 2017-03-27 Richard Biener <rguenther@suse.de>
27301 PR tree-optimization/80170
27302 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
27303 sure DR/SCEV didnt fold in constants we do not see when looking
27304 at the reference base alignment.
27306 2017-03-27 Richard Biener <rguenther@suse.de>
27308 PR middle-end/80171
27309 * gimple-fold.c (fold_ctor_reference): Properly guard against
27310 NULL return value from canonicalize_constructor_val.
27312 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
27315 * config/i386/i386.c (ix86_expand_builtin)
27316 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
27317 flags reg setting and flags reg using instructions.
27318 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
27319 clobbering instructions to zero extend op2.
27321 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
27323 * doc/install.texi (Configuration) <--with-aix-soname>:
27324 Update link to AIX ld.
27326 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
27328 PR rtl-optimization/80160
27329 PR rtl-optimization/80159
27330 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
27331 reg_alternate_class into account.
27333 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
27336 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
27337 to consider in curr_insn_transform.
27339 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27341 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
27342 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
27343 and emit_mode_inner.
27345 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27347 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
27348 argument to the overloaded builtin variants. Use the new flag to
27349 deprecate certain builtin variants.
27350 * config/s390/s390-builtin-types.def: Add new builtin types.
27351 * config/s390/s390-builtins.h: Support new flags field for
27352 overloaded builtins.
27353 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
27354 (s390_macro_to_expand): Enable vector float data type.
27355 (s390_cpu_cpp_builtins_internal): Indicate support of the new
27356 builtins by incrementing the __VEC__ version number.
27357 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
27359 (s390_resolve_overloaded_builtin): Emit error messages depending
27360 on the builtin flags.
27361 * config/s390/s390.c (s390_expand_builtin): Support additional
27362 flags argument. Change error message to match the messages
27363 emitted in s390-c.c.
27364 * config/s390/s390.md: New UNSPEC_* constants.
27365 (op_type): Add new instruction types.
27366 * config/s390/vecintrin.h: Add new builtins and test data class
27368 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
27369 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
27370 (VEC_INEXACT, VEC_NOINEXACT): New constants.
27371 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
27372 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
27373 ("vec_mergel<mode>"): V_HW -> VEC_HW.
27375 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
27376 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
27377 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
27378 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
27380 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
27381 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
27382 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
27383 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
27385 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
27386 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
27387 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
27388 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
27389 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
27390 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
27391 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
27393 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
27394 ("vec_scatter_element<V_HW_4:mode>_DI")
27395 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
27396 ("vec_fpint<mode>", "vflls")
27397 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
27398 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
27399 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
27400 ("*vec_cmphe<mode>_cc"): ... these.
27402 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
27403 mode constant instead of magic value.
27405 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27407 * config/s390/s390.c (s390_expand_vec_compare): Support other
27408 vector floating point modes than just V2DF.
27409 (s390_expand_vcond): Likewise.
27410 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
27411 (s390_cannot_change_mode_class): Prevent mode changes between TF
27412 and V1TF in vector registers.
27413 * config/s390/s390.md (DF, SF): New mode attributes.
27414 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
27415 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
27416 SFmode support for VRs.
27417 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
27419 (VFT, VF_HW): New mode iterators.
27420 (vw, sdx): New mode attributes.
27421 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
27422 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
27423 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
27424 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
27425 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
27426 also the new vector floating point modes. Renaming to ...
27428 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
27429 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
27430 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
27431 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
27432 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
27433 ("vec_unordered<mode>"): ... these.
27435 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
27436 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
27437 ("*vec_extendv2df"): New insn definitions.
27439 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27441 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
27442 ("mulditi3_2", "*muldi3_sign"): New patterns.
27443 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
27444 rename the pattern definition.
27446 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27448 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
27450 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
27452 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27454 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
27455 instruction if possible.
27456 * config/s390/vector.md (vec_halfnumelts): New mode
27458 ("*vec_vllezlf<mode>"): New pattern.
27460 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27462 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
27463 ("popcountv4si2", "popcountv2di2"): Rename to ...
27464 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
27465 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
27467 ("popcount<mode>2_vxe"): New pattern.
27469 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27471 * common/config/s390/s390-common.c (processor_flags_table): Add
27473 * config.gcc: Add arch12.
27474 * config/s390/driver-native.c (s390_host_detect_local_cpu):
27475 Default to arch12 for unknown CPU model numbers.
27476 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
27477 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
27478 PROCESSOR_max sanity check.
27479 * config/s390/s390-opts.h (enum processor_type): Add
27481 * config/s390/s390.c (processor_table): Add arch12.
27482 (s390_expand_builtin): Add check for B_VXE flag.
27483 (s390_issue_rate): Add PROCESSOR_ARCH12.
27484 (s390_get_sched_attrmask): Likewise.
27485 (s390_get_unit_mask): Likewise.
27486 (s390_sched_score): Enable z13 scheduling for arch12.
27487 (s390_sched_reorder): Likewise.
27488 (s390_sched_variable_issue): Likewise.
27489 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
27491 (s390_tune_attr): Use z13 scheduling also for arch12.
27492 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
27493 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
27494 (TARGET_VXE_P): New macros.
27495 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
27496 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
27497 * config/s390/s390.opt: Add arch12 as processor_type.
27499 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27501 * config/s390/s390.md
27502 ("fixuns_truncdddi2", "fixuns_trunctddi2")
27503 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
27504 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
27506 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
27507 Rename expanders to ...
27509 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
27510 ("fixuns_truncdddi2_emu"): ... these.
27512 ("fixuns_trunc<mode>si2_emu"): New expander.
27514 ("*fixuns_truncdfdi2_z13"): Rename to ...
27515 ("*fixuns_truncdfdi2_vx"): ... this.
27517 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27519 * config/s390/2964.md: Remove the single element vector compare
27520 instructions which are no longer used.
27521 * config/s390/s390.c (s390_select_ccmode): Remove handling of
27523 (s390_canonicalize_comparison): Remove handling of DFmode
27525 (s390_expand_vec_compare_scalar): Remove function.
27526 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
27527 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
27529 ("*cmp<mode>_ccs"): Add wfcdb instruction.
27531 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27533 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
27535 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
27536 will anyway by matched by mov<mode>_64dfp.
27538 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27540 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
27541 vlef/vstef. Add missing operand to vleif.
27543 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27545 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
27546 pair for all vector types with 64 bit elements.
27547 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
27548 * config/s390/vector.md (V_HW_64): ... here.
27549 (V_128_NOSINGLE): New mode iterator.
27550 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
27551 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
27552 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
27553 ("*vec_load_pairv2di"): Change to ...
27554 ("*vec_load_pair<mode>"): ... this one.
27556 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27558 * config/s390/constraints.md: Add comments.
27559 (jKK): Reject element sizes > 8 bytes.
27560 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
27562 * config/s390/s390.md: Add the s_operand checks formerly in
27563 s390_split_ok_p to various splitters where they are still
27565 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
27566 for 128 bit vectors. Plus two splitters.
27568 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27570 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
27573 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27576 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
27577 error if the boundary argument is not constant.
27579 2017-03-24 Jakub Jelinek <jakub@redhat.com>
27581 PR rtl-optimization/80112
27582 * loop-doloop.c (doloop_condition_get): Don't check condition
27583 if cmp isn't SET with IF_THEN_ELSE src.
27585 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27587 PR tree-optimization/80158
27588 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
27589 replacing a candidate statement, also replace it for the
27590 candidate's alternate interpretation.
27591 (replace_rhs_if_not_dup): Likewise.
27592 (replace_one_candidate): Likewise.
27594 2017-03-24 Richard Biener <rguenther@suse.de>
27596 PR tree-optimization/80167
27597 * graphite-isl-ast-to-gimple.c
27598 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
27600 (translate_isl_ast_to_gimple::get_rename): Likewise.
27602 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
27604 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
27605 handling of certain combinations of target options, including the
27606 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
27607 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
27609 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27612 * config/arm/arm.md (*load_multiple): Add reload_completed to
27613 matching condition.
27615 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27616 Richard Biener <rguenth@suse.de>
27618 PR tree-optimization/79908
27619 PR tree-optimization/80136
27620 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27621 been cast away, gimplify_and_add suffices.
27623 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
27625 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
27627 2017-03-23 Richard Biener <rguenther@suse.de>
27629 PR tree-optimization/80032
27630 * gimplify.c (gimple_push_cleanup): Forced unconditional
27631 cleanups still have to go to the conditional_cleanups
27634 2017-03-22 Jakub Jelinek <jakub@redhat.com>
27636 PR tree-optimization/80072
27637 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
27639 (next_operand_entry_id): Change type to unsigned int.
27640 (sort_by_operand_rank): Make sure to return the right return value
27641 even if unsigned fields are bigger than INT_MAX.
27642 (struct oecount): Change cnt and id type to unsigned int.
27643 (oecount_hasher::equal): Formatting fix.
27644 (oecount_cmp): Make sure to return the right return value
27645 even if unsigned fields are bigger than INT_MAX.
27646 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
27649 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
27650 TREE_READONLY on result if writing it more than once.
27653 * doc/invoke.texi (-fsanitize=thread): Document that with
27654 -fnon-call-exceptions atomics are not able to throw
27658 * tsan.c: Include tree-eh.h.
27659 (instrument_builtin_call): Call maybe_clean_eh_stmt or
27660 maybe_clean_or_replace_eh_stmt where needed.
27661 (instrument_memory_accesses): Add cfg_changed argument.
27662 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
27663 if it returned true.
27664 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
27666 PR rtl-optimization/63191
27667 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
27668 wrapper function, moved the whole old content into ...
27669 (ix86_delegitimize_address_1): ... this. New inline function.
27670 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
27671 true as last argument instead of ix86_delegitimize_address.
27673 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27675 * config/aarch64/aarch64.c (generic_branch_cost): Copy
27676 cortexa57_branch_cost.
27678 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
27680 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
27682 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27685 * doc/md.texi (Constraints): Document wA constraint.
27686 * config/rs6000/constraints.md (wA): New.
27687 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
27688 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
27689 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
27690 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
27692 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
27695 * gimplify.c (is_oacc_declared): New function.
27696 (oacc_default_clause): Use it to set default flags for acc declared
27697 variables inside parallel regions.
27698 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
27699 declared variables.
27700 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
27701 declare attribute to any decl as necessary.
27703 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
27706 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
27707 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
27708 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
27709 (arm_arch_lpae): This.
27710 * config/arm/arm.c (arm_arch7ve): Rename into ...
27711 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
27712 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
27715 2017-03-22 Martin Liska <mliska@suse.cz>
27718 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
27719 error message instead of an ICE.
27721 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27723 * doc/extend.texi (6.11 Additional Floating Types): Revise.
27725 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
27727 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27729 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
27732 2017-03-21 Martin Sebor <msebor@redhat.com>
27734 * doc/extend.texi: Use "cannot" instead of "can't."
27735 * doc/hostconfig.texi: Same.
27736 * doc/install.texi: Same.
27737 * doc/invoke.texi: Same.
27738 * doc/loop.texi: Same.
27739 * doc/md.texi: Same.
27740 * doc/objc.texi: Same.
27741 * doc/rtl.texi: Same.
27742 * doc/tm.texi: Same.
27743 * doc/tm.texi.in: Same.
27744 * doc/trouble.texi: Same.
27746 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
27749 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
27750 (collect_checksum_attributes): Set it.
27751 (die_checksum_ordered): Use it.
27753 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27755 PR tree-optimization/79908
27756 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
27757 change: For a VA_ARG whose LHS has been cast away, use
27758 force_gimple_operand to construct the side effects.
27760 2017-03-21 David Malcolm <dmalcolm@redhat.com>
27762 PR translation/80001
27763 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
27764 more amenable to translation.
27765 (oacc_loop_auto_partitions): Likewise.
27767 2017-03-21 Marek Polacek <polacek@redhat.com>
27768 Martin Sebor <msebor@redhat.com>
27770 PR tree-optimization/80109
27771 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
27772 on INTEGRAL_TYPE_P.
27774 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27775 Segher Boessenkool <segher@kernel.crashing.org>
27778 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
27779 check reg_used_between_p between insn and one of succ or succ2
27780 depending on if succ is artificial insn not inserted into insn
27783 2017-03-21 Martin Liska <mliska@suse.cz>
27785 PR gcov-profile/80081
27786 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
27787 * doc/gcc.texi: Include gcov-dump stuff.
27788 * doc/gcov-dump.texi: New file.
27790 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
27792 PR rtl-optimization/79150
27793 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
27794 conditional jump, if the jump is the last insn of the loop.
27796 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27797 Richard Biener <rguenth@suse.de>
27799 PR tree-optimization/79908
27800 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
27801 been cast away, use force_gimple_operand to construct the side
27804 2017-03-21 Martin Liska <mliska@suse.cz>
27806 PR libfortran/79956
27807 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
27810 2017-03-21 Brad Spengler <spender@grsecurity.net>
27813 * plugin.c (htab_hash_plugin): New function.
27814 (add_new_plugin): Use it and adjust.
27815 (parse_plugin_arg_opt): Adjust.
27816 (init_one_plugin): Likewise.
27818 2017-03-21 Richard Biener <rguenther@suse.de>
27820 PR tree-optimization/80032
27821 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
27822 if set force the cleanup to happen unconditionally.
27823 (gimplify_target_expr): Push inserted clobbers with force_uncond
27824 to avoid them being removed by control-dependent DCE.
27826 2017-03-21 Richard Biener <rguenther@suse.de>
27828 PR tree-optimization/80122
27829 * tree-inline.c (copy_bb): Do not expans va-arg packs or
27830 va_arg_pack_len when the inlined call stmt requires pack
27832 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
27834 2017-03-21 Jakub Jelinek <jakub@redhat.com>
27837 * tsan.c (instrument_builtin_call): If the memory model argument
27838 is not a constant, assume it is valid.
27841 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
27844 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27846 PR rtl-optimization/79910
27847 * combine.c (can_combine_p): Do not allow combining an I0 or I1
27848 if its dest is used by an insn before I2 (other than the combined
27849 insns themselves, which are properly handled already).
27851 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
27854 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27856 * combine.c (record_used_regs): New static function.
27857 (try_combine): Handle situations where there is an additional
27858 instruction between I2 and I3 which needs to have a LOG_LINK
27862 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27864 * combine.c (try_combine): Delete redundant i1 test. Call
27865 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27867 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27870 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
27871 alternatives 13/14.
27873 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27875 PR tree-optimization/80054
27876 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
27877 the optimization if a PHI or any of its arguments is not dominated
27878 by the candidate's basis. Use gphi* rather than gimple* as
27880 (replace_profitable_candidates): Clean up a gimple* variable that
27881 should be a gphi* variable.
27883 2017-03-20 Martin Sebor <msebor@redhat.com>
27886 * doc/extend.texi (attribute constructor): Document present limitation.
27888 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
27891 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
27892 __POWER9_VECTOR__ #ifdef control, change template definition to
27893 use Power9-specific built-in function.
27894 (vec_any_eq): Likewise.
27895 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
27896 to control outcomes from this test.
27897 (vector_ae_<mode>p): For VEC_F modes, likewise.
27899 2017-03-20 Ian Lance Taylor <iant@google.com>
27901 * config/i386/i386.c (ix86_function_regparm): Save an extra
27902 register for -fsplit-stack with DECL_STATIC_CHAIN.
27904 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27907 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
27908 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
27910 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
27912 * config/riscv/riscv.c (riscv_print_operand): Use "fence
27914 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
27917 2017-03-20 Marek Polacek <polacek@redhat.com>
27920 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
27922 2017-03-20 Richard Biener <rguenther@suse.de>
27924 PR tree-optimization/80113
27925 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
27926 allocate extra SSA name for PHI def.
27927 (add_close_phis_to_outer_loops): Likewise.
27928 (add_close_phis_to_merge_points): Likewise.
27929 (copy_loop_close_phi_args): Likewise.
27930 (copy_cond_phi_nodes): Likewise.
27932 2017-03-20 Martin Liska <mliska@suse.cz>
27934 PR middle-end/79753
27935 * tree-chkp.c (chkp_build_returned_bound): Do not build
27936 returned bounds for a LHS that's not a BOUNDED_P type.
27938 2017-03-20 Martin Liska <mliska@suse.cz>
27942 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
27943 COMPLEX_CST and VECTOR_CST.
27945 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27948 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
27949 target operand. A new splitter adds the clobber statement in case
27950 the target operand is dead anyway.
27952 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
27954 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
27955 to age-old versions of binutils and glibc.
27957 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
27959 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
27961 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27963 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
27965 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
27967 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
27968 requirement for binutils 2.13.
27970 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
27972 * combine.c (try_combine): Delete redundant i1 test. Call
27973 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
27975 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
27977 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
27978 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
27980 <riscv64-*-elf>: Re-arrange section
27981 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
27982 <riscv32-*-linux>: Likewise.
27983 <riscv64-*-elf>: Likewise
27984 <riscv64-*-linux>: Likewise.
27986 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
27989 * aarch64.opt(verbose-cost-dump): Fix typo.
27991 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
27994 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
27995 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
27997 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
27999 * reload.c (find_reloads): When reloading a nonoffsettable address,
28000 use RELOAD_OTHER for it and its address reloads.
28002 PR rtl-optimization/79910
28003 * combine.c (record_used_regs): New static function.
28004 (try_combine): Handle situations where there is an additional
28005 instruction between I2 and I3 which needs to have a LOG_LINK
28008 2017-03-17 Jeff Law <law@redhat.com>
28010 PR tree-optimization/71437
28011 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
28012 conditional in the hash table first.
28013 (vrp_dom_walker::before_dom_children): Extract condition from
28014 ASSERT_EXPR. Record condition, its inverion and any implied
28015 conditions as well.
28017 2017-03-17 Marek Polacek <polacek@redhat.com>
28018 Markus Trippelsdorf <markus@trippelsdorf.de>
28020 PR tree-optimization/80079
28021 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
28024 2017-03-17 Richard Biener <rguenther@suse.de>
28026 PR middle-end/80075
28027 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
28028 Properly verify the LHS before the RHS possibly claims to be
28030 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
28033 2017-03-17 Martin Jambor <mjambor@suse.cz>
28035 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
28036 (List of -O2 options): Likewise.
28037 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
28040 2017-03-17 Tom de Vries <tom@codesourcery.com>
28042 * gcov-dump.c (print_usage): Print bug_report_url.
28044 2017-03-17 Richard Biener <rguenther@suse.de>
28046 PR middle-end/80050
28047 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
28048 (parser::peek): Likewise.
28050 2017-03-17 Richard Biener <rguenther@suse.de>
28052 PR tree-optimization/80048
28053 * sese.c (free_sese_info): Properly release rename_map and
28054 copied_bb_map elements.
28056 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
28058 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
28059 Add linked-list forward and backlinks. Insert on
28060 construction, remove on destruction.
28061 (class pass_store_merging): Add m_stores_head field.
28062 (pass_store_merging::terminate_and_process_all_chains):
28063 Iterate over m_stores_head list.
28064 (pass_store_merging::terminate_all_aliasing_chains):
28066 (pass_store_merging::execute): Check for debug stmts first.
28067 Push new chains onto the m_stores_head stack.
28069 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
28072 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
28073 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
28074 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
28076 2017-03-16 Jeff Law <law@redhat.com>
28078 PR tree-optimization/71437
28079 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
28080 member function. Implementation moved into after_dom_children
28081 member function and into the threader's thread_outgoing_edges
28083 (dom_opt_dom_walker::after_dom_children): Simplify by moving
28084 some code into new thread_outgoing_edges.
28085 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
28086 definition. Simplify marker handling (do it here). Assume we always
28087 have the available expression and the const/copies tables.
28088 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
28090 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
28091 * tree-vrp.c (equiv_stack): No longer file scoped.
28092 (vrp_dom_walker): New class.
28093 (vrp_dom_walker::before_dom_children): New member function.
28094 (vrp_dom_walker::after_dom_children): Likewise.
28095 (identify_jump_threads): Setup domwalker. Use it rather than
28096 walking edges in a random order by hand. Simplify setup/finalization.
28097 (finalize_jump_threads): Remove.
28098 (vrp_finalize): Do not call identify_jump_threads here.
28099 (execute_vrp): Do it here instead and call thread_through_all_blocks
28102 PR tree-optimization/71437
28103 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
28105 (simplify_stmt_for_jump_threading): Add basic_block argument. All
28107 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
28108 (dom_opt_dom_walker::thread_across_edge): Remove
28109 handle_dominating_asserts argument. All callers changed.
28110 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
28111 changes. Remove calls to lhs_of_dominating_assert. Other
28112 uses of handle_dominating_asserts turn into unconditional code
28113 (simplify_control_stmt_condition_1): Likewise.
28114 (simplify_control_stmt_condition): Likewise.
28115 (thread_through_normal_block, thread_across_edge): Likewise.
28116 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
28117 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
28118 object if it is not an SSA_NAME.
28119 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
28120 before calling into the VRP specific simplifiers.
28121 (identify_jump_threads): Remove handle_dominating_asserts
28124 2017-03-16 Jakub Jelinek <jakub@redhat.com>
28127 * tree-diagnostic.c (default_tree_printer): No longer static.
28128 * tree-diagnostic.h (default_tree_printer): New prototype.
28130 2017-03-16 Tamar Christina <tamar.christina@arm.com>
28132 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
28133 Change ins into fmov.
28135 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28137 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
28138 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
28139 Use h_con constraint for operand 1.
28140 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
28141 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
28143 2017-03-15 Jeff Law <law@redhat.com>
28145 PR tree-optimization/71437
28146 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
28147 (record_temporary_equivalences): Use it.
28149 PR tree-optimization/71437
28150 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
28151 tree-ssa-scopedtables.
28152 (lookup_avail_expr, build_and_record_new_cond): Likewise.
28153 (record_conditions, record_cond, vuse_eq): Likewise.
28154 (record_edge_info): Adjust to API tweak of record_conditions.
28155 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
28156 (record_temporary_equivalences, optimize_stmt): Likewise.
28157 (eliminate_redundant_computations): Likewise.
28158 (record_equivalences_from_stmt): Likewise.
28159 * tree-ssa-scopedtables.c: Include options.h and params.h.
28160 (vuse_eq): New function, moved from tree-ssa-dom.c
28161 (build_and_record_new_cond): Likewise.
28162 (record_conditions): Likewise. Accept vector of conditions rather
28163 than edge_equivalence structure for first argument.
28164 for the first argument.
28165 (avail_exprs_stack::lookup_avail_expr): New member function, moved
28166 from tree-ssa-dom.c.
28167 (avail_exprs_stack::record_cond): Likewise.
28168 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
28169 from tree-ssa-dom.c.
28170 (avail_exprs_stack): Add new member functions lookup_avail_expr
28172 (record_conditions): Declare.
28174 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
28177 * lra-constraints.c (process_alt_operands): Increase reject for
28178 reloading an input/output operand.
28180 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
28183 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
28184 insns to convert from signed/unsigned char/short to IEEE 128-bit
28186 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
28188 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
28191 * config/i386/i386.c (ix86_vector_duplicate_value): Create
28192 subreg of inner mode for values already in registers.
28194 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
28196 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
28197 iteration reg is used after the loop.
28199 2017-03-14 Martin Sebor <msebor@redhat.com>
28201 PR tree-optimization/79800
28202 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
28203 precision in negative-positive range.
28204 (format_floating): Call non-const overload with adjusted precision.
28206 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
28209 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
28212 2017-03-14 Martin Sebor <msebor@redhat.com>
28214 PR middle-end/80020
28215 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
28216 * builtins.def (aligned_alloc): Use it.
28219 * Makefile.in (GTFILES): Add calls.c.
28220 * calls.c: Include "gt-calls.h".
28222 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
28224 PR rtl-optimization/79728
28225 * regs.h (struct target_regs): New field
28226 x_contains_allocatable_regs_of_mode.
28227 (contains_allocatable_regs_of_mode): New macro.
28228 * reginfo.c (init_reg_sets_1): Initialize it, and change
28229 contains_reg_of_mode so it includes global regs as well.
28230 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
28231 rather than contains_regs_of_mode.
28233 2017-03-14 Martin Liska <mliska@suse.cz>
28235 * doc/invoke.texi: Document options that can't be combined with
28236 -fcheck-pointer-bounds.
28238 2017-03-14 Martin Liska <mliska@suse.cz>
28240 PR middle-end/79831
28241 * doc/invoke.texi (-Wchkp): Document the option.
28243 2017-03-14 Martin Liska <mliska@suse.cz>
28245 * Makefile.in: Install gcov-dump.
28247 2017-03-14 Martin Liska <mliska@suse.cz>
28249 * multiple_target.c (expand_target_clones): Bail out for
28250 an invalid attribute.
28252 2017-03-14 Richard Biener <rguenther@suse.de>
28254 * alias.c (struct alias_set_entry): Pack properly.
28255 * cfgloop.h (struct loop): Likewise.
28256 * cse.c (struct set): Likewise.
28257 * ipa-utils.c (struct searchc_env): Likewise.
28258 * loop-invariant.c (struct invariant): Likewise.
28259 * lra-remat.c (struct cand): Likewise.
28260 * recog.c (struct change_t): Likewise.
28261 * rtl.h (struct address_info): Likewise.
28262 * symbol-summary.h (function_summary): Likewise.
28263 * tree-loop-distribution.c (struct partition): Likewise.
28264 * tree-object-size.c (struct object_size_info): Likewise.
28265 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
28266 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
28267 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
28268 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
28269 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
28270 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
28271 (struct _stmt_vec_info): Likewise.
28273 2017-03-14 Martin Liska <mliska@suse.cz>
28276 * multiple_target.c (create_dispatcher_calls): Check that
28277 a target can create a function dispatcher.
28279 2017-03-14 Martin Liska <mliska@suse.cz>
28282 * multiple_target.c (expand_target_clones): Drop local.local
28283 flag for default implementation.
28285 2017-03-14 Richard Biener <rguenther@suse.de>
28287 PR tree-optimization/80030
28288 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
28290 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
28292 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
28293 gcc_fallthrough() instead of __attribute__((fallthrough));
28295 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28297 * doc/gcc.texi: Remove "up" link to (DIR).
28298 * doc/gccint.texi: Ditto.
28300 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
28302 * doc/install.texi (Specific) <avr>: Remove reference to
28305 2017-03-13 Jeff Law <law@redhat.com>
28307 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
28308 attribute rather than comments.
28310 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
28311 match_scratch operand is highest.
28313 2017-03-13 Martin Liska <mliska@suse.cz>
28315 PR middle-end/78339
28316 * ipa-pure-const.c (warn_function_noreturn): If the declarations
28317 is a CHKP clone, use original declaration.
28319 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28321 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
28322 (arc_conditional_register_usage): Use a different allocation order
28323 when optimizing for size.
28324 * common/config/arc/arc-common.c (arc_option_optimization_table):
28325 Section anchors default on when optimizing for size.
28327 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28329 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
28331 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28333 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
28334 * config/arc/arc.md (cpu_facility): Add cd variant.
28335 (*movqi_insn): Add code density variant.
28336 (*movhi_insn): Likewise.
28337 (*movqi_insn): Likewise.
28338 (*addsi3_mixed): Likewise.
28339 (subsi3_insn): Likewise.
28341 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28343 * config/arc/arc.md (movsi_cond_exec): Update constraint.
28345 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
28347 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
28348 expressions with MINUS and UNARY ops.
28350 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28353 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
28355 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
28356 between vec_select and vector argument.
28357 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
28358 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
28359 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
28360 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
28361 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
28362 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
28364 2017-03-13 Richard Biener <rguenther@suse.de>
28367 * params.def (vect-max-peeling-for-alignment): Fix typo.
28369 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28371 * doc/install.texi (Specific) <mips-*-*>: Remove description of
28372 issue that only occurred with binutils below 2.18.
28374 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28376 * doc/install.texi (Specific) <cris-axis-elf>: No longer
28377 refer to binutils 2.11/2.12 minimum.
28379 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
28381 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
28382 ftp.kernel.org and simplify binutils requirement.
28384 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
28386 * doc/invoke.texi (Warning Options): Fix spelling of link-time
28388 (Optimize Options): Ditto. Also remove redundancy.
28390 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28392 PR translation/79848
28393 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
28395 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
28396 to G_ to avoid double translation.
28398 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28400 PR translation/79923
28401 * auto-profile.c (get_combined_location): Convert leading
28402 character of diagnostics to lower case and remove trailing period.
28403 (read_profile): Likewise for various diagnostics.
28404 * config/arm/arm.c (arm_option_override): Remove trailing period
28405 from various diagnostics.
28406 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
28407 (msp430_expand_delay_cycles): Likewise.
28409 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28412 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
28413 full command-line argument, rather than just "str".
28414 (aarch64_validate_march): Likewise.
28415 (aarch64_validate_mtune): Likewise.
28417 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
28419 PR rtl-optimization/78911
28420 * lra-assigns.c (must_not_spill_p): New function.
28421 (spill_for): Use it.
28423 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28425 PR tree-optimization/79981
28426 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
28427 ATOMIC_COMPARE_EXCHANGE ifn result.
28428 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
28429 IFN_ATOMIC_COMPARE_EXCHANGE.
28431 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28434 * opts.c (parse_sanitizer_options): Add missing question mark to
28435 "did you mean" message.
28437 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28439 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
28441 (VMULEUH_UNS): Likewise.
28442 (VMULOUB_UNS): Likewise.
28443 (VMULOUH_UNS): Likewise.
28444 * config/rs6000/rs6000.c (builtin_function_type): Remove
28445 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
28447 2017-03-10 David Malcolm <dmalcolm@redhat.com>
28450 * read-rtl-function.c (function_reader::read_rtx_operand): Update
28451 x with result of extra_parsing_for_operand_code_0.
28452 (function_reader::extra_parsing_for_operand_code_0): Convert
28453 return type from void to rtx, returning x. When reading
28454 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
28455 larger size containing struct block_symbol.
28457 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
28459 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
28460 -mfloat128-hardware without -m64.
28462 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
28465 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
28466 entries to the case statement that marks unsigned arguments to
28467 overloaded functions.
28469 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
28471 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
28472 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
28474 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
28477 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
28478 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
28480 2017-03-10 Martin Liska <mliska@suse.cz>
28484 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
28485 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
28488 2017-03-10 Olivier Hainque <hainque@adacore.com>
28490 * tree-switch-conversion (array_value_type): Start by resetting
28491 candidate type to it's main variant.
28493 2017-03-10 Jakub Jelinek <jakub@redhat.com>
28495 PR rtl-optimization/79909
28496 * combine.c (try_combine): Use simplify_replace_rtx on individual
28497 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
28498 of the whole CALL_INSN_FUNCTION_USAGE.
28500 PR tree-optimization/79972
28501 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
28502 get_range_info on SSA_NAMEs. Formatting fixes.
28504 2017-03-10 Richard Biener <rguenther@suse.de>
28505 Jakub Jelinek <jakub@redhat.com>
28507 PR tree-optimization/77975
28508 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
28509 edge to be constant.
28510 (get_val_for): For constant x return it. Formatting fix.
28511 (loop_niter_by_eval): Avoid pointless looping if the next iteration
28512 would use the same bases as the current one.
28514 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28516 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
28517 instead of vec_select for V1TImode.
28518 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
28520 (VSX_LE_128): Add V1TI to this mode iterator.
28521 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
28522 (*vsx_le_perm_store_<mode>): Likewise.
28523 (pre-reload splitter for VSX stores): Likewise.
28524 (post-reload splitter for VSX stores): Likewise.
28525 (*vsx_xxpermdi2_le_<mode>): Likewise.
28526 (*vsx_lxvd2x2_le_<mode>): Likewise.
28527 (*vsx_stxvd2x2_le_<mode>): Likewise.
28529 2017-03-09 Michael Eager <eager@eagercon.com>
28531 Correct failures with --enable-checking=yes,rtl.
28533 * config/microblaze/microblaze.c (microblaze_expand_shift):
28534 Replace GET_CODE test with CONST_INT_P and INTVAL test with
28535 test for const0_rtx.
28536 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
28537 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
28539 2017-03-09 Richard Biener <rguenther@suse.de>
28541 PR tree-optimization/79977
28542 * graphite-scop-detection.c (scop_detection::merge_sese):
28543 Handle the case of extra exits to blocks dominating the entry.
28545 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
28547 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
28550 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
28552 PR rtl-optimization/79949
28553 * lra-constraints.c (process_alt_operands): Check memory when
28554 trying to predict a cycle. Print about the overall increase.
28556 2017-03-09 Richard Biener <rguenther@suse.de>
28558 PR middle-end/79971
28559 * gimple-expr.c (useless_type_conversion_p): Preserve
28560 TYPE_SATURATING for fixed-point types.
28562 2017-03-09 Richard Biener <rguenther@suse.de>
28565 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
28566 alignment of BLKmode params.
28568 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28571 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
28572 (VALL_NO_V2Q): Likewise.
28574 * config/aarch64/aarch64-simd.md
28575 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
28577 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
28578 VALL_NO_V2Q mode iterator.
28579 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
28581 2017-03-09 Martin Liska <mliska@suse.cz>
28583 PR tree-optimization/79631
28584 * tree-chkp-opt.c (chkp_is_constant_addr): Call
28585 tree_int_cst_sign_bit just for INTEGER constants.
28587 2017-03-09 Martin Liska <mliska@suse.cz>
28591 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
28594 2017-03-09 Marek Polacek <polacek@redhat.com>
28597 * tree.c (inchash::add_expr): Handle TREE_VEC.
28599 2017-03-09 Martin Liska <mliska@suse.cz>
28602 (chkp_narrow_size_and_offset): New function.
28603 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
28604 (void chkp_parse_bit_field_ref): New function.
28605 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
28606 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
28608 2017-03-09 Martin Liska <mliska@suse.cz>
28611 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
28612 (chkp_find_bounds_1): Remove gcc_unreachable.
28614 2017-03-09 Jakub Jelinek <jakub@redhat.com>
28617 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
28618 BUILT_IN_SYNC*, determine the access type from the size suffix and
28619 always build a MEM_REF with that type. Handle forgotten
28620 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
28623 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
28624 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
28625 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
28626 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
28627 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
28628 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
28629 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
28630 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
28631 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
28632 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
28633 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
28634 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
28635 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
28636 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
28637 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
28638 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
28639 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
28640 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
28641 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
28642 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
28643 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
28644 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
28645 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
28646 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
28647 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
28648 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
28649 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
28650 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
28651 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
28652 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
28653 definitions outside of __OPTIMIZE__ guarded section.
28656 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
28657 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
28658 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
28659 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
28662 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28664 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
28665 ("vfenez<mode>"): Add missing constraints.
28667 2017-03-08 Martin Sebor <msebor@redhat.com>
28670 * config/nds32/nds32.c (nds32_option_override):
28671 Fix misspelled diagnostic.
28673 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28676 * gimplify.c (gimplify_omp_for): Replace index var in outer
28677 taskloop statement with an artificial variable and add
28678 OMP_CLAUSE_PRIVATE clause for it.
28680 2017-03-08 Richard Biener <rguenther@suse.de>
28682 PR tree-optimization/79955
28683 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
28684 for accesses that are completely outside of the variable.
28686 2017-03-08 Andrew Haley <aph@redhat.com>
28688 PR tree-optimization/79943
28689 * tree-ssa-loop-split.c (compute_new_first_bound): When
28690 calculating the new upper bound, (END-BEG) should be added, not
28693 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28695 * config/avr/avr.md (setmemhi): Make sure match_dup
28696 operand number comes before match_scratch.
28698 2017-03-08 Richard Biener <rguenther@suse.de>
28700 PR tree-optimization/79920
28701 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
28702 with ncopies == 1 to ...
28703 (vect_transform_slp_perm_load): ... here. Properly compute
28704 all element loads by iterating VF times over the group. Do
28705 not handle ncopies (computed in a broken way) in
28706 vect_create_mask_and_perm.
28708 2017-03-08 Jakub Jelinek <jakub@redhat.com>
28711 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
28712 is a uniform vector, use uniform_vector_p return value instead of
28713 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
28715 2017-03-07 Marek Polacek <polacek@redhat.com>
28717 PR middle-end/79809
28718 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
28719 (alloca_call_type): Likewise.
28721 2017-03-07 Martin Liska <mliska@suse.cz>
28723 * gcov.c (process_args): Put comment to correct location.
28725 2017-03-07 Martin Liska <mliska@suse.cz>
28727 PR middle-end/68270
28728 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
28729 Use array_at_struct_end_p instead of DECL_CHAIN (field).
28730 (chkp_narrow_bounds_for_field): Likewise.
28731 (chkp_parse_array_and_component_ref): Pass one more argument to
28734 2017-03-07 Richard Biener <rguenther@suse.de>
28736 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
28739 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
28741 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
28742 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
28744 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28747 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
28748 to end of description.
28749 (PARAM_MAX_STORES_TO_MERGE): Likewise.
28751 2017-03-07 Jakub Jelinek <jakub@redhat.com>
28753 PR rtl-optimization/79901
28754 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
28756 (*avx512f_<code><mode>3<mask_name>): ... this.
28757 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
28758 iterator instead of VI8_AVX2_AVX512BW.
28760 PR rtl-optimization/79901
28761 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
28762 min/max expander, expand it using expand_vec_cond_expr.
28765 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
28768 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28771 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
28772 to void * for PCH reasons.
28773 * dwarf2out.c (output_loc_operands, output_die): Cast
28774 v.val_vec.array to unsigned char *.
28776 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
28779 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
28782 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
28784 PR rtl-optimization/79571
28785 * lra-constraints.c (process_alt_operands): Calculate static
28786 reject and subtract it from overall when only addresses will be
28789 2017-03-06 Julia Koval <julia.koval@intel.com>
28792 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
28793 incoming stack boundary to 128 for 64-bit targets.
28795 2017-03-06 Richard Biener <rguenther@suse.de>
28797 PR tree-optimization/79894
28798 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
28799 to NULL after folding it.
28801 2017-03-06 Richard Biener <rguenther@suse.de>
28803 PR tree-optimization/79824
28804 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
28805 check disabling peeling for gaps.
28807 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
28809 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
28810 attributes): Document gettimeofday.
28812 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28814 * config/s390/s390.c (s390_option_override_internal): Set
28815 PARAM_MIN_VECT_LOOP_BOUND
28817 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
28819 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
28820 * config/s390/s390.md: Likewise.
28822 2017-03-06 Jakub Jelinek <jakub@redhat.com>
28825 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
28826 (<avx2_avx512>_perm<mode>): Rename to ...
28827 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
28829 (<avx512>_perm<mode>_mask): Rename to ...
28830 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
28832 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
28833 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
28834 instead of VI8F_256_512.
28835 (avx512f_perm<mode>): New define_expand.
28836 (avx512f_perm<mode>_mask): Likewise.
28837 (avx512f_perm<mode>_1<mask_name>): New define_insn.
28838 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
28840 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28842 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
28843 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
28845 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
28847 2017-03-06 Martin Liska <mliska@suse.cz>
28850 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
28851 when having a SSA NAME w/o VAR_DECL assigned to it.
28853 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28855 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
28856 msa_dpsub_<su>_d): Fix MODE for vec_select.
28858 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
28860 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
28862 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
28864 2017-03-06 Richard Biener <rguenther@suse.de>
28866 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
28867 * plugin.c (register_plugin_info): Likewise.
28868 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
28870 2017-03-05 Jakub Jelinek <jakub@redhat.com>
28872 * config/i386/sse.md (sse_storehps, sse_storelps,
28873 avx_<castmode><avxsizesuffix>_<castmode>,
28874 avx512f_<castmode><avxsizesuffix>_<castmode>,
28875 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
28876 in condition that at least one operand is not a MEM.
28878 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28880 PR middle-end/79805
28881 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
28882 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
28884 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
28885 gimple_call_nothrow_p flag based on whether original builtin can throw.
28886 If it can, emit following stmts on the fallthrough edge.
28887 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
28888 don't create new bb if inserting just debug stmts on the edge, try to
28889 insert them on the fallthru bb or just reset debug stmts.
28891 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
28894 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
28895 restore recog_data (including the operand rtxes inside it) around
28896 the call to get_insn_template.
28898 2017-03-03 Martin Sebor <msebor@redhat.com>
28900 PR tree-optimization/79699
28901 * context.c (context::~context): Free MPFR caches to avoid
28902 a memory leak on program exit.
28904 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28906 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
28907 Use wide_int::ulow () instead of .elt (0).
28909 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28911 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
28912 (*pushxf): Limit oF constraint to 32bit targets and add oC
28913 constraint for 64bit targets.
28914 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
28915 (*pushdf): Change rmF constraint to rmC.
28917 2017-03-03 Martin Liska <mliska@suse.cz>
28919 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
28920 Remove unused variable.
28922 2017-03-03 Jakub Jelinek <jakub@redhat.com>
28925 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
28926 is a memory operand, increase num_memory.
28927 (ix86_expand_args_builtin): Likewise.
28929 2017-03-03 Jan Hubicka <jh@suse.cz>
28932 * ipa-devirt.c (maybe_record_node): Properly handle
28933 __cxa_pure_virtual visibility.
28935 2017-03-03 Martin Liska <mliska@suse.cz>
28937 PR tree-optimization/79803
28938 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
28940 (pass_loop_prefetch::execute): Disabled optimization if an
28941 assumption about L1 cache size is not met.
28943 2017-03-03 Martin Liska <mliska@suse.cz>
28945 PR rtl-optimization/79574
28946 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
28947 (hash_scan_set): Likewise.
28948 (dump_hash_table): Likewise.
28949 (hoist_code): Likewise.
28951 2017-03-03 Richard Biener <rguenther@suse.de>
28953 * fixed-value.c (fixed_from_string): Restore use of elt (1)
28954 in place of uhigh ().
28955 (fixed_convert_from_real): Likewise.
28957 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
28960 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
28962 2017-03-03 Richard Biener <rguenther@suse.de>
28964 PR middle-end/79818
28965 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
28966 TYPE_OVERFLOW_UNDEFINED check.
28968 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28970 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
28972 (vector_ae_<mode>_p): Likewise.
28973 (vector_nez_<mode>_p): Likewise.
28974 (vector_ne_v2di_p): Likewise.
28975 (vector_ae_v2di_p): Likewise.
28976 (vector_ne_<mode>_p): Likewise.
28977 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
28979 (vsx_tsqrt<mode>2_fe): Likewise.
28981 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
28984 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
28986 2017-03-02 Jakub Jelinek <jakub@redhat.com>
28988 PR rtl-optimization/79780
28989 * cprop.c (one_cprop_pass): When second and further conditional trap
28990 in a single basic block is turned into an unconditional trap, turn it
28991 into a deleted note to avoid RTL verification failures.
28993 2017-03-02 Richard Biener <rguenther@suse.de>
28995 * fold-const.c (const_binop): Use ulow () instead of elt (0).
28997 2017-03-02 Richard Biener <rguenther@suse.de>
28999 PR tree-optimization/79345
29001 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
29002 param and abort the walk, returning -1 if it is hit.
29003 (walk_aliased_vdefs): Take a limit param and pass it on.
29004 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
29005 defaulting to 0 and return a signed int.
29006 * tree-ssa-uninit.c (struct check_defs_data): New struct.
29007 (check_defs): New helper.
29008 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
29009 about uninitialized memory.
29010 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
29011 bogus uninitialized warning.
29012 (fixed_convert_from_real): Likewise.
29014 2017-03-02 Bin Cheng <bin.cheng@arm.com>
29016 PR tree-optimization/66768
29017 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
29018 iv_use if base object can't be determined.
29020 2017-03-02 Jakub Jelinek <jakub@redhat.com>
29022 PR tree-optimization/79345
29023 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
29024 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
29025 (get_pattern_stats): Initialize it.
29026 * genemit.c (gen_expand): Verify match_scratch numbers come after
29027 match_operand/match_dup numbers.
29028 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
29029 match_scratch numbers.
29030 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
29032 * config/s390/s390.md (trunctdsd2): Likewise.
29034 2017-03-02 Richard Biener <rguenther@suse.de>
29036 * wide-int.h (wide_int_storage::operator=): Implement in terms
29039 2017-03-02 Richard Biener <rguenther@suse.de>
29041 PR tree-optimization/79777
29042 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
29043 the to insert expression to sth existing.
29045 2017-03-01 Martin Sebor <msebor@redhat.com>
29047 PR middle-end/79692
29048 * gimple-ssa-sprintf.c
29049 (directive::known_width_and_precision): New function.
29050 (format_integer): Use it.
29051 (get_mpfr_format_length): Consider the full range of precision
29052 when computing %g output with the # flag. Set the likely byte
29053 count to 3 rather than 1 when precision is indeterminate.
29054 (format_floating): Correct the lower bound of precision.
29056 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29058 * doc/invoke.texi: Document default code model for 64-bit Linux.
29060 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29063 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
29064 udiv rather than div since input pattern is unsigned.
29066 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
29068 * config/i386/i386.c (print_reg): Warn for values of
29069 unsupported size in integer register.
29071 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
29074 * config/rs6000/predicates.md (current_file_function_operand): Do
29075 not allow self calls to be local if the function is replaceable.
29077 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
29080 * config/rs6000/altivec.h (vec_ctz and others): Change the
29081 preprocessor macro that controls conditional compilation from
29082 _ARCH_PWR9 to __POWER9_VECTOR__.
29083 (vec_all_ne): Change parameterization of __altivec_scalar_pred
29084 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
29085 control (instead of _ARCH_PWR9 control) so that template
29086 definition uses power9-specific function.
29087 (vec_any_eq): Likewise.
29088 (vec_all_ne): Change macro definition to use a power9-specific
29089 expansion under #ifdef __POWER9_VECTOR__ control (instead of
29090 _ARCH_PWR9 control).
29091 (vec_any_eq) Likewise.
29092 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
29093 expansion for CMPNEF to remove support for xvcmpnesp instruction.
29094 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
29095 support for xvcmpnedp instruction.
29096 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
29097 macro expansion so that Power9 implementation of vec_all_ne does
29098 not use the AltiVec predicate framework.
29099 (VCMPNEH_P): Likewise.
29100 (VCMPNEW_P): Likewise.
29101 (VCMPNED_P): Likewise.
29102 (VCMPNEFP_P): Likewise.
29103 (VCMPNEDP_P): Likewise.
29104 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
29105 implementation of vec_any_eq to not use AltiVec predicate
29107 (VCMPAEH_P): Likewise.
29108 (VCMPAEW_P): Likewise.
29109 (VCMPAED_P): Likewise.
29110 (VCMPAEFP_P): Likewise.
29111 (VCMPAEDP_P): Likewise.
29112 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
29113 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
29114 not use the AltiVec predicate framework.
29115 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
29116 of vec_any_eq to not use AltiVec predicate framework.
29117 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
29118 support for predefined __POWER9_VECTOR__ macro to indicate that
29119 Power9 instruction selection is enabled.
29120 (altivec_overloaded_builtins): Remove extraneous
29121 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
29122 function argument types RS6000_BTI_bool_V16QI and
29123 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
29124 entry for overloaded function argument types RS6000_BTI_bool_V4SI
29125 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
29126 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
29127 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
29128 Power9 for implementations of vec_cmpne. Change the signature for
29129 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
29130 (representing vec_all_ne) to remove the previously described first
29131 argument of type RS6000_BTI_INTSI, as this was an artifact of
29132 reliance on the AltiVec predicate framework, which is no longer
29133 used in the implementation of these functions. Add
29134 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
29135 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
29136 since, unlike the AltiVec predicate framework implementation, we
29137 do not share function descriptors between vec_alle and vec_anyeq.
29138 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
29139 set of modes that receive special treatment even when
29140 TARGET_P9_VECTOR is true. The special treatment emits code that
29141 does not depend on Power9 instructions.
29142 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
29143 define_expand to not rely on AltiVec predicate framework.
29144 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29146 (vector_ne_v2di_p): Change this define_expand to not rely on
29147 AltiVec predicate framework.
29148 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
29150 (vector_ne_<mode>_p): Change this define_expand to not rely on
29151 AltiVec predicate framework.
29152 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
29154 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
29155 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
29156 define_insn pattern.
29157 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
29158 define_insn pattern because the xvcmpne<VSs>. instruction is not
29160 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
29161 instruction is not supported.
29163 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29165 * config/nvptx/nvptx.c: Include intl.h.
29167 2017-03-01 Martin Jambor <mjambor@suse.cz>
29170 * ipa-prop.h (ipa_bits): Removed field known.
29171 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
29172 to pointers. Adjusted their comments to warn about their sharing.
29173 (ipcp_transformation_summary): Change bits to a vector of pointers.
29174 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
29175 (ipa_get_ipa_bits_for_value): Declare.
29176 * tree-vrp.h (value_range): Mark as GTY((for_user)).
29177 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
29178 (ipa_bits_hash_table): Likewise.
29179 (ipa_vr_ggc_hash_traits): Likewise.
29180 (ipa_vr_hash_table): Likewise.
29181 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
29182 being pointers and vr_known being removed.
29183 (ipa_set_jf_unknown): Likewise.
29184 (ipa_get_ipa_bits_for_value): New function.
29185 (ipa_set_jfunc_bits): Likewise.
29186 (ipa_get_value_range): New overloaded functions.
29187 (ipa_set_jfunc_vr): Likewise.
29188 (ipa_compute_jump_functions_for_edge): Use the above functions to
29189 construct bits and vr parts of jump functions.
29190 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
29191 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29193 (ipcp_grow_transformations_if_necessary): Also allocate
29194 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
29196 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
29197 them. Fix too long lines.
29198 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
29199 vr_known being removed.
29200 (ipa_read_jump_function): Use new setter functions to construct bits
29201 and vr parts of jump functions or set them to NULL.
29202 (write_ipcp_transformation_info): Adjust for bits being pointers.
29203 (read_ipcp_transformation_info): Likewise.
29204 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
29206 Include gt-ipa-prop.h.
29207 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
29209 (ipcp_store_bits_results): Likewise.
29210 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
29211 Do not write to existing jump functions but use a temporary instead.
29213 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29216 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
29217 attempt to use its first operand as BIT_FIELD_REF base.
29219 2017-03-01 Richard Biener <rguenther@suse.de>
29221 PR middle-end/79721
29222 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
29223 interpolating formula in wrapping arithmetic.
29224 (chrec_apply): Convert chrec_evaluate return value to wanted type.
29226 2017-03-01 Jakub Jelinek <jakub@redhat.com>
29228 PR tree-optimization/79734
29229 * tree-vect-generic.c (expand_vector_condition): Optimize
29230 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
29231 Handle VEC_COND_EXPR where comparison has different inner width from
29232 type's inner width.
29234 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
29236 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
29237 markup, and similar issues. Remove @opindex entries for things
29238 that aren't options. Add missing -mmpy-option entries.
29240 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29242 PR tree-optimization/79737
29243 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
29244 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
29245 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
29246 instead of byte_size. Formatting fix.
29247 (shift_bytes_in_array_right): Formatting fix.
29249 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
29252 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
29253 condition on optimize for the leaf function test.
29255 2017-02-28 Martin Liska <mliska@suse.cz>
29258 * read-rtl-function.c (function_reader::handle_unknown_directive):
29259 Bail out when one uses -flto.
29261 2017-02-28 Martin Liska <mliska@suse.cz>
29263 * common.opt: Replace space with tabular for options of <number>
29265 * config/i386/i386.opt: Show <number> value for
29266 -mlarge-data-threshold.
29267 * opts.c (print_filtered_help): Do not display number in hexadecimal
29270 2017-02-28 Martin Liska <mliska@suse.cz>
29272 * common.opt: Fix --help=option -Q for options which are of
29275 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
29277 * config/i386/i386.c (print_reg): Error out for values
29278 of 8-bit size in invalid integer register.
29280 2017-02-28 Martin Sebor <msebor@redhat.com>
29282 PR tree-optimization/79691
29283 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
29285 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29288 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
29289 gcc_unreachable with output_operand_lossage.
29291 2017-02-28 Richard Biener <rguenther@suse.de>
29293 PR tree-optimization/79740
29294 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
29296 (visit_nary_op): Insert the nary into the hashtable if we
29297 pattern-matched sth.
29298 * tree-ssa-pre.c (eliminate_insert): Robustify.
29300 2017-02-28 Richard Biener <rguenther@suse.de>
29302 PR middle-end/79731
29303 * fold-const.c (decode_field_reference): Reject out-of-bound
29306 2017-02-28 Jakub Jelinek <jakub@redhat.com>
29308 * config/i386/i386.c: Include intl.h.
29309 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
29310 instead of just cond ? "..." : "...".
29311 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
29312 * coverage.c (read_counts_file): Likewise.
29313 * omp-offload.c: Include intl.h.
29314 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
29315 of just cond ? "..." : "...".
29316 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
29317 of just cond ? "..." : "...".
29319 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
29322 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
29324 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
29325 'tune for' CPU name.
29326 * config/arm/arm-cpu-data.h: Regenerated.
29328 2017-02-28 Richard Biener <rguenther@suse.de>
29330 PR tree-optimization/79732
29331 * tree-inline.c (expand_call_inline): Do not shadow var.
29333 2017-02-28 Richard Biener <rguenther@suse.de>
29335 PR tree-optimization/79723
29336 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
29337 address-space properly.
29339 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
29341 * doc/optinfo.texi (Optimization groups): Fix option used for
29343 * doc/invoke.texi (-fopt-info): Document "omp".
29344 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
29345 (OPTGROUP_ALL): Add OPTGROUP_OMP.
29346 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
29347 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
29348 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
29350 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
29352 * dumpfile.c (optgroup_options): Instead of "openmp", associate
29353 OPTGROUP_OMP with "omp".
29355 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
29358 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
29359 for arithmetic shift of unsigned V2DI.
29361 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
29363 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
29364 arc/linux.h headers.
29365 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
29366 (LINK_SPEC): Likewise.
29367 (ARC_TLS_EXTRA_START_SPEC): Likewise.
29368 (EXTRA_SPECS): Likewise.
29369 (STARTFILE_SPEC): Likewise.
29370 (ENDFILE_SPEC): Likewise.
29371 (LIB_SPEC): Likewise.
29372 (TARGET_SDATA_DEFAULT): Likewise.
29373 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
29374 (MULTILIB_DEFAULTS): Likewise.
29375 (DWARF2_UNWIND_INFO): Likewise.
29376 * config/arc/big.h: New file.
29377 * config/arc/elf.h: Likewise.
29378 * config/arc/linux.h: Likewise.
29379 * config/arc/t-uClibc: Remove.
29381 2017-02-27 Bin Cheng <bin.cheng@arm.com>
29383 PR tree-optimization/77536
29384 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
29385 (tree_transform_and_unroll_loop): Use above function to compute the
29386 estimated niter of unrolled loop and use it when scaling profile.
29387 Also use count info rather than frequency if it's non-zero.
29388 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
29389 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
29390 (vect_transform_loop): Call above function.
29392 2017-02-27 Richard Biener <rguenther@suse.de>
29394 PR tree-optimization/45397
29395 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
29396 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
29397 (visit_nary_op): Add pattern matching for CSEing sign-changed
29398 or truncated operations with wider ones.
29400 2017-02-27 Richard Biener <rguenther@suse.de>
29402 PR tree-optimization/79690
29403 * tree-vect-stmts.c (vectorizable_store): Use vector type
29404 built from the DR with address-space.
29406 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
29408 * doc/invoke.texi (Optimize Options): Refine the description
29409 of asan-use-after-return.
29411 2017-02-25 Alan Modra <amodra@gmail.com>
29413 PR rtl-optimization/79584
29414 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
29415 base, not ad->base_term, the reg within base. Remove assertion
29416 that ad->base == ad->base_term. Replace gen_int_mode using
29417 bogus mode with const0_rtx.
29419 2017-02-25 Jakub Jelinek <jakub@redhat.com>
29421 PR middle-end/79396
29422 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
29423 FMA_EXPR like tcc_binary or tcc_unary.
29425 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
29428 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
29430 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
29431 (output_loc_operands): Handle DW_OP_call_ref and
29432 DW_OP_GNU_variable_value.
29433 (struct variable_value_struct): New type.
29434 (struct variable_value_hasher): Likewise.
29435 (variable_value_hash): New variable.
29436 (string_types): Remove.
29437 (copy_loc_descr): New function.
29438 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
29439 (prepend_loc_descr_to_each): New function.
29440 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
29441 instead of add_loc_descr_to_each if the first argument is single
29442 location list and the second has multiple.
29443 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
29444 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
29445 when looking for variable value which doesn't have other location info.
29446 (loc_list_from_tree): Formatting fix.
29447 (gen_array_type_die): Simplify DW_AT_string_length handling.
29448 (adjust_string_types): Remove.
29449 (gen_subprogram_die): Don't call adjust_string_types nor test/set
29450 string_types. Call resolve_variable_values.
29451 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
29452 (resolve_addr_in_expr): Likewise. Add A argument.
29453 (copy_deref_exprloc): Remove deref argument. Adjust for the
29454 original expression being DW_OP_GNU_variable_value with optionally
29455 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
29456 optionally after it.
29457 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
29458 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
29459 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
29460 (variable_value_hasher::hash, variable_value_hasher::equal): New
29462 (resolve_variable_value_in_expr, resolve_variable_value,
29463 resolve_variable_values, note_variable_value_in_expr,
29464 note_variable_value): New functions.
29465 (dwarf2out_early_finish): Call note_variable_value on all toplevel
29468 2017-02-24 Jakub Jelinek <jakub@redhat.com>
29471 * opts.h (handle_generated_option): Add GENERATED_P argument.
29472 * opts-common.c (handle_option): Adjust function comment.
29473 (handle_generated_option): Add GENERATED_P argument, pass it to
29475 (control_warning_option): Pass false to handle_generated_option
29477 * opts.c (maybe_default_option): Pass true to handle_generated_option
29479 * optc-gen.awk: Likewise.
29481 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29483 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
29484 a REG, look at the REG it is a SUBREG of.
29485 (splitter for cmpeqsi_t): Ditto.
29487 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29489 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
29490 the special USEs with the pattern of the insn, not the insn itself.
29492 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
29495 * doc/invoke.texi: Document -mload-store-pairs.
29497 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29498 Sandra Loosemore <sandra@codesourcery.com>
29500 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
29501 argument isn't a CONST_INT.
29502 (nios2_alternate_compare_const): Assert op is a CONST_INT.
29503 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
29504 (nios2_validate_compare): Bypass alternate compare logic if *op2
29505 is not a CONST_INT.
29506 (ldstwm_operation_p): Return false if first_base is not a REG or
29507 if first_offset is not a CONST_INT.
29509 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29511 * config/cris/cris.md: Use correct operand in a define_peephole2.
29513 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29515 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
29517 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
29519 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
29520 this_insn if it is an INSN or JUMP_INSN.
29521 (force_offsettable): Look at base, not at addr.
29522 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
29523 on things that aren't necessarily CONST_INTs.
29525 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
29527 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
29528 -mfpmath=sse is the default also for x86-32 targets with SSE2
29529 instruction set when @option{-ffast-math} is enabled
29531 2017-02-24 Jeff Law <law@redhat.com>
29533 PR rtl-optimizatoin/79286
29534 * ira.c (update_equiv_regs): Drop may_trap_p exception to
29537 2017-02-24 Richard Biener <rguenther@suse.de>
29539 PR tree-optimization/79389
29540 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
29543 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
29545 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
29546 function comment to reflect reality.
29547 (loop_exits_before_overflow): Fix typo in function description.
29549 2017-02-24 Richard Biener <rguenther@suse.de>
29551 PR tree-optimization/79389
29552 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
29553 properly that a threading opportunity exists. Detect conditional
29554 copy/constant propagation opportunities.
29556 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
29558 * config/visium/visium.md (type): Add trap.
29559 (b): New mode attribute.
29560 (*btst): Rename into...
29561 (*btst<mode>): ...this and adjust.
29562 (*cbranchsi4_btst_insn): Rename into...
29563 (*cbranch<mode>4_btst_insn): ...this and adjust.
29564 (trap): New define_insn.
29566 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29568 PR tree-optimization/79389
29569 * ifcvt.c (struct noce_if_info): Add rev_cond field.
29570 (noce_reversed_cond_code): New function.
29571 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
29572 reversed_comparison_code. Formatting fix.
29573 (noce_try_store_flag): Test rev_cond != NULL in addition to
29574 reversed_comparison_code.
29575 (noce_try_store_flag_constants): Likewise.
29576 (noce_try_store_flag_mask): Likewise.
29577 (noce_try_addcc): Use rev_cond if non-NULL instead of
29578 reversed_comparison_code.
29579 (noce_try_cmove_arith): Likewise. Formatting fixes.
29580 (noce_try_minmax, noce_try_abs): Clear rev_cond.
29581 (noce_find_if_block): Initialize rev_cond.
29582 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
29583 instead of false as last argument never attempt to reverse it
29586 2017-02-23 Bin Cheng <bin.cheng@arm.com>
29588 PR tree-optimization/79663
29589 * tree-predcom.c (combine_chains): Process refs in reverse order
29590 only for ZERO length chains, and add explaining comment.
29592 2017-02-23 Jeff Law <law@redhat.com>
29594 PR tree-optimization/79578
29595 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
29596 in call to operand_equal_p.
29598 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
29601 * config/i386/cpuid.h: Fix another undefined behavior.
29603 2017-02-23 Richard Biener <rguenther@suse.de>
29605 PR tree-optimization/79683
29606 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
29607 vector types for data-refs.
29609 2017-02-23 Martin Liska <mliska@suse.cz>
29611 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
29613 2017-02-23 Jakub Jelinek <jakub@redhat.com>
29615 PR middle-end/79665
29616 * internal-fn.c (get_range_pos_neg): Moved to ...
29617 * tree.c (get_range_pos_neg): ... here. No longer static.
29618 * tree.h (get_range_pos_neg): New prototype.
29619 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
29620 are known to be in between 0 and signed maximum inclusive, try to
29621 expand both unsigned and signed divmod and use the cheaper one from
29624 2017-02-22 Jeff Law <law@redhat.com>
29626 PR tree-optimization/79578
29627 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
29628 to compare base operands.
29630 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
29633 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
29634 gpc_reg_operand instead of fpr_reg_operand.
29636 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
29638 * config/mips/mips.c (mips_return_in_memory): Force FP
29639 vector types to be returned in memory for o32 ABI.
29641 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29643 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
29644 instead of DW_TAG_member for static data member declarations and don't
29645 set no_linkage_name for static inline data members.
29646 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
29649 2017-02-22 Martin Liska <mliska@suse.cz>
29651 * doc/invoke.texi: Replace inequality signs with square brackets
29654 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
29657 * lra-constraints.c (simplify_operand_subreg): Handle
29658 WORD_REGISTER_OPERATIONS targets.
29660 2017-02-22 Jakub Jelinek <jakub@redhat.com>
29663 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
29664 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
29665 elimination by swapping fld*.
29667 2017-02-22 Richard Biener <rguenther@suse.de>
29669 PR tree-optimization/79673
29670 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
29671 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
29672 irrelevant address-space qualifiers and avoiding a
29673 ADDR_SPACE_CONVERT_EXPR from fold_convert.
29675 2017-02-22 Richard Biener <rguenther@suse.de>
29677 PR tree-optimization/79666
29678 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
29679 to not symbolically negate if that may introduce undefined
29682 2017-02-22 Martin Liska <mliska@suse.cz>
29685 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
29686 * data-streamer-out.c (streamer_write_gcov_count_stream):
29688 * value-prof.c (stream_out_histogram_value): Make assert more
29689 precise based on type of counter.
29691 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
29694 * config/i386/i386.md (standard_x87sse_constant_load splitter):
29695 Use nonimmediate_operand instead of memory_operand for operand 1.
29696 (float-extend standard_x87sse_constant_load splitter): Ditto.
29698 2017-02-21 Jeff Law <law@redhat.com>
29700 PR tree-optimization/79621
29701 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
29702 blocks with edges to themselves.
29704 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29707 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
29708 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
29709 Use gimple_call_builtin_p.
29712 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
29713 on temporarily removed DEBUG_INSNs.
29715 PR tree-optimization/79649
29716 * tree-loop-distribution.c (classify_partition): Give up on
29717 non-generic address space loads/stores.
29719 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
29721 * doc/loop.texi (Loop manipulation): Remove nonexistent
29722 tree_ssa_loop_version from the documentation.
29723 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
29725 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29728 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
29729 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
29730 * config/rs6000/rs6000.c: Include except.h.
29731 (rs6000_expand_split_stack_prologue): Call
29732 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
29734 2017-02-21 Martin Jambor <mjambor@suse.cz>
29737 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
29738 have been analyzed.
29740 2017-02-21 Martin Jambor <mjambor@suse.cz>
29742 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
29743 for backward compatibility only.
29744 * doc/invoke.texi (Option Summary): Remove all references to
29745 -fipa-cp-alignment.
29747 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
29751 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29753 * lra-constraints.c (curr_insn_transform): Handle
29754 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29756 2017-02-21 Martin Liska <mliska@suse.cz>
29758 * config/i386/i386.opt: Replace -masm-dialect with -masm.
29760 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
29762 PR translation/79638
29763 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
29765 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
29768 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
29769 (arm_function_ok_for_sibcall): Return false for an indirect call by
29770 descriptor if all the argument registers are used.
29771 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
29772 alignment of the function.
29774 2017-02-21 Jakub Jelinek <jakub@redhat.com>
29776 PR tree-optimization/61441
29777 * simplify-rtx.c (simplify_const_unary_operation): For
29778 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
29779 the sNaN unmodified.
29781 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
29783 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
29784 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
29785 instead of SYSTEM_HEADER_DIR.
29787 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
29788 Martin Liška <mliska@suse.cz>
29790 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
29791 Fix typos and grammar, use active voice, and clarify.
29793 2017-02-20 Marek Polacek <polacek@redhat.com>
29795 PR middle-end/79537
29796 * gimplify.c (gimplify_expr): Handle unused *&&L;.
29799 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
29801 2017-02-20 Jakub Jelinek <jakub@redhat.com>
29804 * config/i386/i386.c (ix86_expand_builtin): Handle
29805 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
29806 ix86_builtins_isa[fcode].isa as a requirement of those
29807 flags and any other flag in the bitmask.
29808 (ix86_init_mmx_sse_builtins): Use 0 instead of
29809 ~OPTION_MASK_ISA_64BIT as mask.
29810 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
29811 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
29812 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
29813 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
29815 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29818 * lra-constraints.c (split_reg): Check requested split mode
29819 is supported by the register.
29821 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29823 * lra-constraints.c (simplify_operand_subreg): Remove early
29826 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29829 * lra-constraints.c (curr_insn_transform): Tighten condition
29830 for converting SUBREG reloads from OP_OUT to OP_INOUT.
29832 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
29835 * lra-constraints.c (curr_insn_transform): Handle
29836 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
29838 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
29841 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
29843 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
29845 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
29848 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
29851 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29853 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
29856 2017-02-18 Jakub Jelinek <jakub@redhat.com>
29858 * final.c (last_columnnum, override_columnnum): New variables.
29859 (final_start_function): Set last_columnnum, pass it to begin_prologue
29860 hook and pass 0 to dwarf2out_begin_prologue.
29861 (final_scan_insn): Update override_columnnum. Pass last_columnnum
29862 to source_line debug hook.
29863 (notice_source_line): Compute last_columnnum and for debug_column_info
29864 return true on column changes.
29865 * debug.h (struct gcc_debug_hooks): Add column argument to
29866 source_line and begin_prologue hooks.
29867 (debug_nothing_int_charstar_int_bool): Remove prototype.
29868 (debug_nothing_int_int_charstar,
29869 debug_nothing_int_int_charstar_int_bool): New prototypes.
29870 (dwarf2out_begin_prologue): Add column argument.
29871 * debug.c (do_nothing_debug_hooks): Adjust source_line and
29872 begin_prologue hooks.
29873 (debug_nothing_int_charstar_int_bool): Remove.
29874 (debug_nothing_int_int_charstar,
29875 debug_nothing_int_int_charstar_int_bool): New functions.
29876 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
29877 through to dwarf2out_source_line.
29878 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
29879 (dwarf2out_source_line): Add column argument, emit it if requested.
29880 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
29882 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29883 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
29884 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
29885 through to dwarf2out_begin_prologue.
29886 (vmsdbgout_source_line): Add column argument, pass it through to
29887 dwarf2out_source_line.
29888 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
29889 dbxout_source_line caller.
29890 (dbxout_source_line): Add column argument.
29892 * common.opt (gno-column-info, gcolumn-info): New options.
29893 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
29894 (check_die): Also test for multiple DW_AT_decl_column attributes.
29895 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
29896 DW_AT_decl_column if requested.
29897 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
29899 (gen_variable_die): Likewise.
29900 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
29901 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
29904 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
29905 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
29906 (ix86_handle_option): Handle OPT_m3dnowa.
29907 * doc/invoke.texi (-m3dnowa): Document.
29908 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
29909 -m3dnowa instead of -m3dnow -march=athlon.
29912 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
29913 instead of gcc_assert for K, r and R code checks. Formatting fixes.
29915 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29918 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
29919 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
29920 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
29921 generator for vsx_xxpermdi_<mode>_be.
29922 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
29923 force big-endian semantics.
29924 (vsx_xxpermdi_<mode>_be): New define_expand with same
29925 implementation as previous version of vsx_xxpermdi_<mode>.
29927 2017-02-17 Jakub Jelinek <jakub@redhat.com>
29929 PR tree-optimization/79327
29930 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
29931 variable, its initialization and use.
29933 2017-02-17 Julia Koval <julia.koval@intel.com>
29935 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
29936 (OPTION_MASK_ISA_PKU_UNSET): New.
29937 (ix86_handle_option): Handle -mrdpid.
29938 * config/i386/cpuid.h (bit_RDPID): New.
29939 * config/i386/driver-i386.c (host_detect_local_cpu):
29940 Detect RDPID feature.
29941 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
29942 * config/i386/i386-c.c (ix86_target_macros_internal):
29944 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
29945 (ix86_valid_target_attribute_inner_p): Add "rdpid".
29946 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
29947 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
29948 * config/i386/i386.md (define_insn "rdpid"): New.
29949 * config/i386/i386.opt Add -mrdpid.
29950 * config/i386/immintrin.h (_rdpid_u32): New.
29952 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
29954 PR rtl-optimization/79541
29955 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
29956 instead of transforming it into USE.
29958 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
29960 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
29961 If HONOR_SNANS (SFmode) force the input to a register.
29962 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
29963 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
29964 an frsp or similar insn.
29966 2017-02-17 Martin Liska <mliska@suse.cz>
29968 PR rtl-optimization/79577
29969 * params.def (selsched-max-sched-times): Increase minimum to 1.
29971 2017-02-17 Martin Liska <mliska@suse.cz>
29973 PR rtl-optimization/79574
29974 * gcse.c (want_to_gcse_p): Prevent integer overflow.
29976 2017-02-17 Martin Liska <mliska@suse.cz>
29978 PR tree-optimization/79529
29979 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
29980 ssa_defined_default_def_p to handle cases which are implicitly
29982 * tree-ssa.c (ssa_defined_default_def_p): New function.
29983 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
29984 which are implicitly defined.
29985 * tree-ssa.h (ssa_defined_default_def_p): Declare.
29987 2017-02-17 Richard Biener <rguenther@suse.de>
29989 PR middle-end/79576
29990 * params.def (max-ssa-name-query-depth): Limit to 10.
29992 2017-02-17 Richard Biener <rguenther@suse.de>
29994 PR tree-optimization/79552
29995 * tree-ssa-structalias.c (visit_loadstore): Properly verify
29998 2017-02-17 Richard Biener <rguenther@suse.de>
30001 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
30003 2017-02-17 Marek Polacek <polacek@redhat.com>
30005 PR middle-end/79536
30006 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
30007 (fold_negate_expr): New wrapper.
30009 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
30011 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
30012 Correct terminology and de-emphasize pre-standard behavior.
30014 2017-02-16 Alan Modra <amodra@gmail.com>
30016 PR rtl-optimization/79286
30017 * ira.c (def_dominates_uses): New function.
30018 (update_equiv_regs): Don't create an equivalence for insns that
30019 may trap where the register def does not dominate the use.
30021 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
30023 PR rtl-optimization/78127
30024 * lra.c (lra): Call lra_eliminate before finish the loop after
30027 2017-02-16 Richard Biener <rguenther@suse.de>
30029 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
30031 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
30032 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
30033 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
30034 (isl_val_int_from_wi): New function.
30035 (extract_affine_gmp): Rename to ...
30036 (extract_affine_wi): ... this, take a widest_int.
30037 (extract_affine_int): Just wrap extract_affine_wi.
30038 (add_param_constraints): Use isl_val_int_from_wi.
30039 (add_loop_constraints): Likewise, and extract_affine_wi.
30041 2017-02-15 Jeff Law <law@redhat.com>
30043 PR middle-end/79521
30044 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
30045 ira_init_register_move_cost_if_necessary.
30047 2017-02-15 Martin Sebor <msebor@redhat.com>
30049 PR middle-end/32003
30050 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
30051 removed in a prior commit.
30053 2017-02-15 Bin Cheng <bin.cheng@arm.com>
30055 PR tree-optimization/79347
30056 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
30057 counters during peeling.
30059 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
30061 * Makefile.in (site.exp): Remove "set ISLVER".
30063 2017-02-15 Jakub Jelinek <jakub@redhat.com>
30066 * real.c (real_from_integer): Call real_convert even for decimal.
30068 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
30071 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
30073 2017-02-14 Andrew Pinski <apinski@cavium.com>
30075 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
30076 cores and change the partno/implementer to be correct.
30077 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
30078 the 'B" as the implementer.
30079 * config/aarch64/aarch64-tune.md: Regenerate.
30081 2017-02-14 Carl Love <cel@us.ibm.com>
30083 * config/rs6000/rs6000.c: Add case statement entry to make the
30084 xvcvuxdsp built-in argument unsigned.
30085 * config/rs6000/vsx.md: Fix the source and return operand types so they
30086 match the instruction definitions from the ISA document. Fix typo
30087 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
30090 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
30093 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
30094 member early_clobber_alts.
30095 * lra-lives.c (reg_early_clobber_p): New.
30096 (process_bb_lives): Use it.
30097 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
30098 (debug_operand_data): Initialize early_clobber_alts.
30099 (setup_operand_alternative): Set up early_clobber_alts.
30100 (collect_non_operand_hard_regs): Ditto. Pass early clobber
30101 alternatives to new_insn_reg.
30102 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
30104 (lra_update_insn_regno_info): Pass the new arg.
30106 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30108 PR middle-end/79505
30109 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
30110 (new_oacc_loop_raw): Don't clear already cleared fields.
30113 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
30114 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
30115 _mm512_prefetch_i64gather_ps): New inline functions and macros.
30117 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
30120 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
30122 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
30125 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
30126 the extra instruction to the right place to store 128-bit constant
30129 2017-02-14 Martin Sebor <msebor@redhat.com>
30131 PR middle-end/79448
30132 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
30133 warning for strings of unknown length.
30135 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
30137 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
30139 2017-02-14 Jeff Law <law@redhat.com>
30142 * ira-costs.c (scan_one_insn): Initialize register move costs
30143 for pseudos seen in USE/CLOBBER insns.
30145 PR tree-optimization/79095
30146 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
30147 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
30148 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
30149 if the operands are known to be not equal, then the resulting range
30151 (intersect_ranges): If the new range is ~[0,0] and the old range is
30152 wide, then prefer ~[0,0].
30153 * tree-vrp.c (overflow_comparison_p_1): New function.
30154 (overflow_comparison_p): New function.
30155 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
30156 if NAME is used in an overflow test.
30157 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
30158 overflow check that can be expressed as an equality test, then adjust
30159 ops to be that equality test.
30161 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30163 * config/s390/s390-builtin-types.def: Remove flags argument.
30164 * config/s390/s390.c (s390_init_builtins): Likewise.
30166 2017-02-14 Martin Liska <mliska@suse.cz>
30168 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
30169 vector. Fix trailing white spaces.
30171 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
30173 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
30176 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30178 PR rtl-optimization/68664
30179 * config/arm/arm.c (arm_sched_can_speculate_insn):
30180 New function. Declare prototype.
30181 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30183 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30185 PR rtl-optimization/68664
30186 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
30188 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
30190 2017-02-14 Amit Pawar <amit.pawar@amd.com>
30192 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
30193 max skip bytes for function, loop and jump.
30195 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
30197 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
30198 ABS_EXPR for gimple dump.
30200 2017-02-14 Jakub Jelinek <jakub@redhat.com>
30203 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
30205 PR tree-optimization/79408
30206 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
30207 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
30208 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
30209 also if rhs1 is INTEGER_CST.
30211 2017-02-14 Richard Biener <rguenther@suse.de>
30213 PR middle-end/79432
30214 * tree-into-ssa.c (insert_phi_nodes): When the function can
30215 have abnormal edges rewrite SSA names with broken use-def
30216 dominance out of SSA and register them for PHI insertion.
30218 2017-02-13 Martin Sebor <msebor@redhat.com>
30220 PR middle-end/79496
30221 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
30222 clearing info.nowrite flag when snprintf size argument is a range.
30224 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30226 * cprop.c (cprop_jump): Add missing space in string literal.
30227 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
30228 (get_constraint_for_component_ref): Likewise.
30229 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
30230 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
30231 * lra-constraints.c (process_alt_operands): Likewise.
30232 * ipa-inline.c (inline_small_functions): Likewise.
30233 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
30234 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
30235 * trans-mem.c (diagnose_tm_1_op): Likewise.
30236 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
30237 (grid_parallel_clauses_gridifiable): Likewise.
30239 * config/nvptx/mkoffload.c (process): Add space in between
30242 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
30243 "MOD4_SSE_REGS" and "ALL_REGS".
30245 * spellcheck.c (test_data): Add , in between "foo" and "food".
30247 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30250 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
30251 boundary crossing check and subsequent code generation agree.
30253 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30255 * config/aarch64/aarch64.c (has_memory_op): Delete.
30256 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
30259 2017-02-13 Jakub Jelinek <jakub@redhat.com>
30261 PR rtl-optimization/79388
30262 PR rtl-optimization/79450
30263 * combine.c (distribute_notes): When removing TEM_INSN for which
30264 corresponding dest has last value recorded, invalidate that last
30267 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30269 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
30270 of explicit '@'. Add missing assembly comment marker on branch costs
30273 2017-02-13 Nathan Sidwell <nathan@acm.org>
30275 * gengtype-lex.l (<in_struct>): Add '/'.
30277 2017-02-13 Martin Liska <mliska@suse.cz>
30280 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
30282 2017-02-13 Richard Biener <rguenther@suse.de>
30284 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
30286 * configure: Re-generate.
30287 * config.in: Likewise.
30288 * graphite-dependences.c: Simplify as if
30289 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
30290 * graphite-isl-ast-to-gimple.c: Likewise.
30291 * graphite-optimize-isl.c: Likewise.
30292 * graphite-poly.c: Likewise.
30293 * graphite-sese-to-poly.c: Likewise.
30294 * graphite.h: Likewise.
30295 * toplev.c: Include isl/version.h and use isl_version () for
30296 printing the ISL version.
30297 * doc/install.texi: Update ISL requirement.
30299 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30301 * doc/standards.texi (Standards): Update reference to
30304 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
30306 * doc/extend.texi (Named Address Spaces): sourceware.org now
30308 * doc/install.texi (Binaries): Ditto.
30311 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30313 * doc/cpp.texi: Replace "stringify"/"stringification" with C
30314 standard terminology "stringize"/"stringizing" throughout.
30315 * doc/cppinternals.texi: Likewise.
30317 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
30319 * doc/extend.texi: Fix some spelling mistakes and typos.
30320 * doc/invoke.texi: Likewise.
30322 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30325 * params.def (inline-min-speedup) Change from 10 to 8.
30327 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30329 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
30332 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
30335 * ipa-inline-analysis.c (get_minimal_bb): New function.
30336 (record_modified): Use it.
30337 (remap_edge_change_prob): Handle also ancestor functions.
30339 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
30341 * doc/contrib.texi (Contributors): Remove broken link into
30342 the Mauve CVS repository.
30344 2017-02-11 Jakub Jelinek <jakub@redhat.com>
30346 PR middle-end/79454
30347 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
30348 result computation whenever lhs doesn't have vector mode, not
30349 just when it has BLKmode.
30351 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30353 * doc/makefile.texi (profiledbootstrap): Refer to the
30354 installation instructions only in textual form.
30356 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30359 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
30361 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
30363 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
30364 (Specific): Update mingw-w64 reference.
30366 (Specific): Remove broken link to Renesas RX processor.
30368 2017-02-10 Richard Biener <rguenther@suse.de>
30370 * toplev.c (process_options): Do not mention obsolete graphite
30371 options when printing sorry message about missing graphite support.
30372 Mention -floop-nest-optimize.
30374 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
30376 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
30377 (vtst_p16): Likewise.
30378 (vtstq_p8): Likewise.
30379 (vtstq_p16): Likewise.
30381 (vtstq_p64): Likewise.
30382 * config/arm/arm_neon.h (vgetq_lane_p64): New.
30383 (vset_lane_p64): New.
30384 (vsetq_lane_p64): New.
30386 2017-02-10 Jakub Jelinek <jakub@redhat.com>
30388 PR tree-optimization/79411
30389 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
30390 stmt operands are SSA_NAMEs used in abnormal phis.
30391 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
30394 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30397 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
30400 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
30402 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
30404 2017-02-09 Jakub Jelinek <jakub@redhat.com>
30406 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
30410 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
30411 not arbitrary TREE_CONSTANT.
30414 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
30415 "omp declare target link" attribute unless is_global_var.
30416 * omp-offload.c (find_link_var_op): Likewise.
30418 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
30419 Chung-Lin Tang <cltang@codesourcery.com>
30421 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
30423 (gimplify_adjust_omp_clauses): Don't delete TILE.
30424 (gimplify_omp_for): Deal with TILE.
30425 * internal-fn.c (expand_GOACC_TILE): New function.
30426 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
30428 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
30429 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
30431 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
30432 avoid DIV for outermost collapse var.
30433 (expand_oacc_for): Insert tile element loop as needed. Adjust.
30434 Remove out of date comments, fix whitespace.
30435 * omp-general.c (omp_extract_for_data): Deal with tiling.
30436 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
30437 adjust OLF_DIM_BASE value.
30438 (struct omp_for_data): Add tiling field.
30439 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
30440 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
30441 for auto loops. Remove default auto determining, moved to
30442 oacc_loop_fixed_partitions.
30443 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
30444 stmts, add e_mask field.
30445 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
30446 (oacc_thread_numbers): Use oacc_dim_call.
30447 (oacc_xform_tile): New.
30448 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
30449 (finish_oacc_loop): Adjust for ifns vector.
30450 (oacc_loop_discover_walk): Append loop abstraction sites to list,
30451 add case for GOACC_TILE fns.
30452 (oacc_loop_xform_loop): Delete.
30453 (oacc_loop_process): Iterate over call list directly, and add
30454 handling for GOACC_TILE fns.
30455 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
30457 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
30458 vector partitioning to outer loops. Assign 2 partitions to loops
30459 when available. Add TILE handling.
30460 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
30461 (execite_oacc_device_lower): Process GOACC_TILE fns,
30462 ignore unknown specs.
30463 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
30464 * tree.c (omp_clause_num_ops): Adjust TILE ops.
30465 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
30467 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
30469 * configure.ac (ACX_BUGURL): Update.
30470 * configure: Regenerate.
30472 2017-02-09 Richard Biener <rguenther@suse.de>
30474 PR tree-optimization/69823
30475 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
30476 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
30478 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30480 * config/arc/arc-c.def: Add __NPS400__ definition.
30481 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
30482 (TARGET_NPS400): Define.
30484 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
30486 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
30488 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
30489 pointer, arch_info.
30490 (arc_cpu_types): Fill the arch_info field with a pointer into the
30491 arc_arch_types table.
30492 (arc_selected_cpu): Declare.
30493 * config/arc/arc.c (arc_selected_cpu): Make global.
30494 (arc_selected_arch): Delete.
30495 (arc_base_cpu): Delete.
30496 (arc_override_options): Remove references to deleted variables,
30497 update access to arch information.
30498 (ARC_OPT): Update access to arch information.
30499 (ARC_OPTX): Likewise.
30500 * config/arc/arc.h (arc_base_cpu): Remove declaration.
30501 (TARGET_ARC600): Update access to arch information.
30502 (TARGET_ARC601): Likewise.
30503 (TARGET_ARC700): Likewise.
30504 (TARGET_EM): Likewise.
30505 (TARGET_HS): Likewise.
30506 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
30509 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
30512 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
30513 condition/operands for integer GE/LE/GEU/LEU operations.
30515 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
30517 PR translation/79397
30518 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
30521 2017-02-08 Martin Jambor <mjambor@suse.cz>
30524 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
30525 whether allocation happened.
30526 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
30527 nothing was allocated.
30529 2017-02-08 Jakub Jelinek <jakub@redhat.com>
30531 PR tree-optimization/79408
30532 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
30533 constant, but SSA_NAME with a known integer range, use the minimum
30534 of that range instead of op1 to determine if modulo can be replaced
30535 with its first operand.
30537 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30539 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
30541 2017-02-08 Richard Biener <rguenther@suse.de>
30543 PR tree-optimization/71824
30544 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
30545 Check all loops contained in the merged region.
30547 2017-02-07 Andrew Pinski <apinski@cavium.com>
30549 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
30551 2017-02-07 Andrew Pinski <apinski@cavium.com>
30553 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
30554 (thunderxt88): Likewise.
30555 (thunderxt81): Disable LSE and change v8.1 to v8.
30556 (thunderxt83): Likewise.
30558 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30559 Richard Biener <rguenther@suse.de>
30561 PR middle-end/79399
30562 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
30563 type from int to size_t.
30564 * ira-costs.c (struct_costs_size): Change type from int to size_t.
30566 2017-02-07 Jakub Jelinek <jakub@redhat.com>
30568 PR rtl-optimization/79386
30569 * cprop.c (bypass_conditional_jumps): Initialize
30570 bypass_last_basic_block already before splitting bbs after
30571 unconditional traps...
30572 (bypass_conditional_jumps): ... rather than here.
30575 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
30576 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
30577 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
30578 fix -masm=intel patterns.
30580 2017-02-07 Richard Biener <rguenther@suse.de>
30582 PR tree-optimization/79256
30583 PR middle-end/79278
30584 * builtins.c (get_object_alignment_2): Use min_align_of_type
30585 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
30586 and ADJUST_FIELD_ALIGN.
30588 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
30590 * doc/tm.texi: Regenerate.
30591 * stor-layout.c (layout_decl): Adjust.
30592 (update_alignment_for_field): Likewise.
30593 (place_field): Likewise.
30594 (min_align_of_type): Likewise.
30595 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
30596 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
30597 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
30598 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
30599 * config/frv/frv.c (frv_adjust_field_align): Likewise.
30600 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
30601 * config/i386/i386.c (x86_field_alignment): Likewise.
30602 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
30603 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
30604 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
30605 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
30606 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
30607 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
30611 2017-01-30 Richard Biener <rguenther@suse.de>
30613 PR tree-optimization/79256
30614 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30615 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30618 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
30620 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
30621 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
30622 builtins to SImode and emit a zero-extend, if necessary.
30624 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30626 * docs/invoke.texi (RISC-V Options): Alphabetize.
30628 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30630 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
30633 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
30635 * config/riscv/riscv.c: New file.
30636 * common/config/riscv/riscv-common.c: Likewise.
30637 * config.gcc: Likewise.
30638 * config/riscv/constraints.md: Likewise.
30639 * config/riscv/elf.h: Likewise.
30640 * config/riscv/generic.md: Likewise.
30641 * config/riscv/linux.h: Likewise.
30642 * config/riscv/multilib-generator: Likewise.
30643 * config/riscv/peephole.md: Likewise.
30644 * config/riscv/pic.md: Likewise.
30645 * config/riscv/predicates.md: Likewise.
30646 * config/riscv/riscv-builtins.c: Likewise.
30647 * config/riscv/riscv-c.c: Likewise.
30648 * config/riscv/riscv-ftypes.def: Likewise.
30649 * config/riscv/riscv-modes.def: Likewise.
30650 * config/riscv/riscv-opts.h: Likewise.
30651 * config/riscv/riscv-protos.h: Likewise.
30652 * config/riscv/riscv.h: Likewise.
30653 * config/riscv/riscv.md: Likewise.
30654 * config/riscv/riscv.opt: Likewise.
30655 * config/riscv/sync.md: Likewise.
30656 * config/riscv/t-elf-multilib: Likewise.
30657 * config/riscv/t-linux: Likewise.
30658 * config/riscv/t-linux-multilib: Likewise.
30659 * config/riscv/t-riscv: Likewise.
30660 * configure.ac: Likewise.
30661 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
30662 Waterman as RISC-V maintainers.
30663 * doc/install.texi: Add RISC-V entries.
30664 * doc/invoke.texi: Add RISC-V options section.
30665 * doc/md.texi: Add RISC-V constraints section.
30666 * configure: Regenerated.
30668 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
30671 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
30672 false values to be constant vectors with all 0 or all 1 bits set.
30673 (vcondu<mode><mode>): Likewise.
30674 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
30676 (fpmask_comparison_operator): Update comment.
30677 (vecint_comparison_operator): New predicate.
30678 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
30679 vector conditionals when the true and false values are constant
30680 vectors with all 0 bits or all 1 bits set.
30682 2017-02-06 Martin Sebor <msebor@redhat.com>
30684 PR tree-optimization/79376
30685 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
30687 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
30689 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
30690 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
30691 to simplify split condition.
30693 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30695 * omp-expand.c (oxpand_omp_atomic_fetch_op,
30696 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
30699 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
30701 PR rtl-optimization/68664
30702 * target.def (can_speculate_insn): New hook.
30703 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
30704 * doc/tm.texi: Regenerate.
30705 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
30706 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
30707 (rs6000_sched_can_speculate_insn): New function.
30709 2017-02-06 Jakub Jelinek <jakub@redhat.com>
30711 PR tree-optimization/79284
30712 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
30713 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
30714 vectorizable_mask_load_store, vectorizable_operation,
30715 vect_is_simple_cond, get_same_sized_vectype): Use it instead
30716 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
30717 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
30718 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
30719 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
30720 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
30721 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
30722 is_gimple_assign (stmt). Replace another such test with
30723 is_gimple_assign (stmt).
30725 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30728 * config/avr/avr.c (rtl-iter.h): Include it.
30729 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
30730 (avr_legitimate_combined_insn): ...and implementation.
30732 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
30734 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
30735 * config/s390/s390.c (s390_const_operand_ok)
30736 (s390_canonicalize_comparison, s390_extract_part)
30737 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
30738 (s390_contiguous_bitmask_p, s390_rtx_costs)
30739 (legitimize_pic_address): Likewise.
30740 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
30741 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
30742 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
30743 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
30744 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
30746 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
30748 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
30749 REGNO($0) == REGNO($1).
30751 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30753 * config/s390/linux.h(SIZE_TYPE): Add comment.
30755 2017-02-06 Julian Brown <julian@codesourcery.com>
30756 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30757 Virendra Pathak <virendra.pathak@broadcom.com>
30759 * config/aarch64/aarch64-cores.def: Change the scheduler
30761 * config/aarch64/aarch64.md: Include thunderx2t99.md.
30762 * config/aarch64/thunderx2t99.md: New file.
30764 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30766 * doc/standards.texi (Go Language): Update link to language
30769 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30771 * tree-eh.c (lower_resx): Sanitize profile.
30772 (cleanup_empty_eh_move_lp): Likewise.
30774 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
30777 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
30779 * cfgloopmanip.h (loop_version): Update prototype.
30780 * modulo-sched.c (sms_schedule): Update call of loop_version.
30781 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
30782 * tree-parloops.c (gen_parallel_loop): Likewise.
30783 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
30784 * tree-ssa-loop-split.c (split_loop): Likewise.
30785 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
30786 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
30788 2017-02-05 Martin Liska <mliska@suse.cz>
30791 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
30793 (print_operand_address): Initialize a struct to zero.
30795 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30797 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
30798 garbage collector only in textual form.
30800 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
30802 * doc/extend.texi (x86 specific memory model extensions for
30803 transactional memory): Simplify a phrase.
30805 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
30808 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
30809 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
30810 (atomic_storedi_1): Likewise.
30812 2017-02-04 Jakub Jelinek <jakub@redhat.com>
30814 PR tree-optimization/79338
30815 * tree-parloops.c (gather_scalar_reductions): Don't call
30816 vect_analyze_loop_form for loop->inner before destroying loop's
30819 2017-02-03 Martin Sebor <msebor@redhat.com>
30821 PR tree-optimization/79327
30822 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
30823 when precision has resulted in leading zeros.
30824 (format_integer): Adjust the likely counter to assume an unknown
30825 argument that may be zero is non-zero.
30827 2017-02-03 Jason Merrill <jason@redhat.com>
30830 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
30831 avoid copying non-taken branch.
30833 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30835 PR tree-optimization/79340
30836 * tree-vect-loop.c (vectorizable_reduction): Release
30837 vec_defs elements after safe_splicing them into other vectors.
30840 PR tree-optimization/79327
30841 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
30842 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
30844 (format_integer): Use wide_int_to_tree instead of build_int_cst
30845 + to_?hwi. If argmin is NULL, just set argmin and argmax to
30846 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
30847 of shortest and longest sequence.
30849 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
30851 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
30852 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
30854 2017-02-03 Walter Lee <walt@tilera.com>
30857 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
30858 after initial stackframe link reg save.
30859 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
30861 2017-02-03 Jakub Jelinek <jakub@redhat.com>
30864 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
30865 wu for stxssp alternative.
30867 2017-02-03 Martin Sebor <msebor@redhat.com>
30869 PR tree-optimization/79352
30870 * gimple-fold.c (get_range_strlen): Add argument.
30871 (get_range_strlen): Change return type to bool.
30872 (get_maxval_strlen): Pass in a dummy argument.
30873 * gimple-fold.h (get_range_strlen): Change return type to bool.
30874 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
30875 * tree.h (array_at_struct_end_p): Add argument.
30876 * tree.c (array_at_struct_end_p): Handle it.
30878 2017-02-03 Martin Liska <mliska@suse.cz>
30881 * multiple_target.c (create_dispatcher_calls): Redirect edge
30882 from a caller of a dispatcher.
30883 (expand_target_clones): Make the clones local.
30884 (ipa_target_clone): Do both target clones and resolvers.
30885 (ipa_dispatcher_calls): Remove the pass.
30886 (pass_dispatcher_calls::gate): Likewise.
30887 (make_pass_dispatcher_calls): Likewise.
30888 * passes.def (pass_target_clone): Put as very first IPA early
30891 2017-02-03 Martin Liska <mliska@suse.cz>
30893 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
30894 in case of a function with ifunc attribute.
30896 2017-02-03 Martin Liska <mliska@suse.cz>
30898 * cgraph.c (cgraph_node::dump): Dump function version info.
30899 * symtab.c (symtab_node::dump_base): Add missing new line.
30901 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30903 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
30904 (ifcombine_ifandif): Use it.
30906 2017-02-03 Martin Liska <mliska@suse.cz>
30908 * doc/invoke.texi: Document default value for
30909 use-after-scope-direct-emission-threshold.
30911 2017-02-03 Martin Liska <mliska@suse.cz>
30913 PR tree-optimization/79339
30914 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
30915 (format_floating): Likewise.
30917 2017-02-03 Martin Liska <mliska@suse.cz>
30920 * ipa-prop.c (ipa_node_params_t::insert): Remove current
30922 (ipa_node_params_t::remove): Likewise.
30923 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
30924 initialization from removed ipa_node_params_t::insert.
30925 (ipa_node_params::~ipa_node_params): Move from removed
30926 ipa_node_params_t::release.
30927 * symbol-summary.h (symbol_summary::m_released): New member.
30928 Do not release a summary twice. Do not allow to call finalizer
30929 for types of a summary that live in GGC memory.
30931 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
30933 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
30936 2017-02-02 Martin Sebor <msebor@redhat.com>
30938 PR middle-end/79275
30939 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
30940 (format_string): Tighten up the range of output for non-constant
30941 strings and correct the expected range for wide non-constant strings.
30943 2017-02-02 Martin Sebor <msebor@redhat.com>
30945 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
30947 PR middle-end/32003
30948 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
30950 (-fdump-tree-@var): Add to index and document how to come up
30951 with pass-specific option and dump file names.
30952 (-fdump-passes): Clarify where to look for output.
30954 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
30956 PR middle-end/77445
30957 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
30958 statistics of the analyzed path; allow threading for speed when
30959 any of BBs along the path are optimized for speed.
30961 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
30963 PR middle-end/78468
30964 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
30965 settings of the virtual registers.
30968 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
30970 * explow.c (get_dynamic_stack_size): Take known alignment of stack
30971 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
30974 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30976 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
30977 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
30979 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30981 * config/s390/s390.md: Add missing comments with the expanded
30983 * config/s390/vector.md: Likewise.
30984 * config/s390/vx-builtins.md: Likewise.
30986 2017-02-02 Jakub Jelinek <jakub@redhat.com>
30989 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
30990 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
30991 conditions on a single line.
30993 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30995 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30996 __S390_VX__ to __VX__.
30998 2017-02-01 Andrew Pinski <apinski@cavium.com>
31000 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
31001 stmt_info to record_stmt_cost.
31002 (vect_get_known_peeling_cost): Pass stmt_info if known to
31004 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
31005 cpu_vector_cost field into
31006 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
31007 field into vec_int_stmt_cost and vec_fp_stmt_cost.
31008 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
31009 splitting of scalar_stmt_cost and vec_stmt_cost.
31010 (thunderx_vector_cost): Likewise.
31011 (cortexa57_vector_cost): LIkewise.
31012 (exynosm1_vector_cost): Likewise.
31013 (xgene1_vector_cost): Likewise.
31014 (thunderx2t99_vector_cost): Improve after the splitting of the two
31016 (aarch64_builtin_vectorization_cost): Update for the splitting of
31017 scalar_stmt_cost and vec_stmt_cost.
31019 2017-02-01 Torvald Riegel <triegel@redhat.com>
31020 Richard Henderson <rth@redhat.com>
31022 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
31023 conditional on existance of a fast atomic load.
31024 * optabs-query.c (can_atomic_load_p): New function.
31025 * optabs-query.h (can_atomic_load_p): Declare it.
31026 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
31027 no fast atomic load is available for the particular size of access.
31028 (expand_atomic_compare_and_swap): Likewise.
31029 (expand_atomic_load): Likewise.
31030 (expand_atomic_store): Likewise.
31031 (expand_atomic_fetch_op): Likewise.
31032 * testsuite/lib/target-supports.exp
31033 (check_effective_target_sync_int_128): Remove x86 because it provides
31034 no fast atomic load.
31035 (check_effective_target_sync_int_128_runtime): Likewise.
31037 2017-02-01 Richard Biener <rguenther@suse.de>
31039 * graphite.c: Include tree-vectorizer.h for find_loop_location.
31040 (graphite_transform_loops): Provide opt-info for optimized nests.
31041 * tree-parloop.c (parallelize_loops): Provide opt-info for
31042 parallelized loops.
31044 2017-02-01 Richard Biener <rguenther@suse.de>
31046 PR middle-end/79315
31047 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
31048 was not set before.
31050 2017-02-01 Richard Biener <rguenther@suse.de>
31052 PR tree-optimization/71824
31053 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
31054 Verify the loops are valid in the merged SESE region.
31055 (scop_detection::can_represent_loop_1): Check analyzing the
31056 evolution of the number of iterations in the region succeeds.
31058 2017-01-31 Ian Lance Taylor <iant@golang.org>
31060 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
31061 REG_ARGS_SIZE note to 32-bit push insns and call insn.
31063 2017-01-31 David Malcolm <dmalcolm@redhat.com>
31065 PR preprocessor/79210
31066 * input.c (get_substring_ranges_for_loc): Replace line_width
31067 assertion with error-handling.
31069 2017-01-31 Richard Biener <rguenther@suse.de>
31071 PR tree-optimization/77318
31072 * graphite-sese-to-poly.c (extract_affine): Fix assert.
31073 (create_pw_aff_from_tree): Take loop parameter.
31074 (add_condition_to_pbb): Pass loop of the condition to
31075 create_pw_aff_from_tree.
31077 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31079 * config/s390/s390.c (s390_asan_shadow_offset): New function.
31080 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
31082 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
31086 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
31088 (convert_int_to_float128): Likewise.
31089 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
31090 (convert_int_to_float128): Likewise.
31091 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
31092 (UNSPEC_IEEE128_CONVERT): Likewise.
31093 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
31094 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
31095 Use local variables for IBM extended format.
31096 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
31097 (fix_trunc<mode>si2_fprs): Likewise.
31098 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
31099 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
31100 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
31101 to know that we can now have integers of all sizes in vector
31103 (fix<uns>_<mode>di2_hw): Likewise.
31104 (float<uns>_<mode>si2_hw): Likewise.
31105 (fix_<mode>si2_hw): Likewise.
31106 (fixuns_<mode>si2_hw): Likewise.
31107 (float<uns>_<mode>di2_hw): Likewise.
31108 (float_<mode>di2_hw): Likewise.
31109 (float_<mode>si2_hw): Likewise.
31110 (floatuns_<mode>di2_hw): Likewise.
31111 (floatuns_<mode>si2_hw): Likewise.
31112 (xscvqp<su>wz_<mode>): Delete, no longer used.
31113 (xscvqp<su>dz_<mode>): Likewise.
31114 (xscv<su>dqp_<mode>): Likewise.
31115 (ieee128_mfvsrd_64bit): Likewise.
31116 (ieee128_mfvsrd_32bit): Likewise.
31117 (ieee128_mfvsrwz): Likewise.
31118 (ieee128_mtvsrw): Likewise.
31119 (ieee128_mtvsrd_64bit): Likewise.
31120 (ieee128_mtvsrd_32bit): Likewise.
31122 2017-01-31 Martin Liska <mliska@suse.cz>
31125 * ipa-prop.c (ipa_free_all_node_params): Call release method
31126 instead of ~sumbol_summary to not to trigger double times
31129 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
31131 PR tree-optimization/71691
31132 * bitmap.h (class auto_bitmap): New.
31133 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
31134 is_maybe_undefined instead of ssa_undefined_value_p.
31136 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31138 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
31139 __S390_ARCH_LEVEL__ to __ARCH__.
31141 2017-01-31 Jakub Jelinek <jakub@redhat.com>
31143 PR tree-optimization/79267
31144 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
31145 if should_remove_lhs_p is true.
31147 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
31150 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
31151 (add_alignment_attribute): New.
31152 (base_type_die): Add alignment attribute.
31153 (subrange_type_die): Likewise.
31154 (modified_type_die): Likewise.
31155 (gen_array_type_die): Likewise.
31156 (gen_descr_array_type_die: Likewise.
31157 (gen_enumeration_type_die): Likewise.
31158 (gen_subprogram_die): Likewise.
31159 (gen_variable_die): Likewise.
31160 (gen_field_die): Likewise.
31161 (gen_ptr_to_mbr_type_die): Likewise.
31162 (gen_struct_or_union_type_die): Likewise.
31163 (gen_subroutine_type_die): Likewise.
31164 (gen_typedef_die): Likewise.
31165 (base_type_cmp): Compare alignment attribute.
31167 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31170 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
31171 (setb_unsigned) New pattern for setb with CCUNS.
31172 * config/rs6000/rs6000.c (expand_block_compare): Use a different
31173 subfc./subfe sequence to avoid overflow problems. Generate a
31174 shorter sequence with cmpld/setb for power9.
31175 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
31176 for generating subfc. instruction.
31177 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
31178 now uses this instruction.
31180 2017-01-30 Ian Lance Taylor <iant@google.com>
31183 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
31184 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
31186 2017-01-30 Martin Sebor <msebor@redhat.com>
31188 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
31189 Move constant to the right of a relational operator.
31190 (get_mpfr_format_length, format_character, format_string): Ditto.
31191 (should_warn_p, maybe_warn): Same.
31193 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
31195 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
31198 * asan.c (get_translation_unit_decl): Remove function.
31199 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
31201 2017-01-30 Martin Liska <mliska@suse.cz>
31203 PR gcov-profile/79259
31204 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
31205 -fprofile-generate.
31207 2017-01-30 Martin Liska <mliska@suse.cz>
31210 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
31211 Initialize variables with NULL value.
31213 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
31216 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
31218 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
31220 2017-01-30 Richard Biener <rguenther@suse.de>
31222 PR tree-optimization/79276
31223 * tree-vrp.c (process_assert_insertions): Properly adjust common
31224 when removing a duplicate.
31226 2017-01-30 Richard Biener <rguenther@suse.de>
31228 PR tree-optimization/79256
31229 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
31230 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
31232 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
31234 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
31237 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
31238 ("*r<noxa>sbg_<mode>_sll_bitmask")
31239 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
31240 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
31241 Use contiguous_bitmask_nowrap_operand.
31243 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31246 * config/rs6000/altivec.h (vec_xl): Revise #define.
31247 (vec_xst): Likewise.
31249 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
31251 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
31253 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
31255 PR rtl-optimization/79194
31256 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
31257 traps before call to bypass_conditional_jumps.
31259 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31261 PR tree-optimization/71374
31262 * lra-constraints.c (check_conflict_input_operands): New.
31263 (match_reload): Use it.
31265 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
31268 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
31269 account to calculate conflict_set.
31271 2017-01-27 Bin Cheng <bin.cheng@arm.com>
31273 PR rtl-optimization/78559
31274 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
31275 other_insn in combine.
31277 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
31279 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
31280 uint16_type_node for BT_UINT16.
31282 2017-01-27 David Malcolm <dmalcolm@redhat.com>
31284 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
31285 "RTL Tests" to menu.
31286 (GIMPLE Tests): New node.
31287 (RTL Tests): New node.
31289 2017-01-27 Richard Biener <rguenther@suse.de>
31291 PR tree-optimization/79245
31292 * tree-loop-distribution.c (distribute_loop): Apply cost
31293 modeling also to detected patterns.
31295 2017-01-27 Richard Biener <rguenther@suse.de>
31297 PR tree-optimization/71433
31298 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
31299 (compare_assert_loc): New function.
31300 (process_assert_insertions): Sort and optimize assert locations
31301 to remove duplicates and push down identical assertions on
31302 edges to their destination block.
31304 2017-01-27 Richard Biener <rguenther@suse.de>
31306 PR tree-optimization/79244
31307 * tree-vrp.c (remove_range_assertions): Forcefully propagate
31308 out SSA names even if abnormal.
31310 2017-01-27 Jakub Jelinek <jakub@redhat.com>
31312 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
31313 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
31314 instead of MPFR_RNDN.
31316 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
31319 * arm.c (arm_option_override): Don't call build_target_option_node
31320 until after doing all option overrides.
31321 (arm_valid_target_attribute_tree): Likewise.
31323 2017-01-27 Martin Liska <mliska@suse.cz>
31325 * doc/invoke.texi (-fprofile-arcs): Document profiling support
31326 for {cd}tors and C++ {cd}tors.
31328 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
31330 * config/s390/s390.md ("*setmem_long_and")
31331 ("*setmem_long_and_31z"): Use zero_extend instead of and.
31333 2017-01-26 Martin Sebor <msebor@redhat.com>
31335 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
31338 2017-01-26 Martin Sebor <msebor@redhat.com>
31340 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
31341 HAVE_DFmode before using XFmode or DFmode.
31342 (parse_directive): Avoid using the z length modifier to avoid
31343 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
31345 PR middle-end/78703
31346 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
31347 to accept adjustment as an array.
31348 (get_int_range): New function.
31349 (struct directive): Make width and prec arrays.
31350 (directive::set_width, directive::set_precision): Call get_int_range.
31351 (format_integer, format_floating): Handle width and precision ranges.
31352 (format_string, parse_directive): Same.
31354 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31357 * dwarf2out.c (generate_skeleton_bottom_up): For children with
31358 comdat_type_p set, just clone them, but keep the children in the
31362 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
31363 which have direct callers with -fvar-tracking-assignments enabled
31365 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
31366 inside of type units.
31368 2017-01-26 Martin Sebor <msebor@redhat.com>
31370 PR middle-end/78703
31371 * gimple-ssa-sprintf.c (struct result_range): Add likely and
31373 (struct format_result): Replace number_chars, number_chars_min,
31374 and number_chars_max with a single member of struct result_range.
31376 (format_result::operator+=): Adjust.
31377 (struct fmtresult): Remove bounded. Handle likely and unlikely
31379 (fmtresult::adjust_for_width_or_precision): New function.
31380 (fmtresult:type_max_digits): New function.
31381 (bytes_remaining): Handle likely and unlikely counters.
31382 (min_bytes_remaining): Remove.
31383 (format_percent): Simplify.
31384 (format_integer, format_floating): Set likely and unlikely counters.
31385 (get_string_length, format_character, format_string): Same.
31386 (format_plain, should_warn_p): New function.
31387 (maybe_warn): Call should_warn_p. Update diagnostic messages
31388 and handle those for all directives, including plain strings.
31389 (format_directive): Handle likely and unlikely counters.
31390 Remove unnecessary quoting from diagnostics. Add an informational
31392 (add_bytes): Remove.
31393 (pass_sprintf_length::compute_format_length): Simplify.
31394 (try_substitute_return_value): Handle likely and unlikely counters.
31396 2017-01-26 Carl Love <cel@us.ibm.com>
31398 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
31399 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
31401 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
31404 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
31405 endianess for subregs into account.
31406 * lra-constraints.c (lra_constraints): Do risky transformations
31407 always on the first iteration.
31408 * lra-lives.c (check_pseudos_live_through_calls): Add arg
31409 last_call_used_reg_set.
31410 (process_bb_lives): Define and use last_call_used_reg_set.
31411 * lra.c (lra): Always continue after lra_constraints on the first
31414 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
31416 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
31418 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
31420 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31422 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
31423 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
31424 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
31425 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
31426 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
31427 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
31428 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
31429 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
31430 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
31432 2017-01-26 Marek Polacek <polacek@redhat.com>
31435 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
31436 for the third operand.
31438 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31440 PR middle-end/79236
31441 * omp-low.c (struct omp_context): Add simt_stmt field.
31442 (scan_omp_for): Return omp_context *.
31443 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
31444 context to the _simt_ SIMD stmt.
31445 (lower_omp_for): For combined SIMD with sibling _simt_
31446 SIMD, make sure to use the same decls in _looptemp_
31447 clauses as in the sibling.
31449 2017-01-26 David Sherwood <david.sherwood@arm.com>
31451 PR middle-end/79212
31452 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
31455 2017-01-26 Jakub Jelinek <jakub@redhat.com>
31458 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
31459 emit fld b; fld a; if possible.
31461 * brig-builtins.def: Update copyright years.
31462 * config/arm/arm_acle_builtins.def: Update copyright years.
31464 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
31467 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
31468 constraint instead of o for the stxsd instruction.
31470 2017-01-25 Carl Love <cel@us.ibm.com>
31472 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
31473 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
31475 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
31477 * doc/invoke.texi (C++ Dialect Options): Fix typo.
31479 2017-01-25 Richard Biener <rguenther@suse.de>
31481 PR tree-optimization/69264
31482 * target.def (vector_alignment_reachable): Improve documentation.
31483 * doc/tm.texi: Regenerate.
31484 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
31486 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
31487 earlier changes with respect to TYPE_USER_ALIGN.
31488 (vector_alignment_reachable_p): Likewise. Improve dumping.
31490 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31493 * config/arm/arm.md (xordi3): Force constant operand into a register
31496 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31498 * doc/invoke.texi (-fstore-merging): Correct default optimization
31499 levels at which it is enabled.
31500 (-O): Move -fstore-merging from list to...
31503 2017-01-25 Richard Biener <rguenther@suse.de>
31506 * omp-expand.c: Include debug.h.
31507 (expand_omp_taskreg): Make sure to generate early debug before
31508 outlining anything from a function.
31509 (expand_omp_target): Likewise.
31510 (grid_expand_target_grid_body): Likewise.
31512 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
31515 * asan.c (get_translation_unit_decl): New function.
31516 (asan_add_global): Extract modules file name from globals
31517 TRANSLATION_UNIT_DECL name.
31519 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
31522 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
31523 for long calls with APCS frame and VFP.
31525 2017-01-24 David Malcolm <dmalcolm@redhat.com>
31527 * cfg.c (original_copy_tables_initialized_p): New function.
31528 * cfg.h (original_copy_tables_initialized_p): New decl.
31529 * cfgrtl.c (relink_block_chain): Guard the call to
31530 free_original_copy_tables with a call to
31531 original_copy_tables_initialized_p.
31532 * cgraph.h (symtab_node::native_rtl_p): New decl.
31533 * cgraphunit.c (symtab_node::native_rtl_p): New function.
31534 (symtab_node::needed_p): Don't assert for early assembly output
31535 for __RTL functions.
31536 (cgraph_node::finalize_function): Set "force_output" for __RTL
31538 (cgraph_node::analyze): Bail out early for __RTL functions.
31539 (analyze_functions): Update assertion to support __RTL functions.
31540 (cgraph_node::expand): Bail out early for __RTL functions.
31541 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
31543 * function.h (struct function): Update comment for field
31545 * gimple-expr.c: Include "tree-pass.h".
31546 (gimple_has_body_p): Return false for __RTL functions.
31547 * Makefile.in (OBJS): Add run-rtl-passes.o.
31548 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
31550 (gcc::pass_manager::get_clean_slate): New accessor.
31551 * passes.c: Include "insn-addr.h".
31552 (should_skip_pass_p): Add logging. Update logic for running
31553 "expand" to be compatible with both __GIMPLE and __RTL. Guard
31554 property-provider override so it is only done for gimple passes.
31556 (skip_pass): New function.
31557 (execute_one_pass): Call skip_pass when skipping passes.
31558 * read-md.c (md_reader::read_char): Support filtering
31559 the input to a subset of line numbers.
31560 (md_reader::md_reader): Initialize fields
31561 m_first_line and m_last_line.
31562 (md_reader::read_file_fragment): New function.
31563 * read-md.h (md_reader::read_file_fragment): New decl.
31564 (md_reader::m_first_line): New field.
31565 (md_reader::m_last_line): New field.
31566 * read-rtl-function.c (function_reader::create_function): Only
31567 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
31568 curr_properties. Set DECL_INITIAL to a dummy block.
31569 (read_rtl_function_body_from_file_range): New function.
31570 * read-rtl-function.h (read_rtl_function_body_from_file_range):
31572 * run-rtl-passes.c: New file.
31573 * run-rtl-passes.h: New file.
31575 2017-01-24 Jeff Law <law@redhat.com>
31577 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
31580 2017-01-24 Bin Cheng <bin.cheng@arm.com>
31582 PR tree-optimization/79159
31583 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
31584 (record_nonwrapping_iv): Improve boundary using above function if no
31585 value range information.
31587 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
31588 Martin Jambor <mjambor@suse.cz>
31590 * brig-builtins.def: New file.
31591 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
31592 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
31593 (DEF_HSAIL_SAT_BUILTIN): Likewise.
31594 (DEF_HSAIL_INTR_BUILTIN): Likewise.
31595 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
31596 * builtin-types.def (BT_INT8): New.
31597 (BT_INT16): Likewise.
31598 (BT_UINT8): Likewise.
31599 (BT_UINT16): Likewise.
31600 (BT_FN_ULONG): Likewise.
31601 (BT_FN_UINT_INT): Likewise.
31602 (BT_FN_UINT_ULONG): Likewise.
31603 (BT_FN_UINT_LONG): Likewise.
31604 (BT_FN_UINT_PTR): Likewise.
31605 (BT_FN_ULONG_PTR): Likewise.
31606 (BT_FN_INT8_FLOAT): Likewise.
31607 (BT_FN_INT16_FLOAT): Likewise.
31608 (BT_FN_UINT32_FLOAT): Likewise.
31609 (BT_FN_UINT16_FLOAT): Likewise.
31610 (BT_FN_UINT8_FLOAT): Likewise.
31611 (BT_FN_UINT64_FLOAT): Likewise.
31612 (BT_FN_UINT16_UINT32): Likewise.
31613 (BT_FN_UINT32_UINT16): Likewise.
31614 (BT_FN_UINT16_UINT16_UINT16): Likewise.
31615 (BT_FN_INT_PTR_INT): Likewise.
31616 (BT_FN_UINT_PTR_UINT): Likewise.
31617 (BT_FN_LONG_PTR_LONG): Likewise.
31618 (BT_FN_ULONG_PTR_ULONG): Likewise.
31619 (BT_FN_VOID_UINT64_UINT64): Likewise.
31620 (BT_FN_UINT8_UINT8_UINT8): Likewise.
31621 (BT_FN_INT8_INT8_INT8): Likewise.
31622 (BT_FN_INT16_INT16_INT16): Likewise.
31623 (BT_FN_INT_INT_INT): Likewise.
31624 (BT_FN_UINT_FLOAT_UINT): Likewise.
31625 (BT_FN_FLOAT_UINT_UINT): Likewise.
31626 (BT_FN_ULONG_UINT_UINT): Likewise.
31627 (BT_FN_ULONG_UINT_PTR): Likewise.
31628 (BT_FN_ULONG_ULONG_ULONG): Likewise.
31629 (BT_FN_UINT_UINT_UINT): Likewise.
31630 (BT_FN_VOID_UINT_PTR): Likewise.
31631 (BT_FN_UINT_UINT_PTR: Likewise.
31632 (BT_FN_UINT32_UINT64_PTR): Likewise.
31633 (BT_FN_INT_INT_UINT_UINT): Likewise.
31634 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
31635 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
31636 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
31637 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
31638 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
31639 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
31640 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
31641 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
31642 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
31643 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
31644 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
31645 * doc/frontends.texi: List BRIG FE.
31646 * doc/install.texi (Testing): Add BRIG tesring requirements.
31647 * doc/invoke.texi (Overall Options): Mention BRIG.
31648 * doc/standards.texi (Standards): Doucment BRIG HSA version.
31650 2017-01-24 Richard Biener <rguenther@suse.de>
31652 PR translation/79208
31653 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
31655 2017-01-24 Martin Jambor <mjambor@suse.cz>
31658 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
31659 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
31660 and known_contexts.
31662 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
31664 PR middle-end/79123
31665 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
31666 casts from signed to unsigned really don't have a range.
31668 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
31670 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
31671 GMP_RNDx for compatiblity.
31673 2017-01-24 Martin Liska <mliska@suse.cz>
31676 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
31677 that would prevent us to call alloca with -1 as argument.
31679 2017-01-24 Jakub Jelinek <jakub@redhat.com>
31681 * dwarf2out.c (output_compilation_unit_header, output_file_names):
31682 Avoid -Wformat-security warning.
31684 2017-01-23 Andrew Pinski <apinski@cavium.com>
31686 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
31689 2017-01-23 Martin Sebor <msebor@redhat.com>
31691 PR middle-end/78703
31692 * gimple-ssa-sprintf.c (warn_level): New global.
31693 (format_integer): Use it here and throughout the rest of the file.
31694 Use the same switch to compute sign as base.
31695 (maybe_warn): New function.
31696 (format_directive): Factor out warnings into maybe_warn.
31697 Add debugging output. Use warn_level.
31698 (add_bytes): Use warn_level.
31699 (pass_sprintf_length::compute_format_length): Add debugging output.
31700 (try_substitute_return_value): Same.
31701 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
31703 PR middle-end/78703
31704 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
31705 (struct fmtresult, format_integer, format_floating): Adjust.
31706 (fmtresult::fmtresult): Set max correctly in two argument ctor.
31707 (get_string_length, format_string,format_directive): Same.
31708 (pass_sprintf_length::compute_format_length): Same.
31709 (try_substitute_return_value): Simplify slightly.
31711 PR middle-end/78703
31712 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
31713 (fmtresult::operator+=): Outlined.
31714 (struct fmtresult): Add ctors.
31715 (struct conversion_spec): Rename...
31716 (struct directive): ...to this. Add and remove data members.
31717 (directive::set_width, directive::set_precision): New functions.
31718 (format_percent): Use fmtresult ctor.
31719 (get_width_and_precision): Remove.
31720 (format_integer): Make naming changes. Avoid computing width and
31722 (format_floating): Same. Adjust indentation.
31723 (format_character, format_none): New functions.
31724 (format_string): Moved character handling to format_character.
31725 (format_directive): Remove arguments, change return type.
31726 (parse_directive): New function.
31727 (pass_sprintf_length::compute_format_length): Move directive
31728 parsing to parse_directive.
31730 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31732 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
31733 (assign_assembler_name_if_needed): ... this.
31734 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
31735 (assign_assembler_name_if_needed): ... this.
31736 (free_lang_data_in_cgraph): Adjust callers.
31737 * cgraphunit.c (cgraph_node::analyze): Likewise.
31738 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
31741 2017-01-23 Richard Biener <rguenther@suse.de>
31743 PR tree-optimization/79088
31744 PR tree-optimization/79188
31745 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
31746 resetting loop bounds after last path deletion. Reset loop
31747 bounds of the target loop, make code match the comments.
31748 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
31749 Make sure loops need no fixups.
31751 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
31753 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
31754 exponent support with double type for first argument.
31755 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
31756 type returned by __builtin_vec_extract_sig,
31757 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
31758 functions from "vector int" to "vector unsigned int" or from
31759 "vector long long int" to "vector unsigned long long int".
31760 Changed type returned by __builtin_vec_extract_exp,
31761 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
31762 functions from "vector int" to "vector unsigned int" or from
31763 "vector long long int" to "vector unsigned long long int".
31764 Changed return type of __builtin_vec_test_data_class,
31765 __builtin_vec_test_data_class_sp, and
31766 __builtin_vec_test_data_class_dp from "vector int" to
31767 "vector bool int" or from "vector long long int" to "vector bool
31768 long long int" and changed second argument type from "unsigned
31769 int" to "int". Added new overloaded function forms "vector float
31770 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
31771 "vector float __builtin_vec_insert_exp_sp (vector float, vector
31772 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
31773 double, vector unsigned long long int)" and "vector double
31774 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
31775 long int)". Changed return type of
31776 __builtin_scalar_test_data_class and
31777 __builtin_scalar_test_data_class_sp and
31778 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
31779 int" and changed second argument from "unsigned int" to "int".
31780 Changed type returned by __builtin_scalar_test_neg,
31781 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
31782 from "int" to "bool int". Added new overloaded function form
31783 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
31784 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
31785 exponent double-precision with floating point first argument.
31786 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
31787 documentation of scalar_test_data_class, scalar_test_neg,
31788 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
31789 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
31790 vec_test_data_class built-in functions to reflect refinements in
31791 their type signatures.
31793 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
31795 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
31797 (aarch64_elf_asm_destructor): Likewise.
31799 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
31801 PR rtl-optimization/78634
31802 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
31803 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
31804 * ifcvt.c (noce_try_cmove): Add missing cost check.
31806 PR rtl-optimization/71724
31807 * combine.c (if_then_else_cond): Look for situations where it is
31808 beneficial to undo the work of one of the recursive calls.
31810 2017-01-23 Bin Cheng <bin.cheng@arm.com>
31812 PR tree-optimization/70754
31813 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
31814 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
31815 combined stmt before it if not NULL.
31816 (combine_chains): Process refs reversely and compute dominance point
31819 2017-01-23 Martin Liska <mliska@suse.cz>
31821 PR tree-optimization/79196
31822 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
31823 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
31825 (strlen_optimize_stmt): Call the renamed function.
31827 2017-01-23 Michael Matz <matz@suse.de>
31829 PR tree-optimization/78384
31830 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
31832 2017-01-23 Richard Biener <rguenther@suse.de>
31834 PR tree-optimization/79186
31835 * tree-vrp.c (register_new_assert_for): Make sure we've seen
31836 both incoming edges before moving an assert.
31838 2017-01-23 Martin Jambor <mjambor@suse.cz>
31840 * ipa-prop.c (load_from_param_1): Removed.
31841 (load_from_unmodified_param): Bits from load_from_param_1 put back
31843 (load_from_param): Removed.
31844 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
31845 with stmt. Reverted back to use of load_from_unmodified_param.
31847 2017-01-23 Martin Jambor <mjambor@suse.cz>
31850 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
31851 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
31852 field a pointer to garbage collected vector, mark lattices and
31853 ipcp_orig_node with GTY((skip)).
31854 (ipa_get_param_count): Adjust to descriptors being a pointer.
31855 (ipa_get_param): Likewise.
31856 (ipa_get_type): Likewise.
31857 (ipa_get_param_move_cost): Likewise.
31858 (ipa_set_param_used): Likewise.
31859 (ipa_get_controlled_uses): Likewise.
31860 (ipa_set_controlled_uses): Likewise.
31861 (ipa_is_param_used): Likewise.
31862 (ipa_node_params_t): Move into garbage collector. New methods insert
31864 (ipa_node_params_sum): Annotate wth GTY(()).
31865 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
31867 (ipa_load_from_parm_agg): Adjust declaration.
31868 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
31869 * ipa-profile.c (ipa_profile): Likewise.
31870 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
31871 (ipa_populate_param_decls): Make descriptors parameter garbage
31873 (ipa_dump_param): Adjust to descriptors being a pointer.
31874 (ipa_alloc_node_params): Likewise.
31875 (ipa_initialize_node_params): Likewise.
31876 (load_from_param_1): Make descriptors parameter garbage collected.
31877 (load_from_unmodified_param): Likewise.
31878 (load_from_param): Likewise.
31879 (ipa_load_from_parm_agg): Likewise.
31880 (ipa_node_params::~ipa_node_params): Removed.
31881 (ipa_free_all_node_params): Remove call to delete operator.
31882 (ipa_node_params_t::insert): New.
31883 (ipa_node_params_t::remove): Likewise.
31884 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
31885 copy known_csts and known_contexts vectors.
31886 (ipa_read_node_info): Adjust to descriptors being a pointer.
31887 (ipcp_modif_dom_walker): Make m_descriptors field garbage
31889 (ipcp_transform_function): Make descriptors variable garbage
31892 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
31894 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
31895 * config/i386/avx512dqintrin.h: Ditto.
31896 * config/i386/avx512fintrin.h: Ditto.
31897 * config/i386/i386.c: Handle new builtins.
31898 * config/i386/i386-builtin.def: Add new builtins.
31899 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
31900 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
31902 2017-01-23 Jakub Jelinek <jakub@redhat.com>
31903 Martin Liska <mliska@suse.cz>
31905 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
31906 * asan.c (asan_expand_poison_ifn): Support stores and use
31907 appropriate ASAN report function.
31908 * internal-fn.c (expand_ASAN_POISON_USE): New function.
31909 * internal-fn.def (ASAN_POISON_USE): Declare.
31910 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
31911 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
31912 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
31913 ASAN_POISON calls w/o LHS.
31914 * tree-ssa.c (execute_update_addresses_taken): Create clobber
31915 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
31916 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
31917 * gimplify.c (asan_poison_variables): Add attribute
31918 use_after_scope_memory to variables that really needs to live
31920 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
31921 having the attribute.
31923 2017-01-23 Martin Liska <mliska@suse.cz>
31925 * asan.c (create_asan_shadow_var): New function.
31926 (asan_expand_poison_ifn): Likewise.
31927 * asan.h (asan_expand_poison_ifn): New declaration.
31928 * internal-fn.c (expand_ASAN_POISON): Likewise.
31929 * internal-fn.def (ASAN_POISON): New builtin.
31930 * sanopt.c (pass_sanopt::execute): Expand
31931 asan_expand_poison_ifn.
31932 * tree-inline.c (copy_decl_for_dup_finish): Make function
31934 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
31935 * tree-ssa.c (is_asan_mark_p): New function.
31936 (execute_update_addresses_taken): Rewrite local variables
31937 (identified just by use-after-scope as addressable) into SSA.
31939 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31941 * doc/install.texi (Specific): opensource.apple.com uses https
31942 now. Remove trailing slash.
31944 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
31946 * README.Portability: Remove note on an Irix compatibility issue.
31948 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
31950 * gcov.c (INCLUDE_ALGORITHM): Define.
31951 (INCLUDE_VECTOR): Define.
31952 No longer include <vector> and <algorithm> directly.
31954 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
31956 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
31958 * doc/invoke.texi (Code Gen Options): Ditto.
31960 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
31963 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
31965 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
31967 rtl-optimization/79125
31968 * cprop.c (local_cprop_pass): Handle cases where we make an
31969 unconditional trap.
31971 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
31975 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
31976 read from, for big endian.
31978 2017-01-20 Jiong Wang <jiong.wang@arm.com>
31980 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
31981 register pauth builtins for LP64 only.
31983 2017-01-20 Marek Polacek <polacek@redhat.com>
31986 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
31989 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
31991 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
31993 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
31994 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
31995 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
31997 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32000 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
32001 in big-endian mode when they are not a single duplicated value.
32003 2017-01-20 Richard Biener <rguenther@suse.de>
32005 * BASE-VER: Bump to 7.0.1.
32007 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
32009 * omp-low.c (omplow_simd_context): New struct. Use it...
32010 (lower_rec_simd_input_clauses): ...here and...
32011 (lower_rec_input_clauses): ...here to hold common data. Adjust all
32012 references to idx, lane, max_vf, is_simt.
32014 2017-01-20 Graham Markall <graham.markall@embecosm.com>
32016 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
32019 2017-01-20 Martin Jambor <mjambor@suse.cz>
32021 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
32022 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
32024 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
32025 (GTFILES): Rename hsa.c to hsa-common.c.
32026 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
32027 * hsa-dump.c: Likewise.
32028 * hsa-gen.c: Likewise.
32029 * hsa-regalloc.c: Likewise.
32030 * ipa-hsa.c: Likewise.
32031 * omp-expand.c: Likewise.
32032 * omp-low.c: Likewise.
32033 * toplev.c: Likewise.
32035 2017-01-20 Marek Polacek <polacek@redhat.com>
32038 * doc/invoke.texi: Document -Wduplicated-branches.
32039 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
32040 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
32041 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
32042 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
32043 return 0 only when not OEP_LEXICOGRAPHIC.
32044 (fold_build_cleanup_point_expr): Use the expression
32045 location when building CLEANUP_POINT_EXPR.
32046 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
32047 * tree.c (add_expr): Handle error_mark_node.
32049 2017-01-20 Martin Liska <mliska@suse.cz>
32052 * tree-profile.c (init_ic_make_global_vars): Do not call
32054 (gimple_init_gcov_profiler): Likewise.
32056 2017-01-20 Martin Liska <mliska@suse.cz>
32059 * cgraph.h (maybe_create_reference): Remove argument and
32061 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
32063 * ipa-cp.c (create_specialized_node): Likewise.
32064 * symtab.c (symtab_node::maybe_create_reference): Handle
32065 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
32067 2017-01-20 Martin Liska <mliska@suse.cz>
32069 * read-rtl-function.c (function_reader::create_function): Use
32070 build_decl instread of build_decl_stat.
32072 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
32074 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
32075 * config/i386/avx512dqintrin.h: Ditto.
32076 * config/i386/avx512fintrin.h: Ditto.
32077 * config/i386/i386-builtin-types.def: Add new types.
32078 * config/i386/i386.c: Handle new types.
32079 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
32080 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
32081 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
32082 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
32083 (__builtin_ia32_kshiftridi): New.
32084 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
32086 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
32090 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
32091 define to rs6000_init_stack_protect_guard.
32092 (rs6000_init_stack_protect_guard): New function.
32094 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32095 Yunqiang Su <yunqiang.su@imgtec.com>
32097 * config.gcc (supported_defaults): Add madd4.
32098 (with_madd4): Add validation.
32099 (all_defaults): Add madd4.
32100 * config/mips/mips.opt (mmadd4): New option.
32101 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32103 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32105 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
32106 (ISA_HAS_FUSED_MADD4): Likewise.
32107 * doc/invoke.texi (-mmadd4): Document the new option.
32108 * doc/install.texi (--with-madd4): Document the new option.
32110 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32112 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
32113 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
32114 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
32115 (aarch64_init_pauth_hint_builtins): New.
32116 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
32117 (aarch64_expand_builtin): Expand new builtins.
32119 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32121 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
32122 * combine-stack-adj.c (no_unhandled_cfa): Handle
32123 REG_CFA_TOGGLE_RA_MANGLE.
32124 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
32125 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
32126 info for return address signing.
32127 (aarch64_expand_epilogue): Likewise.
32129 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32131 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
32132 * config/aarch64/aarch64-protos.h
32133 (aarch64_return_address_signing_enabled): New declaration.
32134 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
32136 (aarch64_expand_prologue): Sign return address before it's pushed onto
32138 (aarch64_expand_epilogue): Authenticate return address fetched from
32140 (aarch64_override_options): Sanity check for ILP32 and ISA level.
32141 (aarch64_attributes): New function attributes for "sign-return-address".
32142 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
32143 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
32144 ("*do_return"): Generate combined instructions according to key index.
32145 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
32146 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
32148 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
32149 * config/aarch64/aarch64.opt (msign-return-address=): New.
32150 * doc/extend.texi (AArch64 Function Attributes): Documents
32151 "sign-return-address=".
32152 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
32154 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
32156 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
32157 overall option summary.
32159 2017-01-19 Jiong Wang <jiong.wang@arm.com>
32161 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
32162 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
32163 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
32164 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
32166 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
32168 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
32169 -mpower9-minmax by default for -mcpu=power9.
32170 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
32171 128-bit floating point.
32173 2017-01-20 Alan Modra <amodra@gmail.com>
32175 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
32176 optimizing for size.
32178 2017-01-20 Alan Modra <amodra@gmail.com>
32181 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
32182 for strcmp and strncmp from corresponding builtin decl.
32184 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32186 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
32187 instead of i386/rtems-64.h.
32188 * config/i386/rtems-64.h: Remove.
32190 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
32194 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
32196 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
32198 2017-01-19 Tamar Christina <tamar.christina@arm.com>
32200 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
32201 Change int to HOST_WIDE_INT.
32202 * config/aarch64/aarch64-protos.h
32203 (aarch64_simd_gen_const_vector_dup): Likewise.
32204 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
32206 2017-01-19 David Malcolm <dmalcolm@redhat.com>
32208 * langhooks-def.h (lhd_type_for_size): New decl.
32209 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
32210 * langhooks.c (lhd_type_for_size): New function, taken from
32213 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
32215 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
32216 define_bypass for CR latency.
32217 (power9-cracked-alu): Update bypass latency and remove power9-branch.
32218 (power9-alu2): Add define_bypass for CR latency.
32220 (power9-mul): Update insn latency.
32221 (power9-mul-compare): Update insn latency, bypass latency and remove
32224 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32226 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
32228 * config/aarch64/aarch64.md
32229 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
32230 aarch64_nopcrelative_literal_loads.
32231 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
32233 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
32235 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
32236 TARGET_LOONGSON_3A.
32237 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
32239 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
32242 * config.gcc (supported_defaults): Add lxc1-sxc1.
32243 (with_lxc1_sxc1): Add validation.
32244 (all_defaults): Add lxc1-sxc1.
32245 * config/mips/mips.opt (mlxc1-sxc1): New option.
32246 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
32248 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
32249 __mips_no_lxc1_sxc1.
32250 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
32251 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
32252 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
32254 2017-01-19 Richard Biener <rguenther@suse.de>
32256 PR tree-optimization/72488
32257 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
32258 sure to restore SSA info.
32259 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
32261 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
32263 PR rtl-optimization/79121
32264 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
32265 of the inner type when shifting an extended value.
32267 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32270 * symtab.c (symtab_node::equal_address_to): Fix comparing of
32271 interposable aliases.
32273 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
32276 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
32277 Use the evmergelohi instruction.
32278 (mov_si<mode>_e500_subreg4_2_le): Likewise.
32279 (mov_sitf_e500_subreg8_2_be): Likewise.
32280 (mov_sitf_e500_subreg12_2_le): Likewise.
32281 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
32282 (mov_si<mode>_e500_subreg4_2_be): Likewise.
32283 (mov_sitf_e500_subreg8_2_le): Likewise.
32284 (mov_sitf_e500_subreg12_2_be): Likewise.
32286 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32288 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
32289 attribute from vecsimple to vecperm.
32290 (altivec_vbpermq2): Likewise.
32292 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32295 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
32297 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
32298 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
32299 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
32300 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
32301 case where N arg is SIZE_MAX.
32302 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
32303 (cmpstrsi): Add pattern.
32305 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
32307 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
32308 __builtin_vec_revb builtins.
32309 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
32310 built-in functions to support generation of the ISA 3.0 XXBR<x>
32311 vector byte reverse instructions.
32312 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
32313 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
32314 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
32315 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
32316 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
32317 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
32318 (P9V_BUILTIN_VEC_REVB): Likewise.
32319 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
32320 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
32321 (p9_xxbrq_v16qi): Likewise.
32322 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
32323 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
32324 (p9_xxbrh_v8hi): Likewise.
32325 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
32326 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
32327 vec_revb built-in functions.
32329 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
32331 PR rtl-optimization/78952
32332 * config/i386/i386.md (any_extract): New code iterator.
32333 (*insvqi_2): Use any_extract for source operand.
32334 (*insvqi_3): Use any_shiftrt for source operand.
32336 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
32338 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
32340 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
32342 2017-01-18 Matthias Klose <doko@ubuntu.com>
32344 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
32346 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32348 * config/rs6000/altivec.h (vec_bperm): Change #define.
32349 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
32350 (altivec_vbpermq2): New define_insn.
32351 (altivec_vbpermd): Likewise.
32352 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
32353 function interface.
32354 (VBPERMD): Likewise.
32355 (VBPERM): New polymorphic function interface.
32356 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
32357 Add entries for P9V_BUILTIN_VEC_VBPERM.
32358 * doc/extend.texi: Add interfaces for vec_bperm.
32360 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32362 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
32363 first letter of error messages.
32364 (s390_resolve_overloaded_builtin): Likewise.
32365 * config/s390/s390.c (s390_expand_builtin): Likewise.
32366 (s390_invalid_arg_for_unprototyped_fn): Likewise.
32367 (s390_valid_target_attribute_inner_p): Likewise.
32368 * config/s390/s390.md ("tabort"): Likewise.
32370 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
32372 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
32373 (ISA_AVOID_DIV_HILO): New macro.
32374 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
32375 (ISA_HAS_DDIV): Likewise.
32377 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32379 * doc/invoke.texi (fabi-version): Correct number of occurrences.
32381 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32383 * doc/invoke.texi (fabi-version): Spelling fix.
32385 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32388 * doc/invoke.texi (fabi-version): Mention mangling fix for
32391 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
32394 * doc/invoke.texi (fabi-version): Document discriminator mangling.
32396 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
32399 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
32400 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
32402 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
32404 (stack_protect_test): Ditto.
32405 * config/rs6000/rs6000.opt (mstack-protector-guard=,
32406 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
32408 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
32409 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
32410 -mstack-protector-guard-offset=.
32411 (RS/6000 and PowerPC Options): Ditto.
32413 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32415 * config/i386/i386.h (MASK_CLASS_P): New define.
32416 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
32417 there are no registers from different register sets also when
32418 mask registers are used. Update function comment.
32419 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
32420 to (*k/*r) and (*k/*km) alternatives.
32422 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
32424 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
32425 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
32426 (EH_RETURN_HANDLER_RTX): New define.
32427 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
32428 Force frame pointer in EH return functions.
32429 (aarch64_expand_epilogue): Add barrier for eh_return.
32430 (aarch64_final_eh_return_addr): Remove.
32431 (aarch64_eh_return_handler_rtx): New function.
32432 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
32434 (aarch64_eh_return_handler_rtx): New prototype.
32436 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32438 * config/rs6000/altivec.h (vec_rlmi): New #define.
32439 (vec_vrlnm): Likewise.
32440 (vec_rlnm): Likewise.
32441 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
32442 (UNSPEC_VRLNM): Likewise.
32443 (VIlong): New mode iterator.
32444 (altivec_vrl<VI_char>mi): New define_insn.
32445 (altivec_vrl<VI_char>nm): Likewise.
32446 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
32448 (VRLDNM): Likewise.
32449 (RLNM): New polymorphic function entry.
32450 (VRLWMI): New monomorphic function entry.
32451 (VRLDMI): Likewise.
32452 (RLMI): New polymorphic function entry.
32453 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
32454 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
32455 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
32458 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32461 * dwarf2out.c (field_byte_offset): Restore the
32462 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
32463 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
32464 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
32467 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32470 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
32472 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32475 * dwarf2out.c (add_data_member_location_attribute): For constant
32476 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
32477 instead of DW_AT_data_member_location, DW_AT_bit_offset and
32478 DW_AT_byte_size attributes.
32480 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
32482 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
32483 after forcing to constant memory when the code model is medium.
32485 2017-01-17 Julia Koval <julia.koval@intel.com>
32488 * config/i386/avx512fintrin.h
32489 (_mm512_i32gather_ps): Change __addr type to void const*.
32490 (_mm512_mask_i32gather_ps): Ditto.
32491 (_mm512_i32gather_pd): Ditto.
32492 (_mm512_mask_i32gather_pd): Ditto.
32493 (_mm512_i64gather_ps): Ditto.
32494 (_mm512_mask_i64gather_ps): Ditto.
32495 (_mm512_i64gather_pd): Ditto.
32496 (_mm512_mask_i64gather_pd): Ditto.
32497 (_mm512_i32gather_epi32): Ditto.
32498 (_mm512_mask_i32gather_epi32): Ditto.
32499 (_mm512_i32gather_epi64): Ditto.
32500 (_mm512_mask_i32gather_epi64): Ditto.
32501 (_mm512_i64gather_epi32): Ditto.
32502 (_mm512_mask_i64gather_epi32): Ditto.
32503 (_mm512_i64gather_epi64): Ditto.
32504 (_mm512_mask_i64gather_epi64): Ditto.
32505 (_mm512_i32scatter_ps): Change __addr type to void*.
32506 (_mm512_mask_i32scatter_ps): Ditto.
32507 (_mm512_i32scatter_pd): Ditto.
32508 (_mm512_mask_i32scatter_pd): Ditto.
32509 (_mm512_i64scatter_ps): Ditto.
32510 (_mm512_mask_i64scatter_ps): Ditto.
32511 (_mm512_i64scatter_pd): Ditto.
32512 (_mm512_mask_i64scatter_pd): Ditto.
32513 (_mm512_i32scatter_epi32): Ditto.
32514 (_mm512_mask_i32scatter_epi32): Ditto.
32515 (_mm512_i32scatter_epi64): Ditto.
32516 (_mm512_mask_i32scatter_epi64): Ditto.
32517 (_mm512_i64scatter_epi32): Ditto.
32518 (_mm512_mask_i64scatter_epi32): Ditto.
32519 (_mm512_i64scatter_epi64): Ditto.
32520 (_mm512_mask_i64scatter_epi64): Ditto.
32521 * config/i386/avx512pfintrin.h
32522 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
32523 (_mm512_mask_prefetch_i32gather_ps): Ditto.
32524 (_mm512_mask_prefetch_i64gather_pd): Ditto.
32525 (_mm512_mask_prefetch_i64gather_ps): Ditto.
32526 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
32527 (_mm512_prefetch_i32scatter_ps): Ditto.
32528 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
32529 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
32530 (_mm512_prefetch_i64scatter_pd): Ditto.
32531 (_mm512_prefetch_i64scatter_ps): Ditto.
32532 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
32533 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
32534 * config/i386/avx512vlintrin.h
32535 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
32536 (_mm_mmask_i32gather_ps): Ditto.
32537 (_mm256_mmask_i32gather_pd): Ditto.
32538 (_mm_mmask_i32gather_pd): Ditto.
32539 (_mm256_mmask_i64gather_ps): Ditto.
32540 (_mm_mmask_i64gather_ps): Ditto.
32541 (_mm256_mmask_i64gather_pd): Ditto.
32542 (_mm_mmask_i64gather_pd): Ditto.
32543 (_mm256_mmask_i32gather_epi32): Ditto.
32544 (_mm_mmask_i32gather_epi32): Ditto.
32545 (_mm256_mmask_i32gather_epi64): Ditto.
32546 (_mm_mmask_i32gather_epi64): Ditto.
32547 (_mm256_mmask_i64gather_epi32): Ditto.
32548 (_mm_mmask_i64gather_epi32): Ditto.
32549 (_mm256_mmask_i64gather_epi64): Ditto.
32550 (_mm_mmask_i64gather_epi64): Ditto.
32551 (_mm256_i32scatter_ps): Change __addr type to void*.
32552 (_mm256_mask_i32scatter_ps): Ditto.
32553 (_mm_i32scatter_ps): Ditto.
32554 (_mm_mask_i32scatter_ps): Ditto.
32555 (_mm256_i32scatter_pd): Ditto.
32556 (_mm256_mask_i32scatter_pd): Ditto.
32557 (_mm_i32scatter_pd): Ditto.
32558 (_mm_mask_i32scatter_pd): Ditto.
32559 (_mm256_i64scatter_ps): Ditto.
32560 (_mm256_mask_i64scatter_ps): Ditto.
32561 (_mm_i64scatter_ps): Ditto.
32562 (_mm_mask_i64scatter_ps): Ditto.
32563 (_mm256_i64scatter_pd): Ditto.
32564 (_mm256_mask_i64scatter_pd): Ditto.
32565 (_mm_i64scatter_pd): Ditto.
32566 (_mm_mask_i64scatter_pd): Ditto.
32567 (_mm256_i32scatter_epi32): Ditto.
32568 (_mm256_mask_i32scatter_epi32): Ditto.
32569 (_mm_i32scatter_epi32): Ditto.
32570 (_mm_mask_i32scatter_epi32): Ditto.
32571 (_mm256_i32scatter_epi64): Ditto.
32572 (_mm256_mask_i32scatter_epi64): Ditto.
32573 (_mm_i32scatter_epi64): Ditto.
32574 (_mm_mask_i32scatter_epi64): Ditto.
32575 (_mm256_i64scatter_epi32): Ditto.
32576 (_mm256_mask_i64scatter_epi32): Ditto.
32577 (_mm_i64scatter_epi32): Ditto.
32578 (_mm_mask_i64scatter_epi32): Ditto.
32579 (_mm256_i64scatter_epi64): Ditto.
32580 (_mm256_mask_i64scatter_epi64): Ditto.
32581 (_mm_i64scatter_epi64): Ditto.
32582 (_mm_mask_i64scatter_epi64): Ditto.
32583 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
32584 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
32585 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
32586 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
32587 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
32588 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
32589 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
32590 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
32591 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
32592 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
32593 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
32594 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
32595 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
32596 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
32597 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
32598 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
32599 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
32600 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
32601 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
32602 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
32603 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
32604 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
32605 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
32606 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
32607 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
32608 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
32609 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
32610 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
32611 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
32612 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
32613 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
32614 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
32615 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
32616 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
32617 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
32618 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
32619 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
32620 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
32621 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
32622 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
32623 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
32624 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
32625 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
32626 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
32627 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
32628 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
32629 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
32630 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
32631 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
32632 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
32633 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
32634 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
32635 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
32636 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
32637 definitions accordingly.
32639 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
32640 Kuan-Lin Chen <kuanlinchentw@gmail.com>
32643 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
32646 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
32649 * ira-conflicts.c (ira_build_conflicts): Update total conflict
32650 hard regs for inner regno.
32652 2017-01-17 Martin Liska <mliska@suse.cz>
32655 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
32656 assumption and add comment.
32658 2017-01-17 Nathan Sidwell <nathan@acm.org>
32660 * ipa-visibility.c (localize_node): New function, broken out of ...
32661 (function_and_variable_visibility): ... here. Call it.
32663 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
32665 PR middle-end/77445
32666 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
32667 correctly set frequency of oudgoing edge.
32668 (duplicate_thread_path): Fix profile updating.
32670 2017-01-17 Jakub Jelinek <jakub@redhat.com>
32673 * configure.ac: Add GCC_BASE_VER.
32674 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
32675 version from BASE-VER file.
32676 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
32677 (gcc.o): Depend on $(BASEVER).
32678 * common.opt (dumpfullversion): New option.
32679 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
32680 * doc/invoke.texi: Document -dumpfullversion.
32681 * doc/install.texi: Document --with-gcc-major-version-only.
32682 * configure: Regenerated.
32684 2017-01-17 Richard Biener <rguenther@suse.de>
32686 PR tree-optimization/71433
32687 * tree-vrp.c (register_new_assert_for): Merge same asserts
32688 on all incoming edges.
32689 (process_assert_insertions_for): Handle insertions at the
32692 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
32694 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
32695 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
32697 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
32700 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
32703 2017-01-17 Alan Modra <amodra@gmail.com>
32706 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
32707 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
32708 symbolic stack limit when pic.
32710 2017-01-16 Martin Sebor <msebor@redhat.com>
32712 PR tree-optimization/78608
32713 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
32715 2017-01-16 Jeff Law <law@redhat.com>
32718 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
32719 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
32720 for several include directories that may be relative to sysroot.
32721 * config/i386/x-mingw32 (gplus_includedir): Define.
32722 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
32723 (native_system_includedir): Likewise.
32724 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
32725 override if TARGET_SYSTEM_ROOT is defined.
32726 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
32728 PR tree-optimization/79090
32729 PR tree-optimization/33562
32730 PR tree-optimization/61912
32731 PR tree-optimization/77485
32732 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
32733 and computed trims into the dump file.
32735 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
32737 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
32739 2017-01-16 Jakub Jelinek <jakub@redhat.com>
32742 * gimplify.c (gimplify_init_constructor): If want_value and
32743 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
32747 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
32748 sequence. Formatting fixes.
32749 (doloop_optimize): Formatting fixes.
32752 * gcc.c (debug_level_greater_than_spec_func): New function.
32753 (static_spec_functions): Add debug-level-gt spec function.
32754 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
32756 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32757 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
32758 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
32759 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
32760 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
32761 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
32763 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
32765 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
32766 QImode fixups to general and mask registers only.
32768 2017-01-16 Carl Love <cel@us.ibm.com>
32770 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
32771 for built-in functions
32772 vector signed char vec_nabs (vector signed char)
32773 vector signed short vec_nabs (vector signed short)
32774 vector signed int vec_nabs (vector signed int)
32775 vector signed long long vec_nabs (vector signed long long)
32776 vector float vec_nabs (vector float)
32777 vector double vec_nabs (vector double)
32778 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
32780 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
32781 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
32782 * doc/extend.texi: Update the documentation file for the new built-in
32785 2017-01-16 Martin Sebor <msebor@redhat.com>
32787 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
32790 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
32792 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
32793 UNSPEC_VSX__XXSPLTD to require special splat handling.
32795 2017-01-16 David Malcolm <dmalcolm@redhat.com>
32798 * system.h: Poison strndup.
32800 2017-01-16 Alan Modra <amodra@gmail.com>
32803 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
32806 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
32808 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
32810 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
32812 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
32813 call recog here. Assert that INSN_CODE (insn) is non-negative.
32815 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32818 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
32820 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
32821 in the currently scheduled RTL fragment.
32823 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
32825 PR rtl-optimization/78751
32826 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
32829 2017-01-14 Jeff Law <law@redhat.com>
32831 PR tree-optimization/79090
32832 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
32833 variable length stores.
32834 (compute_trims): Delete dead assignment to *trim_tail.
32835 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
32838 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
32840 PR rtl-optimization/78626
32841 PR rtl-optimization/78727
32842 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
32843 of a block, and split such blocks after everything else is finished.
32845 2017-01-14 Alan Modra <amodra@gmail.com>
32848 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
32849 target legitimate_combined_insn.
32850 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
32851 (rs6000_legitimate_combined_insn): New function.
32852 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
32854 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
32855 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
32856 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
32858 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
32860 * doc/frontends.texi (G++ and GCC): Remove references to Java.
32862 2017-01-13 Jeff Law <law@redhat.com>
32864 PR tree-optimization/33562
32865 PR tree-optimization/61912
32866 PR tree-optimization/77485
32867 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
32869 (delete_dead_assignment): Likewise.
32870 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
32871 statement to delete_dead_call and delete_dead_assignment.
32873 2017-01-13 David Malcolm <dmalcolm@redhat.com>
32876 * substring-locations.c (format_warning_va): Strengthen case 1 so
32877 that both endpoints of the substring must be within the format
32878 range for just the substring to be printed.
32880 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
32882 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
32883 * config/i386/i386.c (ix86_target_string): Add missing options
32884 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
32885 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
32886 flags_other and ix86_target_other to flags2_other. Display unknown
32888 (ix86_valid_target_attribute_inner_p): Add missing options and
32889 reorder options by implied ISAs, as in ix86_target_string.
32891 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32893 * hash-table.h (hash_table::too_empty_p): New function.
32894 (hash_table::expand): Use it.
32895 (hash_table::traverse): Likewise.
32896 (hash_table::empty_slot): Use sizeof (value_type) instead of
32897 sizeof (PTR) to convert bytes to elements. Shrink the table
32898 if the current size is excessive for the current number of
32901 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
32903 * ira-costs.c (record_reg_classes): Break from the inner loop
32904 early once alt_fail is known to be true. Update outer loop
32905 handling accordingly.
32907 2017-01-13 Jeff Law <law@redhat.com>
32909 * tree-ssa-dse.c (decrement_count): New function.
32910 (increment_start_addr, maybe_trim_memstar_call): Likewise.
32911 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
32912 when we know the partially dead statement is a mem* function.
32914 PR tree-optimization/61912
32915 PR tree-optimization/77485
32916 * tree-ssa-dse.c: Include expr.h.
32917 (maybe_trim_constructor_store): New function.
32918 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
32920 PR tree-optimization/33562
32921 PR tree-optimization/61912
32922 PR tree-optimization/77485
32923 * doc/invoke.texi: Document new dse-max-object-size param.
32924 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
32925 * tree-ssa-dse.c: Include params.h.
32926 (dse_store_status): New enum.
32927 (initialize_ao_ref_for_dse): New, partially extracted from
32929 (valid_ao_ref_for_dse, normalize_ref): New.
32930 (setup_live_bytes_from_ref, compute_trims): Likewise.
32931 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
32932 (maybe_trim_partially_dead_store): Likewise.
32933 (maybe_trim_complex_store): Likewise.
32934 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
32935 Track what bytes live from the original store. Return tri-state
32936 for dead, partially dead or live.
32937 (dse_dom_walker): Add constructor, destructor and new private members.
32938 (delete_dead_call, delete_dead_assignment): New extracted from
32940 (dse_optimize_stmt): Make a member of dse_dom_walker.
32941 Use initialize_ao_ref_for_dse.
32943 PR tree-optimization/33562
32944 PR tree-optimization/61912
32945 PR tree-optimization/77485
32946 * sbitmap.h (bitmap_count_bits): Prototype.
32947 (bitmap_clear_range, bitmap_set_range): Likewise.
32948 * sbitmap.c (bitmap_clear_range): New function.
32949 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
32951 2017-01-13 Martin Liska <mliska@suse.cz>
32954 * function.c (set_cfun): Add new argument force.
32955 * function.h (set_cfun): Likewise.
32956 * ipa-inline-transform.c (inline_call): Use the function when
32957 strict alising from is dropped for function we inline to.
32959 2017-01-13 Richard Biener <rguenther@suse.de>
32961 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
32962 for dumping GIMPLE INTEGER_CSTs.
32964 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32966 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
32967 to 201112L since C++17.
32969 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
32972 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
32973 if -fsanitize=kernel-address is present.
32975 2017-01-13 Richard Biener <rguenther@suse.de>
32977 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
32978 as _Literal ( type ) number in case usual suffixes do not
32979 preserve all information.
32981 2017-01-13 Richard Biener <rguenther@suse.de>
32983 PR tree-optimization/77283
32984 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
32985 and ssa-iterators.h.
32986 (is_feasible_trace): Implement a cost model based on joiner
32989 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
32992 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
32993 char or short to __float128/_Float128 directly.
32995 2017-01-12 Martin Sebor <msebor@redhat.com>
32997 to -Wformat-overflow.
32998 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
32999 (min_bytes_remaining): Same.
33000 (get_string_length): Same.
33001 (format_string): Same.
33002 (format_directive): Same.
33004 (pass_sprintf_length::handle_gimple_call): Same.
33006 2017-01-12 Jakub Jelinek <jakub@redhat.com>
33008 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
33009 info.nowrite calls with no lhs that can't throw. Return bool
33010 whether gsi_remove has been called or not.
33011 (pass_sprintf_length::handle_gimple_call): Return bool whether
33012 try_substitute_return_value called gsi_remove. Formatting fix.
33013 (pass_sprintf_length::execute): Don't use gsi_remove if
33014 handle_gimple_call returned true.
33017 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
33018 be removed due to side-effects, don't remove following barrier nor
33019 turn the successor edge into fallthru edge.
33021 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33024 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
33025 element-reversing loads and stores as not swappable.
33027 2017-01-12 Nathan Sidwell <nathan@acm.org>
33028 Nicolai Stange <nicstange@gmail.com>
33030 * combine.c (try_combine): Don't ignore result of overlap checking
33031 loop. Combine overlap & asm check into single loop.
33033 2017-01-12 Richard Biener <rguenther@suse.de>
33035 * tree-pretty-print.c (dump_generic_node): Provide -gimple
33036 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
33038 2017-01-12 Richard Biener <rguenther@suse.de>
33040 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
33041 and TS_TARGET_OPTION directly derive from TS_BASE.
33042 * tree-core.h (tree_optimization_option): Derive from tree_base.
33043 (tree_target_option): Likewise.
33045 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
33047 * config/i386/i386.c (memory_address_length): Increase len
33048 only when rip_relative_addr_p returns false.
33050 2017-01-11 Julia Koval <julia.koval@intel.com>
33052 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
33053 (OPTION_MASK_ISA_SGX_SET): New.
33054 (ix86_handle_option): Handle OPT_msgx.
33055 * config.gcc: Added sgxintrin.h.
33056 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
33057 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
33058 * config/i386/i386.c (ix86_target_string): Add -msgx.
33060 (ix86_option_override_internal): Handle new options.
33061 (ix86_valid_target_attribute_inner_p): Add sgx.
33062 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
33063 * config/i386/i386.opt: Add msgx.
33064 * config/i386/sgxintrin.h: New file.
33065 * config/i386/x86intrin.h: Add sgxintrin.h.
33067 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33070 * fold-const.c (maybe_nonzero_address): Return 1 for function
33072 (tree_single_nonzero_warnv_p): Don't handle function local objects
33076 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
33079 2017-01-11 David Malcolm <dmalcolm@redhat.com>
33082 * opts.c: Include "spellcheck.h"
33083 (struct string_fragment): New struct.
33084 (struct edit_distance_traits<const string_fragment &>): New
33086 (get_closest_sanitizer_option): New function.
33087 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
33089 2017-01-11 Jakub Jelinek <jakub@redhat.com>
33091 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
33093 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
33094 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
33095 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
33096 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
33097 for initial die_offset if dwarf_split_debug_info.
33098 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
33099 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
33101 (output_skeleton_debug_sections): Formatting fix. Use
33102 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
33103 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
33105 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
33107 * config/arm/cortex-a53.md: Add bypasses for
33108 cortex_a53_r2f_cvt.
33109 (cortex_a53_r2f): Only use for transfers.
33110 (cortex_a53_f2r): Likewise.
33111 (cortex_a53_r2f_cvt): Add reservation for conversions.
33112 (cortex_a53_f2r_cvt): Likewise.
33114 2017-01-11 Tamar Christina <tamar.christina@arm.com>
33116 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
33117 to all inlined functions, change static to extern.
33119 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
33122 * config/arm/arm.c (legitimize_pic_address): Handle reference to
33124 (arm_assemble_integer): Likewise.
33126 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
33128 * config.gcc: Use new awk script to check CPU, FPU and architecture
33129 parameters for --with-... options.
33130 * config/arm/parsecpu.awk: New file
33131 * config/arm/arm-cpus.in: New file.
33132 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
33134 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
33136 * config/arm/t-arm: Update dependency rules.
33137 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
33138 of processing .def files.
33139 * config/arm/genopt.sh: Deleted.
33140 * config/arm/gentune.sh: Deleted.
33141 * config/arm/arm-cores.def: Deleted.
33142 * config/arm/arm-arches.def: Deleted.
33143 * config/arm/arm-fpus.def: Deleted.
33144 * config/arm/arm-tune.md: Regenerated.
33145 * config/arm/arm-tables.opt: Regenerated.
33146 * config/arm/arm-cpu.h: New generated file.
33147 * config/arm/arm-cpu-data.h: New generated file.
33148 * config/arm/arm-cpu-cdata.h: New generated file.
33150 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
33153 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
33155 (input_varpool_node): Unpack dynamically_initialized bit.
33157 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
33159 PR rtl-optimization/79032
33160 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
33161 the alignment of the adjusted memory reference against that of MODE,
33162 instead of the alignment of the original memory reference.
33164 2017-01-11 Martin Jambor <mjambor@suse.cz>
33166 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
33168 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
33169 decorated functions.
33171 2017-01-11 Richard Biener <rguenther@suse.de>
33173 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
33174 set range/nonnull info for PHI results. Do not set it on
33175 stmts marked for removal.
33177 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
33179 * expr.c (store_field): In the bitfield case, fetch the return value
33180 from the registers before applying a single big-endian adjustment.
33181 Always do a final load for a BLKmode value not larger than a word.
33183 2017-01-10 David Malcolm <dmalcolm@redhat.com>
33186 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33187 that we correctly handle column numbers greater than
33188 LINE_MAP_MAX_COLUMN_NUMBER.
33190 2017-01-10 Martin Sebor <msebor@redhat.com>
33192 PR middle-end/78245
33193 * gimple-ssa-sprintf.c (get_destination_size): Call
33194 {init,fini}object_sizes.
33195 * tree-object-size.c (addr_object_size): Adjust.
33196 (pass_through_call): Adjust.
33197 (pass_object_sizes::execute): Adjust.
33198 * tree-object-size.h (fini_object_sizes): Declare.
33200 2017-01-10 Martin Sebor <msebor@redhat.com>
33202 PR tree-optimization/78775
33203 * builtins.c (get_size_range): Move...
33204 * calls.c: ...to here.
33205 (alloc_max_size): Accept zero argument.
33206 (operand_signed_p): Remove.
33207 (maybe_warn_alloc_args_overflow): Call get_size_range.
33208 * calls.h (get_size_range): Declare.
33210 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
33212 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
33213 from TI's devices.csv file as of September 2016.
33214 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
33216 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
33218 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
33219 * doc/invoke.texi: Likewise.
33220 * doc/md.texi: Likewise.
33221 * doc/objc.texi: Likewise.
33223 2017-01-10 Joshua Conner <joshconner@google.com>
33225 * config/arm/fuchsia-elf.h: New file.
33226 * config/fuchsia.h: New file.
33227 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
33228 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
33230 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
33232 2016-01-10 Richard Biener <rguenther@suse.de>
33234 PR tree-optimization/79034
33235 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
33236 Propagate out degenerate PHIs in the joiner.
33238 2017-01-10 Martin Liska <mliska@suse.cz>
33240 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
33241 (sort_congruence_classes_by_decl_uid): Likewise.
33242 (sort_congruence_class_groups_by_decl_uid): Likewise.
33243 (sem_item_optimizer::merge_classes): Sort class, groups in these
33244 classes and members in the groups by DECL_UID of declarations.
33245 This would make merge operations stable.
33247 2017-01-10 Martin Liska <mliska@suse.cz>
33249 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
33250 usage of m_classes_vec.
33251 (sem_item_optimizer::~sem_item_optimizer): Likewise.
33252 (sem_item_optimizer::get_group_by_hash): Likewise.
33253 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
33254 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
33255 (sem_item_optimizer::verify_classes): Likewise.
33256 (sem_item_optimizer::process_cong_reduction): Likewise.
33257 (sem_item_optimizer::dump_cong_classes): Likewise.
33258 (sem_item_optimizer::merge_classes): Likewise.
33259 * ipa-icf.h (congruence_class_hash): Rename from
33260 congruence_class_group_hash. Remove declaration of m_classes_vec.
33262 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
33264 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
33265 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
33266 * config.gcc: Add avx512vpopcntdqintrin.h.
33267 * config/i386/avx512vpopcntdqintrin.h: New.
33268 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
33269 * config/i386/i386-builtin-types.def: Add new types.
33270 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
33271 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
33272 __builtin_ia32_vpopcountq_v8di_mask): New.
33273 * config/i386/i386-c.c (ix86_target_macros_internal): Define
33274 __AVX512VPOPCNTDQ__.
33275 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
33276 (PTA_AVX512VPOPCNTDQ): Define.
33277 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
33278 TARGET_AVX512VPOPCNTDQ_P): Define.
33279 * config/i386/i386.opt: Add mavx512vpopcntdq.
33280 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
33281 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
33283 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33285 PR middle-end/77484
33286 * predict.def (PRED_CALL): Set to 67.
33288 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
33290 * expr.c (store_field): In the bitfield case, if the value comes from
33291 a function call and is of an aggregate type returned in registers, do
33292 not modify the field mode; extract the value in all cases if the mode
33293 is BLKmode and the size is not larger than a word.
33295 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
33298 * config/i386/cpuid.h: Fix undefined behavior.
33300 2017-01-04 Jeff Law <law@redhat.com>
33302 PR tree-optimization/79007
33303 PR tree-optimization/67955
33304 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
33305 conservative for pt.null when flag_non_call_exceptions is on.
33307 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33309 PR translation/79019
33310 PR translation/79020
33311 * params.def (PARAM_INLINE_MIN_SPEEDUP,
33312 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
33313 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
33315 * config/avr/avr.opt (maccumulate-args): Likewise.
33316 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
33317 * common.opt (freport-bug): Likewise.
33318 * cif-code.def (CIF_FINAL_ERROR): Likewise.
33319 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
33320 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
33321 translatable string.
33322 * config/i386/i386.c (function_value_32): Likewise.
33323 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
33324 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
33326 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
33327 * common/config/msp430/msp430-common.c (msp430_handle_option):
33329 * symtab.c (symtab_node::verify_base): Likewise.
33330 * opts.c (set_debug_level): Likewise.
33331 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
33332 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
33333 missing whitespace to translatable strings.
33334 * config/avr/avr.md (bswapsi2): Fix typo in comment.
33335 * config/sh/superh.h: Likewise.
33336 * config/i386/xopintrin.h: Likewise.
33337 * config/i386/znver1.md: Likewise.
33338 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
33339 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
33340 * double-int.h (struct double_int): Likewise.
33341 * double-int.c (div_and_round_double): Likewise.
33342 * wide-int.cc: Likewise.
33343 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
33344 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
33345 * cfgcleanup.c (crossjumps_occured): Renamed to ...
33346 (crossjumps_occurred): ... this.
33347 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
33350 PR tree-optimization/78899
33351 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
33352 returning bool return struct loop *, NULL for failure and the new
33354 (versionable_outer_loop_p): Don't version outer loop if it has
33355 dont_vectorized bit set.
33356 (tree_if_conversion): When versioning outer loop, ensure
33357 tree_if_conversion is performed also on the inner loop of the
33358 non-vectorizable outer loop copy.
33359 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
33360 LOOP_VECTORIZED in inner loop of the scalar outer loop and
33361 prevent vectorization of it.
33362 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
33363 the outer loop vectorization of the non-scalar version is attempted
33364 before vectorization of the inner loop in scalar version. If
33365 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
33366 vectorization of its inner loop.
33367 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
33368 has 2 inner loops, rename also on edges from bb whose single pred
33369 is outer_loop->header. Fix typo in function comment.
33371 2017-01-09 Martin Sebor <msebor@redhat.com>
33374 * asan.c (asan_emit_stack_protection): Increase local buffer size
33375 to avoid snprintf truncation warning.
33377 2017-01-09 Andrew Pinski <apinski@cavium.com>
33379 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
33380 to reference thunderx2t99 for the tuning structure
33381 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
33383 (thunderx2t99_extra_costs): This.
33384 * config/aarch64/aarch64-tune.md: Regenerate.
33385 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
33386 (vulcan_addrcost_table): This.
33387 (vulcan_regmove_cost): Rename to ...
33388 (thunderx2t99_regmove_cost): This.
33389 (vulcan_vector_cost): Rename to ...
33390 (thunderx2t99_vector_cost): this.
33391 (vulcan_branch_cost): Rename to ...
33392 (thunderx2t99_branch_cost): This.
33393 (vulcan_tunings): Rename to ...
33394 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
33395 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
33397 2017-01-09 Martin Jambor <mjambor@suse.cz>
33401 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
33402 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
33403 (propagate_vr_accross_jump_function): Use the above function for all
33404 value range computations for pass-through jump functions and type
33405 converasion from explicit value range values.
33406 (ipcp_propagate_stage): Do not attempt to deduce types of formal
33407 parameters from TYPE_ARG_TYPES.
33408 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
33409 (ipa_write_node_info): Stream type of the actual argument.
33410 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
33412 2017-01-09 Martin Liska <mliska@suse.cz>
33415 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
33416 (lookup_compiler): Do not show error message with have_E.
33418 2017-01-09 Jakub Jelinek <jakub@redhat.com>
33420 PR tree-optimization/78938
33421 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
33422 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
33423 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
33424 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
33427 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
33429 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
33432 2017-01-09 Richard Biener <rguenther@suse.de>
33434 PR tree-optimization/78997
33435 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
33436 name condition properly.
33438 2017-01-09 Richard Biener <rguenther@suse.de>
33441 * dwarf2out.c (is_cxx): New overload with context.
33442 (is_naming_typedef_decl): Use it.
33444 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33446 * invoke.texi (Option Summary): Correct spacing in option lists
33447 and add line breaks to fix over-long lines.
33449 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33451 PR middle-end/17660
33453 * extend.texi (Common Variable Attributes): Add xref to GCC
33454 Internals manual to explain mode attribute keywords.
33456 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
33459 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
33460 and Preprocessor Options.
33461 (Options for Linking): Document -pthread here....
33462 (RS/6000 and PowerPC Options): ...not here.
33463 (Solaris 2 Options): ...or here.
33464 * doc/cppopts.texi: Document -pthread.
33466 2017-01-08 Martin Sebor <msebor@redhat.com>
33468 PR middle-end/77708
33469 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
33470 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
33471 New member functions.
33472 (format_directive): Used them.
33474 (pass_sprintf_length::handle_gimple_call): Same.
33475 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
33476 to avoid truncation for any argument.
33477 (extract_affine_mul): Same.
33478 * tree.c (get_file_function_name): Same.
33480 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33482 PR middle-end/77484
33483 * predict.def (PRED_INDIR_CALL): Set to 86.
33485 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33487 PR preprocessor/54124
33488 * doc/cppopts.texi: Reformat -d subtable to list the full name
33489 of the options. Add cross-reference to the docs for the general
33490 compiler -d options.
33491 * doc/invoke.texi (Developer Options): Add cross-reference to the
33492 preprocessor-specific -d option documentation.
33494 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33496 PR preprocessor/13498
33497 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
33498 redudant material, and reflect new command-line options.
33499 (System Headers): Likewise.
33501 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
33503 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
33504 -isystem, and -idirafter. Copy-edit.
33505 * doc/cppopts.texi: Copy-edit. Remove contradiction about
33506 default for -ftrack-macro-expansion. Delete obsolete and
33507 badly-formatted implementation details about -fdebug-cpp output.
33508 * doc/cppwarnopts.texi: Copy-edit.
33510 2017-01-07 David Malcolm <dmalcolm@redhat.com>
33513 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
33514 that the transition from a max line width >= 1<<10 to narrower
33515 lines works correctly.
33517 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
33519 * doc/options.texi (PerFunction): New.
33520 * opt-functions.awk (switch_flags): Map both Optimization and
33521 PerFunction to CL_OPTIMIZATION.
33522 * opth-gen.awk: Test for PerFunction flag along with
33524 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
33525 it only when the latter is present. Skip those that don't in
33526 the hash function generator.
33527 * common.opt (fvar-tracking): Mark as PerFunction instead of
33529 (fvar-tracking-assignments): Likewise.
33530 (fvar-tracking-assignments-toggle): Likewise.
33531 (fvar-tracking-uninit): Likewise.
33533 2017-01-07 Jakub Jelinek <jakub@redhat.com>
33535 PR translation/79018
33536 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
33539 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
33542 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
33543 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
33544 TARGET_LONG_JUMP_TABLE_OFFSETS.
33545 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
33546 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
33547 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
33548 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
33549 * config/m68k/m68k.md (tablejump expander): Likewise.
33550 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
33551 TARGET_LONG_JUMP_TABLE_OFFSETS.
33552 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
33553 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
33555 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
33556 David Holsgrove <david.holsgrove@xilinx.com>
33558 * common/config/microblaze/microblaze-common.c
33559 (TARGET_EXCEPT_UNWIND_INFO): Remove.
33560 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
33562 * config/microblaze/microblaze.c (microblaze_must_save_register)
33563 (microblaze_expand_epilogue, microblaze_return_addr): Handle
33565 (microblaze_eh_return): New function.
33566 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
33567 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
33568 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
33569 * config/microblaze/microblaze.md (eh_return): New pattern.
33571 2017-01-06 Jakub Jelinek <jakub@redhat.com>
33573 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
33574 GCC_DIAGNOSTIC_STRINGIFY): Define.
33576 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
33578 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33580 * config/arm/arm.md (<mcrr>): New.
33582 * config/arm/arm.c (arm_arch5te): New.
33583 (arm_option_override): Set arm_arch5te.
33584 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
33586 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
33587 (arm_mcrr_qualifiers): ... this. New.
33588 (MRRC_QUALIFIERS): Define to...
33589 (arm_mrrc_qualifiers): ... this. New.
33590 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
33592 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
33593 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
33594 (MRRCI, mrrc, MRRC): New.
33595 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
33596 VUNSPEC_MRRC2): New.
33598 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33600 * config/arm/arm.md (<mcr>): New.
33602 * config/arm/arm.c (arm_coproc_builtin_available): Add
33603 support for mcr, mrc, mcr2 and mrc2.
33604 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
33605 (arm_mcr_qualifiers): ... this. New.
33606 (MRC_QUALIFIERS): Define to ...
33607 (arm_mrc_qualifiers): ... this. New.
33608 (MCR_QUALIFIERS): Define to ...
33609 (arm_mcr_qualifiers): ... this. New.
33610 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
33612 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
33613 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
33614 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
33615 VUNSPEC_MRC2): New.
33617 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33619 * config/arm/arm.md (*ldc): New.
33623 * config/arm/arm.c (arm_coproc_builtin_available): Add
33624 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
33625 (arm_coproc_ldc_stc_legitimate_address): New.
33626 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
33627 'qualifier_const_pointer'.
33628 (LDC_QUALIFIERS): Define to...
33629 (arm_ldc_qualifiers): ... this. New.
33630 (STC_QUALIFIERS): Define to...
33631 (arm_stc_qualifiers): ... this. New.
33632 * config/arm/arm-protos.h
33633 (arm_coproc_ldc_stc_legitimate_address): New.
33634 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
33635 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
33636 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
33637 stc2, stcl, stc2l): New.
33638 * config/arm/constraints.md (Uz): New.
33639 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
33640 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
33641 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
33642 VUNSPEC_STC2L): New.
33644 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33646 * config/arm/arm.md (<cdp>): New.
33647 * config/arm/arm.c (neon_const_bounds): Rename this ...
33648 (arm_const_bounds): ... this.
33649 (arm_coproc_builtin_available): New.
33650 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
33651 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
33652 (CDP_QUALIFIERS): Define to...
33653 (arm_cdp_qualifiers): ... this. New.
33655 (arm_expand_builtin_args): Add case for 6 arguments.
33656 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
33657 (arm_const_bounds): ... this.
33658 (arm_coproc_builtin_available): New.
33659 * config/arm/arm_acle.h (__arm_cdp): New.
33661 * config/arm/arm_acle_builtins.def (cdp): New.
33663 * config/arm/iterators.md (CDPI,CDP,cdp): New.
33664 * config/arm/neon.md: Rename all 'neon_const_bounds' to
33665 'arm_const_bounds'.
33666 * config/arm/types.md (coproc): New.
33667 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
33668 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
33669 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
33670 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
33672 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33674 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
33675 (UBINOP_QUALIFIERS): New.
33677 (acle_builtin_data): New. Change comment.
33678 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
33679 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
33680 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
33681 arm_acle_builtins.def.
33682 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
33683 (arm_init_acle_builtins): New.
33684 (CRC32_BUILTIN): Remove.
33685 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
33686 crc32cb, crc32ch and crc32cw.
33687 (arm_init_crc32_builtins): Remove.
33688 (arm_init_builtins): Use arm_init_acle_builtins rather
33689 than arm_init_crc32_builtins.
33690 (arm_expand_acle_builtin): New.
33691 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
33692 * config/arm/arm_acle_builtins.def: New.
33694 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
33696 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
33697 (arm_builtin_datum): ... this.
33698 (arm_init_neon_builtin): Rename to ...
33699 (arm_init_builtin): ... this. Add a new parameters PREFIX
33700 and USE_SIG_IN_NAME.
33701 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
33702 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
33703 'arm_builtin_datum'.
33704 (arm_init_vfp_builtins): Likewise.
33705 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
33706 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
33707 (arm_expand_neon_args): Rename to ...
33708 (arm_expand_builtin_args): ... this. Rename builtin_arg
33709 enum values and differentiate between ARG_BUILTIN_MEMORY
33710 and ARG_BUILTIN_NEON_MEMORY.
33711 (arm_expand_neon_builtin_1): Rename to ...
33712 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
33713 values, arm_expand_builtin_args and add bool parameter NEON.
33714 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
33715 (arm_expand_vfp_builtin): Likewise.
33716 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
33718 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33720 PR middle-end/77484
33721 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
33722 * predict.c (tree_estimate_probability_bb): Reverse direction of
33723 polymorphic call predictor.
33725 2017-01-06 David Malcolm <dmalcolm@redhat.com>
33727 * passes.c (execute_one_pass): Split out pass-skipping logic into...
33728 (determine_pass_name_match): ...this new function and...
33729 (should_skip_pass_p): ...this new function.
33731 2017-01-06 Nathan Sidwell <nathan@acm.org>
33733 * ipa-visibility.c (function_and_variable_visibility): Reformat
33734 comments and long lines. Remove extrneous if.
33735 * symtab.c (symtab_node::make_decl_local): Fix code format.
33736 (symtab_node::set_section_for_node): Fix comment typo.
33738 2017-01-06 Martin Liska <mliska@suse.cz>
33741 * lra-constraints.c: Rename invariant to lra_invariant.
33742 * predict.c (set_even_probabilities): Initialize e to NULL.
33744 2017-01-05 Martin Sebor <msebor@redhat.com>
33746 PR tree-optimization/78910
33747 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
33748 (format_integer): Correct off-by-one error in the handling
33749 of precision with negative numbers in signed conversions..
33751 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
33753 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
33755 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33757 PR tree-optimization/71016
33758 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
33759 factor_out_conditional_conversion. Formatting fix.
33760 (factor_out_conditional_conversion): Add cond_stmt argument.
33761 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
33762 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
33765 2017-01-05 David Malcolm <dmalcolm@redhat.com>
33767 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
33768 read-rtl-function.o, and selftest-rtl.o.
33769 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
33770 (selftest::aarch64_test_loading_full_dump): New function.
33771 (selftest::aarch64_run_selftests): New function.
33772 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
33773 selftest::aarch64_run_selftests.
33774 * config/i386/i386.c
33775 (selftest::ix86_test_loading_dump_fragment_1): New function.
33776 (selftest::ix86_test_loading_call_insn): New function.
33777 (selftest::ix86_test_loading_full_dump): New function.
33778 (selftest::ix86_test_loading_unspec): New function.
33779 (selftest::ix86_run_selftests): Call the new functions.
33780 * emit-rtl.c (maybe_set_max_label_num): New function.
33781 * emit-rtl.h (maybe_set_max_label_num): New decl.
33782 * function.c (instantiate_decls): Guard call to
33783 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
33784 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
33786 * gensupport.c (gen_reader::gen_reader): Pass "false"
33787 for new "compact" param of rtx_reader.
33788 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
33789 rather than an empty string for NULL strings.
33790 * read-md.c: Potentially include config.h rather than bconfig.h.
33791 Wrap include of errors.h with #ifdef GENERATOR_FILE.
33792 (have_error): New global, copied from errors.c.
33793 (md_reader::read_name): Rename to...
33794 (md_reader::read_name_1): ...this, adding "out_loc" param,
33795 and converting "missing name or number" to returning false, rather
33797 (md_reader::read_name): Reimplement in terms of read_name_1.
33798 (md_reader::read_name_or_nil): New function.
33799 (md_reader::read_string): Handle "(nil)" by returning NULL.
33800 (md_reader::md_reader): Add new param "compact".
33801 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
33802 (md_reader::read_file): New method.
33803 * read-md.h (md_reader::md_reader): Add new param "compact".
33804 (md_reader::read_file): New method.
33805 (md_reader::is_compact): New accessor.
33806 (md_reader::read_name): Convert return type from void to file_location.
33807 (md_reader::read_name_or_nil): New decl.
33808 (md_reader::read_name_1): New decl.
33809 (md_reader::m_compact): New field.
33810 (noop_reader::noop_reader): Pass "false" for new "compact" param
33812 (rtx_reader::rtx_reader): Add new "compact" param.
33813 (rtx_reader::read_rtx_operand): Make virtual and convert return
33814 type from void to rtx.
33815 (rtx_reader::read_until): New decl.
33816 (rtx_reader::handle_any_trailing_information): New virtual function.
33817 (rtx_reader::postprocess): New virtual function.
33818 (rtx_reader::finalize_string): New virtual function.
33819 (rtx_reader::m_in_call_function_usage): New field.
33820 (rtx_reader::m_reuse_rtx_by_id): New field.
33821 * read-rtl-function.c: New file.
33822 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
33823 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
33824 (selftest::verify_three_block_rtl_cfg): New decl.
33825 * read-rtl-function.h: New file.
33826 * read-rtl.c: Potentially include config.h rather than bconfig.h.
33827 For host, include function.h, memmodel.h, and emit-rtl.h.
33828 (one_time_initialization): New function.
33829 (struct compact_insn_name): New struct.
33830 (compact_insn_names): New array.
33831 (find_code): Handle insn codes in compact dumps.
33832 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
33833 (bind_subst_iter_and_attr): Likewise.
33834 (add_condition_to_string): Likewise.
33835 (add_condition_to_rtx): Likewise.
33836 (apply_attribute_uses): Likewise.
33837 (add_current_iterators): Likewise.
33838 (apply_iterators): Likewise.
33839 (initialize_iterators): Guard usage of apply_subst_iterator with
33840 #ifdef GENERATOR_FILE.
33841 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
33842 (md_reader::read_mapping): Likewise.
33843 (add_define_attr_for_define_subst): Likewise.
33844 (add_define_subst_attr): Likewise.
33845 (read_subst_mapping): Likewise.
33846 (check_code_iterator): Likewise.
33847 (rtx_reader::read_rtx): Likewise. Move one-time initialization
33849 (one_time_initialization): New function.
33850 (rtx_reader::read_until): New method.
33851 (read_flags): New function.
33852 (parse_reg_note_name): New function.
33853 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
33854 Handle reuse_rtx ids.
33855 Wrap iterator lookup within #ifdef GENERATOR_FILE.
33856 Add parsing support for RTL dumps, mirroring the special-cases in
33857 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
33858 values, and calling handle_any_trailing_information.
33859 (rtx_reader::read_rtx_operand): Convert return type from void
33860 to rtx, returning return_rtx. Handle case 'e'. Call
33861 finalize_string on XSTR and XTMPL fields.
33862 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
33863 "(nil)" values were omitted. Call the postprocess vfunc on the
33865 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
33866 class ctor. Initialize m_in_call_function_usage. Call
33867 one_time_initialization.
33868 * rtl-tests.c (selftest::test_uncond_jump): Call
33869 set_new_first_and_last_insn.
33870 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
33871 * selftest-rtl.c: New file.
33872 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
33873 (selftest::get_insn_by_uid): New decl.
33874 * selftest-run-tests.c (selftest::run_tests): Call
33875 read_rtl_function_c_tests.
33876 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
33877 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
33880 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
33882 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
33883 operands in a special way. Assert that pos+len <= mode precision.
33885 2017-01-05 Jakub Jelinek <jakub@redhat.com>
33887 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
33888 3 argument Alias with unlimited for the negative form.
33889 (fno-vect-cost-model): Removed.
33891 2017-01-05 Martin Liska <mliska@suse.cz>
33893 * hsa-gen.c (gen_hsa_divmod): New function.
33894 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
33896 2017-01-05 Martin Liska <mliska@suse.cz>
33899 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
33902 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33904 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
33905 small constant length operands.
33907 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
33909 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
33910 between loop iterations.
33912 2017-01-05 Martin Liska <mliska@suse.cz>
33915 * gimplify.c (gimplify_decl_expr): Compare to
33916 asan_poisoned_variables instread of checking flags.
33917 (gimplify_target_expr): Likewise.
33918 (gimplify_expr): Likewise.
33919 (gimplify_function_tree): Conditionally initialize
33920 asan_poisoned_variables.
33922 2017-01-04 Jeff Law <law@redhat.com>
33924 PR tree-optimizatin/78812
33925 * rtl.h (contains_mem_rtx_p): Prototype.
33926 * ifcvt.c (containts_mem_rtx_p): Move from here to...
33927 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
33928 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
33929 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
33930 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
33932 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33934 * input.c (assert_char_at_range): Default-initialize actual_range.
33936 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33938 * df-scan.c (df_ref_create_structure): Make regno unsigned,
33939 to match the caller.
33941 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33943 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
33944 insns after final jump in test to emit dummy move.
33946 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33948 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
33949 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
33951 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
33953 * multiple_target.c (create_dispatcher_calls): Init e_next.
33954 * tree-ssa-loop-split.c (split_loop): Init border.
33955 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
33958 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
33963 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
33964 (altivec_register_operand): Do not return true if the operand
33965 contains a SUBREG mixing SImode and SFmode.
33966 (vsx_register_operand): Likewise.
33967 (vsx_reg_sfsubreg_ok): New predicate.
33968 (vfloat_operand): Do not return true if the operand contains a
33969 SUBREG mixing SImode and SFmode.
33970 (vint_operand): Likewise.
33971 (vlogical_operand): Likewise.
33972 (gpc_reg_operand): Likewise.
33973 (int_reg_operand): Likewise.
33974 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
33975 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
33976 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
33978 (rs6000_emit_move_si_sf_subreg): New helper function.
33979 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
33980 fixup SUBREGs involving SImode and SFmode.
33981 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
33982 numbers for the new peephole2 optimization.
33983 (peephole2 for SFmode unions): New peephole2 to optimize cases in
33984 the GLIBC math library that do AND/IOR/XOR operations on single
33985 precision floating point.
33986 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
33987 target macros to say whether we need to avoid SUBREGs mixing
33989 (TARGET_ALLOW_SF_SUBREG): Likewise.
33990 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
33991 (UNSPEC_SI_FROM_SF): Likewise.
33992 (iorxor): Change spacing.
33993 (and_ior_xor): New iterator for AND, IOR, and XOR.
33994 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
33995 (movdi_from_sf_zero_ext): Likewise.
33996 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
33997 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
33998 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
33999 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
34000 (fms<mode>4): Likewise.
34001 (fnma<mode>4): Likewise.
34002 (fnms<mode>4): Likewise.
34003 (nfma<mode>4): Likewise.
34004 (nfms<mode>4): Likewise.
34006 2017-01-04 Marek Polacek <polacek@redhat.com>
34009 * doc/invoke.texi: Document -Wpointer-compare.
34011 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34013 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
34016 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
34017 descriptions for -gdwarf-5 and emit them as uleb128 instead of
34020 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
34023 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
34024 documentation of the powerpc_popcntb_ok attribute.
34025 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
34026 code to issue warning messages if a requested CPU configuration is
34027 not supported by the binary (assembler and loader) toolchain.
34028 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
34029 made to define a built-in function that has been disabled.
34030 (paired_init_builtins): Add assertion to prevent ICE if attempt is
34031 made to define a built-in function that has been disabled.
34032 (altivec_init_builtins): Add comment explaining why definition
34033 of the DST built-in functions is not preceded by an assertion
34034 check. Add assertions to prevent ICE if attempts are made to
34035 define an altivec predicate or an abs* built-in function that has
34037 (htm_init_builtins): Add comment explaining why definition of the
34038 htm built-in functions is not preceded by an assertion check.
34040 2017-01-04 Jeff Law <law@redhat.com>
34042 PR tree-optimizatin/67955
34043 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
34044 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
34045 the points-to solution does not include pt_null. Use DECL_PT_UID
34048 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
34050 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
34051 Use gen_int_mode instead of gen_lopwart for const_int operands.
34053 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34055 PR tree-optimization/71563
34056 * match.pd: Simplify X << Y into X if Y is known to be 0 or
34057 out of range value - has low bits known to be zero.
34059 2017-01-04 Alan Modra <amodra@gmail.com>
34061 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
34062 * configure: Regenerate.
34063 * config.in: Regenerate.
34065 2017-01-04 Jakub Jelinek <jakub@redhat.com>
34068 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
34069 a substring of the message, but strcmp with the whole message. Ifdef
34070 ENABLE_NLS, translate the message first using dgettext.
34072 2017-01-03 Jeff Law <law@redhat.com>
34074 PR tree-optimizatin/78856
34075 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
34076 (mark_threaded_blocks): Remove code to truncate thread paths that
34077 cross multiple loop headers. Instead invalidate the cached loop
34078 iteration information and handle case of a thread path walking
34079 into an irreducible region.
34081 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
34084 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
34085 assertions. Add support for doing the signbit if the IEEE 128-bit
34086 floating point value is in a GPR.
34087 * config/rs6000/rs6000.md (Fsignbit): Delete.
34088 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
34089 Update the length attribute if the value is in a GPR.
34090 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
34091 the sign or zero extension instruction, since the value is always 0/1.
34092 (signbit<mode>2_dm2): Delete using <Fsignbit>.
34095 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
34096 extracting SImode to a GPR register so that we can generate a
34097 store, limit the vector to be in a traditional Altivec register
34098 for the vextuwrx instruction.
34100 2017-01-03 Ian Lance Taylor <iant@google.com>
34102 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
34104 2017-01-03 Martin Sebor <msebor@redhat.com>
34106 PR tree-optimization/78696
34107 * gimple-ssa-sprintf.c (format_floating): Correct handling of
34108 precision. Use MPFR for %f for greater fidelity. Correct handling
34110 (pass_sprintf_length::compute_format_length): Set width and precision
34111 specified by asrerisk to void_node for vararg functions.
34112 (try_substitute_return_value): Adjust dump output.
34114 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
34116 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
34118 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
34120 * doc/invoke.texi (SPARC options): Document -mlra as the default.
34121 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
34122 -mlra/-mno-lra was passed to the compiler.
34124 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
34126 PR rtl-optimization/65618
34127 * emit-rtl.c (try_split): Move initialization of "before" and
34128 "after" to just before the call to emit_insn_after_setloc.
34130 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
34132 * doc/md.texi (Standard Names): Remove reference to Java frontend.
34134 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
34136 * dwarf2out.c (gen_enumeration_type_die): When
34137 -gno-strict-dwarf, add a DW_AT_encoding attribute.
34139 2017-01-03 Jakub Jelinek <jakub@redhat.com>
34141 PR tree-optimization/78965
34142 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
34143 Change first argument from const call_info & to call_info &. For %n
34144 set info.nowrite to false.
34146 PR middle-end/78901
34147 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
34148 possibly throwing calls.
34150 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
34151 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
34152 and fns handling, rather than in a separate case SSA_NAME.
34154 2017-01-02 Jeff Law <law@redhat.com>
34156 * config/darwin-driver.c (darwin_driver_init): Const-correctness
34157 fixes for first_period and second_period variables.
34159 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
34162 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
34163 (*insvqi_1): New insn pattern.
34164 (*insvqi_1_mem_rex64): Ditto.
34165 (*insvqi_2): Ditto.
34166 (*insvqi_3): Rename from *insvqi.
34168 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
34170 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
34172 * doc/cfg.texi (Edges): Remove reference to Java.
34173 (Maintaining the CFG): Ditto.
34175 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34177 PR middle-end/77674
34178 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
34179 transparent aliases.
34181 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
34183 PR middle-end/77484
34184 * predict.def (PRED_CALL): Update hitrate.
34185 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
34186 * predict.c (tree_estimate_probability_bb): Split CALL predictor
34187 into direct/indirect/polymorphic variants.
34189 2017-01-01 Jakub Jelinek <jakub@redhat.com>
34191 Update copyright years.
34193 * gcc.c (process_command): Update copyright notice dates.
34194 * gcov-dump.c (print_version): Ditto.
34195 * gcov.c (print_version): Ditto.
34196 * gcov-tool.c (print_version): Ditto.
34197 * gengtype.c (create_file): Ditto.
34198 * doc/cpp.texi: Bump @copying's copyright year.
34199 * doc/cppinternals.texi: Ditto.
34200 * doc/gcc.texi: Ditto.
34201 * doc/gccint.texi: Ditto.
34202 * doc/gcov.texi: Ditto.
34203 * doc/install.texi: Ditto.
34204 * doc/invoke.texi: Ditto.
34206 Copyright (C) 2017 Free Software Foundation, Inc.
34208 Copying and distribution of this file, with or without modification,
34209 are permitted in any medium without royalty provided the copyright
34210 notice and this notice are preserved.